Gene Therapy Progress and Prospects: Fetal gene therapy – first proofs of concept – some adverse effects

Gene Therapy Volume 12, Issue 22 (November 2005)

Gene Therapy Progress and Prospects: Fetal gene therapy – first proofs of concept – some adverse effects

C Coutelle, M Themis, S N Waddington, S M K Buckley, L G Gregory, M S Nivsarkar, A L David, D Peebles, B Weisz and C Rodeck

Gene Ther 12: 1601-1607; advance online publication, September 1, 2005; doi:10.1038/sj.gt.3302632

Research Articles

Oncolytic adenovirus that overproduces ADP and replicates selectively in tumors due to hTERT promoter-regulated E4 gene expression

M Kuppuswamy, J F Spencer, K Doronin, A E Tollefson, W S M Wold and K Toth

Gene Ther 12: 1608-1617; advance online publication, July 7, 2005; doi:10.1038/sj.gt.3302581

AAV2-mediated CLN2 gene transfer to rodent and non-human primate brain results in long-term TPP-I expression compatible with therapy for LINCL

D Sondhi, D A Peterson, E L Giannaris, C T Sanders, B S Mendez, B De, A B Rostkowski, B Blanchard, K Bjugstad, J R Sladek Jr, D E Redmond Jr, P L Leopold, S M Kaminsky, N R Hackett and R G Crystal

Gene Ther 12: 1618-1632; advance online publication, July 28, 2005; doi:10.1038/sj.gt.3302549

Eradication of hepatocellular carcinoma xenografts by radiolabelled, lipiodol-inducible gene therapy

Y Kawashita, A Ohtsuru, F Miki, H Kuroda, M Morishita, Y Kaneda, K Hatsushiba, T Kanematsu and S Yamashita

Gene Ther 12: 1633-1639; advance online publication, August 4, 2005; doi:10.1038/sj.gt.3302531

Optimization of adenovirus-mediated endothelial nitric oxide synthase delivery in rat hindlimb ischemia

J Yan, G L Tang, R Wang and L M Messina

Gene Ther 12: 1640-1650; advance online publication, August 18, 2005; doi:10.1038/sj.gt.3302563

Normal growth and regenerating ability of myoblasts from unaffected muscles of facioscapulohumeral muscular dystrophy patients

J-T Vilquin, J-P Marolleau, S Sacconi, I Garcin, M-N Lacassagne, I Robert, B Ternaux, B Bouazza, J Larghero and C Desnuelle

Gene Ther 12: 1651-1662; advance online publication, June 16, 2005; doi:10.1038/sj.gt.3302565

REVIEW

Gene Therapy Progress and Prospects: Fetal genetherapy – first proofs of concept – some adverseeffects

C Coutelle1, M Themis1, SN Waddington1, SMK Buckley1, LG Gregory1, MS Nivsarkar1, AL David2,

D Peebles2, B Weisz2 and C Rodeck2

1Gene Therapy Research Group, Division of Biomedical Sciences, Imperial College London, London, UK; and 2Department of Obstetrics& Gynaecology, Royal Free and University College Medical School, University College London, London, UK

Somatic gene delivery in utero is a novel approach to genetherapy for genetic disease based on the hypothesis thatprenatal intervention may avoid the development of severemanifestations of early-onset disease, allow targeting ofotherwise inaccessible tissues including expanding stem cellpopulations, induce tolerance against the therapeutic trans-genic protein and thereby provide permanent somatic genecorrection. This approach is particularly relevant in relation toprenatal screening programmes for severe genetic diseasesas it could offer prevention as a third option to families facedwith the prenatal diagnosis of a genetically affected child.Most investigations towards in utero gene therapy have beenperformed on mice and sheep fetuses as model animals forhuman disease and for the application of clinically relevantintervention techniques such as vector delivery by minimallyinvasive ultrasound guidance. Other animals such as dogsmay serve as particular disease models and primates have

to be considered in immediate preparation for clinical trials.Proof of principle for the hypothesis of fetal gene therapy hasbeen provided during the last 2 years in mouse models forCrigler Najjar Disease, Leber’s congenital amaurosis, Pom-pe’s disease and haemophilia B showing long-term postnataltherapeutic effects and tolerance of the transgenic proteinafter in utero gene delivery. However, recently we have alsoobserved a high incidence of liver tumours after in uteroapplication of an early form of third-generation equineinfectious anaemia virus vectors with SIN configuration.These findings highlight the need for more investigations intothe safety and the ethical aspects of in utero gene therapyas well as for science-based public information on risks andbenefits of this preventive gene therapy approach beforeapplication in humans can be contemplated.Gene Therapy (2005) 12, 1601–1607. doi:10.1038/sj.gt.3302632; published online 1 September 2005

Keywords: fetal; genetic disease; in utero; ethics

Correspondence: Dr C Coutelle, Gene Therapy Research Group, Divisionof Biomedical Sciences, Imperial College London, Sir Alexander FlemingBuilding, London SW7 2AZ, UKPublished online 1 September 2005

Gene Therapy (2005) 12, 1601–1607& 2005 Nature Publishing Group All rights reserved 0969-7128/05 $30.00

www.nature.com/gt

In brief

Progress

� First proofs of principle for therapeutic in utero geneapplication and for the induction of tolerance tolifelong expression of a transgenic human proteinhave been provided in mouse models of humangenetic disease.

� Minimally invasive ultrasound guided gene deliveryprocedures applicable to the human fetus have beendeveloped in large animal models.

� The possible adverse effects of prenatal gene delivery,including germ-line transfer of transgenic DNAsequences, developmental aberrations caused byexpression of transgenic proteins and vector-inducedoncogenesis, need to be carefully assessed beforeclinical application can be considered.

� The choice of disease for clinical application requiresa high level of certainty that benefit will be providedand that the procedure will not cause additional harmas termination of pregnancy remains an option todeal with prenatally diagnosed genetic disease.

Prospects

� The rapid development of fetal medicine and imagingwill continue to increase the ease and safety of vectordelivery.

� Research towards understanding the mechanisms ofthe observed oncogenic events will lead to the designof safer and more efficient gene delivery systemswithin the next 3–5 years.

� The development of large animal models of severegenetic diseases would contribute significantly to theselection of potential first diseases for in utero genetherapy in humans over the next 10 years.

� Progress in understanding of the human genome willincrease the demand for genetic screening pro-grammes and preventive approaches for humangenetic disease into which safe, effective in utero genetherapy may be incorporated in the future.

Introduction

The concept of fetal (prenatal or in utero) gene therapy isbased on the aims of avoiding early-onset manifestationof life-threatening genetic conditions, achieving perma-nent correction of such diseases by stable transduction ofrelevant fetal progenitor cell populations and avoidingimmune reactions against the therapeutic vector andtransgene by induction of tolerance. Although fetal genetherapy will not replace postnatal gene therapy, it isessentially a preventive approach to the management ofotherwise predominantly incurable diseases and wouldtherefore – if successful and safe – be most effectivelyconducted in conjunction with prenatal screening pro-grammes.1

Maria Hatzoglou et al in Richard Hanson’s laboratorywere the first to apply a replication-incompetent retrovirusvector intraperitoneally to fetal rats in utero in 1990 andshowed that it mediated expression of human growthfactor. This was significantly more effective than when thesame vector was administered to adult animals via theportal vein after partial hepatectomy. The following yearssaw a large number of in utero marker gene experimentsusing several vectors and different animal models, whichdemonstrated gene delivery to virtually all fetal tissues(for review see David et al2). However, the first reportof a curative gene therapy protocol in utero by prenatalinjection of adenovirus into the amniotic fluid of CFTRknockout (KO) mice, by Larson et al in 1997, wassubstantially criticized in the literature at the time andhas not been reproduced independently since then. Thefirst widely accepted confirmation for the concept of fetalgene therapy has only recently been provided by thecorrection of several different disease phenotypes inrelevant animal models.3–6

Since gene therapy for genetic diseases requireslifelong correction, a vector system with the capabilityof permanent transgene expression such as retro/lentiviruses appears presently as the best choice. How-ever, the recent occurrence of oncogenesis by suchvectors (see below) indicates their present limitations.AAV integration is relatively low and it has the tendencyto be maintained in an episomal state for a prolongedtime. However, it tends to be lost quite considerably inrapidly dividing tissues although novel serotype AAVvectors with much higher efficiency may still be able to

retain enough copies to provide a long-term therapeuticeffect. Indications of a potential AAV oncogenesis in micehas been reported by Donsante et al. in 2001 andreaffirmed at the 2005 American Society of Gene Therapyconference (Embury et al Abstract 423). First-generationadenovirus is well known for its transient maintenanceand expression. However, its broad tissue tropism andhigh infectability has made it an ideal pathfinder vectorfor studies on in utero gene delivery in different animals,and later generations of vector, in particular the helper-dependent adenovirus vectors, are significantly lessimmunogenic and show long-lasting expression proper-ties. As in postnatal gene therapy, further developmentson vector safety and efficiency remain key issues forfurther progress.

First proofs of principle for therapeutic in utero geneapplication and for the induction of tolerance to lifelongexpression of a transgenic human protein have beenprovided in mouse models of human genetic disease(Table 1)The first successful therapeutic application of genetransfer in utero was performed in 2003 by Seppen et al3

by direct injection of a lentiviral vector expressing thehuman bilirubin UDP-glucuronyltransferase (UGT1A1)gene under control of the phosphoglycerate kinasepromoter into the liver of Gunn rat fetuses. The Gunnrat is a model for the very rare autosomal recessivehuman condition Crigler Najjar disease Type I (CN l),caused by mutations in the gene encoding UGT1A1 onhuman chromosome 2. Toxic levels of bilirubin accumu-late in the patient’s blood leading to severe braindamage. Although the expression of human UGT1A1in the livers of treated fetuses was too low to allowdetection by Western blotting, a reduction in the levels ofnonconjugated bilirubin of about 45% was demonstratedover a period of 1 year. In patients with CN I, suchreduction would convert severe CN I disease to themilder CN II form. However, the absolute levels ofunconjugated bilirubin observed in human CN I aresubstantially higher than those found in the Gunn ratand it is therefore not clear if the levels of transductionand expression of the transgenic protein achieved in therat would be sufficient for a therapeutic effect in man.

Table 1 Proof of principle for therapeutic in utero gene transfer (DG¼days of gestation)

Animal model Vector Route of application Outcome Ref.

Gunn rat, model ofCrigler Najjar diseaseType I

HIV lentivirusexpressing humanUDP-glucuronyl-bilirubin transferase

Topical liverinjection 17–19 DG

45% reduction of nonconjugated bilirubin for aperiod of 1 year

3

RPE65�/� mouse modelof Leber’s congenitalamaurosis

AAV 2 expressinghuman RPE65

Subretinal injection14 DG

54% of treated eyes showed improvement in ERGand 70% in photoresponse-sensitivity at 1–2.5months after birth; RPE65 expression detectable upto 6 months

4

KO mouse model ofGSDll deficiency(Pompe’s disease)

AAV2 expressingacid a-glucosidase

Intraperitonealinjection on 15 DG

Restoration of normal GAA levels in diaphragmprevented glycogen accumulation and restoreddiaphragm contractility up to 6 monthspostpartum

5

Factor IX KO mice modelfor haemophilia B

HIV lentivirusexpressing, humanFactor IX

Yolk sac vesselinjection on 15 DG

Permanent therapeutic levels of human factor IX at18–32% of normal, lifetime phenotype correction ofbleeding disorder and tolerance of human factor IX

6

Fetal gene therapyC Coutelle et al

1602

Gene Therapy

Dejneka et al4 reported the successful restorationof rhodoposin synthesis and electrophysiologicallymeasured visual functions in a mouse model of Leber’sCongenital Amaurosis (LCA) after subretinal applicationof an AAV2/1 vector expressing the human RPE65protein in utero. RPE65 is normally expressed in thepigmented retinal epithelium, and mutations in thisprotein are responsible for about 10% of LCA, the mostcommon inherited cause of blindness in children.7 Thegene for RPE65 is located on human chromosome 1, andthe protein functions in a cascade of molecular eventsinvolved in rhodopsin production. Although nearlycomplete normalization of the electro-retinogram (ERG)was only achieved in two of the 13 pups that survived toadulthood, more than 50% showed a therapeutic ERGchange and 70% had an improved photoresponsesensitivity. The limited numbers are most likely due totechnical difficulties of topical gene delivery to the retina.Interestingly, however, these experiments also showedthat the effects on ERG after in utero injection, in manycases, surpassed the response after postnatal injection.It is also noteworthy that RPE65 expression could beobserved after 5–6 months as the AVV vector is known toremain predominantly episomal, potentially resulting insignificant loss of vector in the rapidly dividing fetaltissue. After the reported time of observation, however,the retinal epithelium is virtually nondividing andtherefore any future loss would be due either to loss ofretinal cells or vector shutdown.

Further successful in utero application of an AAVvector was reported by Rucker et al.5 Intraperitonealin utero application of an AAV-2 vector expressing thelysosomal enzyme acid alpha glucosidase (GAA) tofetuses of a mouse model of Pompe’s disease restored theenzyme levels in the diaphragm of these animals andprevented glycogen accumulation in this muscle. Thisglycogen accumulation leads, in the untreated animals,to disruption of the contractile apparatus and in humansto neonatal death from respiratory insufficiency. Near-normal contractile properties of the diaphragm werefound up to 6 months after vector application.

Finally, Waddington et al6 observed therapeutic levelsof human factor IX of between 18 and 32% of normalvalues (Figure 1) and permanent amelioration of thebleeding disorder in factor IX KO mice (a model forhuman haemophilia B). This was achieved by in uteroadministration of an HIV lentiviral vector via the fetalyolk sac vessels, a route that delivers a large proportionof the vector directly to the liver. The factor IX KO micehave no functional factor IX, and restoration of levelsto 5% of normal corresponds in humans to a mildhaemophiliac phenotype and 40% to phenotypic cure.Determination of factor IX blood levels allowed themonitoring of transgene expression in the individualtreated animals over their lifetime. Animals were culledfor molecular analysis between days 321 and 432 aftergene delivery, without detection of any adverse effects.This long observation period also allowed repeatedtesting of antibody levels, with and without exogenouschallenge. Importantly, no antibodies against the humanprotein were found even after adjuvant stimulation,whereas strong antibody responses were observed afterthe same challenge on haemophiliac mice not treatedin utero. In liver biopsies, human factor IX expressionwas detected in groups of neighbouring cells suggesting

clonal propagation of human factor IX-expressing fetalcells with progenitor function, similar to earlier observa-tions with a b-galactosidase expression vector.8 In anearlier experiment using an adenovirus vector,14 we alsoobserved tolerance against the human protein but notagainst the adenovirus vector. This is not unexpected asthe fetus is only exposed to the viral proteins for a veryshort time.

Taken together, these four independent studies indifferent animal models, and using differing transgenes,have shown that prenatal gene delivery can provideearly phenotypic correction, sufficient to reduce or avoidotherwise devastating effects of genetic diseases. Theydemonstrate long-term postnatal therapeutic proteinproduction based on long-term transgene expressionfrom the initially transduced cells and their progeny, andthe study by Waddington et al6 also demonstrates thattolerance to the transgenic protein can be induced byin utero expression.

Minimally invasive ultrasound-guided gene deliveryprocedures applicable to the human fetus have beendeveloped in large animal modelsHow closely do these proof-of-principle studies in micebring us to the goal of human fetal gene therapy? Firstly,species differences need to be taken into account.Obviously, the huge size difference between mice andhumans will dictate different approaches to vectordelivery and require experimentation on larger animalmodels.

To accomplish this, in 1999 our team was the first toadopt minimally invasive ultrasound-guided methods,as already used in human fetal medicine, for genedelivery into the umbilical vein of sheep fetuses.Ultrasound guidance has also been used more recentlyfor delivery to the amniotic cavity or for direct injectionof the parenchyma of the liver and lung in primates bythe teams of Larson and Tarantal, respectively. Primatesare most likely the ultimate animal model that will haveto be used for safety studies in immediate preparationfor a clinical trial, but their breeding conditions and highcosts make them prohibitory for routine use for the

Figure 1 Plasma hFIX:Ag concentrations in haemophilic mice afterprenatal intravascular injection of hFIX-HIV-lentivirus at 16 daysgestation. Monthly blood samples were collected into citrate buffer forplasma analysis of hFIX concentrations.


1603

Gene Therapy

development of novel techniques. Sheep are much easierto breed and maintain and are a well-established animalmodel of human fetal physiology with a consistentgestation period of about 145 days and predominantlysingleton pregnancies. They also tolerate in utero manip-ulations and have anatomical features allowing inter-ventions that are applicable to the human fetus. Further,they show important similarities to humans in thedevelopment of their immune system. Recently, ourteam has developed methods that allow access to mostorgan systems, and at gestational ages that have not yetbeen tried in humans, opening potential new fields ofapplication in fetal medicine.2,9–11 These studies havedefined clear windows of feasibility for the differentapplication routes as summarized in Table 2.

For these procedures, maternal mortality is negligibleand fetal mortality is between 10 and 15%. Over 90%of the fetal mortality was due to iatrogenic infection,usually with known commensals from the sheep fleece,and in only 20% was the cause procedure-related trauma.Exceptions are intracardiac application and umbilicalvein injection at around DG (days of gestation) 55, whichproduce an unacceptably high procedure-related fetalmortality, while umbilical vein injection is achievable atDG 60 but is not sufficiently reliable until DG 70.

The relevant time windows for man for the differentapplication routes still need to be established bothwith respect to technical feasibility and in relationto physiological development. Thus, for instance, thepreimmune period for the sheep fetus is assumed to lastup to days 60–65 of gestation but for the human fetus thisstill needs to be established, and may also require initialprimate experiments.

Regulatory T-cells, involved in the maintenance ofimmune tolerance, have been detected as early as 13weeks in the human fetus, whereas in mice these cellscannot be detected in the periphery until day 3 afterbirth.12 In sheep, the first T-cells are observed in the fetalcirculation by DG 50, although mature neutrophils canonly be found from DG 120 onwards. Another specificdifference between the mouse and human fetus, andindeed also between sheep and humans, is the placentalstructure. Six and three cell layers separate the fetal andmaternal circulations in the ovine synepitheliochorialplacenta and the murine haemotrichorial placenta,respectively. In contrast, only one layer of syncytialtrophoblasts acts as the barrier between maternal bloodand fetal capillaries in the haemomonochorial placenta ofprimates and the guinea-pig. This may be of importance

in assessing the possible risk of vector spread from thefetus to the maternal body,13 although the number oflayers may not be the only factor determining placentalpermeability. We are, therefore, planning studies onvector spread in guinea-pigs, but ultimately primatestudies may again be required.

The possible adverse effects of prenatal gene delivery,including germline transfer of the transgenic DNAsequences, developmental aberrations caused byexpression of the transgenic proteins and vector-induced oncogenesis, need to be carefully assessedbefore clinical application can be consideredIn utero gene delivery carries some specific proceduralrisks not encountered in postnatal gene delivery. As formost obstetric interventions, these concern the mother aswell as the fetus, and include infection, fetal loss andinduction of preterm labour. Other risks, frequentlydiscussed in connection with fetal gene therapy, concerngermline transmission, developmental aberrations andthe possibility of insertional mutagenesis when usingintegrating vectors. Transplacental vector spread to themother is of course also a maternal risk factor foroncogenesis and germline transduction.

Germline transmission. High on the list of possibleadverse effects is the risk of an increased danger ofgermline transmission. As outlined previously,1 at theproposed gestation times for fetal gene therapy, the germcells are well compartmentalized in their definitiveorgans and could thus only be reached from thebloodstream, similar to postnatal life. While the fetalvasculature may be more permeable than in adults,several studies have so far only occasionally detectedvector sequences in the gonads of in utero-treatedanimals. However, this was not associated with detect-able transgene expression and vector has, so far, neverbeen found in purified spermatozoa, or in the offspringof these animals in several studies including our own.14

However, evidence for lentiviral transduction of asubpopulation of gonadal cells isolated by laser capturemicrodissection was recently observed in female rhesusfetuses after intraperitoneal vector administration.15

Germline transmission will remain a general point toconsider in pre-and postnatal gene therapy. However,as pointed out by Kazazian in 1999, the calculatedfrequency of naturally occurring endogenous insertionalmutations in humans of about one in eight individuals is

Table 2 Different days of gestation (DG) and application routes for gene delivery to different organs for the sheep fetus (based onadenovirus gene delivery; sheep gestation period is 145 days)

Route of application Time of application (DG) Target organ

Intraamniotic From DG 33 onwards Skin, fetal membranesIntraperitoneal From DG 50 onwards Peritoneum, liverIntrahepatic From DG 50 onwards LiverIntramuscular From DG 50 onwards MuscleIntraumbilical From about DG 70 onwards Systemic delivery (predominantly liver, adrenal gland)Intrapleural From DG 60 onwards Intercostal and diaphragm musclesIntratracheal Around DG 100 (about 80–115) AirwaysIntragastric From DG 60 onwards Stomach, small and large bowel, liverIntraventricular Around DG 55 Choroid plexus, lateral ventricle and neurocortex


1604

Gene Therapy

substantially higher than the suggested upper tolerablelimit of one insertion event per 6000 sperm, due to a genedelivery protocol. Furthermore, gene therapy is not theonly iatrogenic procedure that carries the risk of germ-line modification, but unlike the undirected germlinedamage, caused for instance by high dose chemotherapy,the introduction of a single normal gene sequence can bedetected relatively easily over the endogenous mutatedvariant. This allows the assessment of this side effectby several generation breeding experiments in mice,and by PCR analysis of sperm from treated animals.Even if germline transmission cannot be completelyavoided, the ethical questions arising should focus on arisk/benefit analysis.

Developmental aberrations and oncogenesis. Thepotential of a therapeutic gene product, which isrequired later in life, to interfere with normal fetaldevelopment is so far a hypothetical risk factor. Based onour present knowledge, effects of a transgenic protein ondevelopmental processes are very difficult to predictand can only be detected or ruled out by carefulmonitoring. As they will be specific for individualproteins and may also depend on the time of gestation,this will have to be examined on a case-by-case basis,and even animal experiments may not reveal all specificpotential dangers.

An established risk factor of integrating viral vectors isinsertional mutagenesis, recently underlined by eventsafter gene therapy with a retroviral vector in very youngchildren treated for X-linked SCID.16 This risk is also ofconcern after in utero vector application. The fetal systemmay be particularly sensitive to such events sinceintegrating vectors prefer to insert their genomes intochromatin in open configuration. Recently, we haveobserved a very high postnatal incidence of livertumours in prenatally treated mice after application ofan early form of third-generation equine infectiousanaemia virus (EIAV) vectors with SIN configuration,but not when using a similar vector with an HIVbackbone.17 Although it is not clear whether insertionalmutagenesis has caused this phenomenon, our observa-tions suggest that the fetus may be particularly sensitiveto adverse effects associated with this vector system.We are presently elucidating the molecular mechanismsof our findings of oncogenesis in mice and anticipatethat this will lead to the design of safer vectors forpre- and postnatal gene therapy for which the mousefetus appears to provide an excellent highly sensitivetest system.

An additional safety feature could be provided byex vivo gene therapy, preferably using autologous fetalstem cells obtained, for example, by fetal liver cellsampling in the first trimester.18 This approach wouldallow the use of lower vector doses and prevent vectorspread within the fetal, and possibly the maternalorganism. It may even allow screening, selection andexpansion of transduced cells to avoid those cells withintegrations into potential oncogenes before reinfusion.Irrespective of this, these risk factors need to bethoroughly investigated in animal studies by carefulmonitoring for signs of birth defects following in uteromanipulation, and by long-term postnatal follow-up,before clinical application can be considered.

The choice of disease for clinical application requires ahigh level of certainty that benefit will be provided andthat the procedure will not cause additional harm, astermination of pregnancy remains an option to dealwith prenatal diagnosed genetic diseaseThe choice of disease(s) for the first clinical applicationsof in utero gene therapy requires consideration of diseaseseverity and onset, the ease of targeting and testing forsuccess in utero, as well as evidence for therapeuticefficiency and improved quality of life over conventionaltherapies (for review see David et al,2 Waddington et al19).This choice also needs to take into consideration thepsychological aspects that may lead a couple faced withthe diagnosis of a severe genetic disease to opt for in uterotherapy rather than for abortion or to take no action.Although we believe that the demand for such pre-ventive approach does exist, we also realize that only asmall proportion of affected families will initially beprepared to take this option even when the procedure isshown in extensive animal experiments to be safe andeffective. Therefore, severe but individually very rareconditions such as lysosomal storage diseases areunlikely to be among the first candidates. More likelyare relatively frequent genetic diseases where sympto-matic treatment can delay disease progression but isunable to cure or prevent premature death. In such cases,parents who would accept and care for an affected childmay be prepared to try in utero therapy in particular if aprenatal end point for therapeutic success or failure,which may then allow late abortion, can be built into theprotocol. Most likely, such an option would be taken upwhen an indicator case already exists in the family and anew fetus has been diagnosed prenatally as affected. Ofcourse, the final decision on the choice of disease willalso be very much influenced by new developments infetal gene therapy research, as well as by the progress orfailure of postnatal gene therapy for the investigatedconditions over the coming years.

Based on our preclinical success, and the possibilityfor easy monitoring of the transgenic protein, in uterohaemophilia B would be an ideal candidate disease. It iswell established that early postnatal prophylaxis of thiscondition is very important but also very expensive, andrequires the insertion of central venous catheters, whichposes a risk of infection. In addition, development ofimmune reactions against the therapeutic factor IX issometimes a major complication. However, as haemo-philia is therapeutically manageable and postnatal genetherapy is showing encouraging results, it is unlikelyto be accepted as the first disease for clinical applicationof in utero gene therapy. The successful correction ofhaemophilia by in utero gene delivery and long-termhepatocyte gene expression, indicates that many geneticdiseases that require gene transfer to the liver such asornithine transcarbamylase deficiency, phenylketonuriaand familial hypercholesterolaemia may have a verygood chance of success.

Using the relevant mouse disease models, we havealso made good progress with lentivirus marker genetransduction of all muscle groups relevant for treatmentof Duchenne muscular dystrophy (including the respira-tory muscles and the heart20) and to the airway epitheliausing first-generation and gutless adenoviruses for cysticfibrosis.


1605

Gene Therapy

Central to any ethical considerations of in utero genetherapy is the understanding that termination of preg-nancy is a reasonably safe maternal option for dealingwith an inherited genetic disease. We see this as the mainspecific ethical issue in fetal gene therapy. In uterogene therapy will, therefore, need to be highly reliablein preventing the disease, while not causing additionaldamage. It will, therefore, require even stricter medicalindications and safety standards than most postnatalapplications, where gene therapy is in most cases thelast hope for treatment of an otherwise lethal condi-tion. During the introductory phase of transferringthis technology to humans, the involved risks maynot be easily ascertained. Particular care will thereforebe required regarding informed maternal consentbased on detailed counselling and the understanding ofrisks versus benefits. Another issue concerns the legaland ethical status of the fetus. As pointed out previously1

by Fletcher and Richter, we would like to emphasizethat the ‘previable fetus is totally dependent on apregnant women’s autonomous decision for its statusin medicine’. This includes the prospective mother’sright to abortion, even of a nonaffected fetus, andtherefore, the availability of fetal gene therapyshould not in any way infringe her autonomy in thesedecisions.

Our knowledge about attitudes of the general public,and of affected patients and their families, towards fetalgene therapy is presently at best anecdotal. Preparationsfor clinical in utero gene therapy trials should thereforeinclude a survey of this important aspect amongpatients and families, health workers and a cross-sectionof the general public, based on a clear explanation of theaims, possibilities and limitations of the proposedclinical application. This may help to improve informa-tion provided, identify and address fears and concernsthat this novel approach to gene therapy may causeand provide a scientific basis for the rational assessmentof the risks and benefits of in utero gene therapy. Wehope that this will lead to the necessary broadsupport and ethical acceptance required for its eventualclinical application.

ProspectsAlthough in utero gene therapy has much to offer interms of prevention of severe genetic disease, itsactual application in humans will crucially depend onour ability to demonstrate safety and efficiency asoutlined above.

The rapid development of fetal medicine and imagingwill continue to increase the ease and safety of vectordelivery. In effect, this may well become safer for themother, than termination of pregnancy or interventionsrelated to preimplantation diagnosis.

As often with new therapeutics, only when theybecome efficient are potential side effects and hazardsobserved. This is certainly the case in relation to theobserved oncogenic events associated with in uteroapplication of one particular lentivirus vector and it isbeneficial that this observation was already made in ouranimal studies, as it reflects a more general problem ofgene therapy vector safety as highlighted by the humanSCID trial.

Research towards understanding the mechanisms ofthese events is presently one of the most intensive

research areas in gene therapy and will certainly lead tothe design of safer vectors within the next 3–5 years. Thefetal mouse system may even emerge as a particularlysensitive test system for such systems. Although themain emphasis will clearly be on the development ofgenuinely safer vectors, other approaches such as theincorporation of conditional suicide genes into the vectorto enable destruction of transduced cells in case ofmalignancy and ex vivo transduction of autologous(stem) cells to reduce vector spread, as well as combina-tions of all these approaches, can be expected.

Animal models of severe genetic diseases in the mouseand hopefully, for example, by using transgenic techni-ques, also in sheep or other larger animals will beurgently needed to demonstrate proof of principle forefficiency and safety in preparation for the first diseasesselected for potential in utero gene therapy in humans.These studies will also have to look carefully at thenormal levels of expression of the therapeutic proteinand exclude any adverse effects by over-expression ofthe transgenic protein.

Finally, progress in understanding of the humangenome will increase the demand for genetic screeningprogrammes and preventive approaches for humangenetic disease into which safe, effective in utero genetherapy may be incorporated in the future.

References

1 Coutelle C, Rodeck C. On the scientific and ethical issues of fetalsomatic gene therapy. Gene Therapy 2002; 9: 670–673.

2 David A et al. The current status and future direction of fetalgene therapy. Gene Ther Mol Bio 2003; 7: 181–209.

3 Seppen J et al. Long-term correction of bilirubin UDPglucur-onyltransferase deficiency in rats by in utero lentiviral genetransfer. Mol Ther 2003; 8: 593–599.

4 Dejneka NS et al. In utero gene therapy rescues vision in a murinemodel of congenital blindness. Mol Ther 2004; 9: 182–188.

5 Rucker M et al. Rescue of enzyme deficiency in embryonicdiaphragm in a mouse model of metabolic myopathy: Pompedisease. Development 2004; 131: 3007–3019.

6 Waddington S et al. Permanent phenotypic correction ofhaemophilia B in immunocompetent mice by prenatal genetherapy. Blood 2004; 104: 2714–2721.

7 Cremers F, van den Hurk JA, den Hollander AI. Moleculargenetics of Leber congenital amaurosis. Hum Mol Genet 2002; 11:1169–1176.

8 Waddington S et al. Long-term postnatal b-galactodsidaseexpression in transduced hepatocytes of immuno-competentmice after in utero delivery of an EAIV derived lentiviral vectorvia the yolk sac vessels. Gene Therapy 2003; 10: 1234–1240.

9 David A et al. Ultrasound guided percutaneous delivery ofadenoviral vectors encoding the b-galactosidase and humanfactor IX genes to early gestation fetal sheep in utero. Hum GeneTher 2003; 14: 353–364.

10 David A et al. Ultrasound-guided injection of the trachea in fetalsheep: a novel percutaneous technique to target the fetal airwaysin utero. Fetal Diag Ther 2003; 18: 385–390.

11 Peebles D et al. Widespread and efficient marker gene expressionin the airway epithelia of fetal sheep after minimally invasivetracheal application of recombinant adenovirus in utero. GeneTherapy 2004; 11: 70–78.

12 Darrasse-Jeze GMG, Salomon BL, Catala M, Klatzmann D.Ontogeny of CD4+CD25+ regulatory/suppressor T cells inhuman fetus. Blood 2005; 105: 4715–4721.


1606

Gene Therapy

13 Carter AM, Enders AC. Comparative aspects of trophoblastdevelopment and placentation. Reprod Biol Endocrinol 2004; 2: 46.

14 Waddington S et al. In utero gene transfer of human factor IX tofetal mice can induce tolerance of the exogenous clotting factor.Blood 2003; 101: 1359–1366.

15 Lee CC et al. gene transfer using lentiviral vectors and thepotential for germ cell transduction in rhesus monkeys (Macacamulatta). Hum Gene Ther 2005; 16: 417–425.

16 Hacein-Bey-Abina S et al. A serious adverse event aftersuccessful gene therapy for X-linked severe combined immuno-deficiency. N Engl J Med 2003; 348: 255–256.

17 Themis M et al. Oncogenesis following delivery of a non-primatelentiviral gene therapy vector to fetal mice. Mol Ther 2005; 12:257–266.

18 Surbek DV et al. Ultrasound-guided stem cell sampling from theearly ovine fetus for prenatal ex vivo gene therapy. Am J ObstetGynecol 2002; 187: 960–963.

19 Waddington S et al. In utero gene therapy: current challengesand perspectives. Mol Ther 2005; 11: 661–676.

20 Gregory L et al. Highly efficient EIAV-mediated in utero genetransfer and expression in the major muscle groups affected byDuchenne muscular dystrophy. Gene Therapy 2004; 11: 1117–1125.


