Antigen-presenting cell engineering. The molecular toolbox.

American Journal of Pathology, Vol. 148, No. 1, January 1996Copynight © American Societyfor Investigative Pathology

ReviewAntigen-Presenting Cell Engineering

The Molecular Toolbox

Mark L. Tykocinski, David R. Kaplan, andM. Edward MedofFrom the Department ofPathology, Case Western Reserve

University, Cleveland, Ohio

Antigen-Presenting Cell Therapeutics: TheConceptThe use of cells in therapeutics traditionally has con-sisted of the administration of cells that mediate de-fined end-cell functions. Within the sphere of theimmunohematopoietic system, the earliest end-cellfunctions to be therapeutically targeted were oxygendelivery (via erythrocyte transfer) and coagulation(via platelet transfer). More complex end-cell func-tions later ensued, including blood cell renewal (viatransfer of hematopoietic stem cells derived frombone marrow, peripheral blood, and cord blood) andtumor cell killing (via transfer of tumor-infiltrating lym-phocytes and lymphokine-activated killer cells). Aless explored cellular therapeutic avenue consists ofthe administration of cells with regulatory rather thanend-cell functions. That is, the cell being introducedinto the host is not itself an effector but insteadmodulates downstream cellular effectors. One suchregulatory cell is the antigen-presenting cell (APC).Such APC transfer forms the basis for a newlyemerging class of APC-centered immunotherapeu-tics.APCs are cells with antigen-processing and -pre-

senting potentials. The core grouping of APCs, oftendesignated "professional APCs," are cells that han-dle antigens as their primary cellular function. Thesecells include macrophages,1 dendritic cells,2 Lange-rhans cells,3 B cells,4'5 and possibly newly recog-nized biphenotypic cells sharing B cell and macro-phage characteristics.6 Other cell types, however,share, to varying extents, antigen-processing and

-presenting capabilities. These cells are referred toby a variety of names including "nonprofessionalAPCs,"7' "semiprofessional APCs,"910'facultativeAPCs,'1"1 or "atypical APCs."12 For these latter APCs,antigen-directed functions are believed to be ancil-lary to their principal recognized functions. Addition-ally, in some instances, these cells acquire theirantigen-presenting capabilities only after stimulationwith proinflammatory cytokines. This latter heteroge-neous class of APCs includes cells of both hemato-poietic and nonhematopoietic origin (eg, activated Tcells, 13-15 eosinophils,12 thymic cortical epithelialcells,16 endothelial cells, 17-19 keratinocytes,20'21 as-trocytes,7 22 24 microglial cells,25'26 Schwann cells inperipheral nerve,27 retinal pigment epithelial cells,26myoblasts,11 vascular smooth muscle cells,8,29chondrocytes,30 enterocytes,31 thyrocytes,32 andkidney tubule cells33). The repertoire of cells thatpotentially can function as APCs has been evenfurther expanded by the intriguing observation thatwhen localized within a lymphoid organ microenvi-ronment, even cells such as fibroblasts can presentantigen.34APC:T cell interactions can culminate in different

functional outcomes. In most instances, T cell acti-vation ensues. However, in the case of some non-professional APC:T cell encounters, the T cell is ei-ther anergized by, or alternatively "ignores," the

Supported in part by National Institutes of Health grants R01A131044 (MLT), P01 DK38181 (MLT; MEM), and R01 A123598(MEM).

This article is based upon the Warner-Lambert/Parke-DavisAward lecture delivered by Mark Tykocinski at the ExperimentalBiology '95 meeting in Atlanta, Georgia in April, 1995.

Accepted for publication November 16, 1995.

Address reprint requests to Mark Tykocinski, Department of Pa-thology, Case Western Reserve University, BRB 925, 10900 EuclidAve., Cleveland, OH 44106.

1

2 Tykocinski et alAJPJantuary 1996, Vol. 148, No. 1

APC. It is presumed that such immunological anergyor ignorance is a consequence of the APCs' lack ofcostimulatory activity working in conjunction withtheir class or class 11 major histocompatibility com-plex (MHC) antigen presentation.8'1933 In the caseof nonprofessional APCs with activating function,their role may simply be, because of their strategictissue localization, to amplify and sustain local im-mune reactions during normal and pathological re-sponses.11 This would contrast with professionalAPCs, which initiate antigen-specific reactions.

To date, short peptides have constituted the par-adigmatic APC antigens. However, there is evidencethat the antigenic repertoire of APCs is more exten-sive. In addition to peptide antigens, certain APCscan present superantigens.7'35 Because superanti-gens can engage large populations of T cells shar-ing particular V. T cell receptor elements,35 they inessence extend the reach of APCs to broader seg-ments of the T cell repertoire. Additionally, there isthe recent indication that in addition to peptides,APCs also may be able to present nonpeptide lipidand glycolipid antigens using the non-polymorphic,,32-microglobulin-associated surface glycoproteinCDi 36,37

In recent years, the concept of using APCs ascellular therapeutics has emerged. As therapeuticagents, APCs offer the unique capacity to displaycomplex antigenic arrays and thereby polyclonallyengage a range of different T cells. By using the APCas immunogen (or tolerogen), one can bypass therequirement of preliminarily identifying the full set ofcritical antigens involved in a curative (or patho-genic) immune response. Attempts to harness theexpanding repertoire of peptide antigen, superanti-gen, and lipid antigen presentation potentials ofAPCs for immunotherapeutic ends are presently intheir relative infancy.

To use an APC in a therapeutic mode, it is notnecessary to use it as it exists naturally. Instead, it ispossible to engineer such a cell in order to modify itsimmunomodulatory properties. Here we will considersome newer molecular tools that have been appliedto APC engineering and to the APC therapeuticsarena.

Protein Painting: A New Addition to theMolecular ToolboxIn recent years, gene transfer has dominated thecellular engineer's toolbox. This broad technologyprovides the means for not only expressing newproteins (or over-expressing existing proteins) within

cells, but also for interfering with the production ofendogenous cellular proteins in a selective manner.The latter can be accomplished by a variety of genetransfer modalities, including antisense gene trans-fer, as well as sense gene transfer of sequencesencoding mediators such as ribozymes, triplex-form-ing polynucleotides, proteins with dominant-negativeactivities, and intracellular antibodies with specifici-ties for endogenous proteins. Major limitations ofgene transfer, however, have stemmed primarilyfrom inefficiencies in gene delivery, degradation ofinternalized gene sequences, and instability of geneexpression.

Cellular engineering tools, however, are not lim-ited to those operating at the genetic level. When itcomes to expressing new proteins in or on cells, it ispossible to bypass altogether the cell's own syn-thetic machinery and instead to directly deliver pre-formed proteins to cells, a process that can be ge-nerically referred to as "protein transfer." Proteintransfer can be applied to both soluble and mem-brane-associating proteins and has a number of po-tential advantages over gene transfer. First, mostprimary cells, whether normal or tumor, grow poorlyin culture and are relatively difficult to stably trans-fect, particularly with non-viral vectors. Protein trans-fer, on the other hand, is not dependent upon cellularproliferative potential or transfectability. Second, co-transfection procedures are rather cumbersome,and hence coordinatedly expressing multiple genesin the same cell is still, in most instances, quitechallenging. In contrast, protein transfer permits thesimultaneous delivery of an essentially limitless num-ber of proteins to cells. Third, protein transfer, unlikegene transfer, is a rapid procedure and is thus par-ticularly well suited for therapeutic applications.A line of investigation dating back to the mid-

1980s has led to an interesting strategy for proteintransfer of membrane-associating proteins, a pro-cess that will be referred to here as "protein paint-ing." The starting point was decay-accelerating fac-tor (DAF), a cell surface-associated complementregulatory factor that functions intrinsically within cellsurface membranes to accelerate the decay of theC3 and C5 amplification convertases of the classicaland alternative pathways of complement, therebyprotecting tissues from autologous complement-me-diated lysis. In the course of analyzing the function ofDAF, it was found that when purified DAF was addedto cells in vitro, it incorporated into their surface mem-branes and was functional. A remarkable physicalproperty of DAF was its solubility in low concentra-tions of nonionic detergents, or even in the absenceof detergents. As a consequence of this solubility

APC Engineering 3A/P january 1996, Vol. 148, No. 1

feature, purified DAF could be added back to cellswithout causing detergent-mediated cell lysis. Thismembrane reincorporability of DAF was first re-ported in 198438; importantly, after membrane rein-corporation, DAF retains its full native functional ca-pacities as a complement regulator.38'39Soon thereafter, the molecular underpinnings of

the unique reincorporability properties of DAF be-came apparent. Unlike "conventional" membraneproteins that are anchored to membranes by hydro-phobic transmembrane peptides, DAF is anchoredto membranes by a covalently appended glycolipid,specifically a glycosylinositol phospholipid (GPI)moiety.40-42 This structural feature of human DAF isshared in common with trypanosome variant surfaceglycoproteins, murine and rat Thy-1 antigen, andhuman placental alkaline phosphatase and erythro-cyte acetylcholinesterase, which had earlier beendocumented to be glycolipid-modified. The array ofproteins now known to have native GPI modificationshas grown considerably.43 The glycolipid moietiespermit this class of proteins to form detergent-de-pleted pseudo-micelles in solution.The ability of isolated GPI-modified proteins in

detergent micelles to reintegrate into cell surfacemembranes is a consequence of their GPI-associ-ated lipid moieties,44 glycerol-associated sn-i alkylor acyl and sn-2 acyl groups of 14 to 22 carbons45-47depending on the source of the proteins. Althoughthe presence of a single lipid moiety, or "foot," issufficient for transfer to cell membranes,44 the pres-ence of both lipid feet is necessary for stable inte-gration and full reconstitution of the extracellularfunction of the protein in the lipid bilayer.44 Isolationof GPI-anchored proteins and reincorporation of theisolated proteins into cells can be readily achievedusing a wide range of nonionic detergents. Methodsthat have proven to be most useful for these pur-poses have recently been reviewed.43

Subsequent cloning of the cDNA of DAF,48'49along with cDNAs corresponding to other GPI-mod-ified proteins, permitted a search for consensus GPImodification signal sequences. Significantly, no sim-ple consensus sequence was identified, with a set of"rules" governing GPI modification eventuallyemerging. In any case, we50 and others51 providedevidence that the GPI modification signal sequenceresides within the 3'-end sequence of DAF; signifi-cantly, when chimerized to other protein sequences,the 3'-end sequence of DAF conferred to them acarboxy-terminal GPI modification. In essence, whatconsequently emerged was a means of conferring aGPI anchor to virtually any protein of interest bymeans of gene chimerization. Since then, investiga-

tors studying a variety of other proteins have takenadvantage of the GPI reanchoring strategy.42'43 In allcases, the GPI-anchored proteins, expressed bygene transfer, exhibited the full extracellular interac-tive properties of their conventionally anchoredcounterparts. That an artificial GPI polypeptideadded to cells can similarly retain full extracellularinteractive function after protein transfer was subse-quently shown by us in 1994 (see below).52 Hence,the GPI chimerization strategy provides a path to anessentially limitless "palette of protein paints."

