CA2251781A1 - A method for identifying cytotoxic t-cell epitopes - Google Patents

Abstract

A method for isolating bio-active molecules from complex, rationally designed oligopeptide libraries which elicit cytolytic activity from cloned cytotoxic T lymphocytes (CTLs). The method allows the simultaneous screening of multiple CTL lines against indexed peptide libraries synthesized on solid phase support. Preferably decoding is not dependent on the presence of residual peptide and does not employ peptide sequencing. The method to identify CTLreactive oligopeptides yields products of therapeutic value such as vaccines in treating cancers, viral diseases, and autoimmune diseases, as well as to identify useful clinical diagnostic reagents with a reduction in assay time and increase in throughput.

Description

CA 022~178l 1998-l0-14 w o s7/~so3s PcTruss7/0447s A ~DETHOD FOR rDENTrFyDNG CYTOTOXIC T-CELL EPITOPES

FrF~ n OF TEnEINnvENTIoN

The invention relates to identifir~qtion of bio-active molecules from a combinatorial library of oligopeptides qtt~hPd to solid phase su~l)o~t~. The peptides attached to a single bead S have Pssenti~lly the same amino acid sequenre. The synthesis history of each peptide bead may be recorded on each solid support in a code of inert molecular tags, such that beads of interest can be rapidly and efficiently decoded. A photocleavable crosslinker allows release of some of the oligopeptide by exposure to UV light. Molecular tags if present, remain covalently bound to the beads for post-assay analysis. The bioactive molecules may be 10 screened in cytotoxic T Iymphocyte screening assays.

BAC~GRO~D OF T~nF~l~nvF~TIoN

Cellular in-l-~ul-ul}lerapy is ~ g as a technologically and intellectually compelling anti-cancer L~ ",r~,l The generation of an immune response against tumors has been demonstrated in several animal models and has been inferred from reports of sponrqn~ollc 15 tumor regression in man (Stotter and Lotze, 1990, Cancer Cells; 2:44-55). Cytotoxic ~ T-lymphocyte (CTL) responses can be directed against antigens specifically presented by tumor cells, both in vivo and in vitro, without the need for prior knowledge of the molecular mPchqni.cm by which the tumor arose. In animal models, established tumors can be era~1ir~tP~d by the adoptive transfer of T-cells that are specifically immune to the 20 m~lignqnt cells (Buen et al., Immunol. Today; 15:11-15). Terhni~luec of adoptive T-cell therapy have recently been applied to the treatment of human viral disease, but the application of similar T-cell therapy for human m~lign~qnry has been hindered in part by the lack of well defined tumor antigens recognizable by ~q,~ltochth-mnus T-cells. Many human prog~ e or m~ r~ tic cancers, such as (li~rl~ ted mqlignqnt melanoma or metqctqtic 25 renal cell carcinoma, are resistant to conventional therapies, inrhlfijng chemotherapy and radiotherapy. In these types of cancers, irnmunotherapy has been tried over the past 10 ~ years and although its success rate has been relatively modest, it remains a promising alternative to the conveMional therapies (Bergmann et al., 1990, Onkologie; 13:137).

, CA 022~1781 1998-10-14 In man, ~onLal1cous destruction of mclanoll~a cells occurs in 15% to 20% of p lesions, int~ ting that host protective m~ch~ni~ms which can selectively destroy m.ol~nt)m~
cells are present (Bystryn et âl., 1993, Heme. Onc. Annals. 1:301). Vaccine immunotherapy with crude or partially purified melanoma vaccines can prevent tumor 5 growth in 50% to 100% of mice i~ d to otherwise lethal doses of melanoma cells.
The protection is specific, in~ ting it is m~ t~d by immune mechanisms. The challenge is to devise vaccine strategies that will induce similar h~ lu~loprul~ctive responses in man.

For immllnotherapy to be improved, epitopes recûgnized by tumor-speci~lc CTLs must be tifit~d. CTL epitopes are 8-10 amino acid peptides derived from cellular proteins that 10 are endocytically processed and pr~se.lled on the tumor cell surface by majorhistocompatability complex (MHC) class I and clâss 11 glycoplùl~hl~. MHC molecules are expressed in virtually all m-cle~ted cells and the colnbillation of peptide and MHC molecule is specifir~lly recognized by the ~pl.,o~,iate T-cell receptors (TCRs). T-cells in the presence of antigen ples~ lg cells and their corresponding antigen proliferate and acquire 15 potent cytolytic activity.

Identiflr~tion of the antigens recognized by these tumor-specific CTLs is vital to the rational development of peptide-based anti-tumor vaccines. A common strategy in the search for tumor antigens is to isolate tumor-specific T-cells and attempt to identify the antigens recognized by the T-cells. In patients with cancer, specific CTLs have been often derived 20 from Iymphocytic infiltrates present at the tumor site (WeidmaM et al., 1994, Cancer lmmlln-ll. Immunother. 39:1-14). These tumor infiltrating Iymphocytes (TILs) are a unique cell population that can be traced back to sites of disease when they are labeled with indium and adoptively transferred.