1607

Gene Therapy

RESEARCH ARTICLE

Oncolytic adenovirus that overproduces ADP andreplicates selectively in tumors due to hTERTpromoter-regulated E4 gene expression

M Kuppuswamy, JF Spencer, K Doronin, AE Tollefson, WSM Wold and K TothDepartment of Molecular Microbiology and Immunology, Saint Louis University Health Sciences Center, St Louis, MO, USA

We have constructed a novel oncolytic adenovirus (Ad)vector, named VRX-011, in which the replication of the vectoris targeted to cancer cells by the replacement of the wild-typeAd E4 promoter with the human telomerase reverse tran-scriptase (hTERT) promoter. Genes in the Ad E4 transcriptionunit are essential for Ad replication; therefore, VRX-011 willgrow efficiently only in cells in which the hTERT promoter isactive, that is, in a wide range of cancer and immortalized cellsbut not in most somatic cells. Consistent with these expecta-tions, VRX-011 replicated efficiently in all cancer cell linesexamined, while its growth was restricted in various primaryand normal cells. VRX-011 overexpresses ADP (also knownas E3-11.6K), an Ad protein required for efficient cell lysisand release of virions from cells at late stages of infection.This overexpression enhances cell-to-cell spread and could

significantly increase antitumor efficacy. In a xenograft modelin nude mice, both intratumoral and intravenous administrationof VRX-011 effectively suppressed the growth of subcuta-neous Hep3B human liver tumors. Also, intravenous deliveryof VRX-011 greatly reduced the number and size of A549human lung cancer cell nodules in a disseminated lung tumormodel in nude mice. Importantly, tail vein administration ofdifferent doses of VRX-011 in C57BL/6 mice showed minimalliver toxicity. Considering its broad range of lytic replicationin cancer cells, its attenuated phenotype in primary cells, itsefficacy in suppressing xenografts, and its low toxicity inmouse liver, VRX-011 is a promising candidate for furtherevaluation as an anticancer therapeutic.Gene Therapy (2005) 12, 1608–1617. doi:10.1038/sj.gt.3302581; published online 7 July 2005

Keywords: adenovirus; cancer; telomerase; promoter; oncolytic

Introduction

Cancer gene therapy using genetically modified viral‘vectors’ is a relatively new field of cancer research. Amajor goal of this research is to develop vectors thatselectively kill tumor cells or inhibit their growth while atthe same time sparing normal cells. Adenovirus (Ad)vectors were originally developed as viral gene transfervectors; they exhibit natural tropism to a wide range ofepithelial tissues, including cancer cells.1 Subsequently,replicating Ad vectors designed to selectively grow andlyse tumor cells were constructed.2,3 Conditionallyreplicating Ad vectors can infect at first a portion oftumor cells, replicate in them, lyse them, and releaseprogeny virus. These newly synthesized virions areexpected to infect neighboring tumor cells and perpe-tuate the process until all tumor cells are lysed. Variousstrategies to achieve tumor-selective vector replicationhave been reported.4 One way to engineer conditionallyreplicating Ad vectors is to place key virus genes underthe regulatory control of specific promoters that areactive only in appropriate target cells. Thus far, a large

number of promoters have been applied to control thevector replication.5,6 Although most of these vectorsperform well, many Ad vectors utilizing tumor-specificor tissue-specific promoters showed some activity innormal cells.7 Also, most of these promoter-substitutedAd vectors are restricted to a certain type of cancer, andcannot be applied to a different variety of tumors.

Here, we report a vector named VRX-011 (Figure 1c)that can replicate effectively in a wide range of cancercell lines and spread efficiently from cell to cell, yet itsreplication is restricted in normal cells. We have reportedpreviously that upon replacement of the wild-type (wt)E4 promoter with the surfactant protein B (SPB)promoter or a synthetic b-catenin responsive element,the replication of the vector was restricted to lung orcolon cancer cells, respectively.8,9 The E4 region encodesproteins that are essential for Ad replication; theyregulate replication of viral DNA, induction of apoptosis,and host cell protein synthesis.10–12 With VRX-011, the wtE4 promoter is replaced by the human telomerasereverse transcriptase (hTERT) promoter. hTERT is thecatalytic subunit of the human telomerase enzyme,which maintains telomere length at the end of thechromosomes. About 85% of human cancers displayelevated levels of telomerase activity, while telomeraseactivity is not detected in normal somatic cells.13,14 Theenzyme is regulated at the transcriptional level; hTERTmRNA expression and telomerase activity are tightly

Received 27 February 2005; accepted 17 May 2005; published online7 July 2005

Correspondence: Dr K Toth, Department of Molecular Microbiologyand Immunology, Saint Louis University Health Sciences Center, 1402S Grand Blvd, St Louis, MO 63104, USA


www.nature.com/gt

associated in human cancers.15 This opens up thepossibility of using the hTERT promoter to restrict geneexpression to cancer cells. Various nonreplicating Advectors applying the hTERT promoter to regulate anti-cancer therapeutic genes have been reported.16–21 Severaloncolytic Ad vectors in which the hTERT promoterregulates the expression of the E1A22–28 or E429,30 geneshave also been described.

Besides having the wt E4 promoter replaced, VRX-011overexpresses the adenovirus death protein (ADP). ADPis expressed at very late times during Ad infection andpromotes the lysis of the infected cell, thus facilitatingvirus release.31,32 Overexpression of ADP speeds up thecell-to-cell spread of the virus,33 which can enhance theanticancer efficacy of an oncolytic Ad vector.

We report that VRX-011 replicates within and efficientlylyses cancer cells, and that its replication is attenuated innoncancerous cells. VRX-011 suppresses the growth oftumor xenografts in nude mice. It produced low toxiceffects after intravenous injection into C57BL/6 mice.

Results

The hTERT promoter is active in various cancer celllines but not in primary or normal cellsTo analyze hTERT promoter activity in various cell lines,plasmid phTERT-CAT was constructed from a 1672 bp

hTERT promoter driving the CAT reporter gene. At 48 hafter transfection, CAT activity of cells transientlytransfected with phTERT-CAT was compared with cellstransfected with pGL2-Basic vector (no promoter) orpCMV-CAT (positive control). pCMV b-galactosidasewas used as an internal control plasmid to normalizefor transfection efficacy. As shown in Figure 2a, thehTERT promoter activity was high in the cancer celllines, namely A549, HeLa, DU145, 293-E4, SW480, andHep3B cell lines, but it was very low or undetectable inHEL299 and human foreskin fibroblast (HFF) normalcells. These results demonstrate hTERT promoter activityin different cancer cell lines but little or no activity in thetwo noncancerous cell types tested.

As most primary human cells are difficult to transfect,we constructed a pair of replication-deficient Ad vectors,named AdhTERT-Luc and AdCMV-Luc, in which theluciferase gene is transcribed from the hTERT or theCMV immediate-early promoters, respectively. Thesevectors were used to evaluate hTERT promoter activityin primary normal human bronchial epithelial (NHBE)cells and human umbilical vein endothelial cells (HU-VEC), normal HEL299 cells, and different cancer celllines. Expression of luciferase was analyzed 24 h postinfection (p.i.). In primary NHBE, HUVEC, and normalHEL299 cells, the hTERT promoter activity was at basallevel (3–4 logs lower than CMV promoter activity)(Figure 2b), whereas it was only 1–2 orders of magnitudeless than the CMV promoter activity in A549, HeLa,Hep3B, and DLD-1 cancer cells (Figure 2c). These datademonstrate that the hTERT promoter is active in cancercells but not in primary cells.

The hTERT promoter selectively regulates proteinexpression in VRX-011To investigate whether the hTERT promoter retained itstumor-selectivity when inserted into the E4 region ofVRX-011, A549 lung cancer and primary HUVEC cellswere infected with 3 plaque forming units per cell (PFU/cell) of VRX-011 or VRX-007. Cell extracts were analyzedfor E1A and late proteins by immunoblot. VRX-011 andVRX-007 expressed the same level of E1A proteins (datanot shown) in both cell lines, indicating the same level ofinfection. VRX-011 and VRX-007 also expressed the samelevels of late proteins in A549 cancer cells (Figure 2d).However, late proteins were barely detectable in VRX-011-infected HUVEC cells when compared to VRX-007(Figure 2d). Similar results were obtained with small air-way epithelial cells (SAEC) (data not shown). Inasmuchas the expression of Ad late proteins is dependent on E4proteins, these data indicate that the hTERT promoterdirecting the expression of E4 genes in VRX-011 is muchmore active in cancer cells than in primary cells.

VRX-011 is attenuated in primary cellsSince E4 proteins are required for Ad replication, VRX-011 should replicate well only in those cells in which thehTERT promoter is active. To test this hypothesis, A549,HUVEC, and HEL299 cells were mock-infected orinfected with 3 PFU/cell of VRX-011 or VRX-007. Thecells were photographed through phase-contrast micro-scopy at 3, 4, and 6 days p.i. for A549, HUVEC, andHEL299 cells, respectively. Both VRX-011 and VRX-007caused severe cytopathic effect on A549 cells at 3 days

Figure 1 Schematic representation of the Ad5, VRX-007, and VRX-011genomes. (a) The horizontal bar for Ad5 shows the double-stranded DNAgenome of 36 kbp. Transcription units are indicated by arrows. The‘immediate-early’ E1A proteins induce the expression of the ‘delayed-early’proteins coded by the E1B, E2, E3, and E4 transcription units. Viral DNAbegins to replicate at about 7 h p.i., then the ‘late’ proteins derived from themajor late transcription unit are synthesized. The major late mRNAs areformed by alternative splicing and polyadenylation of a large pre-mRNAinitiated at the single major late promoter and extending to the right end ofthe genome. All late mRNAs have a tripartite leader (leaders 1–3) at their50-termini that facilitates translation. There are five families of major latemRNAs designated L1–L5. Beginning at about 24 h p.i., virions begin toassemble in the cell nucleus, then after 2–3 days the cells begin to lyse andrelease virions, with lysis complete by about 5–6 days. Efficient cell lysis ismediated by ADP. ADP is a late protein derived from the major latetranscription unit. (b) VRX-007 lacks all E3 genes except 12.5K, a gene ofunknown function. The adp gene is reinserted into the deletion such thatthe ADP major late mRNA is formed abundantly with the tripartite leaderat its 50-terminus. (c) VRX-011 is identical to VRX-007, except that it hasthe E4 promoter replaced by the hTERT promoter.

Telomerase promoter-regulated oncolytic Ad vectorM Kuppuswamy et al

1609

Gene Therapy

p.i. (Figure 3). In contrast, only VRX-007 inducedcytopathic effect in primary HUVEC (at 4 days p.i.)and normal HEL299 cells (at 6 days p.i.); VRX-011 leftthese cells largely intact (Figure 3). These data imply thatthe replication of VRX-011 is significantly less efficient inprimary HUVEC and HEL299 cells than in A549 cells.

VRX-011 induces cell death in different cancer cellsTo demonstrate that VRX-011 has an attenuated pheno-type in normal cells but not in tumor cells, we tested itsability to kill six different cancer cell lines. Cells weremock-infected or infected with 20 PFU/cell of VRX-007 orVRX-011. At the time points indicated in Figure 4a, theplasma membrane integrity (cell viability) was checkedby Trypan blue exclusion. Cancer cell lines HepG2,LNCaP, SW480, SW1116, LS513, and LS174T were killedefficiently by either VRX-011 or VRX-007 (Figure 4a).Primary HFF cells infected with VRX-011 were about 62.1and 38.1% viable at 9 and 16 days p.i., respectively,whereas only 40% of VRX-007-infected cells were alive at9 days p.i., and they were all killed at 16 days p.i. (Figure4a). We conclude from these data that VRX-011 inducesonly limited cytopathic effect in normal cells, but is nearlyas efficient in killing assorted cancer cell lines as VRX-007.

The rate of VRX-011 replication and spread is similarto that of Ad5 in A549 cellsOne way to assess the overall rate of viral replication,virus release, and reinfection is to record the rate atwhich plaques appear (‘plaque development’) on a givencell line. In this experiment, VRX-011 plaque develop-

ment was compared to VRX-007 and Ad5 in A549 cells.VRX-011 formed plaques as rapidly as Ad5 (Figure 4b),which confirms the ability VRX-011 to destroy tumorcells. The parental vector, VRX-007, spreads from cell tocell very quickly,33 and therefore produced a steeperplaque development curve (Figure 4b).

VRX-011 replicates well in cancer cell lines butis restricted in primary cellsTo demonstrate that the tumor-selective cell killingcaused by VRX-011 is due to tumor-selective replication,we compared the replication of VRX-011 to VRX-007 inselected cancer cell lines and primary human cells. A549,DU145, SW480, HUVEC, and NHBE cells were mock-infected or infected with 3 PFU/cell of VRX-011 or VRX-007. The infected cells were harvested at 4 days p.i.; bythis time, virus production reached a plateau (data notshown). The virus yield was established by plaque assayin A549 cells. Both viruses gave similar yields in A549,DU145, and SW480 cells (Figure 5). However, in primaryNHBE and HUVEC cells, VRX-011 produced approxi-mately 100-fold less progeny virus than its parent virus(Figure 5). These results demonstrate that VRX-011grows nearly as efficiently as VRX-007 in cancer celllines but its replication is restricted in primary cells.

Intratumoral administration of VRX-011 suppressesthe growth of subcutaneous Hep3B liver carcinomatumor xenografts in nude miceTo further evaluate the cancer cell killing capability ofVRX-011, we examined the ability of VRX-011 to control

Figure 2 The human telomerase promoter is active in cancer cells but not in normal cells. (a) Cells were transfected with the indicated plasmids, and 48 hlater cell extracts were assayed for CAT protein by ELISA. Each data point represents the mean of two independent experiments, each performed induplicate. (b) Normal human bronchial epithelial (NHBE), human umbilical vein endothelial (HUVEC), and normal HEL299 cells or (c) human lungcancer (A549), cervical cancer (HeLa), hepatocellular carcinoma (Hep3B), and colon cancer (DLD-1) cells were mock-infected or infected with AdhTERT-Luc or AdCMV-Luc vectors at 10 PFU/cell. After 24 h, cell extracts were assayed for luciferase activity. The luciferase activity of AdCMV-Luc-infectedcells was set at 100%, and the mock-infected and AdhTERT-Luc-infected cells were compared to this level. Each data point represents the mean of two orthree independent experiments, each performed in triplicate with similar results. (d) VRX-011 expresses Ad late proteins in human lung cancer but notprimary endothelial cells. A549 or HUVEC cells were mock-infected or infected with 3 PFU/cell of VRX-011 or VRX-007. Proteins were extracted at 1 and2 days p.i., and Ad late proteins were detected by immunoblotting using an antiserum against Ad virions.


1610

Gene Therapy

the growth of pre-established human tumor xenograftsin nude mice. Subcutaneous Hep3B tumors wereinduced in nude mice by injecting Hep3B tumor cellsinto both hind flanks. When the tumors reachedapproximately 400 ml volume, they were injected withvehicle control or with 3� 108 PFU of VRX-011 or VRX-007 for three consecutive days (total dose: 9� 108 PFU).Tumor growth was monitored for 3 weeks postinjection.Hep3B tumor growth was suppressed effectively by bothVRX-011 and VRX-007 (Figure 6a). Tumors injected withcontrol vehicle grew about four-fold during the experi-mental period, whereas tumors injected with VRX-011 orVRX-007 showed no growth at all (Figure 6a). (VRX-011versus mock: P¼ 0.018, VRX-007 versus mock: P¼ 0.026;VRX-007 versus VRX-011: P¼ 0.63).

Intravenous administration of VRX-011 suppressesthe growth of Hep3B human liver cancer xenograftsin nude miceTo evaluate if the route of administration of VRX-011influences its ability to suppress tumor growth, weinjected 5� 106 Hep3B cells into both hind flanks ofnude mice. When the tumors reached about 250 ml, theanimals were mock- (buffer-) injected or injected intra-venously (tail vein) with 3� 107 PFU (total dose) or3� 108 PFU (total dose) of VRX-011 or VRX-007. Thedoses were fractionated into three daily injections.Each group contained nine animals. Both VRX-011 andVRX-007 effectively suppressed the growth of the

tumors; during the experiment, the vehicle-injectedtumors grew about eight-fold while the VRX-011- andVRX-007-injected tumors grew about 2.5-fold (Figure 6b).We did not observe a dose effect (VRX-011 3� 107

PFU versus mock: P¼ 0.027; VRX-011 3� 108 PFU versusmock: P¼ 0.014; VRX-011 3� 107 PFU versus VRX-0073� 107 PFU: P¼ 0.37; VRX-011 3� 108 PFU versus VRX-007 3� 108 PFU: P¼ 0.26). These data show that VRX-011was as effective in suppressing tumor growth followingintravenous administration as was the parental VRX-007vector.

Intravenously administered VRX-011 reduces thenumber of nodules in a disseminated lung cancermodelTo test whether VRX-011 can inhibit the growth ofdisseminated cancer in mouse lungs, we injected 2� 106

A549 cells into the tail vein of nude mice. These cells,derived from a human lung adenocarcinoma, populatedthe lung and formed rapidly growing tumor nodules.After the 10th day, groups of mice received PBS++ or6� 108 PFU (total dose) of VRX-011 or VRX-007, respec-tively, injected into the jugular vein. The doses werefractionated into three injections, given on three con-secutive days. At 3 weeks after cell injection, lungs fromeach mouse were removed and stained with India ink(Figure 6c). Each group of lungs was examined for thesize and number of tumor nodules. Both VRX-007 andVRX-011 reduced the tumor nodules considerably when

Figure 3 VRX-011 does not cause cytopathic effect in primary and normal human cells. Primary endothelial (HUVEC), normal fibroblast (HEL299), andlung carcinoma (A549) cells were mock-infected or infected with 3 PFU/cell of VRX-011 or VRX-007 and photographed at 3 days p.i. for A549 cells, 4 daysp.i. for HUVEC, and 6 days p.i. for HEL299.


1611

Gene Therapy

compared to the control group. These data furtherconfirm the efficacy of VRX-011 in tumor cells in vivo.

VRX-011 displays reduced liver toxicity in C57BL/6miceTo evaluate vector-induced hepatotoxicity, three differentdoses (1�108, 5� 108, and 1�109 PFU) of VRX-011 or

VRX-007 were administered by tail vein injection intoC57BL/6 mice. A control group received vehicle (PBS++)only. Blood was collected at 4, 7, and 14 days p.i., andALT and AST liver enzyme levels were analyzed. In theanimals injected with 1�109 PFU of VRX-011, ALT andAST levels were marginally elevated at day 4, andreturned to normal levels at 7 days p.i. (Figure 7a, onlyALT data are shown; AST data were very similar). At thisdose level, VRX-007 caused pronounced liver toxicity;the animals became moribund and had to be killed(Figure 7b). At the doses of 5� 108 and 1�108 PFU ofVRX-011 or VRX-007, the AST and ALT enzymes levelswere the same as the control group, indicating little or notoxicity induced by VRX-011 or VRX-007. The animalweights did not change significantly during the course ofthe study by VRX-011 (any dose level) and VRX-007 (thetwo lower dose levels) (data not shown). These resultsindicate that VRX-011 induces minimal liver toxicity inC57BL/6 mice.

Discussion

We have constructed VRX-011, a vector with selectiveenhanced oncolytic activity in a wide range of tumors.VRX-011 has the E4 promoter substituted with thehTERT promoter, and it overexpresses ADP due to thedeletion of most of the other E3 genes.33 We have usedthe �1672 to 0 fragment (core promoter plus upstreamregions) of the hTERT promoter; this fragment wasshown to retain strict regulation in normal cells, while

Figure 4 VRX-011 efficiently lyses various cancer cell lines but not primary human fibroblasts. (a) Cells were mock-infected or infected with 20 PFU ofVRX-011 or VRX-007 per cell, and, at the indicated days p.i., cells were assayed for cell lysis by the Trypan blue exclusion assay. (b) VRX-011 plaquesdevelop as quickly as those of wt Ad5. A549 lung cancer cells were infected with the indicated viruses, and the number of plaques on a given day of the assay(X-axis) was plotted as percentage of plaques seen on the final day of the experiment (Y-axis).

Figure 5 VRX-011 replicates well in cancer cells but is restricted innoncancerous primary cells as determined in a single-step growth curveassay. A549, DU145, and SW480 cancer cells as well as HUVEC andNHBE primary cells were infected at 3 PFU/cell. At 4 days p.i., virus wasextracted and titer was estimated by plaque assay on A549 cells. #VRX-011-infected NHBE cells produced 0.57 PFU/cell.


1612

Gene Therapy

shorter versions did not.40 Most laboratories that haveapplied the hTERT promoter to regulate gene expressionin Ad vectors have used only the core hTERT promoter.This might lead to unrestricted replication in some cases;the hTERT core promoter (284 bp) driving the E4 genesfailed to restrict the replication of an oncolytic Ad vectorin telomerase-negative cells.29 The 1672 bp hTERTpromoter utilized by us might have more regulatoryregions that may enhance specificity. Consistent with thishypothesis, the CAT and luciferase reporter genes wereexpressed by the 1672 bp hTERT promoter in a widerange of tumor cell lines, but their expression was tightlyrestricted in primary and normal human cells (Figure2a–c). We deliberately put the emphasis on maintainingstrict regulation as opposed to achieving high activity, aswe have demonstrated earlier that E4 proteins are notneeded in great abundance for Ad replication.8,9 Astissue- or tumor-restricted promoters often lose theirspecific regulation in the context of Ad genomes, it isnoteworthy that we have shown that with VRX-011, virallate proteins were expressed in lung cancer cells butnot in primary endothelial cells (Figure 2d). As Ad lateproteins are expressed only in the late phase of infection,that is, after the replication of the viral genome, thismeans that the replication of the vector is blocked in theearly phase of infection, which is what one would expectin the absence of E4 proteins. As predicted from the lackof late protein expression, VRX-011 displays an attenu-ated phenotype on primary and normal human cells:viral yields were 100-fold less in primary HUVEC andNHBE cells compared to its parent virus VRX-007(Figure 5). Importantly, VRX-011 is almost as potent inkilling most cancer cell lines as its parental vector, VRX-007: in a single-step growth curve assay, VRX-011 had thesame growth characteristic as Ad5 and VRX-007 incancer cells (Figure 5). Taken together, these data indicatethat VRX-011 causes much more cytopathic effect incancer cells than in noncancerous cells due to its near wtreplication in the former and restricted replication in thelatter.

Next, we tested the antitumor activity of VRX-011 in anude mouse human tumor xenograft animal model.Several genetically modified Ad vectors have been testedin different human cancer cell lines.3,41,42 Many of thevectors replicate less efficiently than wt Ad5 in mosttumor cell lines because the features that restrict vector

Figure 7 VRX-011 elicits low toxicity in the livers of immunocompetentC57BL/6 mice. Animals were injected intravenously (into the tail vein) withvehicle or with (a) VRX-011 or (b) VRX-007 at different doses (1�109,5� 108, and 1�108 PFU). Blood was collected at 4, 7, and 14 days afterinjection. The mean (n¼ 5) serum AST levels (a and b) are shown. Thewhiskers represent one standard deviation in each direction. #Animalstreated with 1�109 PFU of VRX-007 died between 4 and 6 days p.i.

Figure 6 VRX-011 suppresses the growth of tumor xenografts in nudemice. For (a) and (b), subcutaneous tumors were established by injectingHep3B cells subcutaneously into the flanks of nude mice. (a) Intratumoralinjection of VRX-011. The tumors were injected with vehicle or3� 108 PFU of VRX-011 or VRX-007 for three consecutive days, startingon experimental day 0 (9� 108 PFU total). The results are plotted asmedian volume over time. (b) Intravenous application of VRX-011 orVRX-007. On experimental day 0, the animals were injected intravenouslywith vehicle or 3� 107 or 3� 108 PFU of VRX-011 or VRX-007. Thedoses were fractionated into three consecutive daily injections. The meantumor volume over time is shown. For (a) and (b), the asterisks indicatethat both VRX-007- and VRX-011-injected animals differ significantly(Po0.05) from vehicle-injected animals. The difference between VRX-007-and VRX-011-injected animals was not significant at any time during theexperiment. (c) Orthotopic lung cancer model. At experimental day 1,2� 106 A549 cells were injected into the tail vein of nude mice. At 10 daysafter the injection of cells, mice were injected with vehicle or 2� 108 PFUof VRX-011 or VRX-007 for three consecutive days (total dose of6� 108 PFU per animal). At 3 weeks after the injection of cells, the lungsfrom the animals were harvested and infused with India ink. Threerepresentative lungs from each treatment group are shown (c). Tumornodules in the lungs appear unstained, whereas the remaining lung tissueappears darkly stained with India ink.


1613

Gene Therapy

growth in normal cells result in an attenuated phenotypein cancer cells as well. We anticipated that VRX-011would perform well in tumor experiments due to its twoimportant features, ADP overexpression and the regula-tion of the low-abundance E4 genes.

When subcutaneous Hep3B tumors in nude micewere injected intratumorally with VRX-011, the growthof tumors was significantly suppressed (Figure 6a).We observed similar reduction in the growth rate ofsubcutaneous Hep3B tumors after systemic deliveryof VRX-011 (Figure 6b). In a different model, the numberof disseminated A549 lung tumor nodules was greatlyreduced after intravenous administration of VRX-011(Figure 6c). In all three experiments, VRX-011 providedapproximately the same level of tumor suppression as itsparental vector, VRX-007. The observation that VRX-011has similar antitumor activity as VRX-007 in animalexperiments was not unexpected, as VRX-011 showedonly slightly decreased replication and cytopathic effectcompared to VRX-007 in a wide range of tumor cells.These data imply that VRX-011 has very robust anti-tumor activity, independent of the delivery route.

Safety is another important question regarding theclinical use of an oncolytic Ad vector. The tissue cultureexperiments discussed above indicate that VRX-011 hasan attenuated phenotype in nontumorous cells. To testthe toxicity of VRX-011 in vivo, C57BL/6 mice wereinjected with either VRX-011 or the parental VRX-007.Earlier research by other groups, for example, Jakubczaket al43 indicated that although mice are not permissive forhuman Ads, intravenous injection of an oncolytic Adresults in liver toxicity in these animals.43 With VRX-011,most encouragingly, a high dose of 1�109 PFU causedonly a slight elevation of serum ALT and AST levelsat day 4 that returned to normal levels at 7 and 14 dayspost-treatment. Two lower doses, 5� 108 and 1�108 PFU, did not cause detectable liver toxicity (Figure7a). These data show that VRX-011 displays lower livertoxicity than its parental virus in C57BL/6 mice.

Our premise when designing VRX-011 was that mostcells in a healthy adult human are telomerase negative.Although the hTERT promoter is silent in the majorityof somatic cells, some types of human normal cells,such as lymphocytes,44 CD34+ hematopoietic progenitorcells,45,46 and keratinocytes,47,48 display telomerase acti-vity. Activation of telomerase activity in these cell lines isnot well understood. Since these cells are not of epithelialorigin and express low levels of the coxsackie-Adreceptor (CAR), VRX-011 (and other vectors derivedfrom Ad5) will have difficulty in infecting them.49,50

Furthermore, VRX-011 contains another safety feature: itlacks the E3 proteins that counteract the host immuno-surveillance.51 Thus, in case of accidental systemic vectordissemination, the host immune response would beexpected to eliminate the vector. These two ‘safety’characteristics of VRX-011, the tropism for epithelial cellsand the E3 deletion, should decrease the risk of extensivevector replication in these populations of telomerase-positive cells.

Summarizing our data, VRX-011 replicated well invarious cancer cell lines but replicated minimally innoncancerous cells. It demonstrated robust antitumoractivity against both localized and disseminated tumorsfollowing either intratumoral or intravenous administra-tion of the vector. It showed low hepatotoxicity in mice.

From these data, we conclude that VRX-011 appearsto be a possible candidate for treating a wide range ofcancers.

Materials and methods

Cell linesHuman cancer cell lines A549 (alveolar carcinoma), HeLa(cervical epithelioid carcinoma), DU145 (prostate carci-noma), Hep3B (hepatocellular carcinoma), SW480 (coloncarcinoma), and MCF-7 (human breast cancer) wereobtained from the American Type Culture Collection(ATCC, Bethesda, MD, USA). Human embryo kidney 293(HEK-293, 293) cells were obtained from Microbix(Toronto, ON, Canada). 293-E4 (VK10-9) cells are 293cells that, in addition to the Ad5 E1A and E1B proteins,express Ad E4 proteins as well as the Ad protein pIX.34

HEL299 (human diploid lung fibroblast) cells were fromATCC, and were grown in DMEM containing 10% fetalbovine serum (FBS) (Hyclone, Logan, UT, USA), 1 mM

sodium pyruvate (Sigma, St Louis, MO, USA), and0.1 mM nonessential amino acids (Sigma). A549, DU145,Hep3B, HEK-293, 293-E4, and MCF-7 cells were culturedin DMEM (Gibco, Carlsbad, CA, USA) containing 10%FBS. SW480 cells were cultured in L15 medium (Gibco)supplemented with 10% FBS. NHBE cells, HUVEC, andhuman primary SAEC were obtained and grown inspecific medium from Cambrex (Walkersville, MD,USA). HFF cells were from Washington University (StLouis, MO, USA).

Construction of the VRX-011 vectorA 1672 bp fragment of the human TERT promoter wasamplified from normal male genomic DNA from humanplacenta (Sigma) with a set of primers with a terminalEcoRV restriction enzyme site by use of the Fail Safe PCRamplification kit (Epicenter, Madison, WI, USA) (forwardprimer sequence 50GCGATATCATCAGCTTTTCAAAGACA; reverse primer sequence 50GCGATATCAGCGCTGCCTGAAACTCGCGC). The amplified hTERT PCR frag-ment was cloned into the EcoRV site of the pBS.SK+(Stratagene, La Jolla, CA, USA) plasmid and the clonedfragment was sequenced. The sequence of the promoterfragment differed from the published hTERT promotersequence at two nucleotides (both C versus T), one at the�372 and another at the �577 nucleotide position(relative to transcription initiation site); these changesdo not affect any consensus transcription factor bindingmotifs.35 The hTERT fragment was transferred to thepdlE4prom plasmid, replacing the E4 promoter in correctorientation to drive transcription of the E4 region(pdlE4hTERT). The plasmid pdlE4prom was describedearlier.8 Cotransfections of DNA from the Ad5 E3deletion mutant dl32736 digested with SpeI and EcoRItogether with the pdlE4hTERT plasmid into 293-E4 cellsgenerated VRX-011 viral plaques by a process ofhomologous recombination. An isolate with the correctgenome structure was propagated and characterized asdescribed earlier.9 The vector stocks were dialyzedagainst a physiological buffer and titered by plaqueassay on 293 and A549 cells. Titers of different prepara-tions varied from 2� 1010 to 1011 PFU/ml.

The construction of the VRX-007 vector was describedelsewhere33 (see Figure 1b). VRX-007 is considered as the


1614

Gene Therapy

parental virus for VRX-011; the only difference betweenthe two Ad vectors is that VRX-011 has the wt E4promoter replaced by the hTERT promoter, while VRX-007 has wt E4 promoter (Figure 1).

Construction of pTERT.CATPlasmid pTERT.CAT was constructed by cloning thehTERT DNA fragment into the pGL2-Basic reporterplasmid (Promega, Madison, WI, USA) in such a mannerthat it directs the transcription of the CAT gene.

Construction of AdTERT-LucThe EcoRV digested hTERT promoter fragment from thepBS.SK+hTERT plasmid was inserted into the EcoRV siteof plasmid p722 (p722-hTERT). The p722 plasmidcontains the SspI–Bst1107 I A fragment of pDE1sp1A(Microbix) cloned between the SspI and PvuII sites ofpBS.SK+ (Stratagene). This plasmid is similar to pDE1-sp1A but it replicates to a higher copy number. Theluciferase fragment from plasmid pGL3 (Promega) wasisolated after NcoI and BamHI digestion and was blunt-ended. The blunt-ended luciferase fragment was insertedin the correct orientation into the blunt-ended HindIIIsite of plasmid p722-hTERT (p722-hTERT-Luc). HEK-293cells were cotransfected with plasmids p722-hTERT-Lucand pBHGKD3E3. Plasmid pBHGKD3E3 was con-structed as follows. The SpeI-NdeI fragment containingthe E3 region of KD337 was used to substitute for thecorresponding SpeI-NdeI fragment in pBHG10 (Micro-bix). Plaques consisting of the AdhTERT-Luc recombi-nant virus were plaque-purified twice and expanded intoa virus stock in 293 cells. This stock yielded a titer of6.9� 108 PFU/ml in 293 cells and was stored at �801C.

Construction of AdCMV-LucThe luciferase fragment from plasmid pGL3 was cleavedat the unique NcoI site, then blunt ended and digestedwith BamHI. The luciferase fragment was inserted in thecorrect orientation between the blunt-ended HindIII andBamHI sites of pcDNA3.1 vector (pcDNA3.1-Luc). TheCMV-luciferase fragment was excised from pcDNA3.1-Luc by BglII and BamHI digestion, blunt-ended, andinserted into the EcoRV site of plasmid p722 to give thep722-CMV-Luc construct. The AdCMV-Luc virus wasgenerated, purified, and titered as the AdTERT-Lucvirus, reaching a titer of 6.8� 108 PFU/ml on 293 cells.

CAT assayExpression of the CAT (chloramphenicol acetyltransfer-ase) reporter gene was determined in cells transfectedwith 2 mg of CAT reporter plasmid constructs and 0.5 mgof pCMV-b-galactosidase control plasmid 48 h aftertransfection with a CAT ELISA kit (Roche Diagnostics,GmbH, Mannheim, Germany). A b-galactosidase assaywas also carried out with the same extracts to standar-dize for transfection efficiency.

Luciferase assayCells were infected with Ad vectors at 10 PFU/cell. At24 h p.i., the cells were harvested and the cell extractswere analyzed using the Luciferase Reporter GeneSystem (Promega, Madison, WI, USA) in a Wallac 1420Victor Multilabel Counter luminometer, according to themanufacturer’s protocol. All experiments were repeatedthree times in triplicate.

Plaque development assayViruses were analyzed in a standard plaque assay onA549 cells as outlined by Tollefson et al.38 The number ofplaques seen on a given day of the experiment (X-axis)was plotted as a percentage of plaques seen on the finalday of the experiment (Y-axis).

Single-step growth curveA549, DU145, SW480, NHBE, and HUVEC cells of50–60% confluency were mock-infected or infectedwith 3 PFU/cell of VRX-011 or VRX-007. The cells andthe medium were harvested at 4 days p.i. and virusyield was determined by plaque assay on A549 cells asdescribed earlier.9

Cell viability assayCells were infected at 20 PFU/cell with different Advectors. At the indicated times, a Trypan blue exclusionassay was carried out as described previously.31 For eachdata point, about 400 cells were examined and thepercentage of blue cells was calculated.

ImmunoblotsCells were mock-infected or infected with viruses at3 PFU/cell. At 1 and 2 days p.i., cells were harvested,proteins were extracted, and Ad late or E1A proteinswere detected by immunoblot using an antiserumagainst Ad virions, as described earlier.9

Subcutaneous tumor modelAll mice were cared for and treated according to theGuide for the Care and Use of the Laboratory Animalsand the institutional guidelines of Saint Louis University.For the evaluation of intratumoral efficacy of vectors,5–6-week-old female nude mice (Harlan Sprague Daw-ley, Indianapolis, IN, USA) were inoculated subcuta-neously into both hind flanks with about 7� 106 Hep3Bcells to establish subcutaneous tumors. After the tumorvolume reached about 400 ml, the animals were injectedintratumorally with 50 ml of vehicle (phosphate-bufferedsaline with 1 mM MgCl2 and 1 mM CaCl2 (PBS++)) aloneor with 3� 108 PFU of VRX-011 or VRX-007 in 50 ml ofvehicle, for three consecutive days (9� 108 PFU total).Each group contained nine animals. Tumor sizes weremeasured twice weekly using computer-linked digitalcalipers (Silvac, Switzerland). Tumor volumes werecalculated with an in-house developed software (Mou-ser). The significance of the results was examined by one-way ANOVA, and pair-wise comparisons were madewith Student’s t-test.

