Repeated intrathecal injections of plasmid DNA encoding interleukin-10 produce prolonged reversal of …

www.elsevier.com/locate/pain

Pain 126 (2006) 294–308

Repeated intrathecal injections of plasmid DNA encodinginterleukin-10 produce prolonged reversal of neuropathic pain

Erin D. Milligan a,*, Evan M. Sloane a, Stephen J. Langer b, Travis S. Hughes b,Brian M. Jekich a, Matthew G. Frank a, John H. Mahoney a, Lindsay H. Levkoff a,

Steven F. Maier a, Pedro E. Cruz c, Terence R. Flotte c, Kirk W. Johnson d,Melissa M. Mahoney e, Raymond A. Chavez d, Leslie A. Leinwand b, Linda R. Watkins a

a Department of Psychology and the Center for Neuroscience, University of CO at Boulder, Boulder, CO 80309, USAb Department of Molecular, Cellular and Developmental Biology, University of CO at Boulder, Boulder, CO 80309, USA

c Genetics Institute, The Powell Gene Therapy Center and Department of Pediatrics, University of FL at Gainesville, Gainesville, FL 32610, USAd Avigen, Inc., Alameda, CA 94502, USA

e Department of Chemical and Biological Engineering, University of CO at Boulder, Boulder, CO 80309, USA

Received 16 January 2006; received in revised form 5 July 2006; accepted 17 July 2006

Abstract

Neuropathic pain is a major clinical problem unresolved by available therapeutics. Spinal cord glia play a pivotal role inneuropathic pain, via the release of proinflammatory cytokines. Anti-inflammatory cytokines, like interleukin-10 (IL-10), suppressproinflammatory cytokines. Thus, IL-10 may provide a means for controlling glial amplification of pain. We recently documentedthat intrathecal IL-10 protein resolves neuropathic pain, albeit briefly (�2–3 h), given its short half-life. Intrathecal gene therapyusing viruses encoding IL-10 can also resolve neuropathic pain, but for only �2 weeks. Here, we report a novel approach that dra-matically increases the efficacy of intrathecal IL-10 gene therapy. Repeated intrathecal delivery of plasmid DNA vectors encodingIL-10 (pDNA-IL-10) abolished neuropathic pain for greater than 40 days. Naked pDNA-IL-10 reversed chronic constriction injury(CCI)-induced allodynia both shortly after nerve injury as well as 2 months later. This supports that spinal proinflammatory cyto-kines are important in both the initiation and maintenance of neuropathic pain. Importantly, pDNA-IL-10 gene therapy reversedmechanical allodynia induced by CCI, returning rats to normal pain responsiveness, without additional analgesia. Together, thesedata suggest that intrathecal IL-10 gene therapy may provide a novel approach for prolonged clinical pain control.� 2006 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.

Keywords: Glia; Rats; Neuropathic pain; Naked DNA; Gene therapy

1. Introduction

Neuropathic pain is a debilitating condition, arisingfrom trauma and inflammation of peripheral nerves.Despite decades of research, neuropathic pain remains

0304-3959/$32.00 � 2006 International Association for the Study of Pain. P

doi:10.1016/j.pain.2006.07.009

* Corresponding author. Tel.: +1 303 735 2295; fax: +1 303 4922967.

E-mail address: [email protected] (E.D. Milligan).

a major clinical problem, unresolved by currently avail-able therapeutics (Watkins et al., 2001; Watkins andMaier, 2003). Because the available drugs were devel-oped to target changes in neuronal function that havebeen documented to occur in neuropathic pain, this rais-es the question of whether a unique non-neuronalapproach to neuropathic pain control may providegreater clinical success.

One such approach may arise from the recent recog-nition that spinal cord glia (microglia and astrocytes)

ublished by Elsevier B.V. All rights reserved.

E.D. Milligan et al. / Pain 126 (2006) 294–308 295

are critically involved in the creation and maintenanceof neuropathic pain (Watkins et al., 2001; McMahonet al., 2005; Tsuda et al., 2005). Peripheral nerve injuryleads to the spinal release of neuronally derived signalsthat, in turn, trigger glial activation. Once activated, gliacontribute to the amplification of pain via the produc-tion and release of neuroexcitatory substances. Keyamongst these glial products are the proinflammatorycytokines, tumor necrosis factor (TNF), interleukin-1(IL-1) and interleukin-6 (IL-6). Each has been implicat-ed as an important contributor to neuropathic pain(Watkins et al., 2001). This suggests that the reductionof proinflammatory products released from activatedglial cells may provide new approach for controllingneuropathic pain.

Glial inhibitors and selective TNF, IL-1, and IL-6antagonists have each successfully prevented and/orreversed various animal models of neuropathic painupon acute administration. However, none are appro-priate for clinical use (Watkins and Maier, 2003, forreview). While the glial metabolic inhibitor fluoroci-trate is centrally active after systemic administration,it is inappropriate for clinical applications as sustainedglial inhibition by fluorocitrate suppresses glial uptakeof excitatory amino acids, resulting in seizures. Themicroglial inhibitor, minocycline, is also centrallyactive after systemic administration but it cannotreverse established neuropathic pain. Selective proin-flammatory cytokine antagonists (e.g., IL-1 receptorantagonist, TNF soluble receptors, IL-6 neutralizingantibody) would be challenging to use clinically, aseach requires chronic intrathecal administration. Noneare blood–brain barrier permeable and none exertprolonged effects following bolus injection into theintrathecal space. Also, prolonged resolution of chronicpain by administration of one proinflammatorycytokine antagonist is unlikely to succeed as otherproinflammatory cytokines act coordinately in theseconditions.

An alternative approach for achieving sustained sup-pression of glial amplification of pain is via intrathecalinterleukin-10 protein (IL-10). IL-10 has been docu-mented to suppress the production and function of allproinflammatory cytokines (Moore et al., 2001). Inaddition, evidence to date suggests that spinal cord neu-rons do not express IL-10 receptors (Ledeboer et al.,2003), thus avoiding disruption of neuronal functionby the presence of IL-10. While a bolus intrathecal injec-tion of IL-10 protein and intrathecal delivery of viralvectors encoding IL-10 can each transiently reverseneuropathic pain, neither approach has been found toproduce sufficiently sustained pain reversal to be clini-cally relevant (Milligan et al., 2005b). Here, we describethe development of a novel intrathecal non-viral genetherapy which produces prolonged reversal of neuro-pathic pain.

2. General materials and methods

2.1. Animals

Pathogen-free adult male Sprague–Dawley rats were usedin all experiments. Rats (350–375 g at the time of arrival;Harlan Labs, Madison, WI) were housed in temperature(23 +/� 3 �C) and light (12:12 light:dark; lights on at 0700 h)controlled rooms with standard rodent chow and water avail-able ad libitum. Behavioral testing was performed during thefirst 6 h of the light cycle. All procedures were approved bythe Institutional Animal Care and Use Committee at the Uni-versity of Colorado at Boulder.

2.2. Plasmid vectors

The plasmid originally used in our prior studies as theexpression cassette for adeno-associated virus-2 (AAV-2)vector encoding rat IL-10 (Milligan et al., 2005b) was injectedas naked DNA in these studies (pDNA-rIL-10). This plas-mid’s transcriptional cassette is flanked by two AAV-2 viralelements, inverted terminal repeat sequences (ITRs), andconsists of the cytomegalovirus enhancer/chicken b-actin pro-moter (CB-actin), an intronic region, the rat interleukin-10(IL-10) gene with a point mutation (F129S) and the SV40polyadenylation signal region (Fig. 1A) (Milligan et al.,2005b). The expression cassette encoding human IL-10(pDNA-hIL-10) (Fig. 1B) was similar to that for the ratexpression cassette, with the exception that the ITR sequenc-es were modified by deletion of 17 bp and the CMV promoterdirected the expression of human IL-10 (only Experiment 3).The control plasmids are analogous plasmid cassettes inwhich the CMV enhancer/CB-actin promoter directs theexpression of either the unmodified (Zolotukhin et al.,1996) or the enhanced reporter gene, Jellyfish green fluores-cent protein (GFP), and are noted as pDNA-GFP(Fig. 1C) (Experiments 1 and 2) and pDNA-eGFP(Fig. 1D) (Experiment 3), respectively.