To paint, one needs enough paint. GPI polypep-tides, native or artificial, can be produced by recom-binant means, but require eukaryotic expression sys-tems capable of GPI modification. Although inprinciple yeast expression systems could be usedfor this purpose, efforts to date have been primarilydirected toward mammalian expression systems.Our recent data indicate that the glutamine syn-thetase amplification/expression system53 is wellsuited for this purpose (reviewed in ref. 43).

Proteins incorporating non-GPI lipid modificationsare also reincorporable into cell membranes.54 Pea-cock and coworkers55'56 have provided data alongthese lines. For example, antibodies have beenchemically palmitoylated, reincorporated into cellsurfaces, and shown to function as surrogate recep-tors for mediating specific cell-cell interactions.55'56Additionally, palmitate-conjugated protein A can beincorporated into membranes and then used as ageneric anchor to attach different antibody mole-cules via their Fc regions.57 The palmitate-conjuga-tion approach has additionally been extended to anon-antibody protein, namely, CD4.57 Palmitoylation,however, unlike GPI modification, requires control-ling the degree of lipid derivatization in order to avoidloss of function. Moreover, the site of lipid attach-ment and the membrane orientation of the incorpo-rated protein usually cannot be controlled. Of note,most native cellular acylated proteins are localized tothe cytoplasmic face of cell membranes.58 Clearly,such approaches need to be more thoroughly ex-plored, and advances are needed, especially relat-ing to site-specific chemical lipidation strategies.

Molding the Antigenic Display of the APCA reasonable starting point for APC engineeringstrategies is antigen presentation. The desired end-point is to have APCs present predefined peptideantigens on their surfaces. This goal is not simple toaccomplish via natural antigen processing pathwaysbecause endogenous processing of proteins gener-

4 Tykocinski et alAJP,January 1996, Vol. 148, No. I

ally results in the generation of multiple antigenicpeptides, and there can be temporal changes inprocessing events. Moreover, parameters such asAPC lineage, APC activation state, and the route ofintracellular processing can affect the final antigenicrepertoire of the APC.59'60 Peptide pulsing could bea way for achieving a predictable antigenic peptidedisplay, but here the problem is inefficiency in pep-tide loading, particularly for class MHC. This is inpart a consequence of the fact that MHC moleculesare pre-engaged with antigenic peptides by the timethey reach the cell surface.

These limitations have spawned efforts to bypassthe intracellular antigen-processing pathways alto-gether. One attempted approach entailed deliveringMHC molecules to cell surfaces via fusion with MHC-bearing liposomes.F We recently developed an al-ternative strategy based on protein painting.2'62This approach involves the painting onto cell sur-faces of well defined, membrane-reincorporableMHC:antigenic peptide complexes. Using HLA-A2as a model class MHC molecule, GPI-modifiedvariants of the HLA-A2 heavy chain were producedby chimerization with the GPI modification signalsequence of DAF. When expressed, these heavychains noncovalently associated with cotransfectedf32m light chains. By producing the recombinant GPI-modified class proteins in a Drosophila S2 Schnei-der cell background, they were prepared withoutpre-engaged peptides. This in turn permitted theGPI-modified class protein to be loaded with apreselected (chemically synthesized) antigenic pep-tide, in the absence of pre-engaged, endogenouslyloaded peptide as a confounding factor.

There are possibilities for further refining the GPI-modified MHC:032m:antigenic peptide complexes tobe used for antigenic painting. For example, theantigenic peptide could be covalently linked to theGPI-modified MHC carrier, in this way doing awayaltogether with the antigen preloading step. The fea-sibility of using this approach at the amino termini oftransmembrane peptide-anchored MHC heavychain polypeptides has been demonstrated.63 Alter-natively, another study has shown that it is possibleto generate, via gene chimerization, unimolecularclass MHC heavy chain:f32m peptide-bearing con-jugates.64 In such conjugates, the MHC heavy andlight chains are arrayed in tandem. This strategycould be used in conjunction with the addition of a 3'GPI modification signal. These different approachescould be combined to generate a "unimolecularMHC* peptide antigen paint" consisting of a singlelarge polypeptide that incorporates antigenic pep-tide, class MHC heavy chain, class MHC light

chain (f32m), and a GPI anchor in a linear array.Class 11 MHC proteins could be similarly engineered.

There are a number of therapeutic contexts inwhich achieving a specific antigenic peptide displaywould be useful. In these various settings, proteintransfer of GPI-modified class MHC proteins bear-ing loaded antigenic peptides could be used to tailorthe antigenic repertoires of APCs and the tailoredAPCs then used for ex vivo T cell amplification. Sig-nificantly, the protein transfer route has the addedpotential advantage of controlling antigenic epitopedensities, a variable that may be important, giventhat recent studies point to a window of optimalantigenic densities favoring T cell activation as op-posed to inhibition.65-67 One potential application ofthe MHC * antigenic peptide painting tool is in theinfectious disease setting. Chisari and coworkers68have provided compelling evidence that progressionto chronic active hepatitis correlates closely with thelack of a cytotoxic T cell response to certain hepatitisB virus antigenic peptides.68 This has prompted thesuggestion that the administration of ex vivo-ampli-fied cytotoxic T lymphocytes (CTLs) with definedhepatitis B virus peptide specificities could be ther-apeutic. Another example lies in the area of tumorimmunotherapy where data here too suggest that theuse of ex vivo-amplified CTL with defined antigenicspecificities could be beneficial. One of the bestdefined tumor systems in this regard is melanoma,where specific antigenic peptides have been shownto dominate in antitumor responses.69 In the case ofovarian carcinoma, a well defined Her2/neu peptidemay be the common target of CTLs.70

Immunogenic Tumor Cell Engineering:The Cancer Cell as APCProfessional APCs antigenically target as well asactivate T cell responders, ie, they are both antigen-specific and immunogenic. These dual capacities ofantigenic targeting and immunogenicity are disso-ciable. This is evidenced by the behavior of manynonprofessional APCs which present antigen but areincapable of activating T cells. APC engineering pro-vides a route for enhancing the immunogenicity ofsuch nonprofessional APCs. In particular, this is truefor "tumor APCs."

It is presumed that one frequently employedmechanism by which tumor cells evade host immuneresponses is via modulation of their APC antigenicand/or immunogenic properties. In some instances,tumor cells display lower MHC levels, permittingthem to escape antigenic recognition. In other in-

APC Engineering 5AJPJanuary 1996, Vol. 148, No. 1

stances, MHC-associated tumor antigens arepresent, and yet, there seems to be a deficit in theirimmunogenic capacity. These latter types of tumorcells can be thought of as APCs bearing tumor classpeptide antigens but possessing poor immuno-

genic potential.71-74 In such cases, because thetumor cells are presenting antigen in the absence ofa second costimulator signal, the tumor cells poten-tially can be tolerogenic.75

From the therapeutic perspective, the existence ofthis functional phenotype invites strategies for en-hancing tumor cell immunogenic potential and usingthe engineered tumor cells as cellular cancer vac-cines. Tumor cell engineering applied in this way canbe viewed as a type of APC engineering. The use ofsuch engineered cellular cancer vaccines may beespecially effective early on in the setting of lowtumor burden before the tumor has induced toler-ance.76 In addition to their potential efficacy bythemselves early in disease, cancer vaccines alsoare likely to be useful as adjuvant therapy later indisease, in conjunction with tumor debulking thera-pies or other immunostimulatory regimens that in-voke other antitumor immune effectors (eg, naturalkiller (NK) cells and macrophages).The starting point for most investigators engaged

in engineering immunogenic tumor cells has beensense gene transfer. Gene sequences encoding avariety of proteins with known immunostimulatoryproperties have been introduced into tumor cells toenhance their immunogenic potential. These pro-teins include cytokines (such as granulocyte macro-phage colony-stimulating factor, interleukin (IL)-2,IL-4, IL-6, IL-7, IL-12, interferon (IFN)-y), chemo-kines, cell surface costimulators (such as B7-1, B7-2), heat shock proteins, and MHC molecules.7177 92Considerable data indicate that tumor transfectantsexpressing at least some of these proteins do in factbehave directly as fully functional APCs themselves,triggering activation of T cell subsets.718290

Sense gene transfer is but one tumor cell engi-neering tool. Another involves inhibition of endoge-nous gene function, eg, by antisense gene transfer.In 1993, we reported that endogenous gene inhibi-tion could be used as a path toward engineeringimmunogenic tumor cells. In those first studies,93 94insulin-like growth factor (IGF)-I was the moleculartarget. Antisense inhibition of IGF-I expression inglioblastoma cells enhanced their immunogenicity;the antisense transfectants, when used as a cellularvaccine, elicited a systemic, curative antitumor T cellresponse. In a subsequent study, similar resultswere obtained for teratocarcinoma.95 The impor-tance of the IGF-I axis was further substantiated in a

subsequent study by Baserga and coworkers96 thatdemonstrated enhancement of tumor immunogenic-ity by inhibition of the IGF receptor. Although thebasis for the IGF effect is still unclear, it is becomingapparent that IGF-I is not simply a mitogenic factor intumor cells, but rather may play a distinct, obligatoryrole in malignant transformation.97

While IGF-I is the first target, it certainly will not bethe last. Other candidates would include moleculessuch as cytokines and transcriptional factors thatcan directly or indirectly modulate the transformedand immunogenic phenotypes of tumor cells, or al-ternatively modulate immunosuppressive factorsproduced by tumor cells themselves. A first stepalong these lines is the recent report that antisenseinhibition of transforming growth factor (TGF)-,B, ananti-inflammatory cytokine, can lead to augmentationof the immunogenic potential of a tumor cell.98 Inboth the IGF-I and TGF-,B studies, antisense inhibi-tion of the endogenous genes was achieved by us-ing episomal (extrachromasomally replicating) ex-pression vectors. We had earlier shown the specialutility of these vectors for high level antisense RNAexpression.99-101 Similar results with other factorsare likely to follow, and will undoubtedly call uponadditional genetic engineering tools such as anti-sense oligonucleotides and ribozymes.