The efficacy of the vectors following intravenousinjection was also studied. Female 5–6-week-old nudemice (nine animals per group) were injected subcuta-neously with 5� 106 Hep3B cells into both hind flanks.When the tumors reached about 250 ml, mice wereinjected intravenously (into the tail vein) with 100 ml ofvehicle or 3� 107 or 3� 108 PFU of VRX-011 or VRX-007in 100 ml of vehicle, respectively. The doses werefractionated into three daily injections. Tumors weremeasured twice weekly and the data were analyzedusing the Mouser software.


1615

Gene Therapy

Disseminated lung tumor modelAt day 1, 2� 106 A549 cells were injected intravenously(into the tail vein) of 5–6-week-old female nude mice(nine mice per group). At 10 days following cell injection,mice were injected for three consecutive days throughthe jugular vein with vehicle or 2� 108 PFU (6� 108 PFUtotal) of VRX-011 or VRX-007 in 100 ml of PBS++. Thecontrol group received PBS++ only. Animals were killed21 days after cell injection, and lungs were stained withIndia ink.39 Tumors on the lungs appear as white(unstained) nodules while the healthy lung tissue stainsblack.

Evaluation of VRX-011- and VRX-007-mediatedhepatotoxicityFemale 5–6-week-old female C57BL/6 mice (Taconic,Germantown, NY, USA) were used. Three different doses(1�108, 5� 108, and 1�109 PFU) of VRX-011 or VRX-007were administered intravenously by tail vein injection onday 1 (n¼ 10/group) in a volume of 100 ml PBS++. Thecontrol group was injected with an equal volume ofPBS++. The mice were prebled through the retro-orbitalsinus on day –1, and the serum was assayed for ALT(alanine transferase) and AST (aspartate transferase)liver enzyme activities. Serum collected on days 4(n¼ 10/group) and 15 (n¼ 5/group) was assayed forthe same enzymes. During the study, animals wereweighed periodically and the weights were recorded.Selected animals were necropsied at days 4 and 15.

Acknowledgements

We thank Dawn Schwartz for help in preparation of themanuscript. This research was supported by GrantsCA81829 and CA71704 to WSMW from the NationalInstitutes of Health.

References

1 Russell WC. Update on adenovirus and its vectors. J Gen Virol2000; 81: 2573–2604.

2 Kanerva A, Hemminki A. Modified adenoviruses for cancergene therapy. Int J Cancer 2004; 110: 475–480.

3 Oosterhoff D, Van Beusechem VW. Conditionally replicatingadenoviruses as anticancer agents and ways to improve theirefficacy. J Exp Ther Oncol 2004; 4: 37–57.

4 Kirn D, Martuza RL, Zwiebel J. Replication-selective virotherapyfor cancer: biological principles, risk management and futuredirections. Nat Med 2001; 7: 781–787.

5 Haviv YS, Curiel DT. Engineering regulatory elements forconditionally-replicative adeno-viruses. Curr Gene Ther 2003; 3:357–385.

6 Dobbelstein M. Replicating adenoviruses in cancer therapy. CurrTop Microbiol Immunol 2004; 273: 291–334.

7 Babiss LE, Friedman JM, Darnell Jr JE. Cellular promotersincorporated into the adenovirus genome. Effect of viral DNAreplication on endogenous and exogenous gene transcription.J Mol Biol 1987; 193: 643–650.

8 Doronin K et al. Tissue-specific, tumor-selective, replication-competent adenovirus vector for cancer gene therapy. J Virol2001; 75: 3314–3324.

9 Toth K et al. An oncolytic adenovirus vector combiningenhanced cell-to-cell spreading, mediated by the ADP cytolytic

protein, with selective replication in cancer cells with deregu-lated Wnt signaling. Cancer Res 2004; 64: 3638–3644.

10 Tauber B, Dobner T. Adenovirus early E4 genes in viraloncogenesis. Oncogene 2001; 20: 7847–7854.

11 Branton P, Roopchand DE. The role of adenovirus E4orf4 proteinin viral replication and cell killing. Oncogene 2001; 20: 7855–7865.

12 Halbert DN, Cutt JR, Shenk T. Adenovirus early region 4encodes functions required for efficient DNA replication, lategene expression, and host cell shutoff. J Virol 1985; 56: 250–257.

13 Kim N et al. Specific association of human telomerase activitywith immortal cells and cancer. Science 1994; 266: 2011–2015.

14 Shay JW, Bacchetti S. A survey of telomerase activity in humancancer. Eur J Cancer 1997; 33: 787–791.

15 Takakura M et al. Expression of human telomerase subunits andcorrelation with telomerase activity in cervical cancer. Cancer Res1998; 58: 1558–1561.

16 Bilsland AE et al. Selective ablation of human cancer cells bytelomerase-specific adenoviral suicide gene therapy vectorsexpressing bacterial nitroreductase. Oncogene 2003; 22: 370–380.

17 Gu J et al. Tumor-specific transgene expression from the humantelomerase reverse transcriptase promoter enables targeting ofthe therapeutic effects of the Bax gene to cancers. Cancer Res2000; 60: 5359–5364.

18 Gu J et al. A novel single tetracycline-regulative adenoviralvector for tumor-specific Bax gene expression and cell killingin vitro and in vivo. Oncogene 2002; 21: 4757–4764.

19 Lin T et al. Targeted expression of green fluorescent protein/tumor necrosis factor-related apoptosis-inducing ligand fusionprotein from human telomerase reverse transcriptase promoterelicits antitumor activity without toxic effects on primary humanhepatocytes. Cancer Res 2002; 62: 3620–3625.

20 Majumdar AS et al. The telomerase reverse transcriptasepromoter drives efficacious tumor suicide gene therapy whilepreventing hepatotoxicity encountered with constitutive pro-moters. Gene Therapy 2001; 8: 568–578.

21 Takeda T et al. Tumor-specific gene therapy for undifferentiatedthyroid carcinoma utilizing the telomerase reverse transcriptasepromoter. J Clin Endocrinol Metab 2003; 88: 3531–3538.

22 Huang Q et al. A novel conditionally replicative adenovirusvector targeting telomerase-positive tumor cells. Clin Cancer Res2004; 10: 1439–1445.

23 Irving J et al. Conditionally replicative adenovirus driven by thehuman telomerase promoter provides broad-spectrum antitu-mor activity without liver toxicity. Cancer Gene Ther 2004; 11:174–185.

24 Kawashima T et al. Telomerase-specific replication-selectivevirotherapy for human cancer. Clin Cancer Res 2004; 10: 285–292.

25 Lanson Jr NA et al. Replication of an adenoviral vectorcontrolled by the human telomerase reverse transcriptasepromoter causes tumor-selective tumor lysis. Cancer Res 2003;63: 7936–7941.

26 Wirth T et al. A telomerase-dependent conditionally replicatingadenovirus for selective treatment of cancer. Cancer Res 2003; 63:3181–3188.

27 Zou W et al. A novel oncolytic adenovirus targeting totelomerase activity in tumor cells with potent. Oncogene 2004;23: 457–464.

28 Kim E et al. Ad-mTERT-delta19, a conditional replication-competent adenovirus driven by the human telomerase promo-ter, selectively replicates in and elicits cytopathic effect in acancer cell-specific manner. Hum Gene Ther 2003; 14: 1415–1428.

29 Hernandez-Alcoceba R, Pihalja M, Qian D, Clarke MF. Newoncolytic adenoviruses with hypoxia- and estrogen receptor-regulated replication. Hum Gene Ther 2002; 13: 1737–1750.

30 Ryan PC et al. Antitumor efficacy and tumor-selective replicationwith a single intravenous injection of OAS403, an oncolyticadenovirus dependent on two prevalent alterations in humancancer. Cancer Gene Ther 2004; 11: 555–569.


1616

Gene Therapy

31 Tollefson AE et al. The E3-11.6kDa adenovirus death protein(ADP) is required for efficient cell death: characterization of cellsinfected with adp mutants. Virology 1996; 220: 152–162.

32 Tollefson AE et al. The adenovirus death protein (E3-11.6K) isrequired at very late stages of infection for efficient cell lysisand release of adenovirus from infected cells. J Virol 1996; 70:2296–2306.

33 Doronin K et al. Overexpression of the ADP (E3-11.6K) proteinincreases cell lysis and spread of adenovirus. Virology 2003; 305:378–387.

34 Krougliak V, Graham FL. Development of cell lines capable ofcomlementing E1, E4, and protein IX defective adenvorius type 5mutants. Human Gene Ther 1995; 6: 1575–1586.

35 Takakura M et al. Cloning of human telomerase catalyticsubunit (hTERT) gene promoter and identification of proximalcore promoter sequences essential for transcriptional activa-tion in immortalized and cancer cells. Cancer Res 1999; 59:551–557.

36 Thimmappaya B, Weinberger C, Schneider RJ, Shenk T.Adenovirus VAI RNA is required for efficient translation ofviral mRNAs at late times after infection. Cell 1982; 31: 543–551.

37 Doronin K et al. Tumor-specific, replication-competent adeno-virus vectors overexpressing the Adenovirus Death Protein.J Virol 2000; 74: 6147–6155.

38 Tollefson AE, Hermiston TW, Wold WSM. Preparation andtitration of CsCl-banded adenovirus stocks. In: Wold NJ (ed).Adenovirus Methods and Protocols. The Humana Press, NJ, 1998,pp 1–9.

39 Kataoka M et al. An agent that increases tumor suppressortransgene product coupled with systemic transgene deliveryinhibits growth of metastatic lung cancer in vivo. Cancer Res 1998;58: 4761–4765.

40 Kawakami Y et al. A longer human telomerase reversetranscriptase promoter-based conditionally replicative adeno-

virus shows oncolysis with less toxicity for ovarian cancertreatment. Mol Ther 2004; 9: S235.

41 McCormick F. Cancer-specific viruses and the development ofONYX-015. Cancer Biol Ther 2003; 2: S157–S160.

42 Nemunaitis J. Selective replicating viral vectors: potential for usein cancer gene therapy. BioDrugs 2003; 17: 251–262.

43 Jakubczak JL et al. An oncolytic adenovirus selective forretinoblastoma tumor suppressor protein pathway-defectivetumors: dependence on E1A, the E2F-1 promoter, and viralreplication for selectivity and efficacy. Cancer Res 2003; 63:1490–1499.

44 Liu K et al. Constitutive and regulated expression of telomerasereverse transcriptase (hTERT) in human lymphocytes. Proc NatlAcad Sci USA 1999; 96: 5147–5152.

45 Hiyama K et al. Activation of telomerase in human lympho-cytes and hematopoietic progenitor cells. J Immunol 1995; 155:3711–3715.

46 Norrback KF, Roos G. Telomeres and telomerase in normal andmalignant haematopoietic cells. Eur J Cancer 1997; 33: 774–780.

47 Harle-Bachor C, Boukamp P. Telomerase activity in theregenerative basal layer of the epidermis inhuman skin andin immortal and carcinoma-derived skin keratinocytes. Proc NatlAcad Sci USA 1996; 93: 6476–6481.

48 Yasumoto S et al. Telomerase activity in normal human epithelialcells. Oncogene 1996; 13: 433–439.

49 Dmitriev I et al. An adenovirus vector with genetically modifiedfibers demonstrates expanded tropism via utilization of acoxsackievirus and adenovirus receptor-independent cell entrymechanism. J Virol 1998; 72: 9706–9713.

50 Shayakhmetov DM, Papayannopoulou T, StamatoyannopoulosG, Lieber A. Efficient gene transfer into human CD34(+) cells bya retargeted adenovirus vector. J Virol 2000; 74: 2567–2583.

51 Lichtenstein DL et al. Functions and mechanisms of action of theadenovirus E3 proteins. Int Rev Immunol 2004; 23: 75–111.


1617

Gene Therapy

RESEARCH ARTICLE

AAV2-mediated CLN2 gene transfer to rodent andnon-human primate brain results in long-term TPP-Iexpression compatible with therapy for LINCL

D Sondhi1,5, DA Peterson2,5, EL Giannaris1, CT Sanders2, BS Mendez1, B De1, AB Rostkowski2,

B Blanchard3, K Bjugstad3, JR Sladek Jr3, DE Redmond Jr4, PL Leopold1, SM Kaminsky1,

NR Hackett1 and RG Crystal1

1Department of Genetic Medicine, Weill Medical College of Cornell University, New York, NY, USA; 2Department of Neuroscience,Rosalind Franklin University of Medicine and Science, The Chicago Medical School, North Chicago, IL, USA; 3Department ofPsychiatry, University of Colorado Health Science Center, Denver, CO, USA; and 4Neural Transplantation and Regeneration Program,School of Medicine, Yale University, New Haven, CT, USA

Late infantile neuronal ceroid lipofuscinosis (LINCL) is a fatal,autosomal recessive disease resulting from mutations in theCLN2 gene with consequent deficiency in its product tripeptidylpeptidase I (TPP-I). In the central nervous system (CNS), thedeficiency of TPP-I results in the accumulation of proteins inlysosomes leading to a loss of neurons causing progressiveneurological decline, and death by ages 10–12 years. Toestablish the feasibility of treating the CNS manifestations ofLINCL by gene transfer, an adeno-associated virus 2 (AAV2)vector encoding the human CLN2 cDNA (AAV2CUhCLN2) wasassessed for its ability to establish therapeutic levels of TPP-Iin the brain. In vitro studies demonstrated that AAV2CUhCLN2expressed CLN2 and produced biologically active TPP-Iprotein of which a fraction was secreted as the pro-TPP-Iprecursor and was taken up by nontransduced cells (ie, cross-correction). Following AAV2-mediated CLN2 delivery to the ratstriatum, enzymatically active TPP-I protein was detected. Byimmunohistochemistry TPP-I protein was detected in striatalneurons (encompassing nearly half of the target structure) for

up to 18 months. At the longer time points following striataladministration, TPP-I-positive cell bodies were also observedin the substantia nigra, frontal cerebral cortex and thalamus ofthe injected hemisphere, and the frontal cerebral cortex of thenoninjected hemisphere. These areas of the brain containneurons that extend axons into the striatum, suggesting thatCNS circuitry may aid the distribution of the gene product.To assess the feasibility of human CNS delivery, a totalof 3.6�1011 particle units of AAV2CUhCLN2 was administeredto the CNS of African green monkeys in 12 distributed doses.Assessment at 5 and 13 weeks demonstrated widespreaddetection of TPP-I in neurons, but not glial cells, at all regionsof injection. The distribution of TPP-I-positive cells was similarbetween the two time points at all injection sites. Together,these data support the development of direct CNS genetransfer using an AAV2 vector expressing the CLN2 cDNA forthe CNS manifestations of LINCL.Gene Therapy (2005) 12, 1618–1632. doi:10.1038/sj.gt.3302549; published online 28 July 2005

Keywords: AAV2; LINCL; Batten; CLN2; TPP-I; brain gene therapy

Introduction

Late infantile neuronal ceroid lipofuscinosis (LINCL) isan autosomal recessive, neurodegenerative lysosomalstorage disease.1–3 Mutations in the CLN2 gene resultin a deficiency in the lysosomal activity of tripeptidylpeptidase I (TPP-I), with consequent accumulationof autofluorescent material resembling lipofuscin inthe lysosomes.4,5 This storage material consists of un-degraded proteins, primarily subunit c of mitochondrialATP synthase.6–8 Although all tissues of individuals with

LINCL are TPP-I deficient, retinal pigmented epithelialcells and central nervous system (CNS) neurons are themost sensitive to decreased activity of TPP-I, resulting intheir progressive destruction.2,3,9,10 The consequences areprogressive loss of vision and neurological decline begin-ning at about age 3 years with cognitive impairment,seizures, and deteriorating motor skills that lead to avegetative state and death in middle to late childhood.1–3

There are compelling data to suggest that direct CNSadministration of an adeno-associated virus (AAV)vector expressing the CLN2 cDNA should be able tomediate expression of TPP-I in a sufficient number ofneurons to slow down, and potentially halt, the progres-sion of the CNS disease. First, AAV gene transfer vectorsare capable of mediating transfer and persistence ofexpression of a variety of genes in the CNS.11–17 Second,mucopolysaccharidosis VII, a related lysosomal storagedisease, has been successfully reversed in a knockout

Received 22 July 2004; accepted 10 February 2005; published online28 July 2005

Correspondence: Dr RG Crystal, Department of Genetic Medicine, WeillMedical College of Cornell University, 515 East 71st Street, Suite 1000,New York, NY 10021, USA5These authors contributed equally to this paper


www.nature.com/gt

mouse model by recombinant AAV2-mediated intra-cranial gene transfer.16,18–20 Third, a significant fractionof newly synthesized TPP-I protein is secreted as pro-TPP-I, a 563 amino-acid inactive form which can cross-correct nearby nontransduced cells through mannose6-phosphate receptor-mediated uptake and subsequentactivation in lysosomes to the 367 amino-acid matureform.21,22 This cross-correction between transducedcells and neighboring cells suggests that it will not benecessary to transfer the normal CLN2 cDNA to all thecells in the CNS, thus extending the effective rangeof gene transfer to broader regions of the brain.17 Finally,comparisons of genotype and phenotype of childrenwith CLN2 mutations reveal that expression as lowas 5% of normal TPP-I levels lead to a much milder formof the disease with a delayed age of onset, less severesymptoms, and a prolonged lifespan, suggesting thatthis level of enzyme activity is a reasonable target fortherapy.23 This target level of expression of the trans-ferred cDNA is within levels that could be achieved withcurrent AAV vector technology.

There are several challenges to the developmentof AAV-mediated CLN2 therapy for LINCL. The CNSmanifestations of LINCL are diffuse, and thus successfulgene therapy will need to provide TPP-I activity overas broad a volume of the brain as possible.1,2 Second,the absence of an experimental animal model for thedisease until very recently24 required that the detectionof gene transfer-mediated TPP-I expression is above thebackground corresponding to normal expression in anaive animal. However, the target level for therapy isonly 5% of the endogenous level and the variousdetection methods for TPP-I are not sensitive enough todetect this amount above the background. Finally, aswith most genetic diseases, the optimal therapy wouldbe to produce long-term expression of the therapeuticgene from a single administration of the vector.

Based on these considerations, this study is focusedon assessment of CNS expression of TPP-I mediated bydirect CNS administration of AAV2CUhCLN2, an AAVserotype 2-based gene therapy vector applicable totreating LINCL. Direct CNS administration of AAV2-CUhCLN2 to rats and African green monkeys resulted inrelatively stable and long-lasting expression of TPP-Iwithin the CNS, which is promising for the developmentof gene therapy for LINCL.

Results

Function of AAV vectors in vitroCells infected with AAV2CUhCLN2 were evaluated forTPP-I expression using both immunoperoxidase andimmunofluorescence detection systems using a mousemonoclonal anti-human TPP-I antibody. Vector-mediatedexpression of TPP-I was evident using both immunoper-oxidase and immunofluorescence detection (Figure 1a–d). The negative controls included cells infected with anAAV2CUNull vector, which did not contain the CLN2transgene (panels b, d) as well as naive cells (not shown).Treatment of CLN2-expressing cells with an irrelevantprimary antibody (mouse anti-shiga toxin) did not showpositive staining (not shown). High power images ofimmunofluorescence staining indicated that the antigenwas distributed in a punctate pattern throughout the

cytoplasm consistent with localization in intracellularorganelles (not shown). High power images alsorevealed the presence of a low level of endogenousTPP-I expression in the human 293 cells, appearingas small groups of organelles restricted to an area ofthe cytoplasm close to the nucleus. Similar results of

Time (min)

Flu

ore

scen

ce x

103

(360

/460

nm

)

AAV2CUhCLN2 (108pu)

AAV2GFP (1010pu)

Uninfected cells

AAV2CUhCLN2 (1010pu)

AAV2CUhCLN2 (109pu)

AAV2CUhCLN2(1010pu) + inhibitor

0

10

20

30

40

40200 60 80

AAV2CUhCLN2a

AAV2CUhCLN2

AAV2CUNull

AAV2CUNull

10 µm

b

c d

e

Figure 1 Function of the AAV2CUhCLN2 vector in vitro. (a–d)Morphological detection of TPP-I following CLN2 gene transfer.293ORF6 were infected with AAV2CUhCLN2 or AAV2CUNull at 103

particles per cell and TPP-I expression was assessed after 2 days.Immunodetection of TPP-I was performed using a mouse anti-humanTPP-I monoclonal antibody. For immunoperoxidase detection (a, b),primary antibody binding was detected by horseradish peroxidase-conjugated anti-mouse antibody followed by precipitation of diaminoben-zidine. For immunofluorescence detection (c, d), the TPP-I primaryantibody was detected with Alexa 488-conjugated anti-mouse antibody(green), the nuclei were stained with DAPI (blue) and the cells visualizedby fluorescent microscopy. (a) AAV2CUhCLN2; (b) AAV2CUNull; (c)AAV2CUhCLN2; and (d) AAV2CUNull. (e) Enzymatic activity of TPP-Isecreted from transfected cells. 293ORF6 cells (106 cells) were infectedwith AAV2CUhCLN2 (108–1010 pu) or AAV2GFP (1010 pu, a controlAAV2 vector with a GFP transgene). After 48 h, the medium wascollected, the pro-TPP-I was converted to active TPP-I at low pH andTPP-I activity was assessed. The time course of increase in fluorescence isshown for various conditions.

AAV-mediated CNS transfer of CLN2D Sondhi et al

1619

Gene Therapy

vector-mediated TPP-I expression were obtained withcells infected with AAV2CUrCLN2 (not shown).

Use of a TPP-I specific fluorogenic substrate showedthat infection of cells by AAV expressing CLN2 resultedin TPP-I enzymatic activity in the cells with secretion intothe media (Figure 1e). The background activity in themedia from uninfected cells was o10 fluorescence units(FU)/min-ml with o10 FU/min-mg in the cell homo-genate. As expected, TPP-I activity was dose-dependentwith infection by 4� 103 particle units (PU)/cell (4�109 pu total dose) of AAVCUhCLN2 resulting in thesecretion of 1.2570.24� 105 FU/min-ml, while infectionby 4� 104 pu/cell (4� 1010 pu total dose) gave 970.6�105 FU/min-ml. The percentage of the total TPP-I activityin the media (as opposed to the cell-associated TPP-Iactivity) was 5.772% at 4� 103 pu/cell and 22.377% at4� 104 pu/cell. Similar results of vector-mediated TPP-Iactivity were obtained with cells infected with AAV2-CUrCLN2 (not shown). Vector-derived TPP-I activity wasdemonstrated by lack of activity in controls including theassay of naive cells and cells transduced with compa-rable vectors with irrelevant transgene (AAV2GFP). Thespecificity of the assay for TPP-I activity was confirmedby inhibition using the TPP-I-specific enzyme inhibitor,Ala-Ala-Phe-chloro-methyl ketone.

In vitro cross-correction of LINCL-deficient fibroblastsmediated by AAV2CUhCLN2Like some other lysosomal storage disorders, LINCL isan attractive candidate for gene therapy due to cross-correction, where a portion of the newly synthesizedpro-TPP-I made by transduced cells may be taken up byneighboring cells, thereby amplifying the number of cellswith TPP-I activity. This pathway was demonstratedin vitro by showing transfer of TPP-I from the media ofcells infected with AAV2CUhCLN2 to fibroblasts derivedfrom an LINCL patient. The specificity of the transferwas indicated by the inhibition observed on additionof 1 mM mannose-6-phosphate (an inhibitor of uptakemediated by mannose-6-phosphate-mediated endocyto-sis), which reduced uptake by the LINCL fibroblastsby 78% (Figure 2). To ensure that the TPP-I activity infibroblasts was not due to carry over of vector,AAV2CUhCLN2 levels were assessed in the media trans-ferred to the LINCL fibroblasts. The level was o1% ofthe input level of vector. In a separate experiment, whenthis level of AAV2CUhCLN2 was used to infect LINCLfibroblasts, it did not yield a level of TPP-I activity overthe background TPP-I activity following exposure ofLINCL fibroblasts to media from naive 293ORF6 cells.

TPP-I distribution following AAV2-mediated CLN2delivery to the rat brainHistological analyses and enzyme activity assays wereused to determine the level, duration, and distributionof TPP-I expression that could be obtained by AAV2-mediated gene transfer in rat brain. Results obtainedfrom transfer of the rat CLN2 gene (observed using therabbit anti-TPP-I antibody) were indistinguishable fromtransfer of the human CLN2 gene (observed using themonoclonal anti-human TPP-I antibody). Results forboth rat and human CLN2 gene delivered by AAV2-based vectors expression are described below.

TPP-I enzymatic activity assays were carried outon tissue homogenates of coronal sections of the brainof naive, AAV2CUNull and AAV2CUrCLN2-injected rats(Figure 3). The TPP-I activity of the sections of theuninjected right striatum was similar across the brainwith no significant difference between the AAV2CUNulland AAV2CUrCLN2 group (P40.25 all locations). How-ever, in the injected hemisphere, in the slice corres-ponding to the injection site, there was a significantlyhigher TPP-I activity in the AAV2CUrCLN2-injected mice(Po0.05 compared to AAV2CUNull group). The meanTPP-I activity (n¼ 3 animals/group) was 55% greaterthan the background for this region. No significantdifferences were observed among the groups at anyother location of the injected hemisphere (P40.05).

The striatum was chosen as a delivery site for itsrelatively homogeneous cellular architecture and largevolume relative to the rat brain as a whole. Immuno-histochemical examination of TPP-I expression at 4weeks following a single striatal injection of 1010 pu ofAAV2CUrCLN2 demonstrated robust TPP-I expressionand the neuron-specific tropism of the AAV2-mediatedgene delivery. The majority of striatal neurons in theinjection region were found to express TPP-I, as deter-mined by coexpression with NeuN, a marker for matureneurons (Figure 4). In contrast, adjacent GFAP-positiveastrocytes did not contain detectable levels of TPP-I, norwere other TPP-I-positive/NeuN-negative cells observedby confocal microscopic analysis when registered focalplane series were evaluated for coexpression of cyto-plasmic TPP-I and nuclear NeuN expression. Endo-genous TPP-I gene expression was not observed at thedilution of the rabbit anti-TPP-I antibody used in thisassay.

The TPP-I expression studies were not formal toxi-cology studies and did not use vector prepared withinthe requirements of Good Laboratory Practice. However,there was no observed behavioral difference betweenthe AAV2CUhCLN2-injected animals and the naive

0

500

1000

1500

Mannose-6-phosphateAAV2CUNull Naive

TP

P-I

act

ivit

y x1

03

(FU

/min

-mg

)

AAV2CUhCLN2-

*

+

Figure 2 In vitro cross-correction of LINCL-deficient fibroblasts mediatedby AAV2CUhCLN2. Duplicate wells of 293ORF6 cells (106) were infectedwith AAV2CUhCLN2 or AAV2CUNull at 3� 104 pu/cell or mock infected.After 24 h, the medium was removed, the cells washed twice with PBS andreplaced with RPMI media which was collected after a further 72 h. Thismedium was collected diluted 1:3, and placed onto LINCL-deficientfibroblasts in the presence or absence of 1 mM mannose-6-phosphate. After24 h, the fibroblasts were collected, lysed, and assayed for TPP-I activityusing a fluorometric assay. The means and standard deviation ofquadruplicates is shown. *¼ not detectable.


1620

Gene Therapy

controls nor was there a statistically significant differencein mortality (P40.3 by Kaplan–Meier analysis). Basedon the triple immunofluorescence studies, there was

some degree of astrocytic hypertrophy observed aroundthe injection site that is comparable with that normallyseen following intracranial injection, but there was

TPP-I

GFAPc

a

d

b NeuN

Merged

50 µm

Figure 4 Preferential TPP-I accumulation in neurons following AAV2CUrCLN2 delivery to the rat striatum. Sections of rat brain were studied byimmunofluorescence 4 week following striatal injection of 1010 pu of AAV2CUrCLN2. (a) Immunofluorescence detection of TPP-I (green) using the rabbitanti-TPP-I antibody and Cy2-conjugated secondary antibody; (b) immunofluorescence staining demonstrating neurons (red) in the same section as in panel(a) using anti-NeuN primary antibody and a Cy3-conjugated secondary antibody; (c) immunofluorescence detection of glia (blue) in the same section as inpanel (a) using anti-glial fibrillary acidic protein (GFAP) primary antibody and a Cy5-conjugated secondary antibody; and (d) superimposition of thesignals of TPP-I-expressing cells (green) with neuron-specific staining (red, NeuN) and glial-specific staining (blue, GFAP) demonstrating neuron-specificexpression mediated by the AAV2CUrCLN2 vector. Smaller neurons show obvious colocalization of cytoplasmic TPP-I and the primarily nuclear NeuNexpression. Two large neurons do not include the nucleus in the confocal optical plane of section and thus do not appear NeuN-positive in this image, butwould demonstrate colocalization in subsequent focal planes. The magnification was identical in all panels; bar¼ 50 mm.

Distance (mm) along rostral-caudal axis

Striatum, right

TP

P-I

act

ivit

y (f

luo

resc

ence

un

its

x103 /

min

-mg

)

0

5

10

15

20

0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18

Striatum, left

0

5

10

15

20

*AAV2rCLN2

AAV2Null

Naive

AAV2rCLN2

AAV2Null

Naive

a b

Figure 3 Spatial distribution of TPP-I enzymatic activity following stereotactic injection of AAV2CUrCLN2. Three rats were injected with 5 ml ofAAV2CUrCLN2 containing 5� 109 pu into the left striatum, and as controls three rats were injected with 5 ml of AAV2CUNull containing 5� 109 pu intothe left striatum. After 4 weeks, the animals were killed and the fresh brain was excised from the six treated and from two age-matched naive (untreated)animals. Each brain was divided into nine sections of 2 mm width by coronal sectioning. Each section was then divided into left and right hemispheres andthe cortex was discarded leaving the internal structures (including the striatum). Sections were separated and homogenized. The TPP-I activity wasdetermined and normalized to protein concentration of homogenate. The plots show the spatial distribution along the rostral/caudal axis for the TPP-Iactivity. The vertical arrow shows the estimated location of vector administration. The section for which the TPP-I activity of three AAV2CUrCLN2-injectedand three AAVCU2Null-injected animals differ (Po0.05) is shown with an asterisk. (a) Left (injected) hemisphere; (b) right (uninjected) hemisphere.


1621

Gene Therapy

no sign of inflammatory cell infiltration or vascularcuffing.

The pathology of LINCL is diffuse throughout theCNS. To establish that the cellular expression of TPP-Icould be observed following delivery to CNS locationsother than the striatum, injections of AAV2CUrCLN2were made into several brain structures including thefrontal cortex, the parietal cortex (Figure 5), and thecerebellum (not shown). Examination of these regions 4weeks following delivery confirmed TPP-I expressionand neurotropism of AAV2 infection with TPP-I expres-sion seen in neuronal populations in all regions. TPP-Iexpression was widespread in the cortical injectionregions with neurons of both pyramidal and interneuronmorphologies showing TPP-I expression. There was noobserved TPP-I expression in astrocytes.

As a goal of the study was to obtain stable and long-term expression of the TPP-I transgene, the histologyof striatal TPP-I expression was examined at intervalsfrom 2 weeks to 18 months following injection ofAAV2CUhCLN2 (Figure 6). Transgene expression fromAAV2 vectors is known to increase gradually overseveral weeks and to persist for months or years.25–27

Consistent with this, at 2 weeks postinjection, there waslittle detectable human TPP-I protein in the rat striatum,but by 1 month, abundant human TPP-I protein wasobserved. The high level, stable AAV2-mediated expres-sion of human TPP-I in the rat striatum persisted atcomparable volume and intensity at 4, 8, 12, and 18months (Figure 5). Quantitative assessment of TPP-Iexpression is a challenge because of the difficulties in

standardizing the anti-TPP-I staining process, includingthe different reagent solutions used over time, and thedifferences in optimal immunostaining in the brains ofrats from different ages. With this caveat, the Cavalieriestimator was used to quantify the TPP-I detectedby immunostaining in the striatum over time (Table 1).The volume of cellular TPP-I expression expressed as apercentage of the striatum was variable ranging from 22to 67% of the striatum at different time points. There wasno correlation of average volume and time (P40.7 bylinear regression).

As has been observed with AAV2 vectors expressingother transgenes,28,29 we observed a high level of TPP-Iexpression in cell populations not directly at the siteof injection, but whose axonal projections were locatedin the site of the injection. As widespread delivery ofthe gene product throughout the brain is a goal ofthe current study, we examined other brain regionsthat project to the striatum to determine if expression ofTPP-I could be detected in projection neuronal popula-tions following single intrastriatal administration ofAAV2CUhCLN2. Interestingly, neurons in the cerebralcortex showed TPP-I expression as early as 2 monthsfollowing delivery, and expression was more widespreadat 4 and 8 months relative to the earlier time points(Figure 7). By 8 months, TPP-I expression was alsoobserved outside of the striatal injection area in neuronalpopulations in the thalamus and substantia nigra.Continued detection of TPP-I in neurons in these regionswas observed at 12 months (not shown) and 18 months.In addition, at 18 months, TPP-I-positive cells were also

Figure 5 Comparison of AAV2CUrCLN2-mediated expression of TPP-I in neurons in different delivery sites. (a) Delivery of AAV2CUrCLN2 to the frontalcortex, striatum, and parietal cortex results in TPP-I expression as observed in a parasagittal section at 4 weeks. Immunoperoxidase-positive cells areobserved in a wide region surrounding each injection site. (b) Higher magnification of the frontal cortex shows TPP-I positive cells with both the pyramidalmorphology of projection neurons and the multipolar morphology of local circuit neurons. (c) A similar pattern of TPP-I staining is observed in cells of theparietal cortex. (d) Low magnification, multiple immunofluorescence appearance of a frontal cortex injection site showing the range of cell distribution andthe morphology of TPP-I-expressing cells (green). (e) Colocalization of TPP-I expression (green) with a fluorescent nuclear marker (propidium iodide, red)and the astrocyte marker (GFAP, blue) shows that TPP-I is expressed by cells with a neuronal nuclear morphology, but not astrocytes in frontal cortexinjection sites. (f) Localization of TPP-I in cells with a neuronal morphology, but not astrocytes is also observed in parietal injection sites. Bars in each panelindicate the size for the different magnifications.


1622

Gene Therapy

observed in the contralateral (noninjected) hemisphere inthe cortical regions that send commisural projections(projections crossing from the other hemisphere) to thestriatum. In control rats, no TPP-I-positive cells weredetected in these areas using the human-specific mono-clonal anti-TPP-I antibody (not shown). Assessment ofthe extent of TPP-I-positive volume calculated as apercentage of the total hemisphere volume showed a

range of values (2.2–17.0%; Table 1). There was nocorrelation of the TPP-I-positive volume calculated as a% of the total hemisphere with time ranging from 1 to 18months (P40.8).