Plasmids pDNA-rIL-10, pDNA-GFP and pDNA-eGFPwere amplified in SURE2 cells (Stratagene, La Jolla, CA) asthe ITR components in the plasmid may be important for geneexpression (Flotte et al., 1993; Philip et al., 1994; Xin et al.,2003; Chikhlikar et al., 2004) but are unstable and frequentlydeleted in conventional E. coli strains (i.e., DH5a) ((Zhiet al., 2004), and unpublished observations). In early studies(Experiments 1 and 2), plasmids were purified by cesium chlo-ride gradient centrifugation methods similar to proceduresdescribed previously (Sambrook et al., 1989). However, thismethod has been documented to co-purify plasmid DNA withendotoxin released from bacterial cells during DNA isolationprocedures (Cotten et al., 1994; Wicks et al., 1995). Thus, forall subsequent experiments (Experiments 3–5), plasmids wereamplified in SURE2 cells with the exception of pDNA-hIL-10 that contains modified ITRs engineered to be resistant todeletions during amplification and was used solely for trans-gene detection; amplified in DH10B bacteria (Invitrogen,Carlsbad, CA) and purified using endotoxin-free Qiagen plas-mid Giga purification kit (Valencia, CA, USA) according tothe manufacturer’s instructions. To screen for possible ITRdeletions and/or other rearrangements, the integrity ofpDNA-rIL-10, pDNA-hIL-10, pDNA-GFP and pDNA-eGFP

pAITR

rIL-10

pDNA-rIL-10AmprIVS

CB pro

5.9 Kb

CMV enh

ITR

rIL-10

IVS

CMV pro

pDNA-hIL-109.6 Kb

Ampr

ITR

ITR

pA

hIL-10

pA

GFP

ITR

ITR

IVS

CB pro

CMV enh

Ampr

pDNA-GFP7.2 Kb

TK pro

neor

ITR

pA

eGFP

ITR

pDNA-eGFPAmprIVS

CB pro

6.1 Kb

CMV enh

ITR

A B

C D

Fig. 1. Plasmid constructs. Panel A: plasmid DNA encoding rat IL-10 that was injected to deliver the rat IL10 gene to examine gene expression or toproduce behavioral pain reversal in all experiments. Panel B: plasmid DNA encoding human IL-10 (pDNA-hIL-10) that was used to examine geneexpression in CSF by protein analysis, as part of the third experiment. Panel C: plasmid DNA encoding human modified Jellyfish green fluorescentprotein (pDNA-GFP) that was used for control injections in pain behavioral assessments as part of the first two experiments. Panel D: plasmid DNAencoding enhanced Jellyfish green fluorescent protein (pDNA-eGFP) that was used for the DNA-treated controls for gene expression in the thirdexperiment. Key: Ampr, ampicillin resistance marker; ITR, AAV inverted terminal repeat; CMV enh, cytomegalovirus immediate early promoterenhancer; CMV pro, cytomegalovirus immediate early promoter; CB pro, chicken b-actin promoter; IVS, intervening sequence (intron); rIL-10, ratinterleukin-10 gene; hIL-10, human interlukin-10 gene; GFP, green fluorescent protein gene; eGFP, enhanced green fluorescent protein gene; pA,polyadenylation signal; TK pro, herpes simplex thymidine kinase promoter; neor, neomycin resistance marker.

296 E.D. Milligan et al. / Pain 126 (2006) 294–308

plasmids was verified by restriction site analysis. Endotoxinlevels (Experiments 3–5) were determined by the photometricLimulus amebocyte lysate (LAL) assay according to the man-ufacturer’s instructions (BioWhittaker Inc., Walkersville, MD,USA). Random sampling of Qiagen-purified plasmids used inthese studies revealed endotoxin levels ranging from undetect-able levels to 0.038 EU/lg DNA, a level considered negligible(Qiagen handbook, 2006).

All plasmid constructs were suspended in sterile Dulbecco’sphosphate-buffered saline (DPBS, 1·, 0.1 lm pore-filtered, pH7.2, cat#14190-144; Gibco, Invitrogen Corp, Grand Island,NY) with 3% sucrose (DPBS-3%). The DPBS-3% vehiclewas prepared using molecular biology grade D (+)-sucrose(b-D-fructofuranosyl-a-D-glucopyranoside; Sigma–Aldrich) inDPBS, 0.2 lm sterile filtered (pyrogen-free syringe filter

unit, cat # 25AS020AS, Life Science Products, Inc., CO) andstored in sterile, 50 ml conical tubes at 4 �C until the time ofuse.

After isolation, cesium chloride-purified plasmids weredialyzed for 1 h against DPBS (1·) and twice each for 2 hagainst sterile DPBS-3% sucrose. Qiagen-purified plasmidswere resuspended in sterile DPBS-3% sucrose. Dialyzed plas-mid preparations were stored as 300 ll aliquots at �20 �C.The concentration of the plasmids was determined by260 nm adsorption, and they were stored at a concentrationof no less than 5 lg/ll. The purity of the plasmid DNAwas determined by 260:280 nm adsorption DNA:RNA ratio,and ranged between 1.95 and 2.0. Adsorption at 320 nm wasexamined for the presence of other impurities and was alwaysbelow detection. Thus, these DNA preparations contained

E.D. Milligan et al. / Pain 126 (2006) 294–308 297

virtually no remnant contaminants from purificationprocedures.

2.3. In vitro verification of rat or human IL-10 protein

production by the plasmid vectors

To ensure that the human and rat IL-10 encoding plasmidconstructs produce their respective IL-10 protein, transienttransfection in human embryonic kidney 293 (HEK293) cellswas performed. HEK293 cells were grown in Dulbecco’s mod-ified Eagle’s medium (Gibco-BRL) supplemented with 5% fetalcalf serum (Hyclone), 2 mM L-glutamine, 100 U/ml penicillinG sodium and 100 U/ml-streptomycin sulfate, in a 37 �Chumidified incubator at 5% CO2. To examine expression ofrat IL-10, HEK293 cells were transfected with 5 mg pDNA-rIL-10 or plasmid DNA lacking the IL-10 gene using a calciumphosphate transfection method (Sambrook et al., 1989).Transfection with pDNA-hIL-10 was carried out in HEK293cells using Fugene (Roche Applied Science, Indianapolis, IN)with 1 lg plasmid DNA following the manufacturer’s proto-col. Supernatants from either transfection were collected24–48 h post-transfection, and were analyzed for the presenceof IL-10 protein by enzyme-linked immunosorbent assays(ELISAs) specific for either human or rat IL-10 (R&DSystems, Minneapolis, MN) following the manufacturer’s pro-tocols. All plasmid constructs retained their necessary elementsand revealed no gross rearrangements as determined by restric-tion site analysis. Supernatants from transfected cell culturesexpressed high levels of rat or human IL-10 (30 ng/ml and5 lg/ml, respectively, data not shown).

2.4. Tissue preparation for microscopic examination

In preparation to examine GFP expression in meninges,rats were deeply anesthetized with sodium pentobarbital priorto transcardial perfusion with ice-cold 0.9% heparinized salinefor 3 min followed by fresh ice-cold 2% paraformaldehyde in0.1 M PB for 5 min. The spinal cord with intact meningeswas exposed by laminectomy, carefully removed from the ver-tebral column and placed in a shallow bath of chilled PBS. Asuperficial incision was made along the ventral midline thatonly penetrated the dura mater while leaving the underlyingspinal cord tissue intact. Following a 5-min incubation inice-chilled PBS, the spinal cord was discarded after carefulseparation from the meninges surrounding the injection siteand within �10 mm caudally, which includes the regionoverlaying the cauda equina referred to as lumbar meninges.The resultant 10-mm long meningial sheet was floated inice-chilled PBS to minimize distortion of its shape. Sheets offloating meninges were stored at 4 �C in 0.1 M PB plus 0.1%sodium azide in the dark until microscopic analysis. Meningealsamples were placed on microscope slides. A drop of Vecta-shield mounting medium containing 4 0,6 diamidino-2-phenyl-indole (DAPI; 1.5 lg/ml, Vector Labs, UK), which isfluorescent upon association with DNA in cell-body nuclei,was placed on the tissue atop each slide. The DAPI provideda reference for the number of cells in the field present andthe proximity of eGFP to DAPI expression could be observed.A #1 coverslip was placed on top of the specimen prior toviewing.

Confocal microscopy was used to spatially localize greenfluorescent-protein positive cells in tissue specimens using aZeiss Pascal LSM microscope. Conventional microscopy wasused to localize DAPI-stained nuclei with green fluorescent-positive cells using an upright Zeiss Axioskop microscope.Confocal images were acquired with 40· Plan NeoFluor(1.3) oil immersion objective. GFP was excited with the 488line of an argon laser and a 520 long pass emission filter. Con-ventional microscopy images were acquired with a 40· PlanNeoFluor (1.3) oil immersion objective. An HBO mercurybulb and filter sets (395/450 Ex/Em DAPI) and (484/525Ex/Em GFP) were used to acquire conventional microscopyDAPI and GFP fluorescence signals from the specimen, asconfocal microscopy filters specific to image DAPI nuclei werenot available.

2.5. Surgery and microinjections

Chronic constriction injury (CCI) surgery was performedunder isoflurane anesthesia (1.5–2.0% vol in oxygen) by looselytying four chromic gut sutures around the sciatic nerve in theleft hindleg as previously described (Bennett and Xie, 1988).The sciatic nerves of sham-operated rats were identicallyexposed but not ligated.

The route of drug delivery for all experiments was i.t. Anacute catheter application method under brief isoflurane anes-thesia (5.0% vol in oxygen) was employed, as described previ-ously (Milligan et al., 2005b), to inject compounds at the levelof the lumbosacral enlargement. No abnormal motor behav-iors were observed following any injection. Injections consistedof 100 lg pDNA in 17–20 ll or equivolume vehicle. Therewere a total of four injections (100 lg pDNA-rIL-10 orpDNA-GFP per injection) during Experiment 1 and a totalof two injections (100 lg pDNA-rIL-10, pDNA-hIL-10,pDNA-eGFP or equivolume vehicle per injection) duringExperiments 2–5 or a single injection (100 lg pDNA-hIL-10)in part of Experiment 3.