Notwithstanding continuing advances in vectorol-ogy, as indicated above, gene transfer itself, whethersense or antisense, has limitations as a tumor cellengineering tool because most primary tumor cellsgrow poorly in culture, and once growing, are poortransfection targets. Consequently, alternatives togene transfer are called for.

Protein Painting with Artificial GPI-ModifiedCostimulatorsCostimulators, such as B7-1 (CD80) and B7-2(CD86), are cell surface proteins of APCs whichprovide critical trans-activating signals to Tcells.75,1021103 Transgenic data indicate that B7 sig-naling via CD28 on T cells constitutes the majorcostimulator pathway for T cell activation. lo4 A seriesof studies have now indicated that gene-transferredcostimulators can function similarly to substantiallyenhance tumor cell immunogenicity.82 91 We52 105and others106 have suggested protein transfer as anexpedient alternative to gene transfer for expressingcostimulators on tumor cell surfaces. In recentlycompleted studies, protein-transferred, GPI-modi-fied B7-1 and B7-2 proteins have been shown toretain the costimulator activity of their native, endo-

6 Tykocinski et alAjP january 1996, Vol. 148, iVA. 1

genously expressed counterparts. This sets thestage for clinical trials in which autochthonous tumorcells obtained via excisional biopsy are modified bypainting them with one or more GPI-modified co-stimulators and then used as cellular immunogens.The set of costimulators that might be painted ontotumor cells could grow to encompass the growinglist of proteins with costimulator functions,75,107 suchas intercellular adhesion molecule (ICAM)-1(CD54)/ICAM-2, vascular cell adhesion molecule-1, lympho-cyte function-associated antigen 3 (CD58), CD40,CD72, and CD24, although for some of these pro-teins, eg, CD58,108 there may also be a potential fornegative signaling. Interestingly, at least two co-stimulators, CD591"9 and heat stable antigen,110 arenaturally GPI-modified. In light of costimulator coop-erativity, 111'112 the simultaneous painting of morethan one costimulator (eg, B7-1 or -2 and CD72) percell could be advantageous. A recent report indi-cated that optimal activation of antitumor cytotoxic Tcells can be attained through simultaneous costimu-lation via the T cell surface proteins CD3, CD28(counter-receptor for B7-1 and B7-2), and CD5(counter-receptor for CD72).1 12 The potential forsynergy through the use of other costimulator com-binations is emerging. 109'111

Protein Painting with a Self-AssociatingCytokine, IL-2

Attempts to harness the potent T cell and NK cellactivating potentials of IL-2 for tumor immunotherapyhave included systemic administration of IL-2 andimmunization with tumor cells after IL-2 gene trans-fer.80 Toxicities associated with administration ofsystemic IL-2 and the technical limitations of genetransfer have restricted the clinical usefulness ofthese two approaches. An alternative means for po-tentially exploiting the immunopotentiating capacityof IL-2 has emerged from our studies of its physio-chemical properties.113-116 IL-2 has the remarkablecapacity to self-associate, ie, to form aggregates ofIL-2 molecules. Such self-association can be docu-mented on solid supports and in situ on cell surfaces.This has suggested the possibility that IL-2 could bepreloaded onto cell surfaces ex vivo, eg, onto sur-faces of tumor cells, and then, for T cell activationpurposes, released from the loaded cells as mono-mers. Preliminary data indicate that such "IL-2 su-percharging" of tumor cells does in fact enhancetheir immunogenicity (D. Kaplan, unpublished obser-vations). It is intriguing to speculate whether othercytokines might share the self-associative propertyof IL-2.

Cellular FusionA straightforward approach for immunogenic tumorcell engineering that is based on cell fusion has beenreported recently.117 The concept is simple: com-bine the tumor antigen repertoire of the tumor cellwith the immunogenic potential of the professionalAPC. According to this approach, tumor cells arefused with syngeneic APCs such as activated pe-ripheral B cells. The fusion products display amerged cell surface phenotype and are able to elicitcurative, systemic antitumor T cell responses. Usinga hepatoma model with this system, it was shownthat tumor cell:activated B cell fusion products couldinduce a curative T cell response. Hence, whereasthe hallmark of protein painting is that it allows forchanging the composition of the cell surface, onemolecule at a time, cell fusion offers the opportunityto do this en masse.

With gene transfer, protein transfer, and cell fusiontools in hand, there are now opportunities for com-binatorial engineering of individual tumor cells. NaiveT cells require signals beyond those of B7 costimu-lators to be fully activated.111'1'9 Recent studiesinvolving simultaneous administration of transfectedtumor cells (B7 transfectants) and exogenous cyto-kines have pointed toward some combinatorial pos-sibilities.90'120 Alternatively, tumor cells engineeredto enhance their immunogenicity in different wayscould be co-administered.121 Optimal combinationsfor specific tumor types should emerge from empir-ical studies.

Artificial Veto Cell Engineering: Remakingthe APC in Other ImagesIn place of augmenting the T cell-activating potentialof APCs, one can consider conferring upon them Tcell inhibitory potential. In this way, one is exploitingthe capacity of the APC to functionally engage anti-gen-specific T cells for inhibitory ends. The notion ofa natural APC capable of inhibiting specific mature Tcells in the periphery has been around for sometime,122`124 and in the literature such cells havebeen designated as "veto cells" because of theircapacity to inhibit specific T cells. However, one cantake this concept a step further. Through APC engi-neering, it should be possible to engineer artificialveto cells (AVCs).One path toward AVC engineering emerged from

a series of studies on the lymphoid cell surface mol-ecule CD8. CD8 traditionally has been thought of asa coreceptor on T cell surfaces that is a key compo-

APC Engineering 7A/P january 1996, Vol. 148, No. 1

nent of their activation pathway; CD8 has also beenconsidered by some to be an "adhesin." However,we,125-129 and subsequently others,u130131 haveshown that CD8 possesses yet another function. Aset of complementary antisense and sense studieshave indicated that when CD8 is present on certainAPC surfaces, it can function as a "coinhibitor." Acoinhibitor has been defined by us as a molecule onthe surface of an APC that induces inhibition of Tcells upon antigenic recognition.129

Although the molecular mechanisms underlyingthe coinhibitor effects of CD8 have not been unrav-eled, one possibility is that CD8 is delivering inhibi-tory signals through class MHC molecules on the Tcells. In past years a series of investigators haveshown that anti-class MHC antibodies can directlyinhibit T cells.132 As a molecule that similarly asso-ciates with class MHC, CD8 could constitute anative functional equivalent of experimental anti-class MHC antibodies. Interestingly, both gene-transferred 127 and protein-transferred (B. Srisuchart,D. R. Kaplan, and M. L. Tykocinski, unpublishedobservation) GPI-modified CD8 retain the coinhibitorfunction of native CD8. These findings provide a pathfor AVC engineering based upon cell-surface paint-ing with GPI-modified CD8. So far, the coinhibitorfunction of CD8 has been studied on only a limitednumber of APC backgrounds. It has been suggestedthat endogenous CD8 on subpopulations of APCssuch as CD8+ dendritic cells133 that bear this sur-face protein might confer veto-like function. The ob-served veto-like activity of CD8+ T cells is consistentwith this idea. 122-124,134-139 By defining the APCbackgrounds on which CD8 can manifest its coin-hibitor function, insights into the interplay betweenthis coinhibitor molecule and various costimulatormolecules should emerge.

It would certainly be useful to identify moleculeswith coinhibitory function in addition to CD8. Onereservoir for such potential coinhibitors are trans-signaling molecules that are known to induce apop-tosis, or programmed cell death,140-142 in T cells. Ofparticular relevance in this regard are moleculessuch as Fas ligand that bind to specific cell deathreceptors on T cells and seem to play a central rolein shaping the T cell repertoire.143 Interestingly, Tcells express both Fas ligand and its receptor, Fas;in fact, Fas ligand expression seems to be ratherrestricted to cells of the T cell lineage, with the high-est levels on CD8+ T cells.144 There is now strongevidence that Fas ligand/Fas receptor-mediated sig-naling is involved in veto-like, fratricidal killing amongactivated T cells which may be important for promot-ing the turnover of activated mature T cells.66'145'146

Another coinhibitor candidate is tumor necrosis fac-tor (TNF), which is not only an inducer of apoptosison its own but also may synergize with Fas ligand inthe induction of apoptosis.147 Hence, Fas ligand andTNF could possibly be used in combination for AVCengineering. Yet another potential source of coinhibi-tors surprisingly could be superantigens. While su-perantigens induce dramatic T cell proliferation invitro, administration of superantigens in vivo resultsfirst in T cell activation and expansion and then inanergy induction and clonal deletion of mature Tcells as a late effect.148'149 Moreover, certain com-binations of superantigen/class 11 molecules can ex-ert a negative effect on T cell proliferation in vitro.150Finally, as the cell surface language between APCsand T cells is further deciphered and as the rules thatdictate induction of anergy, apoptosis, or immuno-logical ignorance151 are defined, additional AVC en-gineering opportunities could emerge. Much is de-pendent on decoding the algorithms of cell surfacecommunication and the underpinnings of APC:T celljuxtacrine signaling.AVC engineering could open up therapeutic op-

portunities for autoimmune diseases. It has beenargued that nonprofessional APCs may contribute topathogenesis in certain disease settings. For exam-ple, in diseases such as thyroiditis32 152 and diabeticinsulitis, class 11 MHC molecules can be expressedon epithelial cells that normally do not express theseproteins and can present antigens to T cells. Theseobservations have led to the concept that aberrantclass 11 MHC expression is important in perpetuatingand perhaps even in initiating autoimmune reac-tions.3267 In principle, such pathogenic nonprofes-sional APCs could be converted into AVCs (either insitu or ex vivo) in order to dampen autoimmune re-sponses.AVC therapeutics potentially could be extended

beyond modulating the predominant af3 T cell sub-classes. The recent finding153 that -y T cells, func-tioning in an NK mode, may be critical mediators ofacute graft-versus-host alloimmune disease makesthem interesting potential AVC targets. Activated a:T cells may be an important APC type responsible fory6 T cell activation15; hence, activated af T cellswould be a relevant APC for reaching out to the y6 Tcell subsets. Moreover, the inhibitory receptors forsome HLA-B and -C proteins present on certain NKand cytolytic T cells potentially offers molecular ac-cess to their inhibitory signaling pathways.154'155AVCs could thus be designed that express ligands(eg, HLA proteins) for inhibitory receptors on -y6 Tcells.