TPP-I distribution after AAV2CUhCLN2 administrationto the non-human primate brainExpression of hCLN2 mediated by AAV2 delivery to theCNS was also observed in the brains of African greenmonkeys following injection of AAV2CUhCLN2 (Figures8 and 9). Detectable levels of TPP-I expression wereobserved using human-specific monoclonal antibody viaimmunoperoxidase staining. The morphology of TPP-I-positive cells in the cortex suggested that by 5 weeksboth interneurons and pyramidal neurons expressedTPP-I, while staining in the caudate nucleus showedexpression in cells with the morphology of interneurons(Figure 8). Delivery to the hippocampus resulted in TPP-I expression in large multipolar hilar cells and, to a lesserextent, in dentate granule cells (Figure 9). Thus, success-ful expression of AAV2-mediated delivery of TPP-I in thenon-human primate brain was not region dependent.The identity of TPP-I-expressing cells was revealed bymultiple immunofluorescence staining using the TPP-Ihuman-specific monoclonal antibody. To discriminatebetween true signal and the contribution of autofluores-cence (predominantly seen at the shorter wavelengths innon-human primate brain tissue), detection of TPP-I wasaccomplished with secondary antibodies conjugated tothe fluorophore Cy5, that, upon far red spectral excita-tion, emits at infrared wavelengths. Consistent withdelivery to the rat brain, examination of TPP-I-expres-sing cells in the non-human primate brain by multipleimmunofluorescence staining revealed that neuronal, butnot glial, populations were infected at all injection sites(Figures 8 and 9). Persistence of expression is a key factorfor developing therapies for disorders such as LINCL. Toevaluate duration of TPP-I expression in the non-humanprimate brain, animals were assessed at 5 and 13 weeksand found to contain equivalent distribution of TPP-I-

Figure 6 Time course of TPP-I protein accumulation following AAV2CU -hCLN2 gene transfer. Rats (n¼ 3/group) were injected (1010 pu) into thestriatum with 1 ml of AAV2CUhCLN2. After various time points, TPP-Idistribution was assessed by anti-TPP-I immunoperoxidase staining onsagittal sections of the striatum of the injected hemisphere. Strongexpression of TPP-I was detected at all time points from 1 to 18 months.(a) Naive; (b) 2 weeks; (c) 1 month; (d) 2 months; (e) 4 months; (f) 8months; (g) 12 months; and (h) 18 months. Magnification bar¼ 1 mm forall panels.

Table 1 Volumetric analysis of TPP-I expression in rat brainresulting from a single intrastriatal injection of AAV2CUhCLN2a,b

Postdeliveryinterval(month)

TPP-I staining(% striatum)

TPP-I staining(% extrastriate

hemisphere)

TPP-Istaining(% total

hemisphere)

1 42.972.3 11.471.0 15.871.12 28.5710.2 10.670.2 13.471.24 39.474.4 13.371.9 17.072.28 67.371.2 11.870.3 16.170.3

12 46.074.0 7.270.5 10.270.718 22.572.5 0.670.5c 2.270.5

aTotal dose of 1010 particles units in 1 ml.bShown are volumes of TPP-I-positive staining as a % of differentreference volumes (striatum, extrastriate hemisphere, total hemi-sphere) 7standard error of the mean at different time pointsfollowing a single intrastriatal delivery of AAV2CUhCLN2; n¼ 3/rats were assessed for each time point.cOne animal in this group had no detectable extrastriate labelingand thus the mean is low and the variance is high.


1623

Gene Therapy

positive cells (Figure 9). As with the rats, other thanminor physical damage along the area of injection,there were no signs of inflammatory cell infiltration orvascular cuffing.

Discussion

LINCL is a fatal childhood disease with no knowntherapy. Several features of LINCL make it a good

Figure 7 Expression of TPP-I in structures far from the injection site. Rats (n¼ 3/group) were injected into the striatum (Str) with 1010 pu in 1 ml ofAAV2CUhCLN2. TPP-I distribution was assessed by immunoperoxidase staining. Cell populations expressing robust levels of TPP-I were observed ina number of distant anatomical projection sites at all time points. Examples from 8 month (a–d) and 18 month (e–h) postinjection times are shown.(a) Representative sagittal section of brain showing location of structures analyzed in subsequent panels. This section is medial to the striatum so that onlya tangential profile of the striatum is present here with sparse TPP-I expression. A section through the center of the striatum confirming TPP-I expressionin this same animal is shown in Figure 6f; (b) high magnification of frontal cortex with TPP-I accumulation in pyramidal projection neurons; (c) thalamus,TPP-I stained multipolar neurons; (d) Substantia nigra, TPP-I stained large multipolar neurons; (e) frontal cerebral cortex, injected hemisphere TPP-Iaccumulation in pyramidal projection neurons at 18 months following gene delivery; (f) thalamus, injected hemisphere, TPP-I stained multipolar neurons;(g) Substantia nigra, injected hemisphere, TPP-I stained large multipolar neurons; (h) cerebral cortex, noninjected hemisphere with TPP-I accumulation inpyramidal projection neurons. Control rats had no TPP-I-positive cells in these areas (not shown). Magnification bar is indicated on panel a (for panel aalone) and b (for panels b–h). Cells indicated by the box are shown in higher magnification in the inset.


1624

Gene Therapy

candidate for gene therapy, including the monogeneticnature of the disease, knowledge of the deficient TPP-Iprotein and the gene (CLN2) that encodes it, a need foronly low target levels of the therapeutic protein, and theability of this protein to be secreted and internalized intoneighboring cells providing potential for cross-correc-tion.4,21,23,30 In regard to technologies available today,gene therapy is the best option for this disorder to protectneural cells from death. The present study describes thesuccessful delivery and long-term expression of CLN2 byan AAV2-based vector, and the assessment of its geneproduct, TPP-I, following delivery to rat and non-humanprimate brain. The data fulfills three critical requirementsfor the preclinical development of therapy for thisdisease: (1) long term duration of expression of the

therapeutic protein; (2) broad distribution of the ther-apeutic protein which may be a composite of infectionand cross-correction; and (3) achieving sufficient levels ofexpression of the protein that should have therapeuticeffect.

DurationFor the treatment of a chronic metabolic disorder such asLINCL, it is necessary to achieve long-term expression ofTPP-I in the CNS. In this respect, the choice of vector iscritical. In agreement with other reports,31–34 we haveshown that an AAV2-based vector can maintain expres-sion of the transgene and hence of the therapeuticprotein for 41 year. In addition to the choice of vector,the choice of promoter is also critical for long-term high

Figure 8 TPP-I expression by neurons in the cortex and caudate nucleus of the non-human primate brain 5 weeks following administration ofAAV2CUhCLN2. (a) Low magnification view of immunoperoxidase-stained TPP-I-positive cells in the head of the caudate nucleus. (b) The same regionshown in (a) at higher magnification showing the widespread TPP-I expression in multipolar cells. (c) Similar view to (b) showing widespread expression ofTPP-I in the caudate nucleus identified by immunofluorescence (green). (d) High magnification of TPP-I expression (green) in the head of the caudatenucleus in neurons stained with calbindin (red), but not astrocytes identified by their expression of GFAP (blue). (e) High magnification view of TPP-I-expressing cells (green) in the cortex, identified as neuronal by their morphological staining with propidium iodide (red) and absence of GFAP staining(blue). (f) Similar cortical region to that shown in (e) in a naive control animal using the same labeling as described in (e), showing that the detection ofTPP-I could be discriminated from background autofluorescence. Bars in each panel indicate the size for the different magnifications.


1625

Gene Therapy

level expression, as some viral promoters have beenlinked to premature decline of transgene expression.35,36

In the present study, we utilized the CMV-enhancedchicken b-actin promoter with the b-actin/rabbitb-globin hybrid intron, a promoter/intron combinationthat has been shown to successfully drive the expressionof an AAV vector-mediated transfer of a gene in thebrain for 41 year.16 The combination of the long-termexpressing vector AAV2 and a strong nonviral promoterled to robust TPP-I expression in the brain of rats for 418

months and for at least 3 months (the longest time periodexamined to date) in the brains of non-human primatesin the present study. Owing to resources needed, there islimited data on long-term transgene expression in non-human primates beyond 90 days as described here,although one report shows persistent expression up to134 days.13 In rats, after 1 month, although there wasvariability, there was no significant time-dependentchange in the extent of transgene expression followingAAV2CUhCLN2 gene transfer as assessed by the TPP-I-

Figure 9 TPP-I expression in the non-human primate hippocampus at 5 and 13 weeks and in the cortex and caudate nucleus at 13 weeks followingadministration of AAV2CUhCLN2. Panels (a–d), 5 weeks. (a) Low magnification view of immunoperoxidase staining for TPP-I at 5 weeks in thehippocampus. (b) Region of the hippocampal formation shown at higher magnification. TPP-I-positive staining is observed in cells of the dentate gyrus, thecurving structure at the left and bottom of the panel. There is robust TPP-I staining of multipolar cells in the dentate hilus. (c) TPP-I-positive cells (green)in the hippocampus detected by multiple immunofluorescence staining (neuronal cells identified by morphological staining with propidium iodide, red andglial cells in blue with GFAP staining). (d) Higher magnification of the region in panel (c) shows punctate cytoplasmic accumulation of TPP-I (green) incells of neuronal morphology of both the hilus and dentate granule cells as shown by their appearance with propidium iodide labeling (red). TPP-I stainingis not seen in astrocytes (GFAP, blue, panels c and d). Panels (e–g). High magnification view of TPP-I expression in various regions followingadministration of AAV2CUhCLN2 to the non-human primate CNS 13 weeks as assessed by immunoperoxidase staining. (e) Caudate nucleus; (f) cortex; and(g) hippocampus. Bars in each panel (a–d) indicate the size for different magnifications. The magnification for panels (e–g) is indicated in panel (e).


1626

Gene Therapy

positive volume expressed as a percentage of the injectedstriatum or the injected hemisphere.

TPP-I distributionRobust transduction of neurons was achieved followingadministration of the AAV2CUhCLN2 vector to the ratand non-human primate brain. The neurotropic specifi-city of AAV2-based vectors has been previously ob-served with AAV2-mediated gene delivery.11,37,38 Theneurotropism of AAV2 makes this vector suitable for useas a gene delivery vehicle for neurons, the target cellpopulation for intervention in LINCL.2,9 In the rat, asingle intrastriatal injection resulted in histologicaldetection of TPP-I throughout both dorsal and ventralextents of the striatum. The extent of TPP-I expressionencompassing, nearly half of the striatum, remainedstable over time following gene delivery.

Of interest relevant to the goal of achieving wide-spread expression of TPP-I in the CNS was theobservation in the rat that distant neurons expressedTPP-I after a lengthy interval following delivery. Thedistribution of these sites of extrastriatal TPP-I expres-sion was highly variable over time but consistent withthe known circuitry of striatal projections. Neurons inboth the dorsal and ventral striatum receive input fromcerebral cortex, thalamus, and the midbrain ventraltegmental area and substantia nigra.39–41 Thus, distantneurons, whose axons terminate in the region of vectordelivery, were able over time to express TPP-I. There aretwo possible explanations for TPP-I expression in distantprojection neurons. Previous reports have demonstratedthat distant neurons can be transduced by retrogradetransport of the viral vector following AAV2 delivery tothe CNS.28 Alternatively, the secreted TPP-I enzymecould be taken up by synaptic terminals and undergoretrograde transport to distant cell bodies. This mechan-ism would be consistent with the in vitro cross-correctiondata showing that TPP-I is secreted by cells transducedby AAV2CUhCLN2 and subsequently taken up bynontransduced cells in a mannose-6-phosphate-depen-dent pathway. It would also be consistent with theextensive intracellular staining for TPP-I observed intransfected cells, where TPP-I was likely present in allcompartments of the secretory and endocytic pathways.Whatever the mechanism of this retrograde expression,these results suggest that it may be possible to utilize thecircuitry of neuronal connections to extend the effectiveexpression of TPP-I to distant regions. This possibilitygreatly augments the therapeutic potential of eachintracerebral injection and suggests that the circuitry ofthe brain should be taken into account when designingtherapeutic delivery.

In a careful survey of sections of the brains of theAAV2CUhCLN2 non-human primates, no TPP-I-positivecells far from the injection site suggestive of retrogradetransport were seen. There may be two reasons for this.First, the retrograde transport observed in rats increasedover time with a maximum at 8 months and monkeyswere assessed at shorter time points due to resourcelimitations. Second, the sensitivity of the immunohisto-chemistry in the non-human primates was titrated downby use of lower levels of the monoclonal anti-TPP-Iantibody than that used in the rats (1:15 000 as comparedto 1:1000). This was done to suppress the endogenous

TPP-I background in the non-human primate tissue butthis also reduces the sensitivity with which retrogradetransport may be seen.

TPP-I protein levelsAt present, there is no defined clinical benchmark for theminimum level of TPP-I expression required for success-ful therapeutic intervention for LINCL. Individuals whoare heterozygous for LINCL have no disease phenotypesuggesting that it will not be necessary to completelyrestore normal TPP-I levels.1–3 Furthermore, individualswith LINCL who have a mutation in CLN2 that results in5–10% of normal TPP-I activity have delayed onset of thesymptoms,23 suggesting that as little as 5% of normalactivity spread throughout the brain represents a reason-able target level for clinical benefit.

While a target of 5% of normal levels makes the goal ofeffective therapy for LINCL easier to achieve with genetherapy, the available analytical methods are not suffi-ciently sensitive to demonstrate an increase in only 5%over a background of normal TPP-I levels in vivo. Forexample, the variance in the TPP-I activity measurementon brain homogenates of naive rats was 24%, a valuemuch greater than the target level. However, a TPP-Ilevel was obtained that was 55% above backgroundfollowing AAV2CUhCLN2 infection demonstrating thatlocally therapeutic levels are easily achieved. As nocharacterized TPP-I-deficient animal model is available,we are only able to evaluate TPP-I levels that aresignificantly increased over the background. There is awide distribution of robust TPP-I-positive cells detectedfollowing AAV2-mediated gene delivery using immuno-histochemistry under conditions that gave no signal forendogenous TPP-I. Although immunohistochemicalstaining does not allow for a linear quantitative esti-mation of protein levels, this technique is also notparticularly sensitive to very low intracellular proteinlevels. Thus, immunohistochemical detection likelyunderestimates the true extent of CLN2 gene productexpression, suggesting that at least the minimum targetlevels of TPP-I expression have been achieved in theidentified regions.

Implications for future studiesThis study supports the notion that AAV2-mediatedCLN2 gene delivery to the CNS can meet the criteria ofwidespread, robust transgene expression of long dura-tion required for the development of a therapeutic genedelivery strategy. The recent characterization of a CLN2-deficient mouse,24 which displays a similar pathologyas LINCL in humans will also allow the potentialtherapeutic benefit of intracranial gene transfer to beassessed more rigorously. There is currently favorableshort-term safety data related to the intracranial transferof AAV2CUhCLN2 in human patients based upon thegrowing clinical experience of AAV2-mediated genetransfer to the brain for the treatment of Canavan andParkinson’s diseases.42,43 Therefore, it seems likely thatthere would be no significant safety issues related tointracranial transfer of AAV2CUhCLN2 in humanpatients. In this context, LINCL is a candidate diseasefor further preclinical development of AAV2-mediatedgene therapy.


1627

Gene Therapy

Methods

Production of AAV2 vectorsThe overall genome structure of all AAV vectors used inthe study included (50–30): the 141 nt inverted terminalrepeat (ITR) from AAV2; an expression cassette thatincluded the CAG promotor (the human cytomegalo-virus immediate/early enhancer; the promoter, the splicedonor, and the left-hand intron sequence from chickenb-actin; the right-hand intron sequence and spliceacceptor from rabbit b-globin);44–46 the cDNA of interestwith an optimized Kozak consensus translation initiationsignal prior to the start codon; the rabbit b-globin polyA/stop sequence; and the 141 nt ITR. The vectors wereproduced using two plasmids: (1) ‘AAV2-ITR/expres-sion’ plasmid containing (50–30): the AAV2 ITR, CAGpromoter, the cDNA of interest, the polyA/stop, and theAAV ITR; and (2) adenovirus/AAV helper gene plasmid(pPAK-MA2) containing (50–30): the mouse mammarytumor virus ITR promoter driving the AAV2 rep gene,followed by the AAV2 cap gene and the adenovirusserotype 5 VA, E2, and E4 genes, each driven by theirown promoter.47,48

The AAV2 vectors used in the study included:AAV2CUhCLN2 (expression cassette includes the humanCLN2 cDNA); AAV2CUrCLN2 (identical to AAV2CU -hCLN2 except that it has the rat CLN2 cDNA substitutedfor the human CLN2 cDNA); and AAV2CUNull (used asa control, identical to AAV2CUhCLN2 except that it hasan untranslatable DNA sequence corresponding tointron 5 of the human vascular endothelial growth factorgene to match the size of the human CLN2 cDNA); andAAV2GFP used as a negative control (see Zolotukhinet al49 for a description of AAV2GFP).

All of the AAV2 vectors were produced in the samemanner using Polyfectt (QIAGEN Sciences, German-town, MD, USA) mediated cotransfection of 500 mg of theexpression cassette plasmid containing the transgene ofinterest and 1 mg of the AAV/adenovirus helper geneplasmid into a 10 Stack Cell Factory (NUNCt BrandProducts, VWR Scientific, West Chester, PA, USA) at 70%confluence of low passage 293 cells. Post transfection(72 h), the cells were harvested and a crude viral lysatemade by three cycles of freeze/thaw, then clarified bycentrifugation and purified by discontinuous iodixanolgradients and heparin agarose chromatography.50

All vectors were assessed by quantitative PCR (Taq-Man), capsid ELISA (Research Diagnostics, Flanders, NJ,USA), and by their ability to replicate in adenovirus-infected C12 cells (a cell line expressing the AAV2 repand cap genes) with detection of the amplified genomeby DNA hybridization.51 Evidence that the CLN2 vectorscould direct the expression of the desired transgeneproduct was established by assessing TPP-I enzymeactivity (as described below). For the in vitro and in vivoexperiments, all vectors were dosed on the basis ofparticle units as assessed by ELISA (Research DiagnosticInc., Flanders, NJ, USA).

Morphological evaluation of TPP-I protein expressionin vitroThe AAV2CUhCLN2 vector was evaluated for its abilityto produce TPP-I by indirect immunofluorescence andimmunoperoxidase detection of CLN2 expression.

293ORF6 cells, from a human embryonic kidney cell lineexpressing adenovirus E1 and E4 genes52 were main-tained in Dulbecco’s modified Eagle’s medium (DMEM,Invitrogen Life Technologies, Carlsbad, CA, USA) sup-plemented with 1 mM sodium pyruvate, 10% fetal bovineserum (FBS), 50 U/ml penicillin, 50 mg/ml streptomycin,2 mM L-glutamine, and 0.25 mg/ml fungizone-ampho-tericin. For the AAV expression studies, the 293ORF6cells (5� 104 cells) were infected with AAV2CUhCLN2or AAV2CUNull (103 particles/cell) in 30 ml serum-freeDMEM at 371C. After 30 min, 2 ml of DMEM supple-mented with 100 mM ZnCl2 were added to the cells toactivate the E4 gene expression.52 After 2 days, the cellswere washed twice with phosphate-buffered saline, pH7.4 (PBS) and fixed with 4% paraformaldehyde (ElectronMicroscopy Sciences, Fort Washington, PA, USA) in PBS.For immunoperoxidase detection of TPP-I, the cellswere then pretreated with 0.03% hydrogen peroxide(Sigma-Aldrich, St Louis, MO, USA) for 5 min, 221C toquench endogenous peroxidase activity. The cells werethen treated (20 min, 221C) with blocking medium (5%goat serum, 1% bovine serum albumin (BSA), in PBS).Primary antibodies were diluted in blocking medium(1:100), including mouse anti-human CLN2 hybridomasupernatant (clone mAb 8C4, provided by KrystynaWisniewski, Institute for Basic Research in Develop-mental Disabilities, Staten Island, NY, USA) and mouseanti-shiga toxin hybridoma supernatant as an irrelevantprimary antibody control (clone 11F11, American TypeCulture Collections, Manassas, VA, USA). The primaryantibodies were applied for 60 min at 371C, removedwith three 5 min washes with 1% BSA in PBS. Horse-radish peroxidase-based immunochemistry was per-formed using the Envision kit (Dako, Carpinteria, CA,USA) according to the manufacturer’s instructions.Briefly, horseradish peroxidase-conjugated secondaryantibody (anti-mouse) was applied to cells for 30 minat 371C, washed three times in PBS, and treated with thediaminobenzidine-activator cocktail for 5 min at 221C.Cells were counterstained using hematoxylin (Sigma-Aldrich) and imaged using a Leaf MicroLumina colorcharge coupled device (CCD) camera (ElectroImage,Great Neck, NY, USA).

For immunofluorescence detection of TPP-I, the cellswere infected and fixed as described above, and then,after three washes in PBS, cells were postfixed in 100%methanol (�201C, 20 min), and washed three times inPBS. The cells were then blocked and treated withprimary antibodies as described above with the excep-tion of hybridoma supernatants being diluted 1:10instead of 1:100 and application at 221C instead of371C. The primary antibodies were applied for 60 minand removed as described above and replaced with asecondary antibody for 45 min at 221C (dilution of 1:100in blocking medium, goat anti-mouse Alexa 488, Mole-cular Probes, Eugene, OR, USA). The secondary antibodywas removed with three 5 min washes with 1% BSA inPBS followed by three rinses in PBS. Nuclei were stainedwith 40-6-diamidino-2-phenylindole (DAPI, 2 mg/ml inPBS, 5 min, 221C, Molecular Probes). Cells were observedusing widefield epifluorescence microscopy using anOlympus IX70 microscope. Images were capturedusing a Photometrics Quantix cooled CCD camera andanalyzed using MetaMorph image analysis program(Universal Imaging, Downingtown, PA, USA).


1628

Gene Therapy

Evaluation of TPP-I enzyme activityThe function of vector-derived TPP-I was demonstratedby assessing TPP-I enzyme activity using a fluorescent-activated specific substrate. In 24-well plates, 106

293ORF6 cells were incubated until 90% confluentand then infected (108–1010 pu) with AAV2CUhCLN2,AAV2GFP (an irrelevant transgene control), or AAV2CU -Null. Medium was collected 48 h postinfection andtransferred to a 96-well plate. The cells were harvestedseparately and suspended in 100 ml of 0.15 M NaCl, 0.1%Triton X-100 to make a lysate. Secreted enzyme in 10 ml ofmedum was activated with 20 ml of an activating solution(150 mM NaCl, 0.1 % Triton X-100, and 50 mM formicacid, pH 3.5) to mimic its natural activation in thelysosomes following reuptake.21 After incubation at 371Cfor 30 min, 250 mM Ala-Ala-Phe-7-amido-4-methylcou-marin (the TPP-I-specific substrate; Sigma-Aldrich) in40 ml of 150 mM NaCl, 0.1% Triton X-100, and 100 mM

sodium acetate, pH 4.0 was added and a fluorescence(360/20 nm excitation, 460/25 emission) time course wasthen determined at 10 min intervals for 80 min at 371C ina Cytofluor 4000 TC plate reader (PerSeptive Biosystems,Foster City, CA, USA).21 The activity of the vector-derived TPP-I was determined by calculating the changein FU per viral particle dose per min over a linear portionof the curve minus the corresponding activity in cellsinfected with a control vector encoding an irrelevanttransgene. This assay provides an intrinsic specificationfor vector potency that can be applied independent ofyield, a parameter that varies between different viralpreparations. In this assay, vector preparations wereaccepted that gave 43� 10�6 FU/(min-ml media-pu). Todemonstrate specificity of the assay, it was performed inthe presence and absence of a TPP-I-specific inhibitor,Ala-Ala-Phe-chloro-methyl ketone (1 mM, Sigma-AldrichCo; see Lin and Lobel21 and Ezki et al53).

In vitro cross-correction of LINCL-deficient fibroblastsmediated by AAV2CUhCLN2To assess the ability of AAV2CUhCLN2 to direct theexpression of a CLN2 product that will cross-correctneighboring cells, the medium of quadruplicate wells of293ORF6 cells infected with 3� 104 pu/cell AAV2CU -hCLN2 or AAV2CUNull was removed 24 h postinfectionand the cells were washed twice with PBS. It was thenreplaced by fibroblast medium (RPMI 1640 mediacontaining 1% glutamine, 10% FBS, 50 U/ml penicillin,50 mg/ml streptomycin, 2 mM L-glutamine and 0.25 mg/ml fungizone-amphotericin) for 72 h. This medium wascollected and assessed for pro-TPP-I content by enzy-matic assay, and for AAV2CUhCLN2 content by Taqman.LINCL fibroblasts (provided by P Lobel, Robert WoodJohnson Medical College) were cultured in fibroblastmedium until 70% confluent at which time 30% of themedium was replaced by the media collected from the293ORF6 cells with or without 1 mM mannose-6-phos-phate as a competitive inhibitor of uptake. After 24 h, thecells were harvested and lysed in 150 mM NaCl, 0.1%Triton X-100 for assessment of TPP-I activity.

To insure that the TPP-I activity in the LINCLfibroblasts did not result from carry over of AAV2CU -hCLN2 in the media from the 293ORF6 cells initiallyinfected by the vector, carry over was assessed in themedia using TaqMan real-time quantitative PCR with

primers (forward: GTCAATGGGTGGAGTATTTACGGand reverse: AGGTCATGTACTGGGCATAATGC) andprobe (CAAGTGTATCATATGCCAAGTACGCCCCCT)specific to the CMV enhancer. The amount of AAV2CU -hCLN2 determined to be carried over by this methodwas used to directly infect the LINCL fibroblasts withassessment of the TPP-I activity in the media after 72 h.

CNS administration of AAV2 vectors in ratsTo evaluate TPP-I expression and distribution followingAAV2-mediated CLN2 delivery in a small animalmodel brain, Fischer 344 male rats (180–200 g; Taconic,Germantown, NY, USA) were used with InstitutionalAnimal Care and Use Committee (IACUC) approval ofall procedures. The rats were deeply anesthetized andpositioned in a stereotaxic frame for unilateral injectioninto the left striatum (AP +0.6 mm, ML +2.8 mm, and DV�5.2 mm relative to Bregma), frontal cortex (AP +3.0, ML+2.8, DV �2.5), parietal cortex (AP �1.3, ML +2.5, DV�2.0), or cerebellum (AP �10.0, ML 2.5, DV �3.7). TheAAV vectors, formulated in PBS (1.0–5.0 ml per site), wereinjected at a rate of 0.2 ml/min using a microprocessor-controlled infusion pump (Model 310, Stoelting, WoodDale, IL, USA). The injection needle was left in positionfor 1 min prior to, and 2 min following delivery beforebeing slowly withdrawn from the brain. Animalsreceived either AAV2CUrCLN2 (109 or 1010 pu) orAAV2CUhCLN2 (109 or 1010 pu). Pilot experimentssuggested that there was no effect of total injectionvolume (at constant vector dose) on the distribution orlevel of TPP-I under these conditions (not shown). Otheranimals received AAV2CUNull (109 or 1010 pu) or PBS tocontrol for the specificity of transgene expression.

Enzymatic assay of TPP-I activity in brain homogenateTo assess TPP-I enzymatic activity in the brain ofAAV2CUCLN2-injected rats, vector injections were per-formed as described above, and at the relevant time, theanimals were euthanized and the brains were harvestedunfixed. The brains were rinsed in ice-cold PBS, placedin a rodent brain matrix, which was on ice (ElectronMicroscopy Sciences, Hatfield, PA, USA), and sliced intonine 2 mm coronal sections using matrix-specific razorblades. All sections were hemisected prior to analysis.Each half of the sections from the left hemisphere wasfurther dissected into cortex and noncortex (striatum).Samples were stored at �801C in microcentrifuge tubesuntil used. TPP-I activity was assayed by homogeniza-tion of the brain tissue in blank solution (150 mM NaCl,1 mg/ml Triton X-100) using a disposable pellet pestleand 1.5 ml matching tube (Kimble-Kontes, Vineland, NJUSA). The homogenate was clarified by centrifugationand the supernatant was transferred onto a 96-well plateand assayed for TPP-I activity as described above. Theactivity of the transgene product was determined bycalculating the change in fluorescence units per min permg of protein (standardized by BCA protein assay, PierceBiotechnology, Rockford, IL, USA).

Morphological evaluation of TPP-I protein expressionin the rat brainTo assess TPP-I distribution and tropism at various timepoints after CNS administration of the vectors, rats weredeeply anesthetized and perfused transcardially with


1629

Gene Therapy

ice-cold PBS followed by ice-cold 4% paraformaldehydein 0.1 M phosphate buffer (200 ml each). The brains werethen harvested and immersed in the same fixativeovernight, equilibrated in 30% sucrose at 41C; serial50 mm thick sagittal sections were later produced byfreezing sliding microtomy.

Multiple immunofluorescence labeling was performedusing antibodies specific for neurons (mouse anti-NeuN,1:2500, Chemicon International, Temecula, CA, USA ormouse anti-calbindin, 1:5000, Swant, Bellinzona, Switzer-land) or astrocytes (guinea pig anti-GFAP, 1:500, Ad-vanced Immunochemical, Long Beach, CA, USA).Expression of the CLN2 transgene product was detectedusing antibodies against TPP-I. For specific detection ofthe human CLN2 transgene product, a mouse anti-human TPP-I antibody was used at a dilution of 1:1000.For detection of either human CLN2 or rat CLN2transgene products, a rabbit anti-TPP-I antibody (pro-vided by P Lobel, Robert Wood Johnson MedicalCollege) was used at a dilution of 1:1000. Whenappropriate, sections were counterstained with thefluorescent nucleic acid dye, propidium iodide (1:5000for 3 min; Molecular Probes). Propidium iodide labels allnuclei within the tissue section and acts as a ‘fluorescentNissl stain’, permitting the phenotypic identity to cellsto be established according to classic criteria used withNissl staining.54 Donkey secondary antibodies specificfor relevant species were conjugated to biotin, Cy2, Cy3,or Cy5 or streptavidin-conjugated fluorophores (1:500; allfrom Jackson ImmunoResearch, West Grove, PA, USA) asappropriate for the specific discrimination of primaryantibodies. Specificity of antibody detection was verifiedby staining uninfected tissue and by omission of primaryantibodies in infected tissue. In separate sections,immunoperoxidase detection of the rat CLN2 transgeneproduct was accomplished by quenching of endogenousperoxidase activity and incubation with rabbit anti-TPP-Ipolyclonal antibody (1:1000) or the human CLN2 trans-gene product by mouse anti-human TPP-I antibody(1:1000), followed by biotin–streptavidin amplification(Vectastain kit, Vector Laboratories, Burlingame, CA,USA) and visualization with a nickel chloride-enhanceddiaminobenzidine reaction. Images were obtained byconfocal microscopy (Olympus FluoView) for the im-munofluorescence-stained material or brightfield imageswere obtained by digital image acquisition (OlympusBX50; Nikon D100) for the immunoperoxidase-stainedmaterial. Volume estimates of the extent of detectableTPP-I expression in the rat brain was carried out usingthe Cavalieri estimator on a one in six series of immuno-peroxidase-stained sections through the entire hemis-phere exclusive of the cerebellum, brainstem, andolfactory bulb.55 Linear regression analysis was per-formed to assess volume changes over time.

CNS administration of AAV2 vectors in non-humanprimatesTo evaluate TPP-I expression and distribution followingAAV2-mediated CLN2 delivery in a large animal modelbrain, non-human primates were used. These studieswere carried out at the St Kitts Biomedical ResearchFoundation, a facility that operates under Public HealthService regulations with IACUC oversight and with anassurance document filed at the National Institute of

Health. African green monkeys (Chlorcebus aethiopssabaeus, male, 4.8–6.2 kg, feral, 5–10 years old) weredeeply anesthetized and restrained in a stereotacticframe equipped with Hamilton microsyringes, whichare controlled by a nano injector, stepper motorizedinjection pump (Stoelting Co., Wood Dale, IL, USA) forinjection to 12 locations through six burr holes (twodepths per burr hole), three per hemisphere of the AAVvector or diluent control. Injection coordinates for theprescribed brain structures in the African green monkeywere determined from previously injected and sectionedbrains and are translated to the stereotactic instrumentframe of reference. Injections were made to the headof the caudate nucleus and overlying cerebral cortex(16.2 mm apart), body of the caudate nucleus and theoverlying cerebral cortex (19.2 mm apart), and thehippocampal formation and overlying cerebral cortex(31.2 mm apart). Before and after each injection, the braintissue was allowed to seal around the syringe needle.The vector or PBS was delivered at each site at 1 ml/minfor a total volume of 180 ml (15 ml at each of 12 locationsdelivered over 15 min at each site). Each non-humanprimate received a total dose of 3.6� 1011 pu of theAAV2CUhCLN2 vector or 3.6� 1011 pu of the controlAAV2CUNull vector or the diluent PBS.

Assessment of TPP-I expression in the non-humanprimate brainTo assess TPP-I distribution and tropism in the non-human primate CNS, at various time points after vectorinjection, the non-human primates were deeply anesthe-tized and transcardially perfused with cold PBS. Thebrains were immediately removed and kept on ice whileuniform 4 mm coronal slabs were cut using a speciallyfabricated brain mold. The right hemisphere of each slabwas placed in 4% paraformaldehyde for 3–7 days forimmersion fixation prior to transfer to 30% sucrose. Theleft hemisphere of the slabs was saved for futureanalysis.

The 4 mm coronal slabs from the right hemispherewere further sectioned (50 mm slices) and used forimmunoperoxidase detection and immunofluorescencein methods similar to that described for the rat tissue.However, to help to discriminate transgene product fromendogenous protein, a 1:15 000 dilution of the primarymonoclonal anti-TPP-I antibody was used for the non-human primate brain rather than the 1:1000 dilutionused for sections of rat brain. To separate detection ofTPP-I immunofluorescence from autofluorescence innon-human primate tissue, secondary antibodies wereconjugated to Cy5, which fluoresces in the far red portionof the visible spectra. To maintain continuity to analysisof the rat tissue, signal detection of TPP-I was mapped tothe green channel for image composition.

Acknowledgements

We thank P Lobel, Robert Wood Johnson MedicalCollege, for providing the anti-TPP-I polyclonal antibodyand the CLN2-deficient fibroblasts; K Wisniewski,Institute for Basic Research in Developmental Disabil-ities, for providing the anti-TPP-I monoclonal antibody;B Ferris, R McKinney, M Lam, B Bergman, J Qiu,E Salvin, E Vassallo, R Abplett, and the staff at the


1630

Gene Therapy

St Kitts Biomedical Research Foundation for technicalassistance; and N Mohamed and T Virgin-Bryan forhelp in preparing this manuscript. These studies weresupported, in part, by Nathan’s Battle Foundation,Indianapolis, IN; and the Will Rogers Memorial Fund,Los Angeles, CA.