2.6. Behavioral measures

Behavioral testing was performed within the sciatic andsaphenous innervation area of the hind paws as previouslydescribed. Briefly, calibrated Semmes–Weinstein monofila-ment fibers (von Frey hairs; Stoelting, Wood Dale, IL) wereapplied randomly to the left and right hind paws to elicitpaw withdrawal responses (Milligan et al., 2000; Milliganet al., 2001). The range of monofilaments used in theseexperiments was 0.407–15.136 g bending force. Assessmentswere made prior to (baseline; BL) and Days 3 and 10 afterCCI surgery, and at indicated times after i.t. injections (seerespective figure captions). Behavioral testing was performedblind with respect to drug administration. The behavioralresponse pattern was used to calculate the 50% paw with-drawal threshold (absolute threshold), by fitting a Gaussianintegral psychometric function using a maximum-likelihoodfitting method (Harvey, 1986), as described in detail previ-ously (Milligan et al., 2000; Milligan et al., 2001). Thisfitting method allows parametric statistical analyses. Dataare presented as both the 50% paw withdrawal threshold(g) and the log10 transformation of that value as log10

(milligrams · 10).

298 E.D. Milligan et al. / Pain 126 (2006) 294–308

2.7. Serum, cerebrospinal fluid (CSF) and spinal cord tissue

collection

Five days following the initial plasmid injection, plasmid-injected rats and naı̈ve controls were overdosed with sodiumpentobarbital (Abbot Laboratories, North Chicago, IL). Cer-vical cerebrospinal fluid (CSF), lumbosacral CSF, lumbosacralspinal cord and/or cauda equina spinal cord tissue regionswere collected prior to sacrifice as previously described(Milligan et al., 2000; Milligan et al., 2001). CSF samples werecollected either 3 days after an initial pDNA injection or, in aseparate group of animals, 2 days after the second pDNAinjection. These samples were flash-frozen in liquid nitrogenand stored at �80 �C until analysis by ELISA or by quantita-tive (real-time) reverse transcriptase polymerase chain reaction(RT-PCR) to detect human or rat IL-10 protein and mRNA,respectively. Immediately following CSF collection, wholeblood was collected by cardiac puncture and allowed to clotat 4 �C. Sera were collected and stored at �20 �C until the timeof assay. Serum IL-10 protein ELISA analysis was conductedaccording to the manufacturer’s instructions. For CSF sam-ples, ELISA methods for small volume samples were used(O’Connor et al., 2004).

2.8. Quantitative RT-PCR

2.8.1. RNA isolation and enrichment

Total RNA was isolated from lumbosacral dorsal spinalcord and cauda equina tissue similarly to methods previouslydescribed (Chomczynski and Sacchi, 1987). Briefly, tissueswere homogenized in 1 ml TRIzol reagent (Invitrogen,Carlsbad, CA) according to the manufacturer’s instructions.Samples were DNase-treated (DNA-free kit, Ambion, Austin,TX) to remove contaminating DNA from total nucleic acidfollowed by spectrophotometric determination of nucleic acidquantity and purity.

2.8.2. cDNA synthesis

Total RNA was reverse transcribed into cDNA using theSuperScript II First Strand Synthesis System for RT-PCRaccording to the manufacturer’s specification (Invitrogen,Carlsbad, CA). Control reactions lacking either reverse trans-criptase or RNA template were included to assess carryover ofgenomic DNA and contamination of RNA, respectively. Theno reverse transcriptase control consisted of pooled RNAequally representing all experimental samples. cDNA wasdiluted twofold in DNase-free water and stored at �20 �C.

2.8.3. Primer specifications

cDNA sequences for rat IL-10 (Accession #: NM_012854),plasmid-derived rat IL-10 and an internal control, rat glyceral-dehyde-3-phosphate dehydrogenase (GAPDH; Accession #:M17701) were obtained from GenBank at the National Centerfor Biotechnology Information (NCBI; www.ncbi.nlm.nih.gov). Plasmid-derived IL-10 lacks the 3 0 UTR sequence ofendogenous rat IL-10. Instead a 182 bp sequence of the plas-mid backbone serves as a surrogate 3 0 UTR of plasmid-derivedIL-10. The reverse primers for rat and plasmid-derived IL-10were designed to these unique 3 0 UTR sequences to differenti-ate IL-10 gene expression driven by endogenous rat IL-10 orplasmid IL-10. Primer sequences for rat IL-10 (forward:

5 0-TAAGGGTTACTTGGGTTGCC-3 0; reverse: 5 0-TATCCAGAGGGTCTTCAGC-3), plasmid-derived rat IL-10 (forward:5 0-CAGTGGAGCAGGTGAAGA-3 0; reverse: 5 0-CCGCCAGTGTGATGGATA -3 0) and GAPDH (forward: 5 0-GTTTGTGATGGGTGTGAACC-3 0; reverse: 5 0-TCTTCTGAGTGGCAGTGATG-3 0) were designed using the Qiagen OligoAnalysis & Plotting Tool (oligos.qiagen.com/oligos/tool-kit.php?) and tested for sequence specificity using the BasicLocal Alignment Search Tool (Altschul et al., 1997) at NCBI.Primer specificity was further verified by melt curve analysis(see Section 2.8.4). All primers were obtained from Proligo(Boulder, CO).

2.8.4. Quantitative real-time PCR

PCR amplification of cDNA was performed using theQuantitect SYBR Green PCR Kit (Qiagen, Valencia, CA).cDNA (1 ll) was added to a reaction master mix (25 ll) con-taining 2.5 mM MgCl2, HotStart Taq DNA polymerase,SYBR Green I, dNTPs, fluorescein (10 nM) and gene-specificprimers (500 nM each of forward and reverse primer). Tripli-cate reactions were conducted for each experimental sample.PCR cycling conditions consisted of a hot-start activation ofHotStart. Taq DNA polymerase (95 �C, 15 min) and 40 cyclesof denaturation (95 �C, 15 s), annealing (58 �C, 30 s), andextension (72 �C, 30 s).

2.8.5. Real-time detection and quantitation of PCR product

Formation of PCR product was monitored in real timeusing the MyiQ Single-Color Real-Time PCR Detection Sys-tem (Bio-Rad). Threshold for detection of PCR product fellwithin the log-linear phase of amplification for each reaction.Threshold cycle (CT; number of cycles to reach threshold ofdetection) was determined for each reaction. Relative geneexpression was determined using the 2�DDCt method (Pfaffl,2001).

2.9. Data analysis

All statistical comparisons were computed using Statview5.0.1 for the Macintosh. Baseline measures for the von Freywere analyzed by one-way ANOVA. Timecourse measuresfor behavioral assessments were analyzed by repeated mea-sures ANOVAs followed by Fisher’s protected least significantdifference post hoc comparisons, where appropriate. Cervicaland lumbosacral CSF IL-10 contents were analyzed by one-way ANOVA, as appropriate, followed by Fisher’s protectedleast significant difference post hoc comparisons. Statistical sig-nificance was determined at p < 0.05.

3. Results

3.1. Intrathecal injections of pDNA-rIL-10 during

concurrent allodynia transiently reverse allodynia

The dose of pDNA chosen for study was based onpreviously documented therapeutic effects of pDNAupon peripheral, intracerebroventricular or i.t. delivery(Schwartz et al., 1996; Daheshia et al., 1997). Thepreviously reported doses of pDNA ranged from 50 to150 lg (Schwartz et al., 1996; Daheshia et al., 1997;

E.D. Milligan et al. / Pain 126 (2006) 294–308 299

Meuli-Simmen et al., 1999). Based on pilot studies, weestablished that 100 lg pDNA-rIL10 in 20 ll was effica-cious. Rats were assessed for responses to the von Freytest at BL and on Days 3 and 10 after CCI or sham sur-gery (n = 5–6/group). CCI produced bilateral mechani-cal allodynia in agreement with prior reports (Milliganet al., 2005b). On Day 10 after CCI surgery followingbehavioral assessments, pDNA-rIL-10 or pDNA-GFP(each 100 lg in 20 ll) was injected i.t. Thresholdresponses were reassessed as indicated (Fig. 2 caption)to define the magnitude and the duration of pDNA-rIL-10-induced effects. A second, third and fourth i.t.injection of pDNA-rIL-10 or pDNA-GFP (each100 lg in 20 ll) was given after the anti-allodynic effectsof pDNA-rIL-10 had resolved. Injections were given ondays indicated (Fig. 2 caption). The fourth injection ofpDNA-rIL-10 or pDNA-GFP was given to examine ifpDNA-rIL-10 remained efficacious after allodynia wasstable for one week. Each i.t. pDNA-rIL-10 injectionreversed bilateral allodynia, albeit transiently, with each

A

B

Fig. 2. Single i.t. injections of pDNA-rIL-10 transiently reverse allodynia. Tindicated by black arrows. Sham-operated (Panel A) and CCI-operated (Pan(each 100 lg in 20 ll; n =3–5/group). Each group received a total of four idendays after surgery. Open symbols, responses of the hindpaw ipsilateral to thethe surgery site.

reversal period lasting longer than the previous. In addi-tion, after rats had remained allodynic for one week,pDNA-rIL-10 revealed no loss in effect after a fourthi.t. injection (Fig. 2). All rats appeared healthy, gainedweight normally, and exhibited typical posture, groom-ing, and locomotion.