8 Tykocinski et alA/P january 1996, Vol. 148, No. 1

The Painter's Palette

This discussion so far has dealt with some of themore obvious candidates for protein paints. Thesecandidates, including MHC proteins, cell surface co-

stimulators, cell surface-associating cytokines, andcell surface coinhibitors, are drawn from the expand-ing repertoire of molecules that are known to transmitsignals to T cells. However, it is possible to envisiona more extensive palette of protein paints that couldbe used in APC engineering. One such class ofproteins are adhesins. By painting APCs with ad-hesins, one potentially could modify their intercellularbinding properties. Along these lines, we haveshown that a cytokine ligand, such as macrophagecolony-stimulating factor, can function as an artificialadhesin when tethered to cell membranes and pro-

mote adhesion to second cells bearing the corre-

sponding cytokine receptors.156 By introducing suchartificial adhesins, or alternatively natural adhesins,onto APC surfaces, there is the potential for aug-

menting APC:T cell interactions. Another class ofproteins that could be painted onto APCs are selec-tins, which could have potential utility in tailoringcellular trafficking properties in vivo. There are stillother possibilities. Single chain antibodies (incorpo-rating VH and VL domains separated by a flexiblelinker that allows physiological folding of the antigen-binding site), with specificities for different T cell or

other cellular surface molecules, could be producedwith a C-terminal GPI anchor and painted onto APCsto influence their functional properties. This wouldparallel the reported use of palmitate-conjugated an-

tibodies to custom design cytotoxic lymphocytes.55There is also the possibility of modulating APCs withproteins that selectively tap into early and late T cellactivation pathways. It is even possible to envisionpainting APCs with certain immunoprotective mole-cules, such as DAF, which has been shown to inter-fere with NK activity157 as well as with complementC3/C5 amplification convertases. This potentially, incertain contexts, could serve to prolong APC life.APCs could also be painted with viral receptors thatpermit targeting of viral expression vectors to thesecells158; this would represent a merging of proteinand gene transfer tools. We have applied this ap-

proach by painting onto cells GPI-modified CD4,a potential receptor for HIV vectors. Clearly, therational design of APC engineering strategies will becontingent upon further advances in our understand-ing of the interplay among the various signaling sys-

tems of T cells.

Neo-APC Engineering: Creating NewImagesIn designing APC therapeutics, it should in principlebe possible to draw upon a rather broad spectrum ofAPCs. The more obvious APCs that come to mindare the professional ones, with both ex vivo (amplify-ing therapeutic T cells) and in vivo therapeutic appli-cations (inducing or abrogating specific T cell re-sponses) envisioned. Different professional APCscould be used to target distinct T cell subpopula-tions, because data indicate that the optimal APCdiffers for naive versus activated T cells, as well as forvarious T cell subclasses (a,3 CD4+ versus a43 CD8+cells versus 'y T cells; Thl versus Th2 helper Tcells).5'154'159-161 However, the APC therapeutic po-tentia, of other cells, including but not necessarilylimited to those encompassed by the nonprofes-sional APC category, might also be tapped. Such"neo-APCs," with tailored APC regulatory properties,could be used in rather unusual ways.

T cells constitute one nonprofessional APC classfor which there is demonstrated immunotherapeuticpotential. "T cell vaccination," ie, the use of T cells ascellular immunogens, is in essence an APC thera-peutic strategy. The goal of T cell vaccination is toelicit T cell receptor-specific immune responsesagainst pathogenic, autoimmune T cell subpopula-tions. In one instance, investigators have reportedusing T cells with reactivity to the human 65-kd heatshock protein as cellular vaccines to generate ananti-T cell receptor response in NOD diabeticmice.162 Other experimental applications in which Tcell vaccines have been used include autoimmuneencephalomyelitis,163 experimental autoimmune thy-roiditis,164 and adjuvant arthritis.165 There are ofcourse other potential applications for T cells as APCtherapeutics. One entails using T cells as veto cellsfor inhibiting alloreactive and autoreactive T cells,building upon their apparent ability to mediate fratri-cidal T-cell against T-cell killing.66'125-131,135138,166,167 T cells might also prove especially usefulfor stimulating y6 T cells with NK activity.154

Looking beyond hematopoietic APCs, the endo-thelial cell represents another type of cell that couldbe used as a "neo-APC therapeutic." The endothelialcell has direct access to the intravascular circulatingT cell pool. Moreover, it is known that endothelialcells when activated with cytokines can directly stim-ulate T cells. The molecular underpinnings of thisAPC function have been partially elucidated.17-19 Byappropriately engineering endothelial cells in situ, itis conceivable that one could selectively call uponthem to modulate T cells migrating into tissue beds,

APC Engineering 9A/P.Jainuary 1996, Vol. 148, No. 1

eg, activated T cells exiting into sites of acute andchronic inflammation. It is also conceivable that au-tologous endothelial cells could be first engineeredex vivo and then reintroduced into vascular struc-tures in vivo, again with the goal of altering the im-munological properties of the vessel.

Indirect Modulation of APC PhenotypesThe APC engineering process can assume yet otherforms. One interesting approach is through the useof soluble agents such as those cytokines that me-diate their functions via the APC.168'169 Certain pro-fessional APCs, including macrophages and B cells,require activation signals themselves to becomecompetent for activating naive T cells; in fact, restingB cells in the absence of such signals have beenreported to tolerize naive T cells.170 One mechanismfor this activation is that cytokines such as IL-4, IL-2,and IFN-y function to upregulate costimulator ex-pression on such APCs.171'172 In contrast, thepleiotrophic IL-10173'174 and TGF-13175 cytokines canbe immunosuppressive and mediate their activities,at least in part, via down-modulating costimulatorand/or MHC expression. It is thought that such cyto-kine-mediated APC phenotypic modulation can haveimmunopathogenic consequences.174 By introduc-ing a cytokine receptor onto an APC, a cytokinemight be used to modulate an APC even when thecell usually does not express a native receptor forthat cytokine. Such engineering of APCs "from adistance" with the use of cytokines or other agentscould be applied to therapeutic advantage.One intriguing cellular engineering strategy for con-

ferring to APCs responsiveness to cytokines of interestemerged from a recent study by Butcher and cowork-ers.176 This study showed that when the myeloid-specific chemoattractant N-formylpeptide (fMLP) re-ceptor is transfected into B lymphoid cells, fMLP ligandinduces activation of the a4f31 integrin very late activa-tion antigen (VLA)-4 on these cells, enabling binding tovascular cell adhesion molecule-1 on other cells andintercellular adhesion. This provides an experimentalpath for enabling targeted adhesion and migration inresponse to locally administered ligands. Such ligand-specific triggering of integrin-mediated adhesion con-ferred by transfection of a chemoattractant receptorcould, in principle, be extended to APCs other than Bcells.

Addition of other agents to APCs could be used tomask or passively block APC functions. One exam-ple of this is the use of agents designed to interferewith APC:T cell costimulator pathways. Interestingly,

there is evidence showing that islet cell allograftscan be coated ex vivo with soluble CTLA-4 *Ig, pre-sumably masking B7 proteins on graft-associatedAPCs in the process and thereby promoting islet cellgraft tolerance.177 This finding builds on a body ofdata indicating that in vivo administration of solubi-lized derivatives of costimulator receptors such asCTLA-4 Ig can promote T cell anergy and interferewith transplant rejection.75178 In principle, suchblockers could even be produced within the APCsthemselves. Furthermore, it may be feasible to tailormore compact versions of these APC masking/blocking agents. Because many of the costimulatorsand their receptors are immunoglobulin supergenefamily (IgSF) proteins, their functional IgSF domainscould be produced and used as independent func-tional units. Our recent data showing that the IgSFdomains of class MHC and CD8 proteins can func-tion as isolated units suggest that IgSF protein engi-neering of this sort is indeed feasible.179'180

Thinking Beyond

The enormous potential of gene therapy has yet tobe realized. As the optimal applications for genetherapy are defined, important niches for non-gene-based cellular and tissue engineering strategies willemerge. Protein transfer, cell fusion, and similar al-ternative tools are likely to occupy important placesin this enlarged molecular toolbox. As discussed,these tools are likely to be relevant to cellular thera-peutics that go beyond engineered APCs. The ap-plication of these tools need not be limited to the exvivo environment. In vivo, cell- or tissue-specific tar-geting strategies may be possible. This might beaccomplished by delivering GPI-anchored proteinslocally via liposomes or other carriers.

From its origins in erythrocyte and platelet trans-fusion, and later hematopoietic stem cell transfer andtumor-infiltrating lymphocyte/lymphokine-activatedkiller tumor immunotherapy, cellular therapeuticsnow extends to other modes of cell therapy. APCtherapy is an example of a class of cell therapyinvolving a regulatory cell. This conceptual paradigmof "regulatory cell transfer" could certainly be ex-tended beyond the immunohematopoietic realm.

Experimental pathologists have traditionally beenobservers of cellular phenotypes and dissectors ofpathogenetic mechanisms. With the emergence oftechnologies such as those described above, thereis now the opportunity for them to learn how to be-come modulators of cellular phenotypes on the roadtoward innovative therapeutics.