References

1 Lysosomal enzymes. In: Scriver CR, Beaudet AL, Sly WS, ValleD, Stanbury JB, Wyngaarden JB, Fredrickson DS (eds). TheMetabolic and Molecular Basis of Inherited Disease. McGraw-Hill,Inc.: New York, 1995, p 2427.

2 Williams RE et al. CLN 2. Classic Late Infantile NCL. In: GoebelHH, Mole SE, Lake BD (eds). The Neuronal Ceroid Lipofuscinoses(Batten Disease). IOS Press: Amsterdam, 1999, p 37.

3 Haltia M. The neuronal ceroid-lipofuscinoses. J Neuropathol ExpNeurol 2003; 62: 1–13.

4 Sleat DE et al. Association of mutations in a lysosomal proteinwith classical late-infantile neuronal ceroid lipofuscinosis.Science 1997; 277: 1802–1805.

5 Vines DJ, Warburton MJ. Classical late infantile neuronal ceroidlipofuscinosis fibroblasts are deficient in lysosomal tripeptidylpeptidase I. FEBS Lett 1999; 443: 131–135.

6 Palmer DN et al. Mitochondrial ATP synthase subunit c storagein the ceroid-lipofuscinoses (Batten disease). Am J Med Genet1992; 42: 561–567.

7 Umehara F et al. Accumulation of mitochondrial ATP synthasesubunit c in muscle in a patient with neuronal ceroidlipofuscinosis (late infantile form). Acta Neuropathol 1997; 93:628–632.

8 Ezaki J, Tanida I, Kanehagi N, Kominami E. A lysosomalproteinase, the late infantile neuronal ceroid lipofuscinosis gene(CLN2) product, is essential for degradation of a hydrophobicprotein, the subunit c of ATP synthase. J Neurochem 1999; 72:2573–2582.

9 Boustany RM. Batten disease or neuronal ceroid lipofuscinosis.In: Moser HW (ed). Handbook of Clinical Neurology (Vol. 22, 66):Neurodystrophies and Neurolipidoses. Elsevier Science B.V.: NewYork, 1996, p 671.

10 Birch DG. Retinal degeneration in retinitis pigmentosa andneuronal ceroid lipofuscinosis: an overview. Mol Genet Metab1999; 66: 356–366.

11 Kaplitt MG et al. Long-term gene expression and phenotypiccorrection using adeno-associated virus vectors in the mamma-lian brain. Nat Genet 1994; 8: 148–154.

12 McCown TJ et al. Differential and persistent expression patternsof CNS gene transfer by an adeno-associated virus (AAV) vector.Brain Res 1996; 713: 99–107.

13 During MJ et al. In vivo expression of therapeutic human genesfor dopamine production in the caudates of MPTP-treatedmonkeys using an AAV vector. Gene Therapy 1998; 5: 820–827.

14 Mandel RJ et al. Characterization of intrastriatal recombinantadeno-associated virus-mediated gene transfer of human tyro-sine hydroxylase and human GTP-cyclohydrolase I in a ratmodel of Parkinson’s disease. J Neurosci 1998; 18: 4271–4284.

15 Davidson BL et al. Recombinant adeno-associated virus type 2, 4,and 5 vectors: transduction of variant cell types and regions inthe mammalian central nervous system. Proc Natl Acad Sci USA2000; 97: 3428–3432.

16 Frisella WA et al. Intracranial injection of recombinant adeno-associated virus improves cognitive function in a murine modelof mucopolysaccharidosis type VII. Mol Ther 2001; 3: 351–358.

17 Haskell RE et al. Viral-mediated delivery of the late-infantileneuronal ceroid lipofuscinosis gene, TPP-I to the mouse centralnervous system. Gene Theapy 2003; 10: 34–42.

18 Skorupa AF et al. Sustained production of beta-glucuronidasefrom localized sites after AAV vector gene transfer results inwidespread distribution of enzyme and reversal of lysosomalstorage lesions in a large volume of brain in mucopolysacchari-dosis VII mice. Exp Neurol 1999; 160: 17–27.

19 Bosch A, Perret E, Desmaris N, Heard JM. Long-term andsignificant correction of brain lesions in adult mucopolysacchar-idosis type VII mice using recombinant AAV vectors. Mol Ther2000; 1: 63–70.

20 Sferra TJ et al. Recombinant adeno-associated virus-mediatedcorrection of lysosomal storage within the central nervoussystem of the adult mucopolysaccharidosis type VII mouse.Hum Gene Ther 2000; 11: 507–519.

21 Lin L, Lobel P. Production and characterization of recombinanthuman CLN2 protein for enzyme-replacement therapy in lateinfantile neuronal ceroid lipofuscinosis. Biochem J 2001; 357:49–55.

22 Lin L, Sohar I, Lackland H, Lobel P. The human CLN2 protein/tripeptidyl-peptidase I is a serine protease that autoactivates atacidic pH. J Biol Chem 2001; 276: 2249–2255.

23 Sleat DE et al. Mutational analysis of the defective protease inclassic late-infantile neuronal ceroid lipofuscinosis, a neurode-generative lysosomal storage disorder. Am J Hum Genet 1999; 64:1511–1523.

24 Sleat DE et al. A mouse model of classical late-infantile neuronalceroid lipofuscinosis based on targeted disruption of the CLN2gene results in a loss of tripeptidyl-peptidase I activity andprogressive neurodegeneration. J Neurosci 2004; 24: 9117–9126.

25 Xiao XF, Li JF, Samulski RJ. Efficient long-term gene transfer intomuscle tissue of immunocompetent mice by adeno-associatedvirus vector. J Virol 1996; 70: 8098–8108.

26 Lo WD et al. Adeno-associated virus-mediated gene transfer tothe brain: duration and modulation of expression. Hum GeneTher 1999; 10: 201–213.

27 Thomas CE, Storm TA, Huang Z, Kay MA. Rapid uncoating ofvector genomes is the key to efficient liver transduction withpseudotyped adeno-associated virus vectors. J Virol 2004; 78:3110–3122.

28 Kaspar BK et al. Targeted retrograde gene delivery for neuronalprotection. Mol Ther 2002; 5: 50–56.

29 Passini MA, Lee EB, Heuer GG, Wolfe JH. Distribution ofa lysosomal enzyme in the adult brain by axonal transportand by cells of the rostral migratory stream. J Neurosci 2002; 22:6437–6446.

30 Sondhi D et al. Feasibility of gene therapy for late neuronalceroid lipofuscinosis. Arch Neurol 2001; 58: 1793–1798.

31 Daly TM et al. Prevention of systemic clinical disease in MPS VIImice following AAV-mediated neonatal gene transfer. GeneTherapy 2001; 8: 1291–1298.

32 Passini MA, Wolfe JH. Widespread gene delivery and structure-specific patterns of expression in the brain after intraventricularinjections of neonatal mice with an adeno-associated virusvector. J Virol 2001; 75: 12382–12392.

33 Klein RL et al. Dose and promoter effects of adeno-associatedviral vector for green fluorescent protein expression in the ratbrain. Exp Neurol 2002; 176: 66–74.

34 Klein RL, King MA, Hamby ME, Meyer EM. Dopaminergic cellloss induced by human A30P alpha-synuclein gene transfer tothe rat substantia nigra. Hum Gene Ther 2002; 13: 605–612.

35 Prosch S et al. Inactivation of the very strong HCMV immediateearly promoter by DNA CpG methylation in vitro. Biol ChemHoppe Seyler 1996; 377: 195–201.

36 Fitzsimons HL, Bland RJ, During MJ. Promoters and regulatoryelements that improve adeno-associated virus transgene expres-sion in the brain. Methods 2002; 28: 227–236.

37 Bartlett JS, Samulski RJ, McCown TJ. Selective and rapid uptakeof adeno-associated virus type 2 in brain. Hum Gene Ther 1998; 9:1181–1186.


1631

Gene Therapy

38 Chamberlin NL, Du B, de Lacalle S, Saper CB. Recombinantadeno-associated virus vector: use for transgene expression andanterograde tract tracing in the CNS. Brain Res 1998; 793: 169–175.

39 Haber SN, Groenewegen HJ, Grove EA, Nauta WJ. Efferentconnections of the ventral pallidum: evidence of a dual striatopallidofugal pathway. J Comp Neurol 1985; 235: 322–335.

40 McGeorge AJ, Faull RL. The organization of the projection fromthe cerebral cortex to the striatum in the rat. Neuroscience 1989;29: 503–537.

41 Berendse HW, Groenewegen HJ. Organization of the thalamo-striatal projections in the rat, with special emphasis on theventral striatum. J Comp Neurol 1990; 299: 187–228.

42 During MJ, Kaplitt MG, Stern MB, Eidelberg D. SubthalamicGAD gene transfer in Parkinson disease patients who arecandidates for deep brain stimulation. Hum Gene Ther 2001; 12:1589–1591.

43 Janson C et al. Clinical protocol. Gene therapy of Canavandisease: AAV-2 vector for neurosurgical delivery of aspartoacy-lase gene (ASPA) to the human brain. Hum Gene Ther 2002; 13:1391–1412.

44 Niwa H, Yamamura K, Miyazaki J. Efficient selection for high-expression transfectants with a novel eukaryotic vector. Gene1991; 108: 193–199.

45 Daly TM et al. Neonatal intramuscular injection with recombinantadeno-associated virus results in prolonged beta-glucuronidaseexpression in situ and correction of liver pathology in muco-polysaccharidosis type VII mice. Hum Gene Ther 1999; 10: 85–94.

46 Daly TM et al. Neonatal gene transfer leads to widespreadcorrection of pathology in a murine model of lysosomal storagedisease. Proc Natl Acad Sci USA 1999; 96: 2296–2300.

47 Grimm D, Kern A, Rittner K, Kleinschmidt JA. Novel tools forproduction and purification of recombinant adenoassociatedvirus vectors. Hum Gene Ther 1998; 9: 2745–2760.

48 Qui JP, Mendez BS, Crystal RG, Hackett NR. Construction andverification of an Ad/AAV helper plasmid designed formanufacturing recombinant AAV vectors for human adminis-tration. Mol Ther 2002; 5: S47–S48.

49 Zolotukhin S et al. A ‘humanized’ green fluorescent proteincDNA adapted for high-level expression in mammalian cells.J Virol 1996; 70: 4646–4654.

50 Zolotukhin S et al. Recombinant adeno-associated virus purifica-tion using novel methods improves infectious titer and yield.Gene Therapy 1999; 6: 973–985.

51 Clark KR, Voulgaropoulou F, Fraley DM, Johnson PR. Cell linesfor the production of recombinant adeno-associated virus. HumGene Ther 1995; 6: 1329–1341.

52 Brough DE et al. A gene transfer vector-cell line system forcomplete functional complementation of adenovirus earlyregions E1 and E4. J Virol 1996; 70: 6497–6501.

53 Ezaki J, Takeda-Ezaki M, Kominami E. Tripeptidyl peptidase I,the late infantile neuronal ceroid lipofuscinosis gene product,initiates the lysosomal degradation of subunit c of ATP synthase.J Biochem (Tokyo) 2000; 128: 509–516.

54 Peterson DA. The use of fluorescent probes in cell countingprocedures. In: Evans SM, Janson AM, Nyengaard JR (eds).Quantitative Methods in Neuroscience-A Stereological Approach.Oxford University Press: Oxford, 2004, p 88.

55 Peterson DA et al. Central neuronal loss and behavioralimpairment in mice lacking neurotrophin receptor p75. J CompNeurol 1999; 404: 1–20.


1632

Gene Therapy

RESEARCH ARTICLE

Eradication of hepatocellular carcinoma xenograftsby radiolabelled, lipiodol-inducible gene therapy

Y Kawashita1, A Ohtsuru2,3, F Miki2, H Kuroda1, M Morishita2,3, Y Kaneda4, K Hatsushiba5,

T Kanematsu1 and S Yamashita2

1Department of Transplantation and Digestive Surgery, Graduate School of Biomedical Sciences, Nagasaki University, Nagasaki, Japan;2Department of Molecular Medicine, Graduate School of Biomedical Sciences, Nagasaki University, Nagasaki, Japan; 3Takashi NagaiMemorial International Hibakusha Medical Center, Nagasaki University Hospital, Nagasaki, Japan; 4Division of Gene Therapy Science,Graduate School of Medicine, Osaka University, Osaka, Japan; and 5Pharmaceutical Chemistry & Technology Research LaboratoriesResearch Center, Daiichi Radioisotope Laboratories, Ltd, Japan

The promoter region of the early-growth response-1(Egr-1)gene has been shown to be activated by external radiation,thus making a selective tumoricidal effect possible. A previousexperiment showed that the Egr-1 promoter can be activatedby internal radiation using radioisotopes as well as externalradiation. Internal radiation using I-131 lipiodol (I-131-Lip) hasbeen established as one of the most useful therapeuticstrategies against hepatoma. We herein linked the Egr-1promoter to the herpes simplex virus-thymidine kinase(HSV-TK) gene, and investigated its efficacy in hepatomagene therapy in combination with I-131-Lip. A luciferaseassay showed the Egr-1-promoter activity to be markedlyincreased in hepatoma tissue specimens in an I-131-dose-dependent manner, whereas a less than two-fold increase inthis activity was observed in other organs. In addition, theradioactivity derived from I-131 was selectively accumulatedin the tumor tissue specimens. To examine the efficacy of

EgrTK/ganciclovir (GCV) gene therapy in vivo, subcutaneoushepatoma xenografts in nude mice were transfected usinga hemagglutinating virus of Japan (HVJ)-liposome vector.Complete tumor regression was observed in all the EgrTK-transfected tumors following combination treatment withI-131-Lip and GCV 42 days after treatment without any sideeffects (n¼ 8). In contrast, the tumors continued to grow in allcontrol mice (n¼ 10). Furthermore, the serum a-fetoproteinlevels decreased in the combination therapy group, while theyincreased in the controls. In conclusion, these data indicatethat Egr-1 promoter-based gene therapy combined withinternal radiation has a selective effect on hepatoma tumorswhile also showing an improved in vivo efficacy. Thiscombination therapy might, therefore, be an effective humanhepatoma gene therapy, even in advanced multiple cases.Gene Therapy (2005) 12, 1633–1639. doi:10.1038/sj.gt.3302531; published online 4 August 2005

Keywords: Egr-1 promoter; hepatoma; HVJ-liposome; internal radiation; mouse model

Introduction

Hepatocellular carcinoma (HCC) is a common humanmalignancy with an extremely poor prognosis.1,2 Genetherapy represents a promising treatment for HCC.Among the potential suicide-gene therapies for cancer,the herpes simplex virus (HSV)-thymidine kinase (TK)/ganciclovir (GCV) system has been widely investi-gated.3–5 HSV-TK converts the nontoxic agent GCV intoa highly toxic phosphorylated GCV, which acts as achain terminator of DNA synthesis and an inhibitorof DNA polymerase.6–8 It is not yet possible to trans-duce therapeutic genes into all of the target tumor cells.However, one considerable advantage of the HSV-TKsystem is the ’bystander effect’, in which HSV-TK-negative tumor cells are destroyed as a result of their

proximity to any HSV-TK-positive tumor cells that havebeen exposed to phosphorylated GCV.9 The tumor-specific activation of the therapeutic gene is crucialin order to minimize the side effects of treatment,and suicide-gene therapy controlled by the a-fetoprotein(AFP) promoter has proved satisfactory in this respect inboth in vitro and in vivo models.10–14 However, the level ofAFP expression varies among cases, and nonmalignanttissue – particularly, chronic inflammatory tissue – canalso express AFP.2,15 In order to overcome the difficultiesof tumor-specific activation, radiation-inducible promo-ters can provide spatial control of irradiation, and mighttherefore be useful in the treatment of hepatomas.16

We previously reported a novel technique for hepato-ma-specific suicide-gene therapy, using a radio-inducibleearly-growth response-1 (Egr-1)-promoter containing sixCArG boxes corresponding to the radiosensitive andserum-responsive elements.17 The Egr-1 gene was pre-ferentially expressed in hepatoma cells compared withthe surrounding nontumor tissue, and its promoter wassignificantly activated after radiation exposure. How-ever, this approach might not be suitable for multiple

Received 17 September 2004; accepted 19 February 2005; publishedonline 4 August 2005

Correspondence: Dr A Ohtsuru, Takashi Nagai Memorial InternationalHibakusha Medical Center, Nagasaki University Hospital, 1-7-1 Sakamo-to, Nagasaki 852-8501, Japan


www.nature.com/gt

hepatomas, because whole-liver irradiation could be fatalto patients: firstly, as a result of direct injury fromirradiation, which produces free radicals that causeDNA-strand breaks; and secondly, through the non-specific activation of suicide genes. Recently, the Egr-1promoter was shown to be activated not only by externalradiation but also by internal radiation from isotopessuch as Tc-99m, Ga-67 and I-131.18 Importantly, I-131-labelled lipiodol (I-131-Lip) concentrates preferentiallywithin hepatomas, where it is retained for significantlylonger periods than in non-neoplastic tissues.19–22 Inaddition, I-131 generates 606-MeV b-rays that produce astrong effect over short distances, which could thereforeenhance the bystander effect through increased intra-cellular communication.23

In the present study, we investigated whether com-bined therapy comprising transduction of the EgrTKgene and internal radiation with I-131 was able to reducetumor development in HCC.

Results

Biodistribution of I-131-Lip in tumor-bearing miceanalyzed using autoradiography after injectioninto the tumor xenograftAutoradiography was carried out after I-131-Lip injec-tion in order to examine whether I-131-Lip accumulatedpreferentially in the tumor. These mice did not receiveiodide-uptake inhibitors such as NaI. As shown in Figure1, I-131-Lip accumulated preferentially at the tumor siteand remained there for up to 7 days, which is consistentwith the observed physiological half-life of I-131. Incontrast, no I-131-Lip accumulation was detected in anyother organs, including the thyroid and gut. A densito-metric analysis using Bio-imaging Analyzer System(BAS5000, Fuji) showed the diphasic accumulation ofradioisotopes.

Softex roentgenograms revealed the iodide uptake,which was visualized as white dots, only in tumorstreated with I-131-Lip. No accumulation in any otherorgans was observed (data not shown).

Egr-1-promoter activity in vivoAn in vivo luciferase assay was performed to examinewhether the Egr-1 promoter could be activated by I-131-Lip in tumor xenografts. EgrLuc-HVJ-liposome wasintroduced into tumors when they reached 8–10 mm insize, along with 0–20 mCi of I-131-Lip, which was injectedat 16 h post-transfection. The tumor tissue specimenswere excised and examined to determine their luciferaseactivity 40 h after transfection. Radiation-induced pro-moter activity was expressed as relative light units, aftercorrection using nonirradiated cell lines. The differencesin transfection efficiencies between tumors were negli-gible. Internal radiation therapy using I-131-Lip signifi-cantly increased the Egr-1-promoter activity in thehepatoma tissue specimens (4.4-fold at 2 mCi and 28.8-fold at 20 mCi), whereas no increase was observed in thenormal liver and lung tissue specimens (Figure 2).

Egr-1 protein expression in tumor tissue after I-131-LiptreatmentEgr-1 immunohistochemistry was carried out in tumorsafter treatment in order to examine the level of Egr-1gene expression. Faint staining was detected in thetumor cell clusters of untreated hepatoma tissue (Figure3a), but not in the absence of the primary antibody(Figure 3b). In contrast, a significant degree of Egr-1activation was observed in the tissue specimens treatedwith I-131-Lip (Figure 3c). These results clearly show thatEgr-1 gene expression can be induced by internalirradiation using I-131, as well as by external radiationexposure.

In vivo EgrTK/GCV treatment of tumor xenograftsin miceThe whole-body effective half-life of the therapeuticactivity, which was measured daily using a radiation-survey meter, showed an exponential decrease with amean of 8 days (data not shown). To further investigate

Figure 1 Biodistribution of I-131-lipiodol. I-131 accumulation wasdetected in mice using autoradiography. I-131-lipiodol (I-131-Lip)accumulated preferentially in the tumor. The iodide uptake was notobserved in other organs, such as the thyroid and the gut.

0

100

200

300

400

Tum

or (

0µC

i)

Tum

or (

2µC

i)

Tum

or (

20µC

i)

Tum

or (

cold

lpd)

Liv

er (

0µC

i)

Liv

er (

20µC

i)

Lun

g (0

µCi)

Lun

g (2

0µC

i)

Luc

ifer

ase

acti

vity

(R

LU

)

*

**

Figure 2 Egr-1-promoter activity in vivo. Egr-1-promoter activity wassignificantly increased by internal irradiation to tumors that weretransfected with the Egr-Luc gene. However, no increase in the activitywas detected in either the liver or lung tissue specimens. *Po0.05,**Po0.01 versus control groups.

Suicide-gene therapy for HCCY Kawashita et al

1634

Gene Therapy

the cytotoxic effects of EgrTK/GCV in vivo, Huh7 cells(1�107) were injected subcutaneously into nude mice.Gene therapy was initiated when the tumor volumereached approximately 200 mm3. No spontaneous tumorregression was observed in the untreated tumors duringthe course of treatment. The data for each treatmentgroup were calculated as a percentage of the initial (day0) volume, and were plotted as the mean fractionalvolume7standard error. I-131-Lip alone showed a minordose-responsive growth-suppressive effect (Figure 4a).In contrast, the EgrTK-transfected animals (n¼ 8) thatreceived irradiation without GCV exhibited no suppres-sion of the tumor volume compared to the untreatedanimals (n¼ 10). A retardation of tumor growth was seenin the EgrTK-transfected group that received GCVwithout irradiation (n¼ 8). Complete tumor regressionwas observed in all the EgrTK-transfected tumors thatreceived I-131-Lip and GCV from 24 days to 42 daysafter gene therapy (Figure 4b). Thereafter, seven ofeight tumors continued to show regression at 80 dayspost gene therapy, although one remaining tumorslowly regrew.

Serum AFP measurementAFP is commonly used as a diagnostic marker for HCCbecause of its high sensitivity, and human Huh7 cellsproduce human AFP. Therefore, in order to estimate theviable volume of implanted hepatoma tumors, the serumAFP levels were measured at the time points indicated inFigure 5. The serum AFP values progressively increasedin the untreated controls, whereas a strong suppressionwas observed in the EgrTK plus I-131-Lip group, andthese findings were consistent with those from the in vivogene-therapy data.

Histological findingsHematoxylin–eosin staining revealed massive apoptoticand partially necrotic cells with intratumoral lympho-cyte-infiltration lesions in the EgrTK/GCV plus I-131-Lipgroup (Figure 6d) compared to the untreated tumor(Figure 6a). In contrast, slight tumor cell damage withnecrosis, but without any inflammatory cell infiltration,was observed in the I-131-Lip-injected group (Figure 6c).

EgrTK/GCV treatment resulted in an increased numberof apoptotic tumor cells, but it did not cause any necrotictumor tissue damage (Figure 6b).

Figure 3 Egr-1 immunohistochemistry of the tumor xenograft. Low levels of Egr-1 gene expression were observed in nonirradiated hepatoma tissue (a), butnot in the absence of the primary antibody (b). Internal irradiation greatly increased Egr-1 gene expression in hepatoma tissue (c).

0.1

1

10

100

1 5 12 18 24 29 35 42

Control

I-131(20)

Egr-TK/GCVEgr-TK/GCV, I-131(20)

f (V

/ V0)

f (V

/ V0)

days

0.1

1

10

100

1 5 12 18 24 29 35 42

cold lpdLacZ / GCV, I-131(2)LacZ / GCV, I-131(20)

days

a

b

Figure 4 In vivo gene-therapy experiments in mice with hepatoma tumorxenografts. The injection of cold lipiodol did not cause tumor retardation.However, slight transient retardation was observed in the I-131-treatedgroup in a radiation dose-dependent manner (a). Minor growth retardationwas also observed in the groups treated with EgrTK/GCV or I-131-Lipalone. In contrast, complete tumor retardation was observed in the grouptreated with EgrTK/GCV plus I-131-Lip (b).


1635

Gene Therapy

Discussion

The lymphographic agent lipiodol (also known asiodized oil) has been shown to specifically localize andpersist within hepatoma tissue.24–27 Although lipiodolitself does not seem to have any significant anticancereffects, it allows anticancer drugs to be deliveredselectively to tumor tissue when applied as an emul-sion.28 Lipiodol contains sterile iodine 127, so thelabelling of lipiodol with radioactive I-131 can beachieved through a simple radioisotopic exchangemethod.29 The therapeutic effectiveness is primarilylimited to areas in which lipiodol accumulates, as I-131generates 606-MeV b-rays that produce strong effectsover distances of less than 2.0 mm.30 These findingsindicate that I-131-Lip has potential advantages for use

in hepatoma therapy.31–33 The selective delivery andretention of lipiodol are thought to result from thedeveloped neovasculature, enhanced permeability andthe poor reticuloendothelial systems observed in hepa-toma tissues. Despite this capacity for the selectiveaccumulation at the tumor site, clinical studies haveshown the effectiveness of such I-131-Lip therapy aloneto only be temporary, and early HCC recurrence wasobserved in many cases. In contrast, recent reports haveshown that I-131 therapy following a curative resectionsignificantly improved both the disease-free intervaland the survival rates.22,34 Furthermore, no significantside effects – such as liver toxicity, renal disorders,bone-marrow suppression, thyroid dysfunction orrespiratory dysfunction – have been reported afterI-131-Lip administration.

We previously reported that a total dose of 20 Gy ofexternal ionizing radiation could selectively and effec-tively activate the Egr-1 promoter, thus resulting intumor regression.17 However, it is not possible to focusspecifically on multiple tumor lesions, even when usingconformal irradiation. Furthermore, high doses (50 Gy orabove) of whole-liver irradiation are known to be toxicto normal livers,15 and livers with hepatitis or cirrhosisare thought to be even more highly radiosensitive.These factors have clear implications regarding thetreatment of HCC.

A recent report demonstrated that the Egr-1 promotercan be activated by both external and internal radiation,which thus prompted our investigation of I-131-Lip asboth an antitumor agent and a molecular switch forradio-inducible promoter Egr-1-based gene therapy. Ourresults revealed a 28.8-fold upregulation of the Egr-1promoter in tumors treated with 20 mCi of I-131-Lipcompared to the controls, thus resulting in the completeregression of treated tumor xenografts. Immunohisto-chemical analyses also showed diffuse Egr-1 activation inthe HCC tissue. Clinical trials have used between 50 and70 mCi of I-131-Lip administered via a tumor-supplying

0

20

40

60

80

100

120

140

160

180

1 10 24

Egr-TK/GCV, I-131(20)

I-131(20)

cold lpd

AF

P (

ng/m

l)

days

Figure 5 Measurements of serum AFP in tumor-bearing mice duringtreatment. The serum AFP values significantly decreased after treatmentin the EgrTK/GCV plus I-131-Lip group, which was consistent with thedata on tumor size. In addition, a slight decrease in serum AFP levels wasobserved in the group treated with I-131-Lip alone.

Figure 6 Histological findings in tumor tissue after treatment. Hematoxylin–eosin staining revealed apoptotic shrinkage in tumor cells (arrows) in theEgrTK/GCV group after treatment (b) compared to the normal control (a). In addition, severe steatotic changes (open triangle) were observed in thecytoplasm of the tumor cells of the I-131-Lip group (c). Both apoptotic cells (arrow) and necrotic area (arrow heads) were found in the group treated withEgrTK/GCV plus I-131-Lip (d), which supported the growth-suppressive effect of the combination therapy reported in the in vivo gene-therapy experiments.


1636

Gene Therapy

artery. The local radioactivity of the 20 mCi dose istherefore estimated to be 10 times lower than that usedin clinical I-131-Lip therapy. The selective uptake ofI-131-Lip into tumor cells was also confirmed byautoradiography.

Many different genes can be activated after irradiation,including c-fos, c-jun, TNF-a and NF-kB. However, theadvantage of the Egr-1 promoter is that it is tightlyregulated by irradiation and its basal activity is relativelylow in nonirradiated tumors. Indeed, our in vivoluciferase assay data showed the leakage of the trans-duced suicide gene to be extremely low, thus suggestingthat I-131-Lip could efficiently regulate therapeutic geneexpression. Although, we have not examined whetherrestimulation of the Egr-1 promoter is possible aftermultiple treatments with internal radiation, our in vivogene-therapy data suggested that HSV-TK gene expres-sion might be prolonged. We have previously shown thatthe Egr-1 promoter was relatively hepatoma tissuespecific.17 More interestingly, the Egr-1 gene has recentlybeen shown to be overexpressed in prostate cancer andto be closely associated with its carcinogenesis.35

Regarding the optimal method for administeringI-131-Lip to patients, arterial chemoembolization isgenerally performed to treat multiple HCC. However,repeated embolization is limited by ischemic damageto normal livers and/or occlusion of the main feedingartery. Importantly, I-131-Lip therapy without emboliza-tion can be repeated in patients with hepatoma, thusmaking internal radiation-controlled suicide-gene ther-apy suitable for use even in regions of tumor recurrence.

Regarding future human clinical trials, a subcutaneousHCC implantation model may not be an ideal situation.If adequate animal models are developed, further studieswill be necessary to establish an adequate treatmentmethod for intrahepatic tumors with this strategy withan ultimate goal of future clinical application.

Up to now, adenoviruses have proven to be the mostefficient gene-transfer vectors. However, the clinicalfindings from gene therapies using these vectors directedat the liver have also revealed fatal side effects that occuras a result of systemic inflammatory responses.36 Safergene-transfer methods, such as nonviral vectors, musttherefore be considered for liver-related gene therapy.In the present study, we used an HVJ-liposome vector forthe transfer of genes, which proved to be efficient, safeand simple to use in vivo. This vector was constructedfrom inactivated HVJ envelope proteins and anionicliposomes. It therefore had a low immunogenicity, whichallowed for repeated administrations without anysignificant tissue damage.37–39 The gene-transfer effi-ciency might thus be further improved by using otherlipid formations or cointroduced proteins. Single appli-cations of HVJ-liposome-mediated gene transfer aretransient, as the transferred gene remains episomal inthe nucleus, similar to adenoviruses and other physicalmethods. However, repeated gene transfer might resultin a long-term gene expression. Moreover, previousstudies have detected no obvious functional or histolo-gical liver damage after gene transfer using the HVJ-liposome vector.39 These preliminary results suggest thatthe HVJ-liposome method has several advantagescompared with other gene-delivery systems for thein situ transduction of cancer cells, even though thevector itself does not have any specificity for tumor cells.

In conclusion, radio-inducible suicide-gene therapyin combination with I-131-Lip treatment was foundto enhance the efficacy of gene therapy for hepatomas.This approach might therefore represent a potentiallyeffective treatment modality for clinically unresectablemultiple HCC.


Plasmid constructionEgrLuc and EgrTK plasmids were constructed accordingto the method reported previously.17 Briefly, a region ofthe murine Egr-1 promoter (�425 to +0) was amplifiedusing the polymerase chain reaction with specificprimers from the pE425 plasmid,40 under the kindguidance of Dr DW Kufe of the Dana-Farber CancerInstitute, Harvard Medical School, USA. The polymerasechain reaction fragment was inserted upstream of fireflyluciferase or the HSV-TK gene, producing the EgrLuc orEgrTK plasmid, respectively.

Cell linesThe human hepatoma cell line Huh7, which was kindlyprovided by Dr H Nakabayashi of the Department ofBiochemistry, Hokkaido University School of Medicine,Japan, was cultured in RPMI 1640 medium (Gibco, LongIsland, NY, USA) supplemented with 5% fetal bovineserum, 100 U/ml penicillin and 100 mg/ml streptomycin.

Preparation of the HVJ-liposome vectorThe HVJ-liposome vector was prepared according to themethod described previously. Briefly, phosphatidylser-ine, phosphatidylcholine and cholesterol were mixed at aweight ratio of 1:4.8:2. The lipid mixture (10 mg) wasdried by the removal of chloroform, and then hydratedin 200 ml balanced salt solution containing DNA–high-mobility group I complex (200:64 mg), which had beenincubated previously at 201C for 1 h. Liposomes wereprepared by shaking and sonication. The liposomesuspension (10 mg lipids) was mixed with HVJ(30 000 HAU) in a total volume of 1 ml balanced saltsolution. The mixture was incubated at 41C for 10 min,and then for 1 h with shaking at 371C. Free HVJ wasremoved from the HVJ-liposome solution using sucrosedensity gradient centrifugation, and the second layerof sucrose containing the HVJ-liposome vector wascollected.

Labelling of lipiodol with I-131I-131-labelled lipiodol, which was produced by replacingthe iodine residue of lipiodol with I-131-C104 underaseptic conditions, was supplied by The First Radio-isotope Co. Ltd (Tokyo, Japan). The labelling index wasconsistently more than 99% and the radioactivity of theI-131-labelled lipiodol was 131 MBq/ml.

In vivo luciferase assayTo correct for transfection efficiencies, the cytomegalo-virus–renilla (sea pansy) luciferase vector was cotrans-fected in each experiment. EgrLuc (196 mg) andcytomegalovirus–Luc (4 mg) plasmids were encapsulatedin the HVJ-liposome and transfected into establishedtumors in nude mice. Based on the findings of previousreports, we administered 2–20 mCi of I-131-labelled


1637

Gene Therapy

lipiodol to each animal in a total volume of 100 ml,diluted with cold lipiodol 24 h after transfection.41

Subsequently, 48 and 72 h after transfection, the tumors,livers and lungs of the animals were excised andimmersed in 1200 ml of 3� reporter lysis buffer, andthen were cut with sterile scissors. The mixture wastransferred to a 15 ml falcon tube and homogenizedusing a sonicator for 5–10 min, followed by centrifuga-tion at 3000 r.p.m. for 5 min. The supernatant was thentransferred into a 1.5 ml Eppendorf tube. Aliquots wereassayed for luciferase activity in a Lumat LB9501luminometer (Berthold Systems, Aliquippa, PA, USA)using 20 ml of supernatant and 100 ml of reconstitutedluciferase-assay reagent (Promega, Madison, WI, USA).The light units produced were measured for 20 s,3–5 min after mixing. The reagents used in this stepwere obtained from the Dual-Luciferases ReporterAssay System (Promega).

In vivo imaging and quantitation of accumulatedradionuclides in tumor xenografts in Balb/c nu/nu miceAutoradiography was carried out in order to visualizeI-131 accumulation in the tumors. The mice were placedunder general anesthesia by the intraperitoneal injectionof a sterile 2.5% solution of 2-2-2-tribromoethanol(Aldrich chemicals, Milwaukee, WI, USA). The anaes-thetized animals were then placed on an imaging plate(Fuji, Tokyo, Japan) in a supine position in a dark roomfor 5 min, followed by a densitometric analysis using aFujix Bioimaging Analyzer BAS2000 (Fuji).