Statistical analyses support these observations.Threshold responses at BL were similar between allgroups for both the ipsilateral and contralateral hind-paws (F1,19 = 3.118E-7, p > 0.99; F1,19 = 0.489,p > 0.49, respectively). The von Frey test on Days 3and 10 after CCI revealed a dramatic bilateral decreasein mechanical response thresholds (Panel B) comparedto sham operated controls (Panel A), (F1,19 =1037.348.422, p < 0.0001; F1,19 = 491.422, p < 0.0001,for both hindpaws). Immediately after behavioraltesting on Day 10, rats received i.t. pDNA. pDNA-rIL-10 transiently reversed bilateral allodynia inducedby CCI (F1,19 = 349.821, p < 0.0001; F1,19 = 471.019,p < 0.0001, for each hindpaw). On Day 15, rats received

he time of CCI or sham surgery and each i.t. injection of pDNA areel B) rats were injected i.t with pDNA-IL-10 or pDNA-GFP (control)tical injections of pDNA with behavioral testing continuing through 72surgery site; closed symbols, responses of the hindpaw contralateral to

300 E.D. Milligan et al. / Pain 126 (2006) 294–308

a second i.t. pDNA injection. Plasmid-DNA-rIL-10again transiently reversed bilateral allodynia inducedby CCI (F1,19 = 270.440, p < 0.0001; F1,19 = 495.968,p < 0.0001, for both hindpaws). After a third i.t. pDNAinjection given on Day 24 after CCI, plasmid-DNA-rIL-10 reversed allodynia (F1,19 = 237.980, p < 0.0001;F1,19 = 272.859, p < 0.0001, for both hind paws). Final-ly, allodynia was reversed after a fourth injection givenon Day 65, as data were analyzed between Days 65and 72 after surgery (F1,12 = 111.206, p < 0.0001;F1,19 = 221.812, p < 0.0001, for both hind paws).

Thus, naked pDNA encoding rat IL-10 produced areversal of CCI-induced neuropathic pain at increasing-ly longer therapeutic intervals with each subsequentpDNA injection. CCI-induced allodynia was reversedbilaterally for approximately 4 weeks after a third i.t.pDNA-rIL-10 injection (Fig. 2). This is the first reportthat has demonstrated a potential role of naked pDNAdelivered i.t. for the treatment of chronic pain.

3.2. Long-term reversal of CCI-induced allodynia by i.t.

pDNA-rIL-10

In the experiment above, mechanical allodynia wasfully re-expressed in pDNA-rIL-10-treated rats priorto each successive pDNA injection. This raised the ques-tion whether the duration of the anti-allodynic effects ofpDNA-rIL-10 could be improved if a second pDNAdose was delivered while mechanical allodynia was fullyresolved. Thus, we examined whether the intervalbetween the first and second pDNA-rIL-10 injectionscould affect the duration of reversal of mechanical allo-dynia. The dose of i.t. pDNA-rIL-10 chosen for studywas identical to that above (100 lg in �20 ll). PlasmidDNA-GFP (100 lg in �20 ll) was administered to thecontrol group. Rats were assessed at BL and on Days3 and 10 after CCI or sham surgery. On Day 10, ratsreceived i.t. pDNA-rIL-10 or pDNA-GFP. The vonFrey was again performed daily for the next 3 days toverify reversal from allodynia, as above. On Day 13after CCI, a second i.t. pDNA-rIL-10 or pDNA-GFPinjection was given while the anti-allodynic effects fromthe first pDNA-rIL-10 injection remained robust. This3-day injection interval of pDNA-rIL-10 produced astable, long-duration reversal of allodynia that remainedeffective for greater than 40 days (Fig. 3). Overall, allo-dynia remained stably reversed after the second i.t.pDNA-rIL-10 injection, compared to control pDNA-GFP. This is the first study to report long-lasting anti-al-lodynic effects caused by neuropathic pain after i.t. drugtreatment (Fig. 3).

A trend (although not statistically reliable) towarda return to BL responses in the pDNA-GFP grouplate in the timecourse (see Day 49, Fig. 3) wasobserved. This may be due to lessening of the effectof CCI over time.

Statistical analyses support these conclusions.Threshold responses at BL were similar between ipsilat-eral (Panel A) and contralateral (Panel B) hindpaws andtreatment groups (F7,40 = 0.568, p > 0.77). CCI againproduced a robust bilateral mechanical allodynia ashindpaw threshold responses decreased on Days 3 and10 after CCI (F1,40 = 403.438, p < 0.0001). An initialinjection of pDNA-rIL-10 reversed bilateral allodynia(F1,28 = 249.474, p < 0.0001). After a second injection,reversal from allodynia was observed (F1,36 = 441.846,p < 0.0001) and remained stable through Day 53.

3.3. pDNA-rIL-10 expression and localization

Based on the intriguing results from above, we soughtto address the mechanism whereby the plasmid DNAinjections were achieving their behavioral effects. Wemeasured IL-10 protein and mRNA gene expressionand determined by microscopy the site of recombinantprotein expression using GFP as a surrogate marker.IL-10 protein and mRNA gene expression was mea-sured from spinal cord tissue and/or CSF. Rats wereinjected i.t. with pDNA delivered as either one(100 lg) or two successive doses (100 lg per injection)spaced 3 days apart each in a volume of 18–20 ll. Eachrat received pDNA-rIL-10, pDNA-hIL-10 or pDNA-eGFP (n = 4–5 /group). One group of rats was sacrificed3 days after a single i.t. pDNA-rIL-10 injection forexamination of samples collected at a point where initialfull reversal of mechanical allodynia was observed(Experiments 1 and 2). The remaining groups of ratswere sacrificed for tissue collection 2 days after a secondpDNA injection. For comparison, tissue samples werecollected from a naı̈ve control group. For each group,blood, cervical and lumbosacral CSF, lumbar spinalcord and/or cauda equina (a 5 mm section beginning5 mm caudal to the lumbar spinal cord section) were col-lected at these timepoints to examine gene expression bydetecting human or rat IL-10 protein and/or mRNA.

The behavioral reversal observed in response to i.t.pDNA-rIL-10 in the preceding studies was concurrentwith robust gene expression in spinal cord CSF and tis-sue. Rat IL-10 mRNA expression was substantiallygreater in pDNA-rIL-10-treated rats compared to eithernaı̈ve or pDNA-eGFP-treated controls 2 days followinga second i.t. pDNA-rIL-10 injection. Additionally, plas-mid-derived rat IL-10 mRNA was greater in the caudaequina than in the lumbar spinal cord (Table 1).

While gene expression as measured by rat IL-10 orhuman IL-10 protein levels in serum was virtually unde-tectable two days following a second i.t. pDNA injection(Fig. 4A), large increases of both rat and human IL-10protein 2 days following a second i.t. injection inlumbosacral CSF were detected. The similarity of theCSF protein release profiles between rat and humanIL-10 indicates that human IL-10 protein, as other

A

B

Fig. 3. Long-term reversal of CCI-induced allodynia by two successive pDNA-IL-10 injections, using a 3-day interval between injections. Eachsham- or CCI-operated group received two i.t. injections of either pDNA-IL-10 or pDNA-GFP (each 100 lg in 20 ll; n =5–7/group). The time ofCCI or sham surgery and each i.t. injection are indicated by black arrows. Open symbols, control (pDNA-GFP) treatment; closed symbols, pDNA-IL-10. Responses of the hindpaw ipsilateral (Panel A) and contralateral (Panel B) to the site of surgery were analyzed through 57 days after surgery.

E.D. Milligan et al. / Pain 126 (2006) 294–308 301

exogenous transgenes expressed in vivo, can act as surro-gate markers for gene expression. In line with the behav-ioral data above, robust rat IL-10 protein release intoCSF was measured 3 days after a single i.t. pDNA-rIL-10 injection compared to naı̈ve and pDNA-eGFP-treated controls (Fig. 4B).

Statistical analyses support these findings. Two daysfollowing a second i.t. injection of pDNA-eGFP or

Table 1Verification of plasmid gene expression by mRNA analysis

Quantitative RT-PCR for plasmid-derived rat IL-10 expression

Treatment and region of spinal cord tissue Rat number

Naı̈ve cauda equina 4peGFP cauda equina 5pDNA-rIL-10 cauda equina 5

pDNA-rIL-10 cauda equina 5pDNA-rIL-10 lumbar 5

Total rat IL-10 mRNA from the cauda equina was robustly expressed in ptreated controls (n =5/group). Additionally, plasmid-derived rat IL-10 mRN

pDNA-rIL-10, serum rat IL-10 protein levels were virtu-ally undetectable, and serum human IL-10 protein wascompletely absent after identical treatment withpDNA-hIL-10 (F2,10 = .903, p > .43) (Panel A). Con-versely, rat IL-10 lumbosacral CSF protein expressionwas congruent with mRNA expression 2 days followinga second i.t. pDNA-rIL-10 injection (see Table 1). Alarge increase of rat IL-10 protein in lumbosacral CSF

Total rat IL-10 expression SE

0 00 0

32.839 24.986

Plasmid-derived rIL-10 expression SE64.251 48.8874.528 2.709

DNA-rIL-10-treated rats compared to either naı̈ve or pDNA-eGFP-A was greater in the cauda equina than in the lumbar spinal cord.