10 Tykocinski et alA/P Janiarlv 1996. Vol. 148. No. I

AcknowledgmentsThe authors acknowledge the members of our labo-ratories, past and present, whose dedicated effortshave contributed to the findings described in thisarticle. We appreciate the support and advice ofMichael Lamm, James R. Bell Jr., Frank Chisari,Thomas Kindt, and Victor Nussenzweig over theyears. We thank Paul Lehmann for critical reading ofthe manuscript and Susan Brill for her expert secre-tarial assistance in manuscript preparation. Finally,MLT extends special thanks to Tobi Kahn for hisfriendship and his artistic visions of "cellular biomor-phisms."

References

1. Unanue ER, Allen PM: The basis for the immunoregu-latory role of macrophages and other accessory cells.Science 1987, 236:551-557

2. Steinman RM: The dendritic cell system and its role inimmunogenicity. Annu Rev Immunol 1991, 9:271-296

3. Teunissen MB: Dynamic nature and function of epi-dermal Langerhans cells in vivo and in vitro: a review,with emphasis on human Langerhans cells. Histo-chem J 1992, 24:697-716

4. Chestnut RW, Grey HM: Studies on the capacity of Bcells to serve as antigen-presenting cells. J Immunol1981, 126:1075-1079

5. Constant S, Schweitzer N, West J, Ranney P, BottomlyK: B lymphocytes can be competent antigen-present-ing cells for priming CD4+ T cells to protein antigensin vivo. J Immunol 1995, 155:3734-3741

6. Borrello MA, Phipps RP: Fibroblasts support out-growth of splenocytes simultaneously expressing Blymphocyte and macrophage characteristics. J lmmu-nol 1995, 15:4155-4161

7. Rott C, Tontsch U, Fleischer B: Dissociation of anti-gen-presenting capacity of astrocytes for peptide-an-tigens versus superantigens. J Immunol 1993, 150:87-95

8. Murray AG, Libby P, Pober JS: Human vascularsmooth muscle cells poorly co-stimulate and activelyinhibit allogeneic CD4+ T cell proliferation in vitro. JImmunol 1995, 154:151-161

9. Hughes CCW, Savage COS, Pober JS: Endothelialcells augment T cell interleukin-2 production by acontact-dependent mechanism involving CD2/LFA-3interaction. J Exp Med 1990, 171:1453-1467

10. Kosaka H, Surh CD, Sprent J: Stimulation of matureunprimed CD8+ T cells by semiprofessional antigen-presenting cells in vivo. J Exp Med 1992, 176:1291-1302

11. Goebels N, Michaelis D, Wekerle H, Hohlfeld R: Hu-man myoblasts as antigen-presenting cells. J Immu-nol 1992, 149:661-667

12. Del Pozo V, De Andres B, Martin E, Cardaba B, Fer-

nandez JC, Gallardo S, Tramon P, Leyva-Cobiari F,Palomino P, Lahoz C: Eosinophil as antigen-present-ing cell: activation of T cell clones and T cell hybrido-mas by eosinophils after antigen processing. Eur JImmunol 1992, 22:1919-1925

13. Hewitt CRA, Feldman M: Human T cell clones presentantigen. J Immunol 1989, 142:1429-1436

14. Sidhu S, Deacock S, Bal V, Batchelor JR, Lombardi G,Lechler RI: Human T cells cannot act as autonomousantigen-presenting cells, but induce tolerance in an-tigen-specific, and alloreacrive responder cells. J ExpMed 1992, 176:875-880

15. Spaner D, Cohen BL, Miller RG, Phillips RA: Antigen-presenting cells for naive transgenic y5 T cells. Potentactivation by activated ac4 T cells. J Immunol 1995,155:3866-3876

16. Lorenz RG, Allen PM: Thymic cortical epithelial cellscan present self-antigens in vivo. Nature 1989, 337:560-562

17. Pober JS, Gimbrone MA, Cotran RS, Reiss CS, Bura-koff SJ, Fiers W, Ault KA: la expression by vascularendothelium is inducible by activated T cells and hu-man y interferon. J Exp Med 1983, 157:1339-1353

18. Geppert TD, Lipsky PE: Antigen presentation by inter-feron-y-treated endothelial cells and fibroblasts: dif-ferential ability to function as antigen-presenting cellsdespite comparable la expression. J Immunol 1985,135:3750-3762

19. St. Louis JD, Lederer JA, Lichtman AH: Costimulatordeficient antigen presentation by an endothelial cellline induces a nonproliferative T cell activation re-sponse without anergy. J Exp Med 1993, 178:1597-1605

20. Bal V, Mclndoe A, Denton G, Hudson D, Lombardi G,Lamb J, Lechler R: Antigen presentation by keratino-cytes induces tolerance in human T cells. Eur J Im-munol 1990, 20:1893-1897

21. Barker JN, Mitra RS, Griffiths CE, Dixit VM, NickoloffBJ: Keratinocytes as initiators of inflammation. Lancet1991, 337(8735):211-214

22. Fierz W, Endler B, Reske K, Wekerle H, Fontana A:Astrocytes as antigen presenting cells. I. Induction ofla antigen expression on astrocytes by T cells viaimmune interferon and its effect on antigen presenta-tion. J Immunol 1985, 134:3785-3793

23. Takiguchi M, Frelinger JA: Induction of antigen pre-sentation ability in purified cultures of astroglia byinterferon-y. J Mol Cell Immunol 1986, 2:269-280

24. Sedgwick J, Mossner R, Schwender D, Ter Meulen V:Major histocompatibility complex-expressing nonhe-matopoietic astroglial cells prime only CD8 Tlymphocytes: astroglial cells as perpetuators but notinitiators of CD4 T cell responses in the central ner-vous system. J Exp Med 1991, 173:1235-1246

25. Frei K, Siepl C, Groscurth P, Bodmer S, Schwerdel C,Fontana A: Antigen presentation and tumor cytotoxic-ity by interferon-y treated microglial cells. Eur J Im-munol 1987, 17:1271-1278

APC Engineering 1 1AJPJanuary 1996, Vol. 148, No. 1

26. Hickey WF, Kimura H: Perivascular microglial cells ofthe CNS are bone-marrow derived and present anti-gen in vivo. Science 1988, 239:290-292

27. Zhang Y, Porter S, Wekerle H: Schwann cells andmyasthenia gravis: preferential uptake of soluble andmembrane bound AchR by normal and immortalizedSchwann cells, and immunogenic presentation toAchR-specific T line lymphocytes. Am J Pathol 1990,136:111-122

28. Percopo CM, Hooks JJ, Shinohara T, Caspi R, DetrickB: Cytokine-mediated activation of a neuronal retinalresident cell provokes antigen presentation. J Immu-nol 1990, 145:4101-4107

29. Fabry Z, Waldschmidt MM, Van Dyk L, Moore SA, HartMN: Activation of CD4+ lymphocytes by syngeneicbrain microvascular smooth muscle cells. J Immunol1990, 145:1099-1104

30. Alsalameh S, Jahn B, Krause A, Kalden JR, BurmesterGR: Antigenicity and accessory cell function of humanarticular chondrocytes. J Rheumatol 1991, 18:414-421

31. Bland PW, Whiting CV: Antigen processing by iso-lated rat intestinal villus enterocytes. Immunology1989, 68:497-502

32. Bottazzo GF, Pujol-Borrell R, Hanafusa T, FeldmannM: Role of aberrant HLA-DR expression and antigenpresentation in induction of endocrine autoimmunity.Lancet 1983, 2(8359):1115-1119

33. Hagerty DT, Evavold BD, Allen PM: Regulation of thecostimulator B7, not class 11 major histocompatibilitycomplex, restricts the ability of murine kidney tubulecells to stimulate CD4+ T cells. J Clin Invest 1994,93:1208-1215

34. Kundig TM, Bachmann MF, DiPaolo C, Simard JJ,Battegay M, Lother H, Gessner A, KuhIcke K, OhashiPS, Hengartner H, Zinkernagel RM: Fibroblasts asefficient antigen-presenting cells in lymphoid organs.Science 1995, 268:1343-1349

35. Marrack P, Kappler J: The staphylococcal enterotox-ins and their relatives. Science 1990, 248:705-711

36. Beckman EM, Porcelli SA, Morita CT, Behar SM, Fur-long ST, Brenner MB: Recognition of a lipid antigen byCD1-restricted ag+ T cells. Nature 1994, 372:691-694

37. Sieling PA, Chatterjee D, Porcelli SA, Prigozy TI, So-riano T, Brenner MB, Kronenberg M, Brennan PJ,Modlin RL: CD1-restricted T-cell recognition of micro-bial lipoglycan antigens. Science 1995, 269:227-230

38. Medof ME, Kinoshita T, Nussenzweig V: Inhibition ofcomplement activation on the surface of cells afterincorporation of decay-accelerating factor (DAF) intotheir membranes. J Exp Med 1984, 160:1558-1578

39. Lederman MM, Purvis SF, Walter El, Carey JT, MedofME: Heightened complement sensitivity of acquiredimmunodeficiency syndrome lymphocytes related todiminished expression of DAF. Proc NatI Acad SciUSA 1989, 86:4205-4209

40. Ferguson MAJ, Williams AF: Cell-surface anchoring of

proteins via glycosyl-phosphatidylinositol structures.Annu Rev Biochem 1988, 57:285-320

41. Thomas JR, Dwek RA, Rademacher TW: Structure,biosynthesis, and function of glycosylphosphatidyli-nositols. Biochemistry 1990, 29:5413-5422

42. Englund PT: The structure and biosynthesis of glyco-syl phosphatidylinositol protein anchors. Annu RevBiochem 1993, 62:121-138

43. Medof ME, Nagarajan S, Tykocinski ML: Cell surfaceengineering with GPI-anchored proteins. FASEB J1996 (in press)

44. Medof ME, Walter El, Rutgers JL, Knowles D, Nussen-zweig V: Identification of the decay-accelerating fac-tor (DAF) on epithelial glandular cells and in bodyfluids. J Exp Med 1987, 165:848-864

45. Walter El, Roberts WL, Rosenberry TL, Ratnoff WD,Medof ME: Structural basis for variations in the sensi-tivity of human decay-accelerating factor to phosphoi-nositol-specific phospholipase C cleavage. J Immu-nol 1990, 144:1030-1036