An in vivo imaging study was also performed in orderto examine the distribution of I-131 in established tumorsin nude mice. After 3–5 weeks, when the tumor diameterhad reached 8–10 mm, 200 ml of EgrLuc-HVJ-liposomesolution was injected into the tumor.

In vivo gene therapyFemale nude mice (Charles-River Japan, Tokyo, Japan)aged 4–6 weeks were injected subcutaneously on bothflanks with 200 ml of an Huh7 cell suspension (1�107

cells) in PBS. A transduction efficiency to the subcuta-neous tumor of more than 50% was attained using theHVJ-liposome method, as reported previously.17 After3–5 weeks, when the tumor diameter had reached8–10 mm, 200 ml of the EgrTK-HVJ-liposome was injected(on days 1 and 8). For the in vivo luciferase assayexperiments, 0–20 mCi I-131-lipiodol was injected into thetumor xenografts on day 2. The doses were selected onthe basis of the clinical data and our pilot study. GCVadministration (20 mg/kg per day injected intraperito-neally) was initiated 3 h after transfection and thencontinued for 2 weeks.

The mice were divided into four groups as follows:untreated animals (Group A; n¼ 10); those receivingEgrTK-vector transfection and irradiation (Group B;n¼ 8); those receiving EgrTK-vector transfection andGCV (Group C; n¼ 8) and those receiving EgrTK-vectortransfection, irradiation and GCV (Group D; n¼ 8). Noneof the mice showed signs of wasting or any otherindications of toxicity. The tumor volume was calculatedusing the formula 1/2� (ab2), where ‘a’ represents thelonger diameter and ‘b’ represents the shorter diameter.The care and treatment of the animals were in accor-

dance with the guidelines of the Nagasaki UniversityInstitutional Ethics Committee, Japan.

Measurement of serum AFPIn parallel with the measurement of tumor size, bloodsamples were collected from representative mice atvarious intervals by retro-orbital bleeding. Serum AFPlevels were analyzed using a radioimmunoassay.42

Hematoxylin–eosin stainingTo evaluate the mechanism causing the synergisticeffects of HSV-TK gene therapy and internal radiationby I-131-Lip, samples of subcutaneous tumors werecarefully dissected and removed at various intervalsduring therapy. The tumor tissues were fixed with 10%neutral formalin, embedded in paraffin and histologi-cally examined.

Egr-1 immunohistochemistryImmunohistochemical testing was performed in thehepatoma xenografts in order to evaluate the activationof the Egr-1 gene after internal radiation. The sampleswere fixed with 4% paraformaldehyde and embedded inparaffin. The tissues were then cut into 3 mm sections,deparaffinized in xylene and rehydrated. The sectionswere preincubated with normal bovine serum to preventnonspecific binding, and then were incubated overnightat 41C with an optimal dilution (5 mg/ml) of the primarymouse monoclonal antibody against human Egr-1(Oncogene Science, Uniondale, NY, USA). The slideswere sequentially incubated with the secondary antibodyand the reaction products were visualized by DAB usinga SIMPLE STAIN MAX-PO (MULTI) kit (Nichirei, Tokyo,Japan). Negative control samples were prepared byreplacing the primary antibody with nonimmune murineserum.

Statistical analysisAll data are presented as the mean7standard devia-tion. Statistical analyses were performed using theMann–Whitney U-test. Probability (P) values o0.05 wereconsidered to be statistically significant.

References

1 Askari F, Wilson J. Provocative gene therapy strategy forthe treatment of hepatocellular carcinoma. Hepatology 1992; 16:273–274.

2 Colombo M. Hepatocellular carcinoma. J Hepatol 1992; 15:225–236.

3 Caruso M et al. Regression of established macroscopic livermetastases after in situ transduction of a suicide gene. Proc NatlAcad Sci USA 1993; 90: 7024–7028.

4 Barba D, Hardin J, Sadelain M, Gage FH. Development of anti-tumor immunity following thymidine kinase-mediated killingof experimental brain tumors. Proc Natl Acad Sci USA 1994; 91:4348–4352.

5 Chen SH et al. Gene therapy for brain tumors: regressionof experimental gliomas by adenovirus-mediated gene transferin vivo. Proc Natl Acad Sci USA 1994; 91: 3054–3057.

6 Moolten FL. Tumor chemosensitivity conferred by insertedherpes thymidine kinase genes: paradigm for a prospectivecancer control strategy. Cancer Res 1986; 46: 5276–5281.


1638

Gene Therapy

7 Moolten FL, Wells JM. Curability of tumors bearing herpesthymidine kinase genes transferred by retroviral vectors. J NatlCancer Inst 1990; 82: 297–300.

8 Moolten FL. Drug sensitivity (‘suicide’) genes for selectivecancer chemotherapy. Cancer Gene Ther 1994; 1: 279–287.

9 Freeman SM et al. The ‘bystander effect’: tumor regression whena fraction of the tumor mass is genetically modified. Cancer Res1993; 53: 5274–5283.

10 Ido A et al. Gene therapy for hepatoma cells using a retrovirusvector carrying herpes simplex virus thymidine kinase geneunder the control of human alpha-fetoprotein gene promoter.Cancer Res 1995; 55: 3105–3109.

11 Kaneko S et al. Adenovirus-mediated gene therapy of hepato-cellular carcinoma using cancer-specific gene expression. CancerRes 1995; 55: 5283–5287.

12 Wills KN et al. Gene therapy for hepatocellular carcinoma:chemosensitivity conferred by adenovirus-mediated transferof the HSV-1 thymidine kinase gene. Cancer Gene Ther 1995; 2:191–197.

13 Su H, Chang JC, Xu SM, Kan YW. Selective killing of AFP-positive hepatocellular carcinoma cells by adeno-associatedvirus transfer of the herpes simplex virus thymidine kinasegene. Hum Gene Ther 1996; 7: 463–470.

14 Kanai F et al. Gene therapy for alpha-fetoprotein-producinghuman hepatoma cells by adenovirus-mediated transfer of theherpes simplex virus thymidine kinase gene. Hepatology 1996; 23:1359–1368.

15 Ohguchi S et al. Expression of alpha-fetoprotein and albumingenes in human hepatocellular carcinomas: limitations in theapplication of the genes for targeting human hepatocellularcarcinoma in gene therapy. Hepatology 1998; 27: 599–607.

16 Hallahan DE et al. Spatial and temporal control of gene therapyusing ionizing radiation. Nat Med 1995; 1: 786–791.

17 Kawashita Y et al. Regression of hepatocellular carcinoma in vitroand in vivo by radiosensitizing suicide gene therapy under theinducible and spatial control of radiation. Hum Gene Ther 1999;10: 1509–1519.

18 Takahashi T, Namiki Y, Ohno T. Induction of the suicide HSV-TKgene by activation of the Egr-1 promoter with radioisotopes.Hum Gene Ther 1997; 8: 827–833.

19 Bretagne JF et al. Hepatic artery injection of I-131-labeledlipiodol. Part II. Preliminary results of therapeutic use inpatients with hepatocellular carcinoma and liver metastases.Radiology 1988; 168: 547–550.

20 Raoul JL et al. Hepatic artery injection of I-131-labeled lipiodol.Part I. Biodistribution study results in patients with hepato-cellular carcinoma and liver metastases. Radiology 1988; 168:541–545.

21 Raoul JL et al. Preoperative treatment of hepatocellular carcino-ma with intra-arterial injection of 131I-labelled lipiodol. Br J Surg2003; 90: 1379–1383.

22 Boucher E et al. Adjuvant intra-arterial injection of iodine-131-labeled lipiodol after resection of hepatocellular carcinoma.Hepatology 2003; 38: 1237–1241.

23 Glover D, Little JB, Lavin MF, Gueven N. Low dose ionizingradiation-induced activation of connexin 43 expression.Int J Radiat Biol 2003; 79: 955–964.

24 Nakakuma K et al. Studies on anticancer treatment with an oilyanticancer drug injected into the ligated feeding hepatic arteryfor liver cancer. Cancer 1983; 52: 2193–2200.

25 Iwai K, Maeda H, Konno T. Use of oily contrast medium forselective drug targeting to tumor: enhanced therapeutic effectand X-ray image. Cancer Res 1984; 44: 2115–2121.

26 Kanematsu T et al. Selective effects of lipiodolized antitumoragents. J Surg Oncol 1984; 25: 218–226.

27 Madsen MT, Park CH, Thakur ML. Dosimetry of iodine-131ethiodol in the treatment of hepatoma. J Nucl Med 1988; 29:1038–1044.

28 Chou FI, Lui WY, Chi CW, Chan WK. I-131-lipiodol cytotoxicityin hepatoma cells. Proc Natl Sci Counc Repub China B 1994; 18:154–160.

29 Park CH et al. Evaluation of intrahepatic I-131 ethiodol on apatient with hepatocellular carcinoma. Therapeutic feasibilitystudy. Clin Nucl Med 1986; 11: 514–517.

30 Kajiya Y, Kobayashi H, Nakajo M. Transarterial internalradiation therapy with I-131 lipiodol for multifocal hepatocel-lular carcinoma: immediate and long-term results. CardiovascIntervent Radiol 1993; 16: 150–157.

31 Raoul JI et al. Internal radiation therapy for hepatocellularcarcinoma. Results of a French multicenter phase II trial oftransarterial injection of iodine 131-labeled Lipiodol. Cancer1992; 69: 346–352.

32 Partensky C et al. Intra-arterial iodine 131-labeled lipiodol asadjuvant therapy after curative liver resection for hepatocellularcarcinoma: a phase 2 clinical study. Arch Surg 2000; 135:1298–1300.

33 Buscombe JR. Interventional nuclear medicine in hepatocellularcarcinoma and other tumours. Nucl Med Commun 2002; 23: 837–841.

34 Lau WY et al. Adjuvant intra-arterial iodine-131-labelled lipiodolfor resectable hepatocellular carcinoma: a prospective rando-mised trial. Lancet 1999; 353: 797–801.

35 Abdulkadir SA et al. Impaired prostate tumorigenesis in Egr1-deficient mice. Nat Med 2001; 7: 101–107.

36 Raper SE et al. Fatal systemic inflammatory response syndromein a ornithine transcarbamylase deficient patient followingadenoviral gene transfer. Mol Genet Metab 2003; 80: 148–158.

37 Morishita R et al. Novel in vitro gene transfer method for study oflocal modulators in vascular smooth muscle cells. Hypertension1993; 21: 894–899.

38 Kato K et al. Expression of hepatitis B virus surface antigen inadult rat liver. Co-introduction of DNA and nuclear protein by asimplified liposome method. J Biol Chem 1991; 266: 3361–3364.

39 Hirano T et al. Persistent gene expression in rat liver in vivo byrepetitive transfections using HVJ-liposome. Gene Therapy 1998;5: 459–464.

40 Seung LP et al. Genetic radiotherapy overcomes tumor resistanceto cytotoxic agents. Cancer Res 1995; 55: 5561–5565.

41 Ho S, Lau WY, Leung TW, Johnson PJ. Internal radiation therapyfor patients with primary or metastatic hepatic cancer: a review.Cancer 1998; 83: 1894–1907.

42 Bellet DH, Wands JR, Isselbacher KJ, Bohuon C. Serum alpha-fetoprotein levels in human disease: perspective from a highlyspecific monoclonal radioimmunoassay. Proc Natl Acad Sci USA1984; 81: 3869–3873.


1639

Gene Therapy

RESEARCH ARTICLE

Optimization of adenovirus-mediated endothelial nitricoxide synthase delivery in rat hindlimb ischemia

J Yan, GL Tang, R Wang and LM MessinaPacific Vascular Research Laboratory, Department of Surgery, Division of Vascular Surgery, University of California San Francisco,San Francisco, CA, USA

Adenovirus-mediated overexpression of endothelial nitricoxide synthase (eNOS) induces collateral artery develop-ment and substantially increases blood flow after induction ofexperimental acute hindlimb ischemia. However, the optimaltechnique of gene delivery for this or any other form of genetherapy in limb ischemia is still unknown. The purpose of thisstudy was to determine the effect of the two most commonlyused techniques, intra-arterial and intramuscular injection,on blood flow recovery, collateral artery development, andpreservation of muscle mass. We compared intra-arterialinjection under vascular isolation, intra-arterial injectionunder transient vascular occlusion, and intramuscular injec-tion of phosphate buffered saline (PBS) or adenovirusencoding either the eNOS (AdeNOS) or LacZ (AdlacZ) geneafter induction of acute hindlimb ischemia. Delivery of

AdeNOS by both intra-arterial injection techniques increasedeNOS activity (22.30 versus 10.56, Po0.01), blood flow(0.9070.02 versus 0.6970.07, Po0.001) and collateralartery development (17.56484 versus 13.74259, Po0.05)more than by intramuscular delivery. Intra-arterial injectionunder transient vascular occlusion led to better preservation ofmuscle mass, muscle architecture, and clinical ischemic index,but led to greater transgene expression in distant organs andcontralateral limb muscles. Intra-arterial injection of AdeNOSunder transient vascular occlusion is the optimal technique toreverse severe hindlimb ischemia in the rat. This is the firstsystematic comparison of different delivery techniques usedin gene therapy of experimental hindlimb ischemia.Gene Therapy (2005) 12, 1640–1650. doi:10.1038/sj.gt.3302563; published online 18 August 2005

Keywords: endothelial nitric oxide synthase; intra-arterial injection with tourniquet; intra-arterial injection without tourniquet;intramuscular injection; limb ischemia

Introduction

Critical limb ischemia causes substantial morbidity andmortality. Despite advances in surgical and endovasculartechniques, it is estimated that 150 000 patients per yearrequire lower limb amputations due to critical limbischemia in the United States.1 Developing an effectivegene therapy to promote collateral artery developmentand increase blood flow to critically ischemic limbswould be of considerable clinical value.

The endothelial isoform of nitric oxide synthase(eNOS) is one of three isoenzymes that convertsL-arginine to L-citrulline and nitric oxide (NO). NO, apotent vasodilator and anti-inflammatory agent, plays animportant role in both collateral artery development andangiogenesis in vivo.2–4 NO regulates blood pressure,platelet aggregation, leukocyte adhesion, and smoothmuscle mitogenesis.5–12 Several key angiogenic factors,including FGF, VEGF, angiotensin II, nerve growth factor,TNF-a, and platelet activating factor 4, have been foundto act via NO.13–27

Adenovirus-mediated overexpression of eNOS (Ade-NOS) has been shown to increase significantly collateral

artery development, leading to increased limb bloodflow in a rat model of hindlimb ischemia.28–30 In contrast,eNOS deficiency can decrease angiogenesis.31–35

In order to realize the promise of adenovirus-mediatedeNOS (AdeNOS) overexpression to treat critical limbischemia, the most effective and safest method of genedelivery must be established first. In this study, we chosethree different delivery techniques for AdeNOS: intra-arterial under vascular isolation (intra-arterial with tour-niquet), intra-arterial under transient arterial and venousocclusion (intra-arterial without tourniquet), and intra-muscular injection. We hypothesized that intra-arterialdelivery under vascular isolation would achieve thehighest rate of gene transfer and the lowest rate oftransgene expression in systemic organs based on ourprevious experience with this technique.36 However, thistechnique is relatively invasive and also may causeincreased tissue injury secondary to effects from thetourniquet. We therefore tested a less invasive intra-arterial technique, intra-arterial delivery under transientarterial and venous occlusion, recognizing that transduc-tion of distant organs was likely to be higher than thatunder vascular isolation. Lastly, we tested intramusculardelivery, the simplest and least invasive technique, and thetechnique that experimentally has been the most widelystudied. We performed a systematic comparison of theeffects of these three techniques of AdeNOS delivery onlimb blood-flow, collateral artery development, capillary

Received 3 November 2004; accepted 5 May 2005; published online18 August 2005

Correspondence: Dr LM Messina, Division of Vascular Surgery, UCSFMedical Center, 505 Parnassus Avenue, San Francisco, CA 94143, USA


www.nature.com/gt

angiogenesis, clinical ischemic index,37 muscle mass andarchitecture, and transgene expression in distant organs.

Results

Local and systemic transgene expressionWe compared the differences in the extent and locationof transgene expression by performing RT-PCR, lacZwhole-mount staining, and eNOS immunohisto-chemistry on harvested muscles and organs 4 days afteradenoviral gene transfer using the three differenttechniques (Figure 1). After intra-arterial delivery, RT-PCR results showed transgene expression primarily inthe calf muscles (mainly in the gastrocnemius muscle,lower expression in the tibialis anterior muscle), withsome expression in the thigh muscles (biceps femoris,gracilis, quadriceps). This expression pattern corre-sponded with previously published results from ourlaboratory.36 The transgene was also expressed in thecontralateral limb after intra-arterial delivery techniquesin the vascular isolation group. Minimal transgene wasexpressed in the brain, heart, lung, liver, spleen, kidney,or testis in either intra-arterial group (Figure 1c).

After intramuscular delivery, substantially less trans-gene was expressed in gastrocnemius muscles thanthat after intra-arterial delivery as detected by RT-PCR(Figure 1c), lacZ staining (Figure 1b), and immunohisto-chemical staining for eNOS (Figure 1a). We alsoperformed semiquantitative analysis for gene expressionin the left gastrocnemius. Our results show that afterintra-arterial injection, AdLacZ expression was six- toseven-fold higher than intramuscular injection (Figure1d). Equivalent levels of transgene were detected in thetibialis anterior muscle after the three different deliverytechniques. Minimal expression was found in the brain,liver, and testis (Figure 1c). Intramuscular delivery ofAdeNOS, AdLacZ, or phosphate-buffered saline (PBS)by any technique did not result in contralateral limbexpression. eNOS transgene expression was detectedup to 14 days after injection, but not at 21 days byimmunostaining (data not shown).

eNOS activityeNOS activity in the gastrocnemius muscles fromdifferent groups was determined by measuring theconversion of L-[14C]arginine to L-[14C]citrulline in thepresence and absence of calcium. The eNOS activity wassignificantly higher in the groups received AdeNOSby intra-arterial techniques than after either intra-muscular delivery or PBS control groups. Intramusulardelivery also increased eNOS activity when comparedwith values with PBS control (Figure 2).

Blood flow. All limbs in all groups were scannedserially over 30 days laser Doppler to document therecovery of hindlimb blood flow. By postinjection day30, intra-arterial delivery of AdeNOS with tourniquet(Figure 3a) or without tourniquet (Figure 3b) increasedthe blood flow to the foot to almost normal levels(0.9070.15 and 0.9070.02 respectively, n¼ 6 rats pergroup); both increased blood flow significantly morethan intramuscular delivery of AdeNOS (0.6970.07,Po0.001), AdlacZ (0.5870.1, Po0.001), or PBS (0.5970.04, Po0.001) (Figure 3d). Although intramuscular

delivery of AdeNOS increased blood flow less thandid intra-arterial delivery, it did increase blood flowrecovery more than delivery of PBS or AdlacZ (Po0.01)(Figure 3c). There was no significant difference betweenblood flow recovery after delivery of PBS or AdlacZ;thus, the delivery of modified adenovirus had no effecton the recovery of hindlimb blood-flow.

Histopathology of ischemic hindlimb muscle. Afterdelivery of PBS, hematoxylin and eosin (HE) staining ofhindlimb muscle showed necrotic and regeneratingmyocytes (central nuclei) as well as inflammatory cellinfiltrates 30 days (Figure 4a). In contrast, eNOS over-expression resulted in less inflammatory cell infiltrationand near-normal muscle architecture (Figure 4a). Recov-ery of muscle architecture correlated with the extent ofeNOS gene transfer; thus, intra-arterial delivery resultedin more normal muscle architecture than did intra-muscular delivery.

Muscle mass. After critical hindlimb ischemia, boththe calf and thigh musculature of the rat hindlimbatrophy. Delivery of AdeNOS reversed this atrophy andled to recovery of muscle mass to near normal levelsby 30 days. Both intra-arterial delivery techniques ofAdeNOS increased muscle mass recovery significantlymore than did intramuscular delivery of AdeNOS(Po0.01), PBS (Po0.001), or AdlacZ (Po0.001) in calfmuscle. Again, muscle mass recovery correlated with theextent of eNOS gene transfer (Table 1); thus, intra-arterialdelivery resulted in higher muscle mass than intra-muscular delivery, which resulted in higher muscle massthan delivery of either PBS or AdlacZ.

Clinical ischemic index. By 30 days after intra-aterialinjection of AdeNOS, the rats’ feet and calves appearedalmost normal (Figure 4b). Their gait also had returnedalmost to normal. However, after intramuscular AdeNOSinjection, the calf and foot muscles still showed atrophy,and the rats walked with an abnormal gait. Muscleatrophy and gait abnormalities were worst in rats thatreceived either PBS or AdlacZ.

Capillary density and capillary/muscle fiber ratio. Tofurther evaluate the effects of different eNOS deliverytechniques on capillary angiogenesis, CD31 staining wasperformed on gastrocnemius muscle. AdeNOS deliveryby intra-arterial injection produced significantly strongerCD31 staining and higher capillary/muscle fiber ratios(1.7970.05 IA+T, 1.7870.07 IA�T) than did intra-muscular eNOS (1.3670.01, Po0.05), PBS (0.6470.07,Po0.001) (Figure 5). Intramuscular delivery of AdeNOSalso increased angiogenesis (Po0.01 versus PBS andAdlacZ), although to a lesser degree than did intra-arterial delivery.

Angiography. Collateral artery development of theischemic limb was significantly increased in all groupsafter AdeNOS delivery (Figure 6a). Intra-arterial deliveryinduced more collateral artery development than didintramuscular delivery (Po0.05). Angioscores were con-sistant with the degree of eNOS gene transfer (Figure 6b).

Optimization of adenovirus-mediated eNOSJ Yan et al

1641

Gene Therapy

Figure 1 Intra-arterial delivery of AdeNOS resulted in greater transgene expression than did intramuscular delivery. (a) eNOS immunostaining of the left(ischemic, top) and right (nonischemic, bottom) gastrocnemius of rats 4 days after receiving PBS injections, AdeNOS intramuscular injections (IM),AdeNOS intra-arterial injections with tourniquet (IA+T), AdeNOS intra-arterial injections without tourniquet (IA�T) (� 20 magnification). (b) lacZstaining of left (ischemic, top) and right (nonischemic, bottom) gastrocnemius of rats 4 days after receiving AdlacZ intramuscular injections (IM), AdlacZintra-arterial injections with tourniquet (IA+T), and AdlacZ intra-arterial injections without tourniquet (IA�T) (� 20 magnification). (c) lacZ expressiondetected in different tissues by RT-PCR after AdlacZ delivery. Lane 1, left tibialis anterior; lane 2, left gastrocnemius; lane 3, left hamstring; lane 4, leftgracilis; lane 5, left quardreps; lanes 6–10, corresponding right-sided muscles; lane 11, brain; lane 12, heart; lane 13, kidney; lane 14, liver; lane 15, lung;lane 16, spleen; lane 17, testis. b-Actin was used as an internal loading control. (d) Semiquantative analysis of LacZ mRNA levels in left gastrocnemiusafter AdeLacZ delivey. *Po0.001 versus AdeLacZ+IM.


1642

Gene Therapy

Discussion

This study shows that intra-arterial AdeNOS deliveryunder vascular isolation achieves the highest rate ofischemic leg blood flow recovery and gene transfer,while causing the least level of muscle damage measuredin a variety of ways in a rat model of hindlimb ischemia.Analysis of regional transgene expression showed thatintra-arterial injection was significantly more efficientthan intramuscular injection. Consequently, eNOS activ-ity was greatly increased after both techniques of intra-arterial AdeNOS delivery than after intramuscularAdeNOS delivery or PBS control groups. There was nosignificant difference in eNOS activity, blood flowrecovery or in the extent of transgene expressionbetween either intra-arterial delivery technique, butdelivery under transient vascular occlusion caused lessmuscle damage and greater recovery of muscle massthan did delivery under vascular isolation. Intramuscu-lar injection showed the lowest level of transgeneexpression, blood flow recovery and muscle mass ofthe three gene delivery techniques.

Intramuscular and, to a much lesser extent, intra-arterial gene delivery are commonly used for transgenedelivery in vascular disease.36,38–48 Vajanto et al47 demon-strated that intramuscular administration of AdVEGF

Figure 3 Blood flow in the foot recovers more by intra-arterial delivery of AdeNOS than by intramuscular delivery, as seen by Laser Doppler perfusionimaging. AdeNOS delivered by intra-arterial injection with (a) and without (b) tourniquet (IA+T, IA�T) significantly increased blood flow recovery of theischemic hindlimb at postinjection day 30 over the PBS and AdlacZ groups (Po0.001 versus PBS and AdlacZ). Intramuscular injection of AdeNOS (IM)also increased blood flow more than PBS or AdlacZ, but to a lesser degree than did intra-arterial delivery (Po0.01 versus PBS and AdlacZ). (c) Among thethree different techniques (d), intra-arterial delivery was most effective at increasing blood flow recovery (Po0.001 versus IM). (*Po0.05 versus IM,**Po0.01, ***Po0.001. #Po0.05, versus PBS and AdlacZ, ##Po0.01, ###Po0.001; zPo0.05 versus IA�T).

Figure 2 Calcium-dependent eNOS activity in gastrocnemius muscle wassignificantly higher in all groups receiving AdeNOS than in PBS group.Although no significant difference was found between the two intra-arterial delivery techniques, both of the intra-arterial delivery techniquescaused a significantly higher eNOS activity than did the intramusculargene delivery technique. n¼ 6. *Po0.001 versus PBS; #Po0.01 versusAdeNOS+IM, �Po0.05 versus PBS.


1643

Gene Therapy

into ischemic rabbit skeletal muscle resulted efficientgene transfer and induced angiogenesis. Similarly,intramuscular delivery of AdeNOS in rat hindlimbischemia model also led to significant increases in blood

perfusion and angiogenesis. In that study, eNOS trans-gene expression could be identified 1 week after genetransfer and could still be weakly detected 4 weeks afterinjection.29 Isner et al showed that after arterial gene

Figure 4 Delivery of AdeNOS restored normal muscle architecture and muscle mass. (a) Hematoxylin and eosin staining of left (ischemic, top) and right(nonischemic, bottom) gastrocnemius after delivery of PBS, AdLacZ and AdeNOS by intramuscular injection (IM), AdeNOS by intra-arterial injectionwith tourniquet (IA+T), and AdeNOS by intra-arterial injection without tourniquet (IA�T) after postinjection day 30 (� 20 magnification). (b) Grossappearance of rat hindlimb shows reversal of hindlimb atrophy after AdeNOS delivery.

Table 1 Muscle mass recovery after eNOS gene transfer

PBS AdLacZ AdeNOS IM AdeNOS IA+T AdeNOS IA-T

Tibialis anterior 0.1370.08 0.2370.06 0.3170.03## 0.7570.08***,### 0.8170.18***,###

Gastrocnemius 0.3770.17 0.3670.06 0.6470.14### 0.8170.07**,### 0.8670.06***,###

Gracilis 0.6170.06 0.6170.05 0.7970.19 0.8070.07# 0.8370.12#

Biceps femoris 0.6670.06 0.7070.16 0.8470.12# 0.9270.05## 0.8470.07#

Quadriceps 0.5270.04 0.6170.08 0.7970.19# 0.7670.09# 0.8170.11#

*Po0.05 versus IM, **Po0.01, ***Po0.001; #Po0.05 versus PBS and AdlacZ, ##Po0.01, ###Po0.001.


1644

Gene Therapy

transfer of phVEGF165 in patients with ischemic limbs incollateral arteries at the knee, mid-tibial, and ankle levelsincreased. Our previous study also showed that intra-arterial delivery of adenovirus encoding interleukin-1 tothe rat hindlimb under vascular isolation led to highrates of gene transfer to the gastrocnemius capillaryendothelium and muscle fiber of 7177 and 2575%transduction rate, respectively.36 However, intra-arterialand intramuscular adenovirus-mediated gene transferhave not been compared in terms of efficiency, safety orefficacy. Therefore, we focused on a systematic compar-ison between intra-arterial and intramuscular delivery ofAdeNOS for the reversal of hindlimb ischemia in rats.

In animal models of hindlimb ischemia, the arterieswill remodel themselves in response to a variety ofstimuli, including hemodynamic forces (shear rate) andtissue metabolic demands (ischemia). Two differentprocesses, ‘angiogenesis’ (or creation of capillaries) and‘arteriogenesis’ (or growth and development of pre-existing collateral arteries) are involved in vascularremodeling. The key role that eNOS plays in angiogen-esis and arteriogenesis of hindlimb ischemia has been

demonstrated in both in vitro and in vivo experi-ments.4,10,29,30 Murohara et al4,34 demonstrated that endo-genous endothelium-derived NO plays an important rolein mediating angiogenesis by supporting endothelial cellmigration and proliferation. Lloyd et al28,35 also showedthe role of NO in collateral development (arteriogenesis).Aicher et al33 demonstrated that the impaired neovascu-larization in mice lacking eNOS is related to a defect inprogenitor cell mobilization. Mice deficient in eNOS(NOS3�/�) show reduced vascular endothelial growthfactor (VEGF)-induced mobilization of endothelial pro-genitor cells (EPCs) and increased mortality aftermyelosuppression. Progenitor mobilization from thebone marrow is also impaired in NOS3�/� mice. Thesestudies provide evidence that eNOS is a critical moleculefor angiogenesis and arteriogenesis in vivo in responseto tissue ischemia. In our experiments, we have provi-ded further evidence that eNOS overexpressiongreatly increases blood flow recovery of ischemic legdue to increased angiogenesis (capillaries per skeletalmuscle fiber) and arteriogenesis (angioscore), especiallyafter intra-arterial delivery. We found that 30 days

Figure 5 AdeNOS delivery increased gastrocnemius capillary density. (a) CD31 staining of left (ischemic, top) and right (nonischemic, bottom)gastrocnemius 30 days after delivery of PBS (PBS), AdlacZ and AdeNOS by intramuscular injection (IM), AdeNOS by intra-arterial injection withtourniquet (IA+T), and AdeNOS by intra-arterial injection without tourniquet (IA�T) (� 20 magnification). (b) Quantification of capillary density.(*Po0.05 versus IM, #Po0.001 versus PBS and AdLacZ, zPo0.01 versus PBS and AdlacZ).


1645

Gene Therapy

after intra-arterial delivery of AdeNOS in rats, theblood flow of ischemic hindlimb almost normalized.Intra-arterial delivery of eNOS led to greater angio-genesis and restoration of muscle mass than did intra-muscular delivery.

We have validated the measurement of the majordependent variables used in this study in prior experi-ments.30,36,37,49 In particular, blood flow measurement bymicrospheres, direct tissue O2 determination, and laserDoppler correlated well with one another. However,laser Doppler has limitations and must be applied in arigorous manner. This includes being aware of the typeand level of anesthetic used for blood flow measure-ments, the need to maintain rigorous control of body

temperature, and the recognizability that the laserDoppler only penetrates to a depth of 1 mm and thuslargely reflects changes in skin blood flow. Nonetheless,when applied properly, we believe laser Dopplerrepresents an accurate though not precise estimation ofhind limb blood flow.

There have not been, to our knowledge, any priorsystemic comparisons of gene delivery techniques usedto reverse experimental hindlimb ischemia. In thisregard, the results of this study suggest that adenovirusmediated gene transfer in human could be accomplishedby a relatively simple and less invasive manner. Aballoon catheter could be introduced into the artery andvein of the affected limb. Both balloons could be inflated

Figure 6 (a) Comparison of arteriogenesis after delivery of PBS (PBS), AdlacZ (lacZ), AdeNOS by intramuscular injection (IM), AdeNOS by intra-arterial injection with tourniquet (IA+T), and AdeNOS by intra-arterial injection without tourniquet (IA�T). (b) Quantification of arteriogenesis.(*Po0.05 versus IM; #Po0.05 versus AdLacZ; ##Po0.01 versus PBS and LacZ).


1646

Gene Therapy

and an appropriate volume and concentration ofAdeNOS could be injected into the arterial catheter.After 30 min, the balloons could be deflated, and thecatheters removed. Vascular isolation of a chronicallyischemic human leg by inflation of a blood pressure cuffover the proximal thigh may prove to be less injuriousthan it has been in small animal models such as thoseused in this study.

In conclusion, this study shows that intra-arterialdelivery of AdeNOS under transient venous and arterialocclusion results in the most efficient level of genetransfer and the best therapeutic effect by measurementof blood flow, collateral artery development, and reversalof muscle atrophy. Intra-arterial delivery may be theoptimal technique to overexpress eNOS in humanischemic limbs to reverse the devastating effect of criticallimb ischemia.


Rat hindlimb ischemia modelAdult male Sprague–Dawley rats (270–300 g), purchasedfrom Charles River, were housed in an environmentallycontrolled room and given chow and water ad libitum.After induction of 2% isoflurane inhalational anesthesia,left hindlimb ischemia was induced by ligation of the leftcommon iliac, internal and external iliac, and superficialfemoral arteries and associated branches as describedpreviously.37 The care of animals complied with theGuide for the Care and Use of Laboratory Animals (Instituteof Laboratory Animal Resources, Commission on LifeSciences, National Research Council, Washington, DC,National Academy Press, 1996). All protocols wereapproved in accordance with the Committee on AnimalResearch at the University of California San Francisco.

Preparation of recombinant adenovirusesAdenoviral recombinants encoding LacZ or bovineeNOS cDNA driven by the Rous sarcoma virus promo-ter,50 generous gifts from Dr Beverly Davidson (Uni-versity of Iowa College of Medicine), were propagatedand titrated by 293 cells as described previously.49 Alladenovirus stock was stored at �801C, then thawedimmediately before use, and diluted in PBS to achievea final concentration of 1�1010 plaque forming units(PFU)/ml.

Study designRats were randomly divided into nine treatment groupsof nine rats each as follows: AdeNOS, AdlacZ, or PBSdelivered intra-arterially under vascular isolation (Ade-NOS+T, AdlacZ+T, or PBS+T, respectively), AdeNOS,AdlacZ, or PBS delivered intra-arterially under transientvascular occlusion (AdeNOS-T, AdlacZ-T, or PBS-Trespectively), and AdeNOS, AdlacZ, or PBS deliveredintramuscularly into the calf (AdeNOS+M, AdlacZ+M,or PBS+M, respectively). All rats in all groups underwentsurgically created hindlimb ischemia 10 days prior togene delivery. Three rats per group were killed at 4 daysafter gene delivery for analysis of transgene expressionby immunohistochemistry, lacZ staining, and RT-PCR.The remaining six rats per group underwent laserDoppler perfusion imaging pre- and postgene transfer,and at 4, 14, 21, and 30 days after gene delivery. A clinical

ischemia index and angiograms were obtained prior tokilling at 30 days. Muscles were harvested and weighedat killing, and then processed histopathologically andimmunohistochemically for capillary-to-muscle-fiberratio analysis.