A Serum

0

50

100

150

200

250

Inte

rleu

kin-

10 (

pg/m

l)

GFP rIL-10 hIL-100

50

100

150

200

250

Inte

rleu

kin-

10 (

pg/m

l)

Lumbosacral CSF

GFP rIL-10 rIL-10(3 day)

None hIL-10

B

C

C D E

20 μm 20 μm 20 μm

Fig. 4. Verification of plasmid gene expression by protein analysis. Each group (except ‘‘None’’) received either one or two successive i.t. injectionsusing a 3-day interval between injections. Each group received two i.t. injections of either pDNA-eGFP, pDNA-rIL-10 or pDNA-hIL-10 (each100 lg in 20 ll). Panel A: two days following the second i.t. injection, serum rat IL-10 protein levels were barely detectable in rats receiving i.t.pDNA-rIL-10, and serum human IL-10 protein was completely absent in rats receiving i.t. pDNA-hIL-10. Panel B: rat IL-10 protein expressionlevels in lumbosacral CSF were at the limits of assay detection in rats receiving either no i.t. pDNA injection (labeled ‘‘None’’) or i.t. pDNA-GFP.Conversely, rat IL-10 protein expression levels in lumbosacral CSF were increased in rats receiving i.t. pDNA-rIL-10 (labeled ‘‘rIL-10’’). Thisincrease was congruent with mRNA expression 2 days following the second i.t. pDNA-rIL-10 injection (see Table 1). A large increase of rat IL-10protein expression levels in lumbosacral CSF was detected 3 days after a single i.t. pDNA-rIL-10 injection (labeled ‘‘rIL-10 (3 day)’’). In addition,substantial human IL-10 protein expression was observed in lumbosacral CSF in rats receiving i.t. pDNA-hIL-10. Panels C–E: confocal and lightmicroscopic verification of plasmid gene expression 5 days following a second i.t. injection of pDNA-eGFP. All animals (N = 2/group) received afirst i.t. injection of pDNA-rIL-10 (100 lg in 20 ll) followed by either pDNA-rIL-10 (control; 50 lg in 10 ll) or pDNA-eGFP (50 lg in 10 ll). Nucleiin the meninges counterstained with DAPI are large, ovoid blue structures (Panel C). GFP expression after i.t. pDNA-GFP injection was observed byconfocal microscopy in lumbar meninges surrounding the injection site (Panel D). These GFP-expressing cells contained darkened ovoid structures,presumably nuclei. The darkened ovoid structures within GFP-expressing cells appear as blue ovoid structures after application of the blue nuclearstain, DAPI (Panel E). These DAPI-stained nuclei are surrounded by GFP-expressing cells. GFP-positive cells appear to be clustered in the lumbarmeninges, here facing the right margin with the rostral-caudal axis of the meninges running longitudinally (Panel E).

302 E.D. Milligan et al. / Pain 126 (2006) 294–308

was detected 3 days after the initial i.t. pDNA-rIL-10injection (Panel B). In addition, substantial expressionof both rat and human IL-10 protein 2 days followinga second i.t. injection of their respective transgene-en-coding plasmids was detected compared to naı̈ve andpDNA-eGFP treated controls (F3,12 = 14.879,p < .001) (Panel B).

We further sought to examine the topographical loca-tion of plasmid DNA gene expression of Jellyfish GFPprotein detected five days after i.t gene delivery. Ratsreceived i.t. 100 lg pDNA-rIL-10 followed three dayslater by i.t. 50 lg pDNA-rIL-10 plus 50 lg pDNA-eGFP or equivolume PBS control. Expression of eGFPwas examined in tissues collected 5 days after i.t. 50 lgpDNA-eGFP.

Examination of the lumbar spinal cord parenchymadid not reveal detectable GFP-positive cells. All GFPexpression was observed in the superficial spinal cord

dorsal horns and the space between the spinal cordand dura mater matrix, referred to here as the lumbarmeninges. Confocal microscopy revealed GFP-express-ing cells within which darkened ovoid structuresappeared (Fig. 4D). These darkened structures appearedas large, ovoid blue structures when counterstained withDAPI. Thus, DAPI-stained nuclei are surrounded byGFP-expressing cells that appear to be clustered in themeninges facing the right margin with the rostral-caudalaxis of the meninges running longitudinally (Fig. 4E).No GFP-positive cells were detected in control rats(Fig. 4C).

3.4. Reversal of long-established CCI-induced mechanical

allodynia

In order to determine the effectiveness of pDNA-rIL-10 spinal cord gene therapy on long-term pain (58 days),

E.D. Milligan et al. / Pain 126 (2006) 294–308 303

two successive injections of pDNA-rIL-10 (100 lg in18–20 ll) or equivolume vehicle (3% sucrose in DPBS)were given to rats with behaviorally verified chronic(2-month) CCI-induced allodynia (Fig. 5A and B).The first i.t. pDNA-IL-10 injection (after behavioraltesting on Day 58) reversed bilateral allodynia within 2days. The second i.t. pDNA-IL-10 injection, given onDay 61 after CCI, produced a prolonged and reliablebilateral reversal of allodynia for approximately 35 days(95 days after CCI) (Fig. 5A and B). The bilateral anti-allodynic effect of this second pDNA-rIL-10 injectionwas reliable through Day 95 after CCI after which fullallodynia was observed through Day 103.

Statistical analyses confirmed these observations.Threshold responses at BL were similar between treat-ment groups and hindpaw laterality, ipsilateralhindpaws (Panel A) and contralateral hindpaws (PanelB) (F1,12 = .014, p > 0.90). CCI produced a robust bilat-eral mechanical allodynia throughout the timecourseprior to pDNA treatment (F8,96 = 3.388, p < 0.01).Immediately after behavioral testing on Day 58, ratsreceived i.t. pDNA, and compared to i.t. vehicle,pDNA-rIL-10 reversed bilateral allodynia (F1,12 =24.910, p < 0.0005) induced by CCI within this 3-day

BL 3 10 22 26 30 34 38 40 58 59 60 61 62

CCI; pDNA-IL-10

CCI; Vehicle

Days aft

pDNA-IL-10 or Vehicle pD

Ipsilateral Hindpaw - i.t. pDNA-IL-10

3.25

3.50

3.75

4.00

4.25

4.50

4.75

5.00

Log

Abs

olut

e T

hres

hold

(mg

x 10

)

CCI Surgery

Contralateral Hindpaw - i.t. pDNA-IL-10

CCI; pDNA-IL-10

CCI; Vehicle

Days aft

Log

Abs

olut

e T

hres

hold

(mg

x 10

)

3.25

3.50

3.75

4.00

4.25

4.50

4.75

5.00

BL 3 10 22 26 30 34 38 40 58 59 60 61 62

pDNA-IL-10 or Vehicle pDCCI Surgery

A

B

Fig. 5. Reliable reversal of persistent (2-month) CCI-induced mechanical asuccessive i.t. injections of either pDNA-IL-10 or pDNA-GFP (control) (eacCCI-operated animals. The time of CCI surgery and each i.t. injection aresucrose); closed symbols, pDNA-IL-10. Responses of the hindpaw ipsilateanalyzed through 103 days after CCI surgery.

time period (F1,24 = 45.596, p < 0.0001). The second i.t.pDNA injection (given on Day 61 after CCI while con-tinued reversal was observed) produced a prolonged andreliable bilateral reversal of a 2-month chronic CCI-in-duced allodynia for approximately 35 days(F12,144 = 5.450, p < 0.0001) (95 days after CCI) (PanelsA and B).

3.5. A shorter inter-injection interval improves reversal by

pDNA-rIL-10 of CCI-induced mechanical allodynia

Examination of CsCl-purified plasmid injected inearlier experiments revealed significant levels of endo-toxin. Because of endotoxin proinflammatory cytokineexpression (Nguyen et al., 2002), we reasoned thatendotoxin contamination of plasmid DNA could affectoverall efficacy of pDNA-rIL-10 therapy on CCI rats.Therefore, we examined both the magnitude and dura-tion of reversal of CCI-induced allodynia by pDNA-rIL-10 with endotoxin-free plasmid. Additionally, wetested whether a shorter inter-injection interval couldincrease the duration of reversal from mechanical allo-dynia, perhaps by temporally altering the characteris-tics of the local anti-inflammatory milieu. Thus,

63 65 67 71 75 79 83 87 91 95 99 103

er CCI

NA-IL-10 or Vehicle

Abs

olut

e T

hres

hold

(gra

ms)

0.17

0.32

0.56

1

1.73

3.16

5.62

10

er CCI

0.17

0.32

0.56

1

1.73

3.16

5.62

10

Abs

olut

e T

hres

hold

(gra

ms)

63 65 67 71 75 79 83 87 91 95 99 103

NA-IL-10 or Vehicle

llodynia after two sequential i.t. injections of pDNA-rIL-10. The twoh 100 lg in 18–20 ll; n =4/group) were delivered at 3 days interval toindicated by black arrows. Open symbols, vehicle (control; PBS-3%

ral (Panel A) and contralateral (Panel B) to the site of surgery were

304 E.D. Milligan et al. / Pain 126 (2006) 294–308

altering the inter-injection interval may improve uponthe effectiveness of pDNA-rIL-10 to reverse CCI-in-duced allodynia.