46. Roberts WL, Myher JJ, Kukis A, Low MG, RosenberryTL: Lipid analysis of the glycoinositol phospholipidmembrane anchor of human erythrocyte acetylcho-linesterase. Palmitoylation of inositol results in resis-tance to phosphoinositol-specific phospholipase C. JBiol Chem 1988, 263:18766-18775

47. Walter El, Ratnoff WD, Long KE, Kazura JW, MedofME: Effect of glycoinositolphospholipid anchor lipidgroups on functional properties of decay-acceratingfactor protein in cells. J Biol Chem 1992, 261:356-362

48. Medof ME, Lublin DM, Holers VM, Ayers DJ, Getty RR,Leykam JF, Atkinson JP, Tykocinski ML: Cloning andcharacterization of cDNAs encoding the complete se-quence of decay-accelerating factor of human com-plement. Proc Natl Acad Sci USA 1987, 84:2007-2011

49. Caras IW, Davitz MA, Rhee L, Weddell G, Martin DW,Nussenzweig V: Cloning of decay-accelerating factorsuggests novel use of splicing to generate two pro-teins. Nature 1987, 325:545-549

50. Tykocinski ML, Shu HK, Ayers DJ, Walter El, Getty RR,Groger RK, Hauer CA, Medof ME: Glycolipid rean-choring of T-lymphocyte surface antigen CD8 usingthe 3' end sequence of decay-accelerating factor'smRNA. Proc Natl Acad Sci USA 1988, 85:3555-3559

51. Caras IW, Weddell GN, Davitz MA, Nussenzweig V,Martin Jr DW: Signal for attachment of a phospholipidmembrane anchor in decay accelerating factor. Sci-ence 1987, 238:1280-1283

52. Huang J-H, Getty RR, Chisari FV, Fowler P, Green-span NS, Tykocinski ML: Protein transfer of preformedMHC-peptide complexes sensitizes target cells to Tcell cytolysis. Immunity 1994, 1:607-613

53. Davis SJ, Ward HA, Puklavec MJ, Willis AC, WilliamsAF, Barclay AN: High level expression in Chinesehamster ovary cells of soluble forms of CD4 T lympho-cyte glycoprotein including glycosylation variants. JBiol Chem 1990, 265:10410-10418

12 Tykocinski et alA/P,january 1996, Vol. 148, No. 1

54. Sherr DH, Cheung N-K, Heghinian KM, Benacerraf B,Dorf ME: Immune suppression in vivo with antigen-modified syngeneic cells. II. T cell-mediated nonre-sponsiveness to fowl -y-globulin. J Immunol 1979,122:1899-1904

55. Colsky AS, Stein-Streilein J, Peacock JS: Surrogatereceptor-mediated cellular cytotoxicity: a method for"custom designing" killer cells of desired target spec-ificity. J Immunol 1988, 140:2515-2519

56. Colsky AS, Peacock JS: Palmitate-derivatized anti-bodies can function as surrogate receptors for medi-ating specific cell-cell interactions. J Immunol Meth-ods 1989, 124:179-187

57. Kim SA, Peacock JS: The use of palmitate-conjugatedprotein A for coating cells with artificial receptorswhich facilitate intercellular interactions. J ImmunolMethods 1993, 158:57-65

58. Wilcox CA, Olson EN: The majority of cellular fatty acidacylated proteins are localized to the cytoplasmicsurface of the plasma membrane. Biochemistry 1987,26:1029-1036

59. Lehmann PV, Sercarz EE, Forsthuber T, Dayan CM,Gammon G: Determinant spreading and the dynam-ics of the autoimmune T cell repertoire. Immunol To-day 1993, 14(5):203-208

60. Sercarz EE, Lehmann PV, Ametani A, Benichou G,Miller, A, Moudgil K: Dominance and crypticity of Tcell antigenic determinants. Annu Rev Immunol 1993,11:729-766

61. Coeshott C, Grey HM: Transfer of antigen-presentingcapacity to la-negative cells upon fusion with la-bear-ing liposomes. J Immunol 1985, 134:1343-1348

62. Huang JH, Greenspan NS, Tykocinski ML: Alloanti-genic recognition of artificial glycosylphosphatidyli-nositol-anchored HLA-A2.1. Mol Immunol 1994, 31:1017-1028

63. Mottez E, Langlade-Demoyen P, Gournier H, MartinonF, Maryanski J, Kourilsky P, Abastado J-P: Cells ex-pressing a major histocompatibility complex classmolecule with a single covalently bound peptide arehighly immunogenic. J Exp Med 1995, 181 :493-502

64. Lee L, McHugh L, Ribaudo RK, Kozlowski S, Margu-lies DH, Mage MG: Functional cell surface expressionby a recombinant single-chain class major histocom-patibility complex molecule with a cis-active ,3 2-mi-croglobulin domain. Eur J Immunol 1994, 24:2633-2639

65. Ferber I, Schbnrich G, Schenkel J, Mellor AL, Ham-merling GJ, Arnold B: Levels of peripheral T cell tol-erance induced by different doses of tolerogen. Sci-ence 1994, 263:674-676

66. Pelfrey CM, Tranquill LR, Boehme SA, McFarland HF,Lenardo MJ: Two mechanisms of antigen-specific ap-optosis of myelin basic protein (MBP)-specific T lym-phocytes derived from multiple sclerosis patients andnormal individuals. J Immunol 1995, 154:6191-6202

67. Eberl G, Roggero MA, Corradin G: A simple mathe-

matical model for the functional peptide/MHC/TCRinteractions. J Immunol 1995, 154:219-225

68. Nayersina R, Fowler P, Guilhot S, Missale G, Cerny A,Schlicht H-J, Vitiello A, Chestnut R, Person JL, Re-deker AG, Chisari FV: HLA A2 restricted cytotoxic Tlymphocyte responses to multiple hepatitis B surfaceantigen epitopes during hepatitis B virus infection. JImmunol 1993, 150:4659-4671

69. Kawakami Y, Eliyahu S, Sakaguchi K, Robbins PF,Rivoltini L, Yannelli JR, Appella E, Rosenberg SA:Identification of the immunodominant peptides of theMART-1 human melanoma antigen recognized by themajority of HLA-A2-restricted tumor infiltrating lym-phocytes. J Exp Med 1994, 180:347-352

70. Peoples GE, Goedegebuure PS, Smith R, LihehanDC, Yoshino I, Eberlein TJ: Breast and ovarian can-cer-specific cytotoxic T lymphocytes recognize thesame HER2/neu-derived peptide. Proc Natl Acad SciUSA 1995, 92:432-436

71. Ostrand-Rosenberg S: Tumor immunotherapy: the tu-mor cell as an antigen-presenting cell. Curr Opin Im-munol 1994, 6:722-727

72. Houghton AN: Cancer antigens: immune recognitionof self and altered self. J Exp Med 1994, 180:1-4

73. Boon T, Cerottini J: Van den Eynde B, van der Brug-gen P, Van Pel A: Tumor antigens recognized by Tlymphocytes. Annu Rev Immunol 1994, 12:337-365

74. Lotze MT, Finn OJ: Cellular immunity and the immu-notherapy of cancer. J Immunotherapy 1993, 14:79-87

75. Boussiotis VA, Gribben JG, Freeman GJ, Nadler LM:Blockade of the CD28 co-stimulatory pathway: ameans to induce tolerance. Curr Opin Immunol 1994,6:797-807

76. Koeppen H, Singh S, Schreiber H: Genetically engi-neered vaccines. Comparison of active versus pas-sive immunotherapy against solid tumors. Ann NYAcad Sci 1993, 690:244-255

77. Pardoll D: New strategies for active immunotherapywith genetically engineered tumor cells. Curr OpinImmunol 1992, 4:619-623

78. Blankenstein T, Rowley DA, Schreiber H: Cytokinesand cancer: experimental systems. Curr Opin Immu-nol 1991, 3:694-698

79. Colombo M, Forni G: Cytokine gene transfer in tumorinhibition and tumor therapy: where are we now? Im-munol Today 1994, 15:48-51

80. Fearon ER, Pardoll DM, Itaya T, Golumbek P, LevitskyHI, Simons JW, Karasuyama H, Vogelstein B, Frost P:Interleukin-2 production by tumor cells bypasses Thelper function in the generation of an immune re-sponse. Cell 1990, 60:397-403

81. Zitvogel L, Tahara H, Robbins PD, Storkus WJ, ClarkeMR, Nalesnik MA, Lotze MT: Cancer immunotherapyof established tumors with IL-12: effective delivery bygenetically engineered fibroblasts. J Immunol 1995,155:1393-1 403

82. Chen L, Ashe S, Brady WA, Hellstrom I, Hellstrom KE,

APC Engineering 13AJPJanuary 1996, Vol. 148, No. 1

Ledbetter JA, McGowan P, Linsley PS: Costimulationof antitumor immunity by the B7 counterreceptor forthe T-lymphocyte molecules CD28 and CTLA-4. Cell1992, 71:1093-1102

83. Baskar S, Ostrand-Rosenberg S, Nabavi N, Nadler L,Freeman G, Glimcher L: Constitutive expression of B7restores immunogenicity of tumor cells expressingtruncated major histocompatibility complex class 11molecules. Proc Natl Acad Sci USA 1993, 90:5687-5690

84. Townsend SE, Allison JP: Tumor rejection after directcostimulation of CD8+ T cells by B7-transfected mel-anoma cells. Science 1993, 259:368-370

85. Townsend, SE, Su FW, Atherton JM, Allison JP: Spec-ificity and longevity of antitumor immune responsesinduced by B7-transfected tumors. Cancer Res 1994,54:6477-6483

86. Chen L, McGowan P, Ashe S, Johnston J, Li Y, Hell-strom 1, Hellstrom K: Tumor immunogenicity deter-mine the effect of B7 costimulation on T cell mediatedtumor immunity. J Exp Med 1994, 179:523-532

87. Li Y, McGowan P, Hellstrom L, Hellstrom K, Chen L:Costimulation of tumor-reactive CD4+ and CD8+ Tlymphocytes by B7, a natural ligand for CD28, can beutilized to treat established mouse melanoma. J Im-munol 1994, 153:421-428 ,

88. Hodge JW, Abrams S, Schlom J, Kantor JA: Inductionof antitumor immunity by recombinant vaccinia vi-ruses expressing 87-1 or 87-2 costimulatory mole-cules. Cancer Res 1994, 54:5552-5555