Gene delivery techniquesIntra-arterial delivery under vascular isolation (intra-arterial with tourniquet). Intra-arterial delivery ofAdeNOS under vascular isolation was performed aspreviously described.30,36 Briefly, a tourniquet wasplaced on the proximal thigh after isolation of thecommon femoral artery, vein, and nerve to preventcollateral inflow or outflow. The saphenous artery wasligated and cannulated in a retrograde manner with a30G needle. After the femoral vein and the proximalsaphenous vein were clamped with microvascularclamps, a venotomy was made in the distal saphenousvein to serve as outflow for flushing the hindlimbvasculature. Warm heparinized sodium (10 ml at100 U/ml) was injected, followed by 5 ml of PBS. Thevenotomy was temporarily clamped, and 1�1010 PFU/ml of AdeNOS, AdlacZ, or PBS (sham) was subsequentlyinfused in a total volume of 0.7 ml and allowed to dwellfor 30 min. Any residual virus vector was flushed out ofthe saphenous venotomy by using 5 ml of PBS, afterwhich the saphenous vein and artery were ligated.

Intra-arterial delivery under temporary vascular occlu-sion (intra-arterial without tourniquet). Intra-arterialdelivery under temporary vascular occlusion was per-formed as above, except that the tourniquet to eliminatecollateral inflow or outflow was omitted, and thehindlimb vasculature was not flushed out prior to orafter delivery of vector.

Intramuscular delivery. A small skin incision wasmade to expose the site of intramuscular delivery.AdeNOS, AdlacZ or PBS at 1�1010 PFU/ml was injectedusing a 28G needle into the tibialis anterior muscle(100 ml) and the gastrocnemius muscle (200 ml into eachof the medial, lateral, and central muscle bellies).

Laser Doppler perfusion imaging of hindlimb blood flowRats were anesthetized using 1% isoflurane with a1 l/min O2 flow rate. Their hindlimbs were shaved andthen depilated using Surgiprep depilatory cream. Therats were put on a homeothermic heating pad andallowed to equilibrate to a rectal temperature of 371C.Laser Doppler perfusion imaging was then performedover the legs and feet by using a laser Doppler perfusionimager (Moor LDI; Moor instruments, Devon, UK). Thesame area of interest (bilateral leg and foot) was scannedthree times, and the results were averaged. To avoid datavariations due to ambient light and temperature,hindlimb blood flow was analyzed as a ratio of left(ischemic) to right (nonischemic).

AngiogramsRats were anesthetized using 1% isoflurane with a1 l/min O2 flow rate. The infrarenal abdominal aortawas isolated and then ligated proximally and clampeddistally. A 20G polyethylene catheter was used tocannulate the aorta, the distal clamp was removed, and


1647

Gene Therapy

2.5 ml heparinized saline (10 U/ml) was injected,followed by 2.5 ml contrast medium (EZPaque, MerryX-Ray, South San Francisco, CA, USA). The aorta andinferior vena cava were then ligated. The skin wasremoved from the hindlimbs to prevent imaging of thedermal vasculature. Images were taken by using a singledeveloped Kodak X-OMAT TL film at 500 mA, 50 kV, and0.5 s exposure. Angioscores were derived by placing agrid over the angiogram in the region of the greatertrochanter to the patella. Three blinded observers countedthe number of contrast-opacified arteries crossing overthe gridlines and the total number of lines encompassingthe region of interest. The angioscore was calculated asthe ratio of overlying opacified arteries divided by thetotal number of lines in the region of interest.

Tissue preparationTibialis anterior muscle (TA), gastrocnemius muscle(GC), biceps femoris (BF), gracilis (Gr), quadreps (Q),brain, heart, lung, liver, spleen, kidney, testis wereharvested and weighed. Middle sections of the gastro-cnemius muscle were either preserved in O.C.T. com-pound (Tissue-Tek, Fisher Scientific, Fairlawn, NJ, USA)and cooled in dry-ice liquid pentane slush for frozensections or fixed in 1.5% glutaraldehyde for 3 h for lacZstaining. The remainder of the gastrocnemius and allother tissues were snap-frozen in liquid nitrogen for laterharvest of RNA for use in RT-PCR.

HistopathologyIn total, 10-mm-thick frozen gastrocnemius muscle sec-tions were stained with Gills HE, and the muscles werestudied to determine qualitatively the extent of damageto the muscle architecture, edema, and inflammatory cellinfiltration in the different experimental groups.

LacZ stainingGastrocnemius muscle was cut into 100 mm sectionsusing a vibratome (VT 1000 S, Leica, Lafayette, Califor-nia). Muscles were stained overnight with 5-bromo-4-chloro-3-indolyl-b-D-galactopyranoside (X-Gal) solution(1 mg/ml X-gal, 5 mM potassium ferrocyanide, 5 mM

potassium ferricyanide, 2 mM magnesium chloride,0.02% NP40 and PBS) at room temperature.

ImmunohistochemistryIn total, 10-mm-thick transverse frozen sections of tissuesamples were used for immunohistochemical analysis.Transgene eNOS expression was detected using a mouseanti-human eNOS monoclonal antibody that specificallystains transgene eNOS, but not endogenous rat eNOS(BD Biosciences, San Diego, CA, USA, 1:50 dilution).Capillary-to-muscle fiber ratios were determined bystaining endothelial cells for platelet endothelial celladhesion molecule-1 (PECAM-1, CD31) using a mouseanti-rat PECAM-1 monoclonal antibody (Serotec, Raleigh,NC, USA, 1:500 dilution) as previously described.37 Thesecondary antibody for all staining was horse anti-mousemonoclonal antibody, rat adsorbed (Vector Laboratories,Burlingame, CA, USA, 1:250 dilution). Negative controlswere created by substituting blocking buffer for primaryantibody. Antibody staining was detected by using theVectastain ABC kit followed by the Vectastain DAB kit(Vector Laboratories, Inc., Burlingame, CA, USA) as perthe manufacturer’s instructions.

RT-PCRTotal RNA was extracted from frozen tissue by homo-genization in 1 ml of Trizol solution (Invitrogen, Carls-bad, CA, USA); 2 mg of RNA was used for reversetranscription. The cDNA was amplified by PCR for 40cycles (941C for 3 min, 941C for 30 s, 571C for 30 s, 721Cfor 45 s, 721C for 10 min) using Taq polymerase. b-Actinwas used as an internal control. Primers were as follows:

lacZ, forward 50-CGCCCGTTGCACCACAGATG-30,reverse 50-CCAGCTGGCGTAATAGCGAAG-30;

b-actin, forward 50-ATGAAGATCCTGACCGAGCG-30,reverse 50-TACTTGCGCTCAGGAGC-30.

PCR products were electrophoresed on a 1% agarosegel and visualized by using ethidium bromide.

Measurement of eNOS activityeNOS activty in the gastrocnemius muscle was measured4 day after gene transfer of eNOS (Calbiochem, SanDiego, CA, USA). The rats were killed and the gastro-cnemius muscles were excised and snap-frozen in liquidnitrogen. To extract total protein, the muscles werehomogenized in ice-cold buffer (250 mM Tris-HCl, pH7.4, 10 mM EDTA, 10 mM EGTA) and centrifuged at14 000 r.p.m. for 5 min. The supernatant was transferredto another microtube, then protein concentration wasmeasured with BCA protein assay reagent (Pierce,Rockford, IL, USA) and diluted to the final concentrationof 5 mg/ml. A measure of 10 ml protein reagent wasincubated in 10 mM NADPH 1 mCi/ml 14C-arginine/6 mM CaCl2/50 mM Tris HCl, pH 7.4/6 mM tetrahydro-biopterin/2 mM FAD/2 mM FMN for 30 min at 371C. Thereaction was stopped with 50 mM HEPES, pH 5.5/5 mM

EDTA. Identical samples were prepared without CaCl2

to determine the amount of calcium-dependent NOSactivity. All samples were run in duplicates. The calcium-dependent NOS activity was calculated by subtractingthe NOS activity measured without calcium. The radio-activity of the sample eluate was measured by liquidscincillation. Enzyme activity was expressed as countsper minutes per microgram (CPM/mg).

Clinical ischemia indexOn postinjection day 30, we assessed the degree ofischemia based on the presence or absence of pressuresores, gangrene, or atrophy, and observed the rat’s gaitas previously described.37 Digital pictures of the rats’hindlimbs were taken to document signs of ischemia.

Statistical analysisData were presented as mean7s.d. Analysis of variance(ANOVA) was performed for multiple comparisonsamong different groups, followed by post hoc analysisusing the Student–Newman–Keuls test. Differences witha P-value less than 0.05 were considered statisticallysignificant.

Acknowledgements

We express our appreciation to Rickmer Braren, YungHaeKim, Yuankai Lin, Yuki Sakai, and Ari Hoffman for theirexcellent assistance. This work was supported by the Natio-nal Institutes of Health Grant HL68042 (GLT), HL75353(LMM) and Pacific Vascular Research Foundation (LMM).


1648

Gene Therapy

References

1 Schainfeld RM, Isner JM. Critical limb ischemia: nothing to give

at the office? Ann Intern Med 1999; 130: 442–444.2 Katusic ZS. Therapeutic angiogenesis: new indication for

endothelial NO synthase gene transfer. Arterioscler Thromb VascBiol 2002; 22: 1254–1255.

3 Tsutsui M et al. Adventitial expression of recombinant eNOS

gene restores NO production in arteries without endothelium.

Arterioscler Thromb Vasc Biol 1998; 18: 1231–1241.4 Murohara T et al. Nitric oxide synthase modulates angiogenesis

in response to tissue ischemia. J Clin Invest 1998; 101: 2567–2578.5 Gurjar MV, Sharma RV, Bhalla RC. eNOS gene transfer inhibits

smooth muscle cell migration and MMP-2 and MMP-9 activity.

Arterioscler Thromb Vasc Biol 1999; 19: 2871–2877.6 Radomski MW, Palmer RM, Moncada S. An L-arginine/nitric

oxide pathway present in human platelets regulates aggregation.

Proc Natl Acad Sci USA 1990; 87: 5193–5197.7 Kubes P, Suzuki M, Granger DN. Nitric oxide: an endogenous

modulator of leukocyte adhesion. Proc Natl Acad Sci USA 1991;

88: 4651–4655.8 Huang PL et al. Hypertension in mice lacking the gene for

endothelial nitric oxide synthase. Nature 1995; 377: 239–242.9 Kullo IJ et al. Expression and function of recombinant endo-

thelial NO synthase in coronary artery smooth muscle cells.

Arterioscler Thromb Vasc Biol 1997; 17: 2405–2412.10 Yogo K et al. Different vasculoprotective roles of NO synthase

isoforms in vascular lesion formation in mice. Arterioscler ThrombVasc Biol 2000; 20: E96–E100.

11 Arnal JF et al. Endothelium-derived nitric oxide and vascular

physiology and pathology. Cell Mol Life Sci 1999; 55: 1078–1087.12 van Haperen R et al. Reduction of blood pressure, plasma

cholesterol, and atherosclerosis by elevated endothelial nitric

oxide. J Biol Chem 2002; 277: 48803–48807.13 Dimmeler S et al. Activation of nitric oxide synthase in

endothelial cells by Akt-dependent phosphorylation. Nature1999; 399: 601–605.

14 Breslin JW et al. VEGF increases endothelial permeability by

separate signaling pathways involving ERK-1/2 and nitric

oxide. Am J Physiol Heart Circ Physiol 2003; 284: H92–H100.15 Morbidelli L et al. Nitric oxide mediates mitogenic effect of

VEGF on coronary venular endothelium. Am J Physiol 1996; 270:

H411–H415.16 Ziche M et al. Nitric oxide synthase lies downstream from

vascular endothelial growth factor-induced but not basic

fibroblast growth factor-induced angiogenesis. J Clin Invest1997; 99: 2625–2634.

17 Wilcox JN et al. Expression of multiple isoforms of nitric oxide

synthase in normal and atherosclerotic vessels. ArteriosclerThromb Vasc Biol 1997; 17: 2479–2488.

18 Da Silva-Azevedo L, Baum O, Zakrzewicz A, Pries AR. Vascular

endothelial growth factor is expressed in endothelial cells

isolated from skeletal muscles of nitric oxide synthase knockout

mice during prazosin-induced angiogenesis. Biochem Biophys ResCommun 2002; 297: 1270–1276.

19 Feng Y et al. VEGF induces nuclear translocation of Flk-1/KDR,

endothelial nitric oxide synthase, and caveolin-1 in vascular

endothelial cells. Biochem Biophys Res Commun 1999; 256:

192–197.20 Papapetropoulos A et al. Nitric oxide synthase inhibitors

attenuate transforming-growth-factor-beta 1-stimulated capil-

lary organization in vitro. Am J Pathol 1997; 150: 1835–1844.21 Silvestre JS et al. Vascular endothelial growth factor-B promotes

in vivo angiogenesis. Circ Res 2003; 93: 114–123.22 Zhang R et al. Nitric oxide enhances angiogenesis via the

synthesis of vascular endothelial growth factor and cGMP after

stroke in the rat. Circ Res 2003; 92: 308–313.

23 Cernadas MR et al. Expression of constitutive and inducible

nitric oxide synthases in the vascular wall of young and aging

rats. Circ Res 1998; 83: 279–286.24 Tamarat R et al. Endothelial nitric oxide synthase lies down-

stream from angiotensin II-induced angiogenesis in ischemic

hindlimb. Hypertension 2002; 39: 830–835.25 Fukumura D et al. Predominant role of endothelial nitric oxide

synthase in vascular endothelial growth factor-induced angio-

genesis and vascular permeability. Proc Natl Acad Sci USA 2001;

98: 2604–2609.26 Leibovich SJ et al. Production of angiogenic activity by

human monocytes requires an L-arginine/nitric oxide-synthase-

dependent effector mechanism. Proc Natl Acad Sci USA 1994; 91:

4190–4194.27 Montrucchio G et al. Nitric oxide mediates angiogenesis induced

in vivo by platelet-activating factor and tumor necrosis factor-

alpha. Am J Pathol 1997; 151: 557–563.28 Lloyd PG, Yang HT, Terjung RL. Arteriogenesis and angiogen-

esis in rat ischemic hindlimb: role of nitric oxide. Am J PhysiolHeart Circ Physiol 2001; 281: H2528–H2538.

29 Smith RS et al. Human endothelial nitric oxide synthase gene

delivery promotes angiogenesis in a rat model of hindlimb

ischemia. Arterioscler Thromb Vasc Biol 2002; 22: 1279–1285.30 Brevetti LS et al. Overexpression of endothelial nitric oxide

synthase increases skeletal muscle blood flow and oxygenation

in severe rat hind limb ischemia. J Vasc Surg 2003; 38: 820–826.31 Lee PC et al. Impaired wound healing and angiogenesis in

eNOS-deficient mice. Am J Physiol 1999; 277: H1600–H1608.32 Kubis N et al. Decreased arteriolar density in endothelial nitric

oxide synthase knockout mice is due to hypertension, not to the

constitutive defect in endothelial nitric oxide synthase enzyme.

J Hypertens 2002; 20: 273–280.33 Aicher A et al. Essential role of endothelial nitric oxide synthase

for mobilization of stem and progenitor cells. Nat Med 2003; 9:

1370–1376.34 Murohara T et al. Role of endothelial nitric oxide synthase in

endothelial cell migration. Arterioscler Thromb Vasc Biol 1999; 19:

1156–1161.35 Buckwalter JB et al. Endogenous vascular remodeling in

ischemic skeletal muscle: a role for nitric oxide. J Appl Physiol2003; 94: 935–940.

36 Welling TH et al. Systemic delivery of the interleukin-1 receptor

antagonist protein using a new strategy of direct adenoviral-

mediated gene transfer to skeletal muscle capillary endo-

thelium in the isolated rat hindlimb. Hum Gene Ther 1996; 7:

1795–1802.37 Paek R et al. Correlation of a simple direct measurement of

muscle pO(2) to a clinical ischemia index and histology in a rat

model of chronic severe hindlimb ischemia. J Vasc Surg 2002; 36:

172–179.38 Rios CD et al. Adenovirus-mediated gene transfer to normal and

atherosclerotic arteries. A novel approach. Arterioscler ThrombVasc Biol 1995; 15: 2241–2245.

39 Tsurumi Y et al. Direct intramuscular gene transfer of naked

DNA encoding vascular endothelial growth factor augments

collateral development and tissue perfusion. Circulation 1996; 94:

3281–3290.40 Majesky MW. A little VEGF goes a long way. Therapeutic

angiogenesis by direct injection of vascular endothelial growth

factor-encoding plasmid DNA. Circulation 1996; 94: 3062–3064.41 Asahara T et al. Synergistic effect of vascular endothelial growth

factor and basic fibroblast growth factor on angiogenesis in vivo.

Circulation 1995; 92: II365–II371.42 Nabel EG. Gene therapy for cardiovascular disease. Circulation

1995; 91: 541–548.43 Pu LQ et al. Enhanced revascularization of the ischemic limb by

angiogenic therapy. Circulation 1993; 88: 208–215.


1649

Gene Therapy

44 Vassalli G et al. A mouse model of arterial gene transfer: antigen-specific immunity is a minor determinant of the early loss ofadenovirus-mediated transgene expression. Circ Res 1999; 85:e25–e32.

45 Taniyama Y et al. Therapeutic angiogenesis induced by humanhepatocyte growth factor gene in rat and rabbit hindlimbischemia models: preclinical study for treatment of peripheralarterial disease. Gene Therapy 2001; 8: 181–189.

46 Cho WK et al. Modulation of Starling forces and muscle fibermaturity permits adenovirus-mediated gene transfer to adultdystrophic (mdx) mice by the intravascular route. Hum GeneTher 2000; 11: 701–714.

47 Vajanto I et al. Evaluation of angiogenesis and side effects inischemic rabbit hindlimbs after intramuscular injection of adeno-viral vectors encoding VEGF and LacZ. J Gene Med 2002; 4: 371–380.

48 Isner JM et al. Clinical evidence of angiogenesis after arterialgene transfer of phVEGF165 in patient with ischaemic limb.Lancet 1996; 348: 370–374.

49 Brevetti LS et al. Administration of adenoviral vectors inducesgangrene in acutely ischemic rat hindlimbs: role of capsidprotein-induced inflammation. J Vasc Surg 2001; 34: 489–496.

50 Ooboshi H et al. Altered vascular function after adenovirus-mediated overexpression of endothelial nitric oxide synthase.Am J Physiol 1997; 273: H265–H270.


1650

Gene Therapy

RESEARCH ARTICLE

Normal growth and regenerating ability of myoblastsfrom unaffected muscles of facioscapulohumeralmuscular dystrophy patients

J-T Vilquin1,5, J-P Marolleau2,5, S Sacconi3,5, I Garcin1, M-N Lacassagne2, I Robert2, B Ternaux2,

B Bouazza4, J Larghero2 and C Desnuelle3

1Inserm U582, Groupe hospitalier Pitie-Salpetriere, Institut de Myologie, Paris, France; 2Laboratoire de Therapie Cellulaire, Hopital SaintLouis, Paris, France; 3Inserm U638, CHU de Nice, Federation des Maladies Neuromusculaires, Nice, France; and 4CNRS UMR 7000,Groupe hospitalier Pitie-Salpetriere, Paris, France

Facioscapulohumeral muscular dystrophy (FSHD) is anautosomal dominant disease characterized by a typicalregional distribution, featuring composed patterns of clinicallyaffected and unaffected muscles. No treatment is availablefor this condition, in which the pathophysiological mechanismis still unknown. Autologous transfer of myoblasts fromunaffected to affected territories could be considered asa potential strategy to delay or stop muscle degeneration.To evaluate the feasibility of this concept, we explored andcompared the growth and differentiation characteristics ofmyoblasts prepared from phenotypically unaffected musclesof five FSHD patients and 10 control donors. According toa clinically approved procedure, 109 cells of a high degree ofpurity were obtained within 16–23 days. More than 80%

of these cells were myoblasts, as demonstrated by labelingof the muscle markers CD56 and desmin. FSHD myoblastspresented a doubling time equivalent to that of control cells;they kept high proliferation ability and did not show earlytelomere shortening. In vitro, these cells were able todifferentiate and to express muscle-specific antigens. Invivo, they participated to muscle structures when injectedinto immunodeficient mice. These data suggest that myo-blasts expanded from unaffected FSHD muscles may besuitable tools in view of autologous cell transplantationclinical trials.Gene Therapy (2005) 12, 1651–1662. doi:10.1038/sj.gt.3302565; published online 16 June 2005

Keywords: facioscapulohumeral muscular dystrophy; myoblast; cell transplantation; muscle regeneration; muscle differentia-tion; telomere

Introduction

With an incidence of 1/20 000 in Occidental countries,facioscapulohumeral muscular dystrophy (FSHD) isconsidered a rare disease, but is the third most commoninherited myopathy.1 It is characterized by typicalregional distribution with progressive weakening offace and shoulder girdle muscles, further extendingto abdominal and pelvic girdle, and to humeral andanterior foreleg muscles. The disease is transmitted as anautosomal dominant trait and is associated with adeletion of an integral number of copies of a 3.3 kbtandem repeated unit termed D4Z4 located at the 4qsubtelomeric region. This deletion results in an EcoRIfragment shorter than 35 kb, providing the molecularmarker for FSHD diagnosis.2,3 While, in general popula-tion, D4Z4 comprises 11–150 U, FSHD patients carry lessthan 11 repeats. The disease is usually much worse

(earlier onset and greater clinical severity) when theshort 4q D4Z4 array is in the smaller size range (1–4repeats).4,5 In spite of the advances in molecular biology,its pathophysiological mechanism has not been eluci-dated yet, nor has its anatomical specificity. Thedifficulty to define this disease exactly is mainly due toits complexity, since a profound misregulation of geneexpression rather than a single gene defect seems to beinvolved.6,7 For this reason, no classical gene therapyapproach can be considered yet (ie gene transfer, generepair, exon skipping), prompting for the identification ofnew therapeutic concepts. No animal model is available,and no treatment can currently be recommended.8

Among potential therapeutic strategies for musculardystrophies, the intramuscular injection of culturedmyoblasts (myoblast transplantation, MT) has beenconsidered for generalized myopathies – mainly Duch-enne muscular dystrophy (DMD).9,10 Unlike FSHD, thisdisease is characterized by diffuse muscle involvementleading to a progressive degeneration of the whole bodymusculature. The pathophysiological mechanism ofDMD involves deficits of the dystrophin gene, andexhaustion of the myogenic reservoir secondary tomultiple cycles of degeneration–regeneration. As aconsequence, the remaining cells cannot sustain the

Received 30 November 2004; accepted 5 May 2005; published online16 June 2005

Correspondence: Dr J-T Vilquin, Inserm U582, Institut de Myologie,Groupe hospitalier Pitie-Salpetriere, 47 Boulevard de l’Hopital, 75651Paris Cedex 13, France5These authors contributed equally to this work


www.nature.com/gt

large-scale expansions required for perpetual muscleregeneration, or use in clinical practice.11–13 For this reason,the DMD myoblasts have been prepared from healthydonors, and MT has been performed in a heterologouscontext. The MT concept has been validated in normal anddystrophic mouse models, where injection of healthymyoblasts increased muscle mass and function.14,15

Despite positive results in these small rodents,16–20 clinicaltrials have presented limited success.21–29 Improvementsof injection and immunosuppression parameters haverecently been achieved in primates9,10,30 and translated tothe human situation with encouraging, although localizedsuccess.31 Nevertheless, one of the most importantproblems of heterologous MT remains the immuneresponse raised by the recipient against donor cells.22,29

This issue is completely avoided in the context ofautologous MT, which has been successfully developedas an experimental treatment for postischemic heartfailure or dilated cardiomyopathy, showing its feasibility,safety and its great clinical interest.32–36 In some myo-pathic condition such as FSHD, characterized by arelatively localized muscle involvement, and the absenceof therapeutic options, autologous MT could be consid-ered as an alternative strategy. In this case, unaffectedmuscles could be the source of myoblasts to betransferred into affected muscles in order to improvetheir regeneration ability.

The validation of this concept requires the completionof four stages: (i) production and in vitro characterizationof FSHD myoblasts from unaffected muscles, andcomparison with myoblasts from muscles of healthydonors. This comparison would allow establishing ifFSHD myoblasts grown from unaffected muscle, but stillcarrying the FSHD molecular defect, would manifest ornot its deleterious effect when submitted to a large-scaleexpansion or in vitro differentiation. (ii) Evaluationin vivo of FSHD myoblast regenerating ability, throughimplantation into the skeletal muscles of relevant animal

models. Upon validation of these first two steps, clinicaltrials should be considered as (iii) set up of a phase Iassay dedicated to the assessment of safety andfeasibility, and (iiii) set up of a phase II clinical trialdedicated to the evaluation of efficacy. This studyregards the first two steps. We evaluated in vitroproliferation and differentiation and in vivo regeneratingability of myoblasts obtained from phenotypicallyunaffected muscles of five FSHD patients and of 10control donors prepared in a previous clinical trial usingthe same procedures.34 We could not observe anydifference between FSHD and control myoblasts.

Results

Patient selectionFive FSHD patients (FSHD 1–5) carrying, respectively,10, 7, 9, 6 and 7 D4Z4 repeats were included after givingtheir informed consent. These patients presented with afamilial form of the disease. According to Brooke–Vignosfunctional scale,37 all patients were ranked 2 for arms,and 1 for legs. Facial and shoulder girdle muscles wereaffected while no clinical involvement of the legs wasevident. Muscle MRI did not show any fibrotic oradipose infiltration in the muscular territory elected formuscle biopsy. The 10 control donors (C 1–10) had beenrecruited in the context of a previous clinical studydedicated to the autologous MT in postinfarction leftventricular dysfunction.34 The mean ages of the FSHDpatients and control donors were, respectively, 5375 and61711 years. All FSHD patients and controls were males.

Muscle biopsy and transmission to the cell culturelaboratoryFSHD patients and control donors underwent a right orleft vastus lateralis muscle biopsy, which mean weightswere, respectively, 1.170.1 and 12.970.4 g. The larger

Table 1 Schedules of FSHD and control culture productions

Identification Initial day/# plates P1 (day, # plates) P2 (day, # plates) P3 (day, # plates) P4 (day, # plates) Harvest (day) Harvest (day)

FSHD1 D1/1 8/2 12/4 15/11 19/21 23FSHD2a D1/1 8/1 15/1 19/2 22/8 25/24 28FSHD3 D1/1 8/1 14/3 16/11 20/26 25FSHD4 D1/1 8/1 12/3 15/14 18/26 22FSHD5 D1/1 8/1 12/4 15/16 18/23 21

Identification Initial day/# plates P1 (day, # plates) P2 (day, # plates) Harvest (day)

C1 D0/6 8/6 13/14 16 — — —C2 D0/3 9/6 13/22 18 — — —C3 D0/4 9/6 15/24 20 — — —C4 D0/4 7/6 10/26 14 — — —C5 D0/4 8/6 13/24 17 — — —C6 D0/4 7/6 10/26 15 — — —C7 D0/4 8/6 11/26 15 — — —C8 D0/4 8/6 11/26 15 — — —C9 D0/4 8/6 11/26 14 — — —C10 D0/4 8/6 11/24 15 — — —

Cell cultures were serially expanded at the indicated times (days). The number of culture plates (#) is given as a function of the expansionstep. C: controls. The large size of control muscle biopsies allowed the initial seeding of a larger number of plates than that used for FSHDbiopsies. From the onset of myoblast cultures to the day of final harvest, FSHD cells underwent four to five passages (P) within 21 to 28 days,and control cells two passages within 14 to 20 days. The large-scale production required 14 to 26 plates on the last expansion.aOwing to storage problem, the FSHD2 schedule was moved forward by one passage.

Production and characterization of FSHD myoblastsJ-T Vilquin et al

1652

Gene Therapy

size of control muscle biopsies was a specific require-ment of the clinical trial dedicated to intramyocardialMT. Biopsies were well tolerated by both groups. Theirhistological evaluation did not reveal any sign of musclepathology. Transmission to the cell culture laboratorywas eventless, except for the biopsy of patient FSHD2,which underwent a storage problem. This adverse eventdelayed the final harvest without affecting the final yield.For this reason, the FSHD2 data were not included in thestatistical calculations.

Cell productionThe schedules of cell expansions are presented in Table 1.Except for the FSHD2 biopsy, the methodology allowedto extract 1.13� 106 and 9.7� 105 cells/g of muscle fromFSHD patients and controls, respectively (Figure 1a). Thecells were then seeded at low density (2000–4000/cm2).Following a lag time (4–6 days), cells grew as isolatedcolonies. This first passage (P) allowed a homogeneousspreading of the cells and was performed before colonieswere reaching confluence. The highest growth rates wereobserved between P1 and P2, with doubling times of 31.6and 29.6 h for FSHD and control cells, respectively.Thereafter, the doubling times increased in both condi-tions within the same proportions (Figure 2a and b).From Day 0 (D0) to P3, the cells underwent 7.0 and 6.6divisions, respectively. At P3, 156� 106 FSHD cells and957� 106 control cells were obtained. Since doublingtimes were equivalent, the difference was attributed tothe starting sizes of biopsies, the control being 12 timeslarger than the FSHD. The threshold number of cellsrequired for potential clinical trials was arbitrarilyestablished at 8� 108 on the basis of previous experi-ence.34 To reach this number, FSHD cells were grownup to P5, for three supplementary divisions, and overallperformed 9.8 divisions. We produced 8.5� 108 to1.6� 109 myoblasts (mean 1.06� 109) within 22.7 days.This result is to be compared to the production of6.5� 108 to 1.2� 109 myoblasts (mean 9.5� 108) within15.8 days in the historical control group. Our results thussuggest that the initial numbers of myogenic progenitorspresent per gram of disease or control muscles wereequivalent. Myoblasts were small, fusiform or polygonalcells, mostly slightly refringent, and not evenly spreadonto the substrate. The nuclear/cytoplasm ratio wasgenerally high, and two or three dark nucleoli wereclearly distinguishable. No morphological difference wasobserved between FSHD and control cultures at differentpassages (Figure 3a–d).

Cell viabilityAs estimated by propidium iodide exclusion, theviability of the cells at each passage and at the time offinal harvest was above 90% in both the FSHD andcontrol groups (Figure 1b).

Phenotypic characterizationMyoblasts express the surface NCAM antigen, pre-viously described as leu-19, 5.1H11, NKH-1 or CD56.Muscle cells (cardiac, smooth and skeletal types) alsoexpress desmin, an intracellular cytoskeleton protein, atan early stage of differentiation.38–41

As presented in Figure 1c and d, the cell culturemethodologies allowed the rapid and reproducibleemergence of myoblasts as a dominant population. The

percentage of CD56- and desmin-positive cells increasedrapidly, reaching a plateau (above 80%) within twopassages and remained elevated throughout theculture. Among the whole-cell population, the specificgrowth rate of myoblasts can be calculated from CD56

Figure 1 Production and characterization of FSHD and controlmyoblasts. Presented values are mean7s.e. Control: n¼ 10; FSHD:n¼ 4. The growth characteristics of FSHD2 are not included. (a) Numberof myoblasts harvested at each passage. Since the initial sizes of controlbiopsies were 12 times bigger than those of FSHD biopsies, twosupplementary passages were required to reach the final expected numberof cells from the FSHD biopsies. (b) Cell viability was assessed by FACSusing propidium iodide. It was above 90%. (c) Percentage of CD56-positive cells was evaluated by FACS. Although this population was notdominant on the initial day, the growth conditions allowed the myoblaststo take over any other cell type and to represent more than 80% of the cellsfrom P2 and ongoing passages. (d) The percentage of desmin-positivecells was evaluated by FACS. The low number of cells did not allow thisestimation on D0. This percentage reflects that of CD56-positive cells,confirming the high degree of purity in myoblasts of the cultures.


1653

Gene Therapy

evaluation. The initial content in CD56-positive cells wasnot assessable in the case of FSHD patients because theinitial number of cells was too low for precise cytofluori-metric evaluation. Nevertheless, our data indicate thatmyoblasts first grew rapidly (Figure 2c and d) andpresented the shortest doubling times between P1 andP2. Subsequently, the doubling times were increased.Over the P1–P3 period, growth rates were comparable(doubling times of 31.2 versus 32.8 h respectively). In thecontrol group, myoblasts underwent more divisions thanthe whole-cell populations during the first phases of theculture (D0-P1, 3.7 versus 1.6; and D0-P3, 8.8 versus 6.6).This may explain their aptitude to become and remainthe dominant cell type in the primary cultures.

Proliferative abilityFollowing the final harvesting (P5 or P6), cells obtainedfrom FSHD patients were further subcultured until thedrop of CD56-positive cells under 50%. The finaltheoretical yield was obtained by multiplication of theexpansion ratios observed at each passage. This calcula-tion allowed extrapolating the proliferative potential ofthese cells (Table 2). These data indicate that the cellsproduced by serial expansions still kept the ability togrow for at least two (patient FSHD1) to seven (patientsFSHD3 and FSHD4) additional passages.

In vitro differentiationWhen placed into differentiation conditions, the cellsfused to form multinucleated myotubes (Figure 3e–h).Their sizes and shapes could vary, both in patient andcontrol cultures. Myotubes may contain two to dozens ofnuclei. Some were branched, and others were straight. Insome instances, large and flat syncitia were observed. Nomorphological difference was noted between FSHD andcontrol myoblasts. The fast and slow isoforms of myosinheavy chain (MyHC) were expressed according to thesame chronological pattern in FSHD and control myo-tubes. No expression was observed at D0. Fast MyHCisoform appeared in myotubes at D1, while slow isoformappeared at D3. Both persisted up to D6 (Figure 4). Nosignificant difference was observed between fusionindexes calculated at different time points (D1, D3, D6)in FSHD and control cultures (data not shown).

Telomere sizeThe telomere sizes were evaluated for three out of fivepatients and two controls. They were estimated to be7.970.1 (FSHD1), 6.870.3 (FSHD4) and 7.970.3(FSHD5). As a mean (7.53), they did not differ from thatof the two controls (6.970.2 and 7.470.3), and alltelomere sizes kept within the range of normal values(6.5–8.5), as published.42–44

Number of D4Z4 repeatsThe number of D4Z4 repeats was assessed by the samelaboratory (Dr M Jeanpierre) first at the time of patientselection as a molecular confirmation of the clinicaldiagnosis, and then on cell cultures at the end of theproduction process (P5 or P6). No difference betweenthese two values was noted (data not shown). These dataexclude the possibility that myoblasts not harboring theFSHD mutation would have taken over mutated myo-blasts, as could result from the expression of a somaticmosaicism.45

Figure 2 Growth characteristics of FSHD and control myoblasts. Thenumber of divisions (a, c) achieved by the cells and, consequently, thedoubling times (b, d) have been estimated from cell counting at eachpassage. These values were calculated from successive and from first andend passages. Filled bars: Control myoblasts; open bars: FSHD myoblasts.The parameters were established for total cells in the culture, and for themyoblast population only, as the proportion of CD56-positive cells could beestablished. The doubling times were the longest during the first phaseof the culture (a), between initial seeding and P1. The first expansiontriggered a round of rapid divisions. The doubling times were the fastestbetween P1 and P2 (b, d) whatever the nature of the culture or of the cells.Thereafter, it increased gradually with the number of passages. Control andFSHD cells underwent 7 and 9.8 divisions, respectively.