The dose of naked pDNA-rIL-10 was chosen basedon therapeutic effects observed above. Rats wereassessed for their responses to the von Frey test at BLand again on Days 3 and 10 after CCI surgery. Afterbehavioral assessment on Day 10, all rats received i.t.pDNA-rIL-10 or equivolume vehicle (3% Sucrose inDPBS). The von Frey test was again performed dailyafter the first injection. That is, Day 11 and 12 for the2-day injection interval group and Day 11–13 for the3-day injection interval group. Equivolume vehicle wasgiven at a 3-day interval. Doses and volumes of the firstand second injections were identical. Von Frey testingwas performed as indicated (Fig. 6 caption).

The anti-allodynic effects of two successive doses ofpDNA-rIL-10 given at a 2-day injection intervalproduced a longer and more robust reversal fromCCI-induced allodynia, compared to giving the samepDNA-rIL-10 doses at a 3-day injection interval. BothpDNA-rIL-10 injection schedules reversed allodynia,compared to i.t. vehicle control (Fig. 6A and B). Thus,both the 3-day and 2-day inter-injection interval effec-tively reversed CCI-induced mechanical allodynia, but

3.25

3.50

3.75

4.00

4.25

4.50

4.75

5.00

BL 3 10 12 14 16 18 20 24 28

Days After CC

Ipsilateral Hindpaw

Log

Abs

olut

e T

hres

hold

(m

g x

10)

3.50

3.75

4.00

4.25

4.50

4.75

5.00

BL 3 10 12 14 16 18 20 24 28

Days After CC

Contralateral Hindpaw

Log

Abs

olut

e T

hres

hold

(m

g x

10)

FirstpDNACCI Surgery

FirstpDNACCI Surgery

A

B

Fig. 6. A 2-day inter-injection interval improves pDNA-rIL-10 reversal of Ci.t. injections of either vehicle (control; PBS-3% sucrose) or pDNA-rIL-10 (e(pDNA-rIL-10, open diamond) or three (pDNA-rIL-10, closed diamond; veinjection are indicated by black arrows. Responses of the hindpaw ipsilateanalyzed through 54 days after CCI surgery.

the magnitude and duration of reversal was determinedby the inter-injection interval.

Statistical analyses support these conclusions. Therewere no group differences in responses to the von Freytest at BL (F5,22 = .150, p > .97). A robust bilateral allo-dynia from the ipsilateral (Panel A) and contralateral(Panel B) hindpaws was observed on Day 3 and 10 afterCCI compared to BL values (F2,44 = 186.336,p < 0.0001). Bilateral allodynia was reversed by day 2after the initial pDNA-rIL-10 injection compared tovehicle-injected rats (F5,22 = 6.828, p < 0.001) (Panel Aand B). On the day of the second i.t. injection,pDNA-rIL-10-treated groups remained reliablyreversed compared to the vehicle-treated group(F2,22 = 15.159, p < 0.0001). The anti-allodynic effectsof pDNA-rIL-10 remained effective through Day 47or 51 post CCI in the 2-day vs. 3-day interval injectedgroups, respectively, compared to vehicle(F12,264 = 14.857, p < 0.0001). Rats treated with i.t.pDNA-rIL-10 at the 2-day interval resulted in reversalfrom allodynia that was greater in magnitude and dura-tion compared to rats treated i.t. with pDNA-rIL-10 atthe 3-day interval in both the ipsilateral and contralater-al hindpaws (F12,72 = 3.876, p = 0.0001; F12,72 = 2.654,p < 0.01, for each hind paw).

32 36 40 44 48 54

I Surgery

CCI; pDNA-IL-10; 2 Day interval

CCI; pDNA-IL-10; 3 Day interval

CCI; Vehicle Day 10; 3 Day interval

0.17

0.32

0.56

1

1.73

3.16

5.62

10

Abs

olut

e T

hres

hold

(gr

ams)

32 36 40 44 48 54

I Surgery

0.17

0.32

0.56

1

1.73

3.16

5.62

10

Abs

olut

e T

hres

hold

(gr

ams)

CCI; pDNA-IL-10; 2 Day interval

CCI; pDNA-IL-10; 3 Day interval

CCI; Vehicle Day 10; 3 Day interval

CI-induced mechanical allodynia. CCI-operated animals received twoach 100 lg in 18–20 ll; n =4/group). Injections were spaced either twohicle, open square) days apart. The time of CCI surgery and each i.t.ral (Panel A) and contralateral (Panel B) to the site of surgery were

E.D. Milligan et al. / Pain 126 (2006) 294–308 305

4. Discussion

These experiments demonstrate that long-term reduc-tion of allodynia can be achieved by i.t. naked pDNAencoding the anti-inflammatory cytokine, IL-10(pDNA-rIL-10). One pDNA-rIL-10 injection inducestransient anti-allodynia (�3 days). Remarkably, thisdramatically enhances the efficacy of a second pDNA-rIL-10 injection 2–3 days later. Two such pDNA-rIL-10 injections reverse allodynia for >40 days, suggestingthat this may provide a novel paradigm enabling pro-longed clinical pain control. Plasmid-DNA-IL-10 wasnot only effective shortly after neuropathic pain onset,but produced anti-allodynia after chronic (2-month)neuropathic pain as well. As expected, pDNA-IL-10increased plasmid-driven mRNA for IL-10 andincreased IL-10 in lumbosacral CSF. Plasmid-DNA-eGFP induced GFP expression in meninges surroundingthe injection site, in keeping with previous reports of i.t.gene delivery (Mannes et al., 1998; Milligan et al.,2005b). Given that neuropathic pain is especially diffi-cult to treat (McQuay et al., 1996), the success of thisnovel gene therapy suggests clinical utility. Thus, itappears that intrathecal non-viral gene therapy to drivethe release of anti-inflammatory cytokines may providea new approach for pathological pain control.

IL-10 is a powerful anti-inflammatory cytokine thatcan suppress proinflammatory cytokines (Moore et al.,2001) implicated in pathological pain (Watkins andMaier, 2004). Intrathecal IL-10 suppresses IL-1-medi-ated pain enhancement by intrathecal dynorphin(Laughlin et al., 2000) and CCI (Milligan et al.,2005a). While studies of spinal IL-10 actions are ongo-ing, it is clear that IL-10 can suppress CSF levels of atleast IL-1 (Milligan et al., 2005a). Importantly, IL-10receptors are not expressed by neurons but are expressedby non-neuronal cells such as glia in spinal cord underbasal or inflammatory states (Ledeboer et al., 2003).This distribution suggests IL-10 gene therapy may con-trol the proinflammatory consequences of glial activa-tion while leaving neuronal function unaltered.

While intrathecal gene therapy to treat neuropathicpain is not unique (Wu et al., 2001a; Wu et al., 2001b;Eaton et al., 2002), IL-10 gene delivery shows exception-al promise. Intrathecal IL-10 gene therapy using viralvectors prevents and/or reverses IL-1-mediated painenhancement induced by morphine withdrawal (John-ston et al., 2004), intrathecal HIV-1 gp120 (Milliganet al., 2001), and sciatic nerve inflammation/trauma(Milligan et al., 2005a; Milligan et al., 2005b). Sucheffects are transient, as IL-10 has �2 h half-life in CSF(Milligan et al., 2005a) and intrathecal IL-10 encodingviral vectors suppress pain for only �2 weeks (Milliganet al., 2005a; Milligan et al., 2005b). Intrathecal genetransfer using viral vectors to treat neuropathic painhas proven challenging at best (Beutler et al., 2005).

Although single i.t. viral vector injections infect cells inthe meninges without expression in the spinal parenchy-ma (Iadarola et al., 1997; Mannes et al., 1998; Milliganet al., 2005b), they are short-lived possibly due to recog-nition by the immune system (Jooss and Chirmule, 2003;Liu and Muruve, 2003), or to inefficient cellular specificpromoter activity (Davidson and Breakefield, 2003).While such studies provide converging lines of evidencethat IL-10 is efficacious for pain treatment, none pro-duce sufficiently prolonged effects to consider clinicalapplication. Thus, alternative methods are needed.

Somewhat more promising for gene delivery werereports that single or multiple injections of pDNA-basedvectors in the periphery, brain or spinal cord parenchy-ma lead to transgene expression for up to 4 weeks with-out toxicity or glial activation (Schwartz et al., 1996;Brooks et al., 1998; Meuli-Simmen et al., 1999; Tanet al., 2001 Shi et al., 2003). Our current studiesemployed naked pDNA doses (100 lg) that are withinthe range of prior reports. High pDNA doses may berequired because of the inefficiency of naked DNAuptake, at least upon the first i.t. injection. Our ongoingstudies indicate that while a high pDNA dose is requiredfor the first injection, the second injection dose can bereduced 100-fold without loss of efficacy (Sloane et al.,2005), thus decreasing the likelihood of adverse side-ef-fects such as toxicity. While conventional applications ofpolymer or liposome-treated DNA improve DNA geneexpression at much lower DNA doses, the DNA com-plexes are inflammatory (Li et al., 1999; Tan et al.,2001), limiting their clinical viability.