89. Ramarathinam L, Castle M, Wu Y, Liu Y: T cell co-stimulation B7/BB1 induces CD8 T cell-dependenttumor rejection: an important role of 87/881 in theinduction, recruitment, and effector function of antitu-mor T cells. J Exp Med 1994, 179:1205-1214

90. Yang G, Hellstrom KE, Mizuno MT, Chen L: In vitropriming of tumor-reactive cytolytic T lymphocytes bycombining IL-10 with B7-CD28 costimulation. J Immu-nol 1995, 155:3897-3903

91. Yang G, Hellstrom KE, Hellstrom I, Chen L: Antitumorimmunity elicited by tumor cells transfected with87-2, a second ligand for CD28/CTLA-4 costimula-tory molecules. J Immunol 1995, 154:2794-2800

92. Lukacs KV, Lowrie DB, Stokes RW, Colston MJ: Tumorcells transfected with a bacterial heat-shock genelose tumorigenicity and induce protection against tu-mors. J Exp Med 1993, 178:343-348

93. Trojan J, Blossey BK, Rudin SD, Tykocinski ML, Ilan J:Loss of tumorigenicity of rat glioblastoma directed byepisome-based antisense cDNA transcription of insu-lin-like growth factor 1. Proc NatI Acad Sci USA 1992,89:4874-4878

94. Trojan J, Johnson TR, Rudin SD, Ilan J, Tykocinski ML,Ilan J: Treatment and prevention of rat glioblastomaby immunogenic C6 cells expressing antisense IGF-1RNA. Science 1993, 259:94-97

95. Trojan J, Johnson TR, Rudin SD, Blossey BK, KelleyKM, Shevelev A, Abdul-Karim FW, Anthony DD, Tyko-

cinski ML, Ilan J, Ilan JO: Gene therapy of murineteratocarcinoma: separate functions for IGF-I andIGF-ll in immunogenicity and differentiation. Proc NatlAcad Sci USA 1994, 91:6088-6092

96. Resnicoff M, Sell C, Rubini M, Coppola D, Ambrose D,Baserga R, Rubin R: Rat glioblastoma cells express-ing an antisense RNA to the insulin-like growth fac-tor-1 (IGF-1) receptor are nontumorigenic and induceregression of wild-type tumors. Cancer Res 1994, 54:2218-2222

97. Baserga R: The insulin-like growth factor receptor: akey to tumor growth? Cancer Res 1995, 55:249-252

98. Dorigo 0, Shawler DL, Royston I, Sobol RE, Fahrai H:Transforming growth factor ,B (TGFj3) inhibits effect ofIL-2 gene therapy in TGF,B producing tumors (SanDiego, CA). AACR Eighty-sixth Annual Meeting,March 18-22, 1995, Toronto, Ontario, Canada

99. Hauer C, Getty RR, Tykocinski ML: Epstein-Barr virusepisome-based expression in human myeloid leuke-mia cells. Nucleic Acids Res 1989, 17:1989-2003

100. Groger RK, Morrow D, Tykocinski ML: Directional an-tisense and sense cDNA cloning using Epstein-Barrvirus episomal expression vectors. Gene 1989, 81:285-294

101. Hambor JE, Hauer CA, Shu HK, Groger RK, KaplanDR, Tykocinski ML: Use of an EBV virus episomalreplication for antisense RNA-mediated gene inhibi-tion in human cytotoxic T cell clone. Proc Natl AcadSci USA 1988, 85:4010-4014

102. Liu Y, Linsley P: Costimulation of cell growth. CurrOpin Immunol 1992, 4:265-270

103. Bellone M, lezzi G, Manfredi AA, Protti MP, DellabonaP, Casorati G, Rugarli C: In vitro priming of cytotoxic Tlymphocytes against poorly immunogenic epitopesby engineered antigen-presenting cells. Eur J Immu-nol 1994, 24:2691-2698

104. Green, JM, Noel PJ, Sperling Al, Walunas TL, GrayGS, Bluestone JA, Thompson CB: Absence of B7-dependent responses in CD28-deficient mice. Immu-nity 1994, 1:501-508

105. Brunschwig EB, Levine E, Trefzer U, Tykocinski ML:Glycosylphosphatidylinositol-modified murine B7-1and B7-2 retain costimulator function. J Immunol1995, 155:5498-5505

106. McHugh RS, Ahmed SN, Wang Y-C, Sell KW, SelvarajP: Construction, purification, and functional incorpo-ration on tumor cells of glycolipid-anchored humanB7-1 (CD80). Proc Natl Acad Sci USA 1995, 92:8059-8063

107. Cayabyab M, Phillips JH, Lanier LL: CD40 preferen-tially costimulates activation of CD4+ T lymphocytes.J Immunol 1994, 152:1523-1531

108. Bell GM, Imboden JB: CD2, and the regulation of Tcell anergy. J Immunol 1995, 155:2805-2807

109. Menu E, Tsai BC, Bothwell ALM, Sims PJ, Bierer BE:CD59 costimulation of T cell activation. CD58 depen-dence, and requirement for glycosylation. J Immunol1994, 153:2444-2456

14 Tykocinski et alA/P January 1996, Vrol. 148, iVo. 1

110. Liu Y, Jones B, Aruffo A, Sullivan KM, Linsley PS,Janeway CA Jr: Heat-stable antigen is a costimulatorymolecule for CD4 T cell growth. J Exp Med 1992,175:437-445

111. Parra E, Wingren AG, Hedlund G, Bjorklund M,Sjogren H-O, Kalland T, Sansom D, Dohlsten M: Co-stimulation of human CD4+ T lymphocytes with B7and lymphocyte function-associated antigen-3 resultsin distinct cell activation profiles. J Immunol 1994,153:2479-2487

112. Kroesen B-J, Bakker A, van Lier RAW, The HT, de LeijL: Bispecific antibody-mediated target cell-specificcostimulation of resting T cells via CD5 and CD28.Cancer Res 1995, 55:4409-4415

113. Kaplan DR: Solid phase interleukin 2. Mol Immunol1991, 28:1255-1261

114. Bergman CA, Gould DL, Kaplan DR: Cytometricallydetected specific binding of interleukin 2 to cells.Cytokine 1992, 4:192-200

115. Kaplan DR: Delivery of interleukin 2 for immunother-apy. J Chromatogr Biomed Appl 1994, 662:315-323

116. Kaplan DR, Smith D, Huang R, Yildirim Z: Self-asso-ciation of interleukin 2 bound to its receptor. FASEB J1995, 9:1096-1102

117. GuoY, Wu M, Chen H, Wang X, Liu G, Li G, MaJ, SyM-S: Effective tumor vaccine generated by fusion ofhepatoma cells with activated B cells. Science 1994,263:518-520

118. Kubin M, Kamoun M, Trinchieri G: Interleukin 12 syn-ergizes with B7/CD28 interaction in inducing efficientproliferation, and cytokine production of human Tcells. J Exp Med 1994, 180:211-222

119. Murphy EE, Terres G, Macatonia SE, Hseih C, Matt-son J, Lanier L, Wysocka M, Trinchieri G, Murphy K,O'Garra A: B7 and IL-12 cooperate for proliferation,and IFN--y production by mouse helper T clones thatare unresponsive to B7 costimulation. J Exp Med1994, 180:223-231

120. Coughlin CM, Wysocka M, Kurzawa HL, Lee WMF,Trinchieri G, Eck SL: B7-1, and interleukin 12 syner-gistically induce effective antitumor immunity. CancerRes 1995, 55:4980-4987

121. Porgador A, Tzehoval E, Vadai E, Feldman M, Eisen-bach L: Combined vaccination with major histocom-patibility class and interleukin 2 gene-transducedmelanoma cells synergizes the cure of postsurgicalestablished lung metastases. Cancer Res 1995, 55:4941-4949

122. Martin DR, Miller RG: In vivo administration of histoin-compatible lymphocytes leads to rapid functional de-letion of cytotoxic T lymphocyte precursors. J ExpMed 1989, 170:679-690

123. Fink PJ, Shimenkovitz RP, Bevan MJ: Veto cells. AnnuRev Immunol 1988, 6:115-137

124. Rammensee H-G: Veto function in vitro and in vivo.Intern Rev Immunol 1989, 4:175-191

125. Kaplan DR, Hambor JE, Tykocinski ML: An immuno-

regulatory function for the CD8 molecule. Proc NatlAcad Sci USA 1989, 86:8512-8515

126. Hambor JE, Weber MC, Tykocinski ML, Kaplan DR:Regulation of allogeneic responses by expression ofCD8 a chain on stimulator cells. Int Immunol 1990,2:879-883

127. Hambor JE, Kaplan DR, Tykocinski ML: CD8 functionsas an inhibitory ligand in mediating the immunoregu-latory activity of CD8+ cells. J Immunol 1990, 145:1646-1652

128. Tykocinski ML, Kaplan DR: A multifunctional perspec-tive of the CD8 molecule. NK Cell MediatedCytotoxicity: Receptors, Signaling and Mechanisms.Edited by E Lotzova, RB Herberman. Ann Arbor, CRCPress, 1992, pp 393-408

129. Tykocinski ML, Kaplan DR: CD8-dependentimmunoregulation: prospects for anti-rejection thera-pies based upon CD8 modification of alloantigen-presenting cells. Kidney Int 1993, 43(Suppl 39):S120-S123

130. Sambhara SR, Miller RG: Programmed cell death of Tcells signaled through the T cell receptor and the a3domain of class MHC. Science 1991, 252:1424-1427

131. Zhang L, Shannon J, Sheldon J, Teh H-S, Mak TW,Miller RG: Role of infused CD8+ cells in the inductionof peripheral tolerance. J Immunol 1994, 152:2222-2228

132. Smith DM, Bluestone JA, Jeyarajah DR, Newberg MH,Engelhard VH, Thistlethwaite JR, Woodle ES: Inhibi-tion of T cell activation by a monoclonal antibodyreactive against the a3 domain of human MHC classmolecules. J Immunol 1994, 153:1054-1067

133. Vremec D, Zorbas M, Scollay R, Saunders DJ, Ar-davin CF, Wu L, Shortman K: The surface phenotypeof dendritic cells purified from mouse thymus andspleen: investigation of the CD8 expression by a sub-population of dendritic cells. J Exp Med 1992, 176:47-58