Indeed, the presence of a large number of T-cells in tumors has been correlated with a 25 prognosti~lly favorable l~ulcoll,c in some cases (Whiteside and pi1~ .I.i~ni, 1994, Cancer ~mmllnol Inlll~uilolher. 39:15-21). Recently it was shown that implantation of poly~ e sponges coll~h~i"g irradiated tumor cells can efficiently trap anti-tumor CTLs (4-times greater than lymph fluid, 50-times greater than spleen or p~ ,h~,lal blood) (Woolley et al., 1995, Immunology, 84: 55-63). Following activation with T-cell cytokines in the prese"ce 30 of their app~uL~ ly presented recognition antigen, TILs proliferate in culture and acquire potent anti-tumor cytolytic properties (Weidmann et. al., 1994, supra). Thus, TILs are a CA 022~1781 1998-10-14 convenient source of Iymphocytes greatly enriched for cells with tumor cell speci~lrity.
Additionally, tumor-specific CTLs have been found in peripheral blood or m~lign~nt ascites of patients with cancer, intli~ting that a systemic response to the tumor may be present or that redistribution of CTLs from the tumor to the periphery might occur (Wallace et al., 1993, Cancer Res. 53:2358-2367). In either case, this is an attractive feature for the i",..,.-.~. lherapeutic treatment of m~t~t~tic or di~ ed cancers.

The reasons why tumor cells may express tumor-specific antigens (TSAs) are beginninE~ t~
be understood. For example, TSAs may be the result of the processes of carcinogenesis, which are generally thought to stem from damage to a large number of genes, some of 10 which have a role in the molecular m~c~-~ni~",~ regulating cell growth and division. This damage results in ul~co~ olled cellular proliferation that defines the transformed cell.
Thus, possible origins of TSAs include self proteins (such as fetal antigens) oncogene products (including fusion proteins), mutated tumor su~)rcssor gene products, other mutated cellular proteins, or foreign proteins such as viral gene products. No,.. ~z ~ed 15 cellular proteins may also be antigenic if they are expressed abel,~"lly (e.g., in an indl)~ropliate subcellular co"l~,all"l~llL) or in supernormal (~ n~itiçs. Given the llum~uu~
steps of cellular transformation and snmPtirnPs bizarre genotypes observed in cancer cells, it could be argued that tumor cells are likely to contain many new antigens potentially recognizable by the immune system.

20 Reports of shared tumor antigens are frequent in the literature. In the case of me!~nom~, there is recent evidence that the same T-cell-defined tumor antigens are ~ ssed by independent human melanoma suggesting that tran~r~ .lion-associated events may give rise to le. Ull~ e~LI,Iession of the same tumor antigen in different tumors of related tissue and cellular origin (Sahasrabudhe e~ al., 1993, J. Immunol., 151:6302-6310; Sh~mqmi~n et 25 al., 1994, Cancer Immunol. T.",.. ,.. tller., 39:73-83; Cox et al., 1994, Science 264:716;
Peoples et al., 1993, J. Tmml-n~l., 151:5481-5491: Jerome et al., 1991, Cancer Res., ~ 51:2908-2916; Morioke et al., 1994, J. Tmmllnol., 153:5650-5658). Previous studies in animal models have, in contrast, suggested that most ch~mic~l and ultraviolet radiation-induced tumors are antigenically diverse and that tumor rejection antigen may be 30 genelaled by random mutation (Srivastava et al., 1986, Proc. Natl. Acad. Sci. USA, 83:3407-3411). However, it is highly i"l~lobable that a completely random process would give rise to shared antigens even in very closely related tumors. This data ~u~oll~ the CA 022~1781 1998-10-14 possibility that specific anti-tumor i,.~ u..olherapies, such as vaccines, may be active against more than one form of cancer and that the same vaccine may be effective against independently derived tumots of the same type.

While isolation, expansion, and retransfusion of TILs is appealing, there are severe adverse 5 cardiol~s~ atury and hemodynamic effects such as tachycardia, increases in cardiac index, systemic vascular resictqnf~e, and pulmonary artery diastolic pressure which appear within two hours post-infusion. These effects are similar to the physiologic changes seen in interleuken-2 (IL-2) therapy and septic shock (Marincola et al., 1993, J. lmmnnol., 13:282-288). These changes are sllctqin~d and a~lgm~llt~l by 5llhse~uent IL-2 10 ~ ldljon (Lee et al., 1989, J. Clin. Oncol., 7:7-20). IL-2 is a T-cell cytokine and its production is among the earliest events following stimlllqtion of the