The present studies are unique not only in applyingnon-viral vectors for intrathecal IL-10 gene transfer,but also in the duration of therapeutic efficacy for paincontrol. Until now, spinal cord pDNA gene transferhas been limited to transient expression either withnaked or complexed pDNA. One intrathecal interleu-kin-2 pDNA injection reduced CCI-induced neuropath-ic pain for 3 days (Yao et al., 2002a), preventedsubcutaneous carrageenan-induced pain (Yao et al.,2002b), and produced spinal cord mRNA and proteinexpression for 3 days. The present study supports thata single intrathecal naked pDNA-IL-10 injection leadsto a similarly transient reversal of CCI-induced neuro-pathic pain. Previous reports suggest that intrathecalinjection interval critically impacts spinal gene expres-sion (Shi et al., 2003). The present study extends theseresults to demonstrate that repeated injections of nakedpDNA-IL-10 at inter-injection intervals of 2–3 daysleads to behavioral phenotypic expression for muchlonger times.

The meninges show the most robust transgene expres-sion following intrathecal non-viral vector delivery(Meuli-Simmen et al., 1999; Yao et al., 2002a). Menin-ges contain macrophages and dendritic cells (Braunet al., 1993) which form a dense, extensive network

306 E.D. Milligan et al. / Pain 126 (2006) 294–308

capable of receptor-mediated endocytosis (McMenamin,1999; McMenamin et al., 2003) and contact the CSF(Guseo, 1977). These cells express a wide range of pat-tern recognition receptors (PRR) that include scavengerreceptors and toll-like receptors (TLR) that initiateimmune activation (Medzhitov and Janeway, 1997;Aderem and Ulevitch, 2000; Gordon, 2002). Ligandsof PRRs include unmethylated cytosine-linked guaninedinucleotide motifs (CpG) prevalent in bacterial DNAs(Krieg, 2002). These CpG motifs are methylated in ver-tebrate genomes and do not activate immune cells. How-ever, unmethylated CpG motifs bind to and activatedendritic cells and macrophages via scavenger and intra-cellular TLR-9 receptors (Stacey et al., 1996; Krieg,2002; Latz et al., 2004; McCoy et al., 2004). CpG-con-taining DNA form intra-cellular endosomes from whichTLR-9-mediated intracellular signaling occurs (Latzet al., 2004). The pDNA used here contains unmethylat-ed CpG motifs suggestive that the effects reported mayinvolve such an uptake and signaling mechanism.

While CpG-containing pDNA is well documented toactivate dendritic cells and macrophages (Krieg, 2002),the profiles of immune-cell activation can vary. Inflam-matory responses of classically activated dendritic cellsand macrophages include proinflammatory cytokines,nitric oxide, reactive oxygen species, and recognitionand killing of pathogens. However, under conditionsthat enhance tissue repair and dampen inflammation,dendritic cells and macrophages suppress inflammation(Goerdt and Orfanos, 1999). This alternative activationprofile includes a preferential expression of macrophagescavenger receptors, increased production and release ofIL-1 receptor antagonist and IL-10, and decreased pro-inflammatory cytokine expression (Goerdt and Orfanos,1999; Gordon, 2003). Moreover, IL-10 itself can inhibitthe generation of classically activated macrophages(Katakura et al., 2004).

It is possible that CpG motifs are detrimental foroptimal IL-10 transgene expression (Schluesener et al.,2001). However, IL-10 itself may rescue transgeneexpression by inducing an anti-inflammatory activationprofile of both dendritic cells and macrophages. Studiesto examine this are ongoing. Sustained alternative acti-vation of dendritic cells and macrophages, in the pres-ence of IL-10 expression, may set the stage for greatertransgene uptake and/or efficacy. Intriguingly, pDNAconstructs containing inverted terminal repeat nucleo-tide sequences (ITR) of the adeno-associated viral gen-ome, similar to that used in these studies (Fig. 1),produced improved transgene expression in vitro andin vivo compared to pDNA constructs without ITRs(Flotte et al., 1993; Philip et al., 1994; Xin et al., 2003;Chikhlikar et al., 2004). Thus, multiple factors such asITR sequences and CpG motifs in pDNA-rIL-10 mayprove critical for transgene uptake upon a secondpDNA exposure.

In conclusion, the present series of studies provideexciting evidence that pathological pain states may becontrolled by intrathecal gene therapy that drives theproduction of the anti-inflammatory cytokine IL-10,using repeated injections of naked plasmid DNA.Naked pDNA- IL-10 provides a safe and simple alterna-tive approach to viral-based spinal cord gene therapy.The intrathecal route employed mimics lumbar punc-tures in clinical routine use. Importantly, naked plas-mid-DNA-IL-10 is therapeutic for pain established forshort-term and established for extended periods of timeshowing strong support to extend this application toclinical pain.

Acknowledgements

This work was supported by NIH Grants DA018156,DA015642, DA015656, and HL56510, and grants fromAvigen.

References

Aderem A, Ulevitch RJ. Toll-like receptors in the induction of theinnate immune response. Nature 2000;406:782–7.

Altschul SF, Madden TL, Schaffer AA, Zhang J, Zhang Z, Miller W,et al. Gapped BLAST and PSI-BLAST: a new generation ofprotein database search programs. Nucleic Acids Res1997;25:3389–402.

Bennett GJ, Xie YK. A peripheral mononeuropathy in rat thatproduces disorders of pain sensation like those seen in man. Pain1988;33:87–107.

Beutler AS, Banck MS, Walsh CE, Milligan ED. Intrathecal genetransfer by adeno-associated virus for pain. Curr Opin Mol therap2005;7:431–9.

Braun JS, Kaissling B, Hir LH, Zenker W. Cellular components of theimmune barrier in the spinal meninges and dorsal root ganglia ofthe normal rat: immunohistochemical (MHC class II) and electron-microscopic observations. Cell Tissue Res 1993;273:209–17.

Brooks AI, Halterman MW, Chadwick CA, Davidson BL, Haak-Frendscho M, Radel C, et al. Reproducible and efficient murineCNS gene delivery using a microprocessor-controlled injector.J Neurosci Methods 1998;80:137–47.

Chikhlikar P, Arruda LB, Agrawal S, Byrne B, Guggino W, AugustJT, et al. Inverted terminal repeat sequences of adeno-associatedvirus enhance the antibody and CD8+ responses to a HIV-1p55Gag/LAMP DNA vaccine chimera. Virology 2004;323:220–32.

Chomczynski P, Sacchi N. Single-step method of RNA isolation byacid guanidinium thiocyanate–phenol–chloroform extraction. AnalBiochem 1987;162:156–9.

Cotten M, Baker A, Saltik M, Wagner E, Buschle M. Lipopolysac-charide is a frequent contaminant of plasmid DNA preparationand can be toxic to primary human cells in the presence ofadenovirus. Gene Ther 1994;1:239–46.

Daheshia M, Kuklin N, Kanangat S, Manickan E, Rouse BT.Suppression of ongoing ocular inflammatory disease by topicaladministration of plasmid DNA encoding IL-10. J Immunol1997;159:1945–52.

Davidson BL, Breakefield XO. Viral vectors for gene delivery to thenervous system. Nature Rev Neurosci 2003;4:353–64.

E.D. Milligan et al. / Pain 126 (2006) 294–308 307

Eaton MJ, Blits B, Ruitenberg MJ, Verhaagen MJ, Oudega M.Amelioration of chronic neuropathic pain after partial nerveinjury by adeno-associated viral (AAV) vector-mediated over-expression of BDNF in the rat spinal cord. Gene Ther2002;9:1387–95.

Flotte TR, Afione SA, Solow R, Drumm ML, Markakis D, GugginoWB, et al. Expression of the cystic fibrosis transmembraneconductance regulator from a novel adeno-associated virus pro-moter. J Biol Chem 1993;268:3781–90.

Goerdt S, Orfanos CE. Other functions, other genes: alternativeactivation of antigen-presenting cells. Immunity 1999;10:137–42.

Gordon S. Pattern recognition receptors: doubling up for the innateimmune response. Cell 2002;111:927–30.

Gordon S. Alternative activation of macrophages. Nat Rev2003;3:23–35.

Guseo A. Classification of cells in the cerebrospinal fluid. Eur Neurol1977;15:169–76.

Harvey LOJ. Efficient estimation of sensory thresholds. Behav ResMeth Instrum Comput 1986;18:623–32.

Iadarola MJ, Lee S, Mannes AJ. Gene transfer approaches to paincontrol. In: Borsook D, editor. Molecular Neurobiology ofPain. Seattle: IASP Press; 1997. p. 337–60.

Johnston IN, Milligan ED, Wieseler-Frank J, Frank MG, Zapata V,Langer S, et al. A role for pro-inflammatory cytokines andfractalkine in analgesia, tolerance and subsequent pain facilitationinduced by chronic intrathecal morphine, 2004;24:7353–7365.

Jooss K, Chirmule N. Immunity to adenovirus and adeno-associatedviral vectors: implication for gene therapy. Gene Ther2003;10:963–95.