1654

Gene Therapy

Figure 3 Morphological aspect of FSHD and control myoblasts during proliferation or in vitro differentiation. Inverted phase contrast microphotographswere performed before the onset of differentiation (a–d) and 6 days after (e–h). Myoblasts are small, polygonal or slightly fusiform cells with a highnucleocytoplasmic ratio. No morphological difference was observed between control (a) and FSHD myoblasts (b–d illustrate FSHD2, FSHD3 and FSHD5,respectively). Myotubes presented various shapes: some were branched (e, f), flattened (g) or straight (e, h). They contained two to dozens of nucleipresenting one or two contrasted nucleoli. These morphologies could be observed both in control (e) and FSHD cultures (f–h illustrate FSHD2, FSHD3 andFSHD5, respectively). Original magnification: � 40.


1655

Gene Therapy

In vivo regenerating abilitySince no FSHD animal model is currently available,immunodeficient Rag2/gc�/� (Rag) mice were used toassess the in vivo regenerating ability of xenogenicmyoblasts. Although they do not present any muscle

pathology, Rag mice were chosen because they do notdevelop any specific immunity or antigen-dependent cellcytotoxicity.46 FSHD myoblasts were injected twiceinto the tibialis anterior muscles of Rag mice preceded,in some instances, by the injection of the myotoxinnotexin (Table 3). At 1 month after MT, the expression ofhuman proteins validated the implantation of humancells into mouse muscle fibers.16,47–49 Human dystrophinwas observed in all cell-injected muscles, and was absentfrom muscles injected with a saline solution (Table 3,Figure 5). No significant differences were observedbetween muscles receiving FSHD or control myoblasts(P40.8). In the absence of notexin, the dystrophin-positive fibers were of small diameter and frequentlytrapped within muscle fascicles. Notexin administrationdid not significantly affect the number of dystrophin-positive fibers (P40.1), but these appeared morescattered within the muscle.

Discussion

The present study has been performed to evaluate theconcept of autologous MT as a therapeutic tool for

Table 2 Maximal theoretical yield of FSHD myoblast cultures

Identification # Cells at harvest,passage

Theoretical yield End passage;CD56-positive

(%)

FSHD1 980� 106, P5 5–10� 109 P7; 54%FSHD2 930� 106, P6 50� 109 P11; 67%FSHD3 850� 106, P5 41000� 109 P12; 75%FSHD4 910� 106, P5 480� 109 P12; 60%FSHD5 1648� 106, P5 42000� 109 P11; 84%

From the harvest (P5 for FSHD1, 3, 4, 5 and P6 for FSHD2), analiquot of cells was further expanded and subcultured to evaluatethe potential maximal yield that could be achieved before the dropof CD56-positive population below 50%. This yield was estimatedby the multiplication of expansion ratios between each passage, bythe number of cells obtained at the time of harvest.

Figure 4 Immunocytofluorescent characterization of control and FSHD myotubes. At 6 days after the onset of differentiation, control (a–f) and FSHD (g–l)cultures were fixed and processed for the analysis of muscle markers expression. Desmin (d, j) stained both myoblasts and myotubes. The parallelobservation of phase contrast and immunofluorescent images (a with d, g with j) shows that most cells express this marker, illustrating the high purity of thecultures. The fast (e, k) and slow (f, l) isoforms of MyHC are expressed in most, if not all myotubes, suggesting a co-expression at this stage of the culture.No difference in the expression pattern nor the chronology of differentiation could be observed between control and FSHD cultures (FSHD4 presented here).Original magnification: � 20.


1656

Gene Therapy

patients presenting with a hereditary muscle disordercharacterized by a composed pattern of affected andunaffected muscles. This approach would avoid theimmunological drawbacks met in heterologous MTclinical trials developed for generalized muscular dys-trophies, and it relies on the assumption that, in regionalmyopathies, clinically unaffected muscles could be asource of myoblasts functionally able to participate inaffected muscles regeneration. To our knowledge, thisconcept has not been proven yet. To establish itspertinence, it was necessary (i) to choose an appropriatepathological model; (ii) to set up and validate a specificcell culture methodology in order to produce, fromunaffected muscles, a large amount of myoblasts in adelay compatible with their clinical use; and (iii) tocharacterize both in vitro and in vivo these cells.

Among muscular dystrophies presenting with loca-lized muscular involvement, we have selected FSHD,the third most common inherited muscular dystrophy,which is characterized by a typical regional distributionof muscle weakness, an established molecular diagnosis,and an absence of therapeutic solution. We exploredproliferation and in vitro and in vivo differentiation ofmyoblasts prepared from unaffected vastus lateralismuscle of five FSHD patients presenting a mildphenotype and D4Z4 repeats number ranging from 6 to10. Beyond the simple large-scale cell production, ourstudy has been designed to evaluate the characteristics ofthe cells at any step of the culture, in order to standardizethe comparison between these cell cultures and that ofcontrol myoblasts prepared from the same muscle of 10healthy donors. We assumed that these controls could beconsidered as a baseline reference, since they haveundergone a previous successful use in large-scaleclinical trials.

Using a validated cell culture methodology, we wereable to produce 109 FSHD myoblasts within 23 days.

They presented normal morphology, high and sustainedviability, thus the methodologies (including enzymatictreatments and mechanical constraints) elected for cellproduction were considered harmless for these cells.Both FSHD and control cultures presented a high degreeof purity. Myoblasts and, upon differentiation, myotubes,expressed specific muscular markers. No differencesbetween FSHD and control cells were observed in termof in vitro differentiation. This study is the firstdemonstration that large-scale production of myoblastsfrom unaffected muscles of FSHD patients is feasiblewithin a short delay, standardized, reproducible andfulfilling the clinical requirements. The production waseven possible in a critical situation (eg FSHD2 cells).

Interestingly, FSHD myoblasts obtained in our studydid not present any morphological aberrations whencompared to controls. This finding contrasts with aprevious report on FSHD myoblasts carrying three tosix D4Z4 repeats, showing a ‘vacuolar-necrotic pheno-type’.50 We cannot conclude about this discrepancybecause many variables could be involved as, forexample, the anatomical location of muscle biopsy, thedegree of muscular dystrophy and the cell cultureparameters. It may also be due to differences in thegenetic status of FSHD patients, since our patients hadlarger (six to 10) D4Z4 repeats.

Our study also underlines some differences betweenFSHD myoblasts prepared from unaffected muscles andmyoblasts prepared from patients presenting with DMDor congenital myotonic dystrophy (CMD). In fact, theproportion of CD56- and desmin-positive FSHD myo-blasts remained high throughout the production, whileDMD myoblasts have been reported to display aninexorable and rapid decline during consecutive pas-sages11–13,51 (personal communication), and to undergopremature senescence, although they generally keepthe ability to form myotubes in vitro. FSHD myoblasts

Table 3 Myoblast transplantation into immunodeficient Rag mice

Identification # Injected muscles # Cells injected CD56-positive labeling (%) Injection protocola Dystrophin-positive fibers

FSHD1 5 5� 106 82 1 237135 4� 106 68

FSHD2 5 5� 106 88 1 18782 3� 106 97

FSHD3 5 5� 106 86 2 407105 5.2� 106 88

FSHD4 5 5� 106 70 1 407265 5� 106 99

FSHD5 5 3� 106 80 2 337185 3.4� 106 81

C1 5 4.5� 106 94 2 557235 4.5� 106 70

C2 5 3.5� 106 82 2 237195 3.3� 106 90

Culture medium 5 — — 2 05

Immunodeficient Rag mice (12 weeks old) were injected with patient or control myoblasts at 1-week interval (n¼ 2–5 muscles for eachcondition). The cells were used at P6 and P7, and the percentage of CD56-positive myoblasts was estimated by FACS on the day of MT.aTwo protocols were followed: (1) one group received 100 ng of notexin at two sites the day before the first cell injection. (2) The other one didnot receive notexin. Then, all mice received two series of myoblast injections at 1-week interval. The specific expression of human dystrophinwas assessed by immunocytochemistry, and pictures of the muscle sections were taken using a digital camera for the quantification ofdystrophin-positive muscle fibers per section (n¼ 3). No such fibers could be observed upon injection of culture medium alone. Dystrophin-positive fibers were identified in the animals injected with control or FSHD myoblasts; however, there was no significant difference betweenthe two groups.


1657

Gene Therapy

also differ from CMD myoblasts, which show reductionin proliferative ability, and delays in fusion anddifferentiation.52 The proliferative ability of FSHDmyoblasts was also explored by the measurement ofthe telomere size. The length of the telomeric DNA is anindicator of cell proliferation, its measurement reflectsthe past divisions and is related to the remaining

proliferation ability. Myoblasts from unaffected FSHDmuscles displayed a normal telomere size, contrastingwith myoblasts from DMD patients or elderly people,42–

44,53 which usually present shortened telomeres andreplicative defects due to a history of cumulative proli-feration and muscle regeneration. As expected, thesedata suggest that the pathophysiological mechanism

Figure 5 Participation of FSHD and control myoblasts to in vivo muscle regeneration. Transverse muscle sections were labeled for human dystrophinexpression. At 1 month after MT, donor myoblasts have contributed to the formation of muscle fibers expressing the human antigen (arrows). (a) Culturemedium; (b, c) control myoblasts; and (d–h) FSHD1 to FSHD5, respectively. In the absence of notexin pretreatment, the human dystrophin-positive fiberswere grouped as clusters (b, f, g). Following notexin pretreatment, they showed a more scattered distribution (d). Original magnification: � 25.


1658

Gene Therapy

leading to muscle degeneration in FSHD would differfrom that involved in Duchenne or congenital myotonicdystrophies, or would not progress along the sameprograms. This assumption is conforted by a previousdemonstration that FSHD myoblasts display a normalcalcium homeostasis as opposed to DMD myoblasts.54

The reason why some muscles are specifically affectedin FSHD is not understood yet, and the role played bythe myoblasts in its pathophysiology remains unknown.Our data suggest that the large-scale expansion did nottrigger an in vitro expression of the FSHD disease, sincethe myoblasts from unaffected muscles maintained anormal morphology and a high proliferative capacity.However, the characteristics of myoblasts prepared fromaffected muscles have not been investigated, and itwould be interesting to explore the in vitro and in vivocapacities of such myoblasts prepared using our stan-dardized cell culture methodology. The direct compar-ison of the growth and differentiation characteristics ofmyoblasts from affected and unaffected muscles couldshed light on the mechanisms leading to the specificity ofmuscular involvement observed in FSHD. These com-plementary studies would allow identifying a relation-ship between the dystrophic status of a muscle and thebiological features of its myogenic progenitors, andcould provide an indication of their role in the progres-sion of the disease. Such a relation has been proposed inother pathological situations, such as DMD, but was notestablished in FSHD. In this disease, the muscle-specificinvolvement of local myogenic programs could lead to arestricted deregulation of myogenesis in some musclegroups.

The complete preclinical validation of autologous MTwould have implied the transplantation of myoblastsfrom unaffected muscles into some affected muscles inFSHD animal model, but none is currently available. Theuse of other dystrophic animal models would not havebeen relevant since most of these diseases are basedon monogenic alterations and their pathophysiologicalmechanisms are unrelated to FSHD.55–58 In the presentstudy, human FSHD myoblasts were injected intoimmunodeficient Rag mice, a validated mouse modelfor the analysis of muscle regeneration in the context ofxenogenic MT. The implantation results we obtainedwere close to that reported previously47–49 and confirmedthat myoblasts prepared from unaffected FSHD muscles,even used at the 6th and 7th passages, take part intoin vivo muscle regeneration. The overall number ofdystrophin-positive fibers was inferior to that reportedpreviously in immunocompetent models of musculardystrophies in the context of syngenic or allogenicMT.17,18,20 In these models, however, dystrophic musclesfrequently undergo spontaneous cycles of degeneration/regeneration that would increase myoblast incorporationwithin host tissue. In the present study, the evaluation oftransplantation outcome has been performed 1 monthafter transplantation, a time course that is of general usein human and animal MT assays.9,16,17,20,31,47,49 Thismodel was not regarded helpful to evaluate the long-term persistence of muscle fibers harboring the humanmutated genes. Indeed, a negative selection of musclefibers expressing the mutated gene could occur with timein the healthy environment of mouse muscle fibers, andthis selection would not be representative of the situationpresented in FSHD.

In conclusion, the present study suggests that myo-blasts prepared from unaffected muscles of FSHDpatients may tolerate the large-scale expansions requiredfor their clinical use, given the large number of cellspresently needed to perform intramuscular injec-tions.31,34 These myoblasts also preserve, at differentpassages, their ability to in vitro differentiate and toin vivo participate to muscle regeneration. The combina-tion of our results and studies on intramuscular injectionprocedures raise the possibility of future clinical trials ofautologous cell therapy for FSHD patients. Moreover,this standardized cell culture methodology could allowin the future the preparation and the assessment ofmyoblasts from patients presenting with other formsof muscular dystrophies, either in view of similarautologous MT approaches or to establish cellularmodels of these diseases.


Patient selectionThis preclinical study was approved by the FrenchRegulatory Health Authorities and our Ethics committee(CCPPRB, Nice, France).

Eligibility for inclusion of FSHD patients was basedon: (1) age (between 18 and 65 years old); (2) clinicalpresentation of FSHD, confirmed by molecular diagnosis(number of D4Z4 repeats between 1 and 10); to avoidthe possibility of somatic mosaicism, only familial cases(being at least the second generation of affected) wereselected; (3) clinical status according to the Brooke andVignos scale (grade 42 for upper limb muscles andgrade 1 for lower limb muscles); and (4) absence ofclinical deficit in the vastus lateralis muscle assessed bymanual muscle testing, and of dystrophic changes at theNMR imaging. A subsequent routine histological exam-ination on muscle biopsy confirmed these findings. Themain exclusion criteria consisted of diabetes, pregnancy,vasculopathy, steroid or immunomodulatory treatment,anticoagulant therapy and positive serological test forHIV or hepatitis B, C virus.

Historical control culture data were collected from aprevious phase I clinical trial dedicated to the evaluationof MT as a therapeutic approach for postischemic heartfailure.34 In this study, 10 patients received intramyocar-dial transplantation of autologous myoblasts preparedaccording to the same validated procedure. The exclu-sion criteria have been described previously,34 one ofwhich being the absence of muscular dystrophy.

Muscle biopsyA 1 g biopsy of the vastus lateralis muscle was performedin FSHD patients under local anesthesia, weighed,placed into a preservation medium containing antibiotics(gentamicin, 50 mg/ml, Panpharma) and sent overnightto the cell therapy laboratory.

The vastus lateralis muscle biopsies from controlpatients were obtained through the same procedure.The initial biopsy retrieved from these patients waslarger (up to 14 g) to comply with shorter produc-tion delays required by the specific schedule of theclinical trial.


1659

Gene Therapy

Cell cultureSet-up of culture. Cell culture procedures were agreedby the French Health Authorities (AFSSAPS) and wereapplied identically to FSHD and control groups. Themuscle biopsies were received and processed in adedicated clean room according to Good ManufacturingPractices. The biopsies were processed 24 h after collec-tion (D1) for FSHD patients, and 4–6 h after collection(D0) for controls, due to a shortened transportationprocedure. The muscle fragments were minced anddigested using collagenase (1 h at 371C, Roche-Boehrin-ger), then trypsin–EDTA (0.25%, 20 min at 371C, Hy-clone-Perbio). The cell suspension was filtered through100 then 40 mm cell strainers (Becton-Dickinson) andpelleted. All subsequent handlings were performed inclose systems using custom-made plastic connectionsand bags. The cells were seeded onto a one-layersubstrate (600 cm2, Nunc) in the proliferation mediumcontaining: 80% of modified MCDB medium (custom-made by Hyclone-Perbio), 20% Defined fetal bovineserum (DFBS, Hyclone), penicillin (100 U/ml, Sarbach)and streptomycin (10 mg/ml, Specia), 10 ng/ml humanrecombinant bFGF (R&D) and 1 mM dexamethasone(Merck).

Culture expansion. The cells were settled for 7 days,then harvested by trypsinization and expanded. Onefraction was reserved for counting and phenotypicanalysis (see below). Expansions were repeated succes-sively, before reaching confluence (Table 1). Up to 26layers were required to reach the expected numberof cells.

The number of divisions (n) between each passage wascalculated as:

n ¼ ½logðCx=Cx�zÞ�= log 2

The doubling time (D) was calculated as:

D ¼ h=n

where Cx is the number of cells harvested at a consideredpassage x and Cx�z is the number of cells at a given,previous (x�z) passage. h is the number of hoursbetween passages x and x�z.

Residual proliferation ability. An aliquot of cells wasfurther subcultured to evaluate the potential maximalyield that could be achieved before the drop of CD56-positive population below the level of 50%. This limitwas fixed as an indicator of the exhaustion of themyogenic potential. Briefly, 106 cells were plated into a25 cm2 flask (Falcon), and then harvested before reachingconfluence. Cells were counted and a phenotypicanalysis was performed. One million cells was furtherplated into a 25 cm2 flask and this operation wasrepeated as long as cell growth was ongoing. The finaltheoretical yield was obtained by the multiplication ofthe expansion ratios observed at each passage.

Cell viability. Cell viability was evaluated by FACS(FACSCalibur, BD-Pharmingen), based on the propertyof exclusion of propidium iodide (0.1 mg/ml, 2 min atroom temperature, Sigma) by living cells.

Microbiological controls. Microbiological controlswere performed complying with the standard protocols

defined for bacteriological controls of blood-derivedproducts prepared in clinical practice at Saint LouisHospital (Paris, France). Samples were tested for thepresence of aerobic or anaerobic pathogens at all steps ofthe production using the Vitale system (Merieux).

Phenotypical characterizationCD56 evaluation. The cells were incubated with thephycoerythrin-coupled anti-CD56 antibody (cloneNCAM 16.2, 1/5, 15 min at 41C, BD-Pharmingen).Nonspecific labeling was evaluated using the controlisotype (IgG2b, BD-Pharmingen). FACS data (at least 104

events) were analysed using the Cell Quest software.

Desmin evaluation. The cells were first fixed andpermeabilized using saponine (0.1%) in PAF (4%,20 min at room temperature), then incubated with theanti-desmin antibody (Clone D33, 1/100, 15 min at roomtemperature, Dako) or the isotypic control (IgG1, 1/25,BD-Pharmingen). The cells were incubated with thesecondary antibody (Goat F (ab0) 2 anti-mouse H+L)coupled to FITC (1/50, 30 min at room temperature,Caltag) and analysed by FACS.

The characteristics of myoblasts (number of divisions,doubling time) were calculated as above. The truenumber of myoblasts (Mp) at a given passage (p) wascalculated as follows:

Mp ¼ Cp Fp

where Cp is the total number of cells at the passage p,and Fp is the fraction of these cells positive for CD56antigen.

Differentiation. Following cell harvest, 5� 105 cellswere seeded into 25 cm2 flasks and allowed to settleand proliferate for 72 h. Differentiation was induced byreplacement of the proliferation medium by the differ-entiation medium containing the modified MCDBmedium (98%), DFBS (2%) and antibiotics. Microphoto-graphs (Olympus microscope equipped with a Photo-metrics Cool snap digital Camera) were performed at theonset of differentiation (D0) and 6 days later. In parallel,2� 104 cells were seeded into 12-well plates, allowed toproliferate for 72 h. Following induction of differentia-tion, cells were permeabilized and fixed by coldmethanol at D0, D1, D3 and D6. The expression ofCD56, desmin and MyHC isoforms was assessed usinganti-CD56, anti-desmin, anti-fast MyHC (clone MY32,1/400, 1 h, Sigma) or anti-slow MyHC (clone NOQ7.5.4D,1/1000, 1 h, Sigma) followed by a goat anti-mouseantibody (Alexa Fluor 568, 1/1000, 1 h, MolecularProbes). The presence or absence of labeled structureswas evaluated qualitatively using an inverted Olympusmicroscope equipped for fluorescence analysis. Thefusion index was calculated on four to seven microscopicfields (at least 1000 nuclei counted) using the formula:

FI ¼ððnumber of nuclei within myotube structuresÞ=ðtotal number of nuclei in the fieldÞÞ�100

Evaluation of telomere size. DNA samples wereextracted from cell pellets, digested by HinfI (Biolabs)for 2 h at 371C to generate terminal restriction fragments


1660

Gene Therapy

(TRFs) containing the telomeric region with several kbpairs of the (TTAGGG) repeat sequence and a subtelo-meric region of non-TTAGGG DNA.42 Digested genomicDNA (3 mg), as well as molecular markers (32P ladder1 kb and HMW, Life Technologies) were resolved byelectrophoresis for 1400 V h in 0.7% agarose gel. TheTRFs were detected by hybridization to a 32P-(TTAGGG)4

probe as described.42 The signals on hybridized gelswere quantified using a phosphorimager (Biorad) andanalyzed by a computer-assisted system43,53 (n¼ 3 foreach determination). The hybridized signals were nor-malized to have equalized signal intensities. The highpurity of myoblasts in the cultures allows attributing thetelomere size to this population.

In vivo implantation. To evaluate their ability to takepart in muscle formation in vivo, the myoblasts wereinjected into the tibialis anterior muscles of immunodefi-cient Rag mice (kindly provided by Dr James Di Santo).46

The cell implantation model had been validated pre-viously.47 Aliquots of cells were thawed and grown inone single plate (600 cm2, Nunc) in proliferation med-ium. The myoblasts were then used at P6 and P7. On theday of injection, they were harvested, counted and analiquot was reserved for the quantification of CD56-positive cells by FACS. Cells were finally suspended inDMEM (Gibco-BRL) containing 0.5% BSA (Sigma) andkept on ice until the injection. Animals were anesthetizedusing ketamine (30 mg/kg) and xylazine (15 mg/kg).The tibialis anterior muscle was exposed by a small skinincision. Two protocols were followed (Table 3). The micereceived two series of cell injections at one week interval.2 to 5� 106 cells were injected at 8 to 10 sites in a volumeof 20 ml using a glass capillary.

The mice were killed 4 weeks after the last series ofcell injections, which is a delay compatible with completemouse muscle regeneration and the expression ofdonor markers.16,47–49 The muscles were snap frozen innitrogen-cooled isopentan and serially sectioned (8 mm)using a cryostat. The participation of human cells inmuscle formation was detected by the specific expressionof human dystrophin beneath the basement membraneof muscle fibers. The Mouse on Mouse kit (Vector) wasused. Sections were incubated with the anti-dystrophinantibody (NCL-Dys3, 1/20, 30 min at room tempera-ture, Novocastra), and then with a goat anti-mouse IgG(H+L) (Alexa Fluor 568, 1/1000, 30 min at roomtemperature, Molecular Probes). Slides were mountedunder Mowiol (Calbiochem) and observed using a Zeissfluorescence microscope. Pictures were performed usingthe Photometric digital camera and used for thequantification of the number of dystrophin-positivefibers. Statistical analysis was performed using thepaired t-test.

Acknowledgements

This work was supported by a grant from the ‘Associa-tion Francaise contre les Myopathies’ (AFM). This studyhas been promoted by the CHU de Nice. The productionof control myoblasts was supported by both AFM andthe ‘Assistance Publique-Hopitaux de Paris’ (France). Wethank Dr James Di Santo (Institut Pasteur, Paris) for thekind gift of the immunodeficient Rag mice, Dr Marc

Jeanpierre for FSHD molecular diagnosis, Dr VincentMouly for telomere evaluation, Drs Gillian Buttler-Browne and Marc Fiszman for critical review of themanuscript, Drs Philippe Chaumet-Riffaud and SolangeSolbes-Latourette for helpful assistance at setting thispreclinical protocol and Miss Jessica Shama for helpfulrevision of the document. We thank the company MyosixSA and Mr Frederic Chereau for giving access to themyoblast culture methodology. Finally, we would alsolike to thank all the patients for their participation tothis study.

References

1 Lunt P. Facioscapulohumeral muscular dystrophy: diagnosticand molecular aspects. In: Deymeer F (ed). NeuromuscularDiseases: From Basic Mechanisms to Clinical Management. Mono-graphs in Clinical Neuroscience, Vol. 18. Karger: Basel, Freiburg,Paris, London, New York, New Delhi, Bangkok, Singapore,Tokyo, Sydney, 2000: 44–60.

2 Wijmenga C et al. Chromosome 4q DNA rearrangementsassociated with facioscapulohumeral muscular dystrophy. NatGenet 1992; 2: 26–33.

3 Hewitt JE et al. Analysis of the tandem repeat locus D4Z4associated with facioscapulohumeral muscular dystrophy. HumMol Genet 1994; 3: 1287–1295.

4 Lunt PW et al. Correlation between fragment size at D4F104S1and age at onset or at wheelchair use, with a possiblegenerational effect, accounts for much phenotypic variation in4q35-facioscapulohumeral muscular dystrophy (FSHD). HumMol Genet 1995; 4: 951–958.

5 Tawil R et al. Evidence for anticipation and association ofdeletion size with severity in facioscapulohumeral musculardystrophy. Ann Neurol 1996; 39: 744–748.

6 Gabellini D, Green MR, Tupler R. Inappropriate gene activationin FSHD: a repressor complex binds a chromosomal repeatdeleted in dystrophic muscle. Cell 2002; 110: 339–348.

7 Winokur ST et al. Expression profiling of FSHD muscle supportsa defect in specific stages of myogenic differentiation. Hum MolGen 2003; 12: 2895–2907.

8 van der Kooi EL et al. Strength training and albuterol infacioscapulohumeral muscular dystrophy. Neurology 2004; 63:702–708.

9 Skuk D, Roy B, Goulet M, Tremblay JP. Successful myoblasttransplantation in primates depends on appropriate cell deliveryand induction of regeneration in the host muscle. Exp Neurol1999; 155: 22–30.

10 Skuk D, Goulet M, Roy B, Tremblay JP. Efficacy of myoblasttransplantation in nonhuman primates following simple intra-muscular cell injections: toward defining strategies applicable tohumans. Exp Neurol 2002; 175: 112–126.

11 Blau HM, Webster C, Pavlath GK. Defective myoblasts identifiedin Duchenne muscular dystrophy. Proc Natl Acad Sci USA 1983;80: 4856–4860.

12 Webster C, Blau HM. Accelerated age-related decline inreplicative life-span of Duchenne muscular dystrophy myo-blasts: implications for cell and gene therapy. Somat Cell MolGenet 1990; 16: 557–565.

13 Wright WE. Myoblast senescence in muscular dystrophy. ExpCell Res 1985; 157: 343–354.

14 Irintchev A et al. Functional improvement of damaged adultmouse muscle by implantation of primary myoblasts. J Physiol1997; 500: 775–785.

15 Arcila ME et al. Mass and functional capacity of regeneratingmuscle is enhanced by myoblast transfer. J Neurobiol 1997; 33:185–198.


1661

Gene Therapy

16 Huard J et al. High efficiency of muscle regeneration after humanmyoblast clone transplantation in SCID mice. J Clin Invest 1994;93: 586–599.

17 Kinoshita I et al. Very efficient myoblast allotransplantation inmice under FK506 immunosuppression. Muscle Nerve 1994; 17:1407–1415.

18 Kinoshita I, Vilquin JT, Roy R, Tremblay JP. Successive injectionsin mdx mice of myoblasts grown with bFGF. Neuromuscul Disord1996; 6: 187–193.

19 Partridge TA et al. Conversion of mdx myofibres fromdystrophin-negative to dystrophin-positive by injection ofnormal myoblasts. Nature 1989; 337: 176–179.

20 Morgan JE, Pagel CN, Sherratt T, Partridge TA. Long-termpersistence and migration of myogenic cells injected into pre-irradiated muscles of mdx mice. J Neurol Sci 1993; 115: 191–200.

21 Gussoni E et al. Normal dystrophin transcripts detected inDuchenne muscular dystrophy patients after myoblast trans-plantation. Nature 1992; 356: 435–438.

22 Huard J et al. Human myoblast transplantation: preliminaryresults of 4 cases. Muscle Nerve 1992; 15: 550–560.

23 Karpati G et al. Myoblast transfer in Duchenne musculardystrophy. Ann Neurol 1993; 34: 8–17.

24 Law PK et al. Dystrophin production induced by myoblasttransfer therapy in Duchenne muscular dystrophy. Lancet 1990;336: 114–115.

25 Mendell JR et al. Myoblast transfer in the treatment ofDuchenne’s muscular dystrophy. N Engl J Med 1995; 333:832–838.

26 Miller RG et al. Myoblast implantation in Duchenne musculardystrophy: the San Francisco study. Muscle Nerve 1997; 20:469–478.

27 Morandi L et al. Lack of mRNA and dystrophin expression inDMD patients three months after myoblast transfer. NeuromusculDisord 1995; 5: 291–295.

28 Neumeyer AM et al. Pilot study of myoblast transfer in thetreatment of Becker muscular dystrophy. Neurology 1998; 51:589–592.

29 Tremblay JP et al. Results of a triple blind clinical study ofmyoblast transplantations without immunosuppressive treat-ment in young boys with Duchenne muscular dystrophy. CellTransplant 1993; 2: 99–112.

30 Skuk D, Goulet M, Roy B, Tremblay JP. Myoblast transplantationin whole muscle of non human primates. J Neuropathol ExpNeurol 2000; 59: 197–206.

31 Skuk D et al. Dystrophin expression in myofibers of Duchennemuscular dystrophy patients following intramuscular injectionsof normal myogenic cells. Mol Therapy 2004; 9: 475–482.

32 Hagege AA et al. Viability and differentiation of autologousskeletal myoblast grafts in ischaemic cardiomyopathy. Lancet2003; 361: 491–492.

33 Menasche P et al. First successful clinical myoblast transplanta-tion for heart failure. Lancet 2001; 357: 279–280.

34 Menasche P et al. Autologous skeletal myoblast transplantationfor severe postinfarction left ventricular dysfunction. J Am CollCardiol 2003; 41: 1078–1083.

35 Pagani FD et al. Autologous skeletal myoblasts transplanted toischemia-damaged myocardium in humans. J Am Coll Cardiol2003; 41: 879–888.

36 Pouly J et al. Does the functional efficacy of skeletal myoblasttransplantation extend to nonischemic cardiomyopathy? Circula-tion 2004; 110: 1626–1631.

37 Vignos PJ, Spencer GE, Archibald KC. Management of progres-sive muscular dystrophy in childhood. JAMA 1963; 184: 89–96.

38 Illa I, Leon-Monzon M, Dalakas MC. Regenerating anddenervated human muscle fibers and satellite cells expressneural cell adhesion molecule recognized by monoclonalantibodies to natural killer cells. Ann Neurol 1992; 31: 46–52.

39 Lanier LL, Testi R, Bindl J, Phillips JH. Identity of Leu-19 (CD56)leukocyte differentiation antigen and neural cell adhesionmolecule. J Exp Med 1989; 169: 2233–2238.

40 Webster C et al. Isolation of human myoblasts with thefluorescence-activated cell sorter. Exp Cell Res 1988; 174: 252–265.

41 Li ZL, Lilienbaum A, Butler-Browne G, Paulin D. Humandesmin-coding gene: complete nucleotide sequence, character-ization and regulation of expression during myogenesis anddevelopment. Gene 1989; 78: 243–254.

42 Allsopp RC et al. Telomere length predicts replicative capacity ofhuman fibroblasts. Proc Natl Acad Sci USA 1992; 89: 10114–10118.

43 Decary S et al. Replicative potential and telomere length inhuman skeletal muscle: implications for satellite cell-mediatedgene therapy. Hum Gene Ther 1997; 8: 1429–1438.

44 Decary S et al. Shorter telomeres in dystrophic muscle consistentwith extensive regeneration in young children. NeuromusculDisord 2000; 10: 113–120.

45 van der Maarel SM et al. De novo facioscapulohumeral musculardystrophy: frequent somatic mosaicism, sex-dependent pheno-type, and the role of mitotic transchromosomal repeat inter-action between chromosomes 4 and 10. Am J Hum Genet 2000; 66:26–35.

46 Goldman JP et al. Enhanced human cell engraftment in micedeficient in RAG2 and the common cytokine receptor gammachain. Br J Haematol 1998; 103: 335–342.

47 Cooper RN et al. A new immunodeficient mouse model forhuman myoblast transplantation. Hum Gene Therapy 2001; 12:823–831.

48 Skuk D et al. Transplantation of human myoblasts in SCID miceas a potential muscular model for myotonic dystrophy.J Neuropathol Exp Neurol 1999; 58: 921–931.

49 Vilquin JT et al. Myoblast transplantations lead to the expressionof the laminin alpha 2 chain in normal and dystrophic (dy/dy)mouse muscles. Gene Therapy 1999; 6: 792–800.

50 Winokur ST et al. Facioscapulohumeral muscular dystrophy(FSHD) myoblasts demonstrate increased susceptibility tooxidative stress. Neuromuscul Disord 2003; 13: 322–333.

51 Blau HM et al. Differentiation properties of pure populations ofhuman dystrophic muscle cells. Exp Cell Res 1983; 144: 495–503.

52 Furling D et al. Defective satellite cells in congenital myotonicdystrophy. Hum Mol Genet 2001; 10: 2079–2087.

53 Renault V et al. Regenerative potential of human skeletal muscleduring aging. Aging Cell 2002; 1: 132–139.

54 Vandebrouck C et al. Normal calcium homeostasis in dystro-phin-expressing facioscapulohumeral muscular dystrophy myo-tubes. Neuromuscul Disord 2002; 12: 266–272.

55 Hoffman EP, Brown Jr RH, Kunkel LM. Dystrophin: the proteinproduct of the Duchenne muscular dystrophy locus. Cell 1987;51: 919–928.

56 Xu H et al. Defective muscle basement membrane and lack ofM-laminin in the dystrophic dy/dy mouse. Proc Natl Acad SciUSA 1994; 91: 5572–5576.

57 Xu H, Wu XR, Wewer UM, Engvall E. Murine musculardystrophy caused by a mutation in the laminin alpha 2 (Lama2)gene. Nat Genet 1994; 8: 297–302.

58 Richard I et al. Loss of calpain 3 proteolytic activity leads tomuscular dystrophy and to apoptosis-associated IkappaBalpha/nuclear factor kappaB pathway perturbation in mice. J Cell Biol2000; 151: 1583–1590.


1662

Gene Therapy

Gene Therapy Progress and Prospects: Fetal gene therapy – first proofs of concept – some adverse effects

Documents