134. Fast LD: Generation and characterization of IL-2-ac-tivated veto cells. J Immunol 1992, 149:1510-1515

135. Vitiello A, Heath WR, Sherman LA: Consequences ofself-presentation of peptide antigen by cytolytic Tlymphocytes. J Immunol 1989, 143:1512-1517

136. Webb SR, Sprent J: Induction of neonatal tolerance toMIsa antigens by CD8+ T cells. Science 1990, 248:1643-1646

137. Martin PJ: Donor CD8 cells prevent allogeneic mar-row graft rejection in mice: potential implications formarrow transplantation in humans. J Exp Med 1993,178:703-712

138. Su M, Walden PR, Eisen HN: Cognate peptide-in-duced destruction of CD8+ cytotoxic T lymphocytesis due to fratricide. J Immunol 1993, 151:658-667

139. Kaufman CL, Colson YL, Wren SM, Watkins S, Sim-mons RL, lldstad ST: Phenotypic characterization of anovel bone marrow-derived cell that facilitates en-

APC Engineering 15A/P january 1996, Vol. 148, No. 1

graftment of allogeneic bone marrow stem cells.Blood 1994, 84:2436-2446

140. Critchfield JM, Boehme SA, Lenardo MJ: The regula-tion of antigen-induced mature T cell apoptosis. Ap-optosis and the Immune Response. Edited by CGregory. New York, John Wiley & Sons, 1994, pp55-114

141. Majno G, Joris I: Apoptosis, oncosis, and necrosis: anoverview of cell death. Am J Pathol 1995, 146:3-15

142. Steller H: Mechanisms and genes of cellular suicide.Science 1995, 267:1445-1449

143. Nagata S, Goldstein P: The Fas death factor. Science1995, 267:1449-1456

144. Suda T, Okazaki T, Naito Y, Yokota T, Arai N, Ozaki S,Nakao K, Nagata S: Expression of the Fas ligand incells of T cell lineage. J Immunol 1995, 154:3806-3813

145. Russell JH, Rush B, Weaver C, Wang R: Mature Tcells of autoimmune Iprllpr mice have a defect inantigen-stimulated suicide. Proc Natl Acad Sci USA1993, 90:4409-4413

146. Ettinger R, Panka DJ, Wang JKM, Stanger BZ, Ju S-T,Marshak-Rothstein A: Fas ligand-mediated cytotoxic-ity is directly responsible for apoptosis of normalCD4+ T cells responding to a bacterial superantigen.J Immunol 1995, 154:4302-4308

147. Wong GHW, Goeddel DV: Fas antigen and p55 TNFreceptor signal apoptosis through distinct pathways.J Immunol 1994, 152:1751-1755

148. Waanders GA, Shakhov AN, Held W, Karapetian 0,Acha-Orbea H, MacDonald HR: Peripheral T cell ac-tivation, and deletion induced by transfer of lympho-cyte subsets expressing endogenous or exogenousmouse mammary tumor virus. J Exp Med 1993, 177:1359-1366

149. Webb SR, Hutchinson J, Hayden K, Sprent J: Expan-sion/deletion of mature T cells exposed to endoge-nous superantigens in vivo. J Immunol 1994,152:586-597

150. Ben-Nun A: Staphylococcal enterotoxin B as a potentsuppressant of T cell proliferative responses in rats.Eur J Immunol 1991, 21:815-818

151. Theobald VA, Lauer JD, Kaplan FA, Baker KB, Rosen-berg M: "Neutral allografts": lack of allogeneic stimu-lation by cultured human cells expressing MHC classand class 11 antigens. Transplantation 1993, 55:128-133

152. Londei M, Lamb JR, Bottazzo GF, Feldmann M: Epi-thelial cells expressing aberrant MHC class 11 deter-minants can present antigen to cloned T cells. Nature1984, 312:639-641

153. Ellison Ca, MacDonald GC, Rector ES, Gartner JG: -yT cells in the pathobiology of murine acute graft-versus-host disease: evidence that y6 T cells mediatenatural killer-like cytotoxicity in the host and that elim-ination of these cells from donors significantly re-duces mortality. J Immunol 1995, 155:4189-4198

154. D'Andrea A, Chang C, Franz-Bacon K, McClanahan

T, Phillips JH, Lanier LL: Molecular cloning of NKB1.A natural killer cell receptor for HLA-B allotypes.J Immunol 1995, 155:2306-2310

155. Rajagopalan S, Winter CC, Wagtmann N, Long EO:The Ig-related killer cell inhibitory receptor binds zincand requires zinc for recognition of HLA-C on targetcells. J Immunol 1995, 155:4143-4146

156. Weber MC, Groger RK, Tykocinski ML: A glyco-sylphosphatidylinositol-anchored cytokine can func-tion as an artificial cellular adhesin. Exp Cell Res1994, 145:1646-1652

157. Finberg RW, White W, Weller N: Decay-acceleratingfactor expression on either effector or target cellsinhibits cytotoxicity by human natural killer cells. JImmunol 1992, 149:2055-2060

158. Brodsky RA, Jane SM, Vanin EF, Mitsuya H, PetersTR, Shimada T, Medof ME, Nienhuis AW: PurifiedGPI-anchored CD4DAF as a receptor for HIV-medi-ated gene transfer. Hum Gene Ther 1994, 5:1231-1239

159. Ashwell JD, Jenkins MK, Schwartz RH: Effect of yirradiation on resting B lymphocytes. II. Functionalcharacterization of the antigen-presenting defect. JImmunol 1988, 141:2536-2544

160. Gajewski TF, Pinnas M, Wong T, Fitch FW: Murine Thland Th2 clones proliferate optimally in response todistinct antigen-presenting cell populations. J Immu-nol 1991, 146:1750-1758

161. Croft M: Activation of naive, memory, and effector Tcells. Curr Opin Immunol 1994, 6:431-437

162. Elias D, Reshef T, Birk OS, Van der Zee R, Walker MD,Cohen IR: Vaccination against autoimmune mousediabetes with a T-cell epitope of the human 65-kDaheat shock protein. Proc Natl Acad Sci USA 1991,88:3088-3091

163. Zhang J, Medaer R, Stinissen P, Hafler D, Raus J:MHC-restricted depletion of human myelin basic pro-tein-reactive T cells by T cell vaccination. Science1993, 261:1451-1454

164. Maron R, Zerubavel R, Friedman A, Cohen IR: T lym-phocyte line specific for thyroglobulin produces orvaccinates against autoimmune thyroiditis in mice. JImmunol 1983, 131:2315-2322

165. Chiocchia G, Manoury-Schwartz B, Boissier MC, Ga-hery H, Marche PN, Fournier C: T cell regulation ofcollagen-induced arthritis in mice. Ill. Is T cell vacci-nation a valuable therapy? Eur J Immunol 1994, 24:2775-2783

166. Ottenhoff TH, Mutis T: Specific killing of cytotoxic Tcells and antigen-presenting cells by CD4+ cytotoxicT cell clones. A novel potentially immunoregulatoryT-T cell interaction in man. J Exp Med 1990, 171:2011-2024

167. Moss DJ, Burrows SR, Baxter GD, Lavin MF: T cell-Tcell killing is induced by specific epitopes: evidencefor an apoptotic mechanism. J Exp Med 1991, 173:681-686

16 Tykocinski et alA/PJanuary 1996, Vol. 148, No. 1

168. Janeway CA Jr, Bottomly K: Signals and signs forlymphocyte responses. Cell 1994, 76:275-285

169. Jenkins MK: The ups and downs of T cell costimula-tion. Immunity 1994, 1:443-446

170. Fuchs EJ, Matzinger P: B cells turn off virgin but notmemory T cells. Science 1992, 258:1156-1159

171. Valle A, Aubry JP, Durand I, Banchereau J: IL-4 andIL-2 upregulate the expression of antigen B7, the Bcell counter-structure to T cell CD28: an amplificationmechanism for T-B interactions. Int Immunol 1991,3:229-235

172. Stack RM, Lenschow DJ, Gray GS, Bluestone JA,Fitch FW: IL-4 treatment of small splenic B cells in-duces costimulatory molecules B7-1 and B7-2. J Im-munol 1994, 152:5723-5733

173. Ding L, Linsley PS, Huang L-Y, Germain RN, ShevachEM: IL-10 inhibits macrophage costimulatory activityby selectively inhibiting the up-regulation of B7 ex-pression. J Immunol 1993, 151:1224-1234

174. Villanueva POF, Reiser H, Stadecker MJ: Regulationof T helper cell responses in experimental murineschistosomiasis by IL-10. Effect on expression of B7and B7-2 costimulatory molecules by macrophages.J Immunol 1994, 153:5190-5199

175. Czarniecki CW, Chiu-HH, Wong-GH, McCabe-SM,

Palladino-MA: Transforming growth factor f-1 modu-lates the expression of class 11 histocompatibility an-tigens on human cells. J Immunol 1988, 140:4217-4223

176. Honda S, Campbell JJ, Andrew DP, Engelhardt B,Butcher VA, Warnock RA, Ye RD, Butcher EC: Ligand-induced adhesion to activated endothelium and tovascular cell adhesion molecule-1 in lymphocytestransfected with the N-formyl peptide receptor. J Im-munol 1994, 152:4026-4035

177. Steurer W, Nickerson PW, Steele AW, Steiger J,Zheng XX, Strom TB: Ex vivo coating of islet cell al-lografts with murine CTLA4/Fc promotes graft toler-ance. J Immunol 1995, 155:1165-1174

178. Yin D, Fathman CG: Induction of tolerance to heartallografts in high responder rats by combining anti-CD4 with CTLA4 Ig. J Immunol 1995, 155:1655-1659

179. Fayen J, Huang JH, Meyerson H, Zhang D, Getty R,Greenspan N, Tykocinski M: Class MHC a 3 domaincan function as an independent structural unit to bindCD8a. Mol Immunol 1995, 32:267-275

180. Meyerson HM, Huang J-H, Fayen JD, Tsao H-M, GettyRR, Greenspan NS, Tykocinski ML: Functional disso-ciation of CD8a's immunoglobulin homologue andconnecting peptide domains. J Immunol 1996 (in press)