Katakura T, Miyazaki M, Kobayashi M, Herndon DN, Suzuki F.CCL17 and IL-10 as effectors that enable alternatively activatedmacrophages to inhibit the generation of classically activatedmacrophages. J Immunol 2004;172:1407–13.

Krieg AM. CpG motifs in bacterial DNA and their immune effects.Annu Rev Immunol 2002;20:709–60.

Latz E, Schoenemeyer A, Visintin A, Fitzgerald KA, Monks BG,Knetter CF, et al. TLR9 signals after translocating from the ER toCpG DNA in the lysosome. Nature Immunol 2004;5:190–8.

Laughlin TM, Bethea JR, Yezierski RP, Wilcox GL. Cytokineinvolvement in dynorphin-induced allodynia. Pain2000;84:159–67.

Ledeboer A, Wierinckx A, Bol JGJM, Floris S, Renardel de LavaletterC, DeVries HE, et al. Regional and temporal expression patternsof interleukin-10, interleukin-10 receptor and adhesion moleculesin the rat spinal cord during chronic relapsing EAE. J Neuroim-munol 2003;136:94–103.

Li S, Wu S-P, Whitmore M, Loeffert EJ, Wang L, Watkins SC, Pitt,BR, Huang L. Effect of immune response on gene transfer to thelung via systemic administration of cationic lipidic vectors. LungCell Mol Physiol 1999;20:L796–804.

Liu Q, Muruve DA. Molecular basis of the inflammatory response toadenovirus vectors. Gene Ther 2003;10:935–40.

Mannes AJ, Caudle RM, O’Connell BC, Iadarola MJ. Adenoviralgene transfer to spinal cord neurons: intrathecal vs. intraparen-chymal administration. Brain Res 1998;793:1–6.

McCoy SL, Kurtz SE, Hausman FA, Trune DR, Bennett RM,Hefeneider SH. Activation of RAW264.7 macrophages bybacterial DNA and lipopolysaccharide increases cell surfaceDNA binding and internalization. J Biol Chem2004;279:17217–23.

McMahon SB, Cafferty WBJ, Marchand F. Immune and glial cellfactors as pain mediators and modulators. Exper Neurol2005;192:444–62.

McMenamin PG. Distribution and phenotype of dendritic cells andresident tissue macrophages in the dura mater, leptomeninges, andchoroid plexus of the rat brain as demonstrated in whole mountpreparations. J Comp Neurol 1999;405:553–62.

McMenamin PG, Wealthall RJ, Deverall M, Cooper SJ, Griffin B.Macrophages and dendritic cells in the rat meninges and choroidplexus: three-dimensional localisation by environmental scanningelectron microscopy and confocal microscopy. Cell Tissue Res2003;313:259–69.

McQuay H, Tramer M, Nye BA, Carroll D, Wiffen P, Moore RA. Asystematic review of antidepressants in neuropathic pain. Pain1996;68:217–27.

Medzhitov R, Janeway CAJ. Innate Immunity: the virtues of anonclonal system of recognition. Cell 1997;91:295–8.

Meuli-Simmen C, Liu Y, Yeo TT, Liggitt D, Tu G, Yang T, et al.Gene expression along the cerebral-spinal axis after regional genedelivery. Hum Gene Ther 1999;10:2689–700.

Milligan ED, Langer SJ, Sloane EM, He L, Wieseler-Frank J,O’Connor KA, et al. Controlling pathological pain by adenovi-rally driven spinal production of the anti-inflammatory cytokine,Interleukin-10. Eur J Neurosci 2005a;21:2136–48.

Milligan ED, Mehmert KK, Hinde JL, Harvey LOJ, Martin D, TraceyKJ, et al. Thermal hyperalgesia and mechanical allodynia pro-duced by intrathecal administration of the Human Immunodefi-ciency Virus-1 (HIV-1) envelope glycoprotein, gp120. Brain Res2000;861:105–16.

Milligan ED, O’Connor KA, Nguyen KT, Armstrong CB, Twining C,Gaykema R, et al. Intrathecal HIV-1 envelope glycoprotein gp120enhanced pain states mediated by spinal cord proinflammatorycytokines. J Neurosci 2001;21:2808–19.

Milligan ED, Sloane EM, Langer SJ, Cruz PE, Chacur M, Spataro L,et al. Controlling neuropathic pain by adeno-associated virusdriven production of the anti-inflammatory cytokine, interleukin-10. Mol Pain 2005b;1:9–22.

Moore KW, deWaal Malefyt R, Coffman RL, O’Garra A. Interleukin-10 and the interleukin-10 receptor. Annu Rev Immunol2001;19:683–765.

Nguyen MD, Julien J-P, Rivest S. Innate immunity: the missing link inneuroprotection and neurodegeneration? Nat Rev 2002;3:216–27.

O’Connor KA, Holguin A, Hansen MK, Maier SF, Watkins LR. Amethod for measuring multiple cytokines from small samples.Brain Behav Immun 2004;18:274–80.

Pfaffl MW. A new mathematical model for relative quantification inreal-time RT-PCR. Nucleic Acids Res 2001;29:e45.

Philip R, Brunette E, Kilinski L, Murugesh D, McNally MA, Ucar K,et al. Efficient and sustained gene expression in primary Tlymphocytes and primary and cultured tumor cells mediated byadeno-associated virus plasmid DNA complexed to cationicliposomes. Mol Cell Biol 1994;14:2411–8.

Qiagen, Inc. Endofree plasmid purification handbook. Qiagen 2006;pp. 38–41.

Sambrook J, Fritsch ER, Maniatis T. Molecular cloning. Cold SpringHarbor Press; 1989.

Schluesener HJ, Saeid K, Deininger M, Schwab J. Transient in vivoactivation of rat brain macrophages/microglial cells and strocytesby immunostimulatory multiple CpG oligonucleotides. J Neuro-immunol 2001;113:89–94.

Schwartz B, Benoist C, Abdallah B, Rangara R, Hassan A, SchermanD, et al. Gene transfer by naked DNA into adult mouse brain.Gene Ther 1996;3:405–11.

Shi L, Tang GP, Gao SJ, Ma YX. Repeated intrathecal administrationof plasmid DNA complexed with polyethylene glycol-graftedpolyethylenimine led to prolonged transgene expression in thespinal cord. Gene Ther 2003;10:1179–88.

Sloane EM, Langer SJ, Jekich J, Mahoney J, Huberty G, Hughes T,et al. Resolution of neuropathic pain by intrathecal (IT) genetherapy to induce interleukin-10 (IL-10): Initial exploration ofmechanisms. Syndney, Australia: International Association forthe Study of Pain; 2005.

Stacey KJ, Sweet MJ, Hume DA. Macrophages ingest and areactivated by bacterial DNA. J Immunol 1996;157:2116–22.

308 E.D. Milligan et al. / Pain 126 (2006) 294–308

Tan Y, Liu F, Li Z, Li S, Huang L. Sequential injection of cationicliposome and plasmid DNA effectively transfects the lung withminimal inflammatory toxicity. Mol Ther 2001;3:673–81.

Tsuda M, Inoue K, Salter MW. Neuropathic pain and spinalmicroglia: a big problem from molecules in ‘small’ glia. TrendsNeurosci 2005;28:101–7.

Watkins LR, Maier SF. Glia: a novel drug discovery target for clinicalpain. Nature Rev Drug Discov 2003;2:973–85.

Watkins LR, Maier SF. Targeting glia to control clinical pain: anidea whose time has come. Drug Discov Today: Ther Strateg2004.

Watkins LR, Milligan ED, Maier SF. Glial activation: a driving forcefor pathological pain. Trends Neurosci 2001;24:450–5.

Wicks IP, Howell ML, Hancock T, Kohsaka H, Olee T, Carson D.Bacterial Lipopolysaccharide copurifies with plasmid DNA: impli-cations for animal models and human gene therapy. Hum GeneTher 1995;6:317–23.

Wu CL, Garry MG, Zollo RA, Yang J. Gene therapy for themanagement of pain. Part II: Mol targets 2001.

Wu CL, Garry MG, Zollo RA, Yang J. Gene therapy for themanagement of pain: Part I: methods and strategies. Anesthes2001b;94:1119–32.

Xin K-Q, Ooki T, Jounai N, Mizukami H, Hamajima K, Kojima Y,et al. A DNA vaccine containing inverted terminal repeats fromadeno-associated virus increases immunity to HIV. J Gene Med2003;5:438–45.

Yao MZ, Gu JF, Wang JH, Sun LY, Lang MF, Liu J, et al.Interleukin-2 gene therapy of chronic neuropathic pain. Neurosci-ence 2002a;112:409–16.

Yao MZ, Wang JH, Gu JF, Sun LY, Liu H, Zhao ZQ, et al.Interleukin-2 gene has superior antinociceptive effects when deliv-ered intrathecally. Clin Neurosci Neuropathol 2002b;13:791–4.

Zhi N, Zadori Z, Brown K, Tijssen P. Construction and sequencing ofan infectious clone of the human parvovirus B19. Virology2004;318:142–52.

Zolotukhin S, Potter M, Hauswirth WW, Guy J, Muzyczka N. A‘humanized’ green fluorescent protein cDNA adapted for high levelexpression in mammalian cells. J Virol 1996;70:4646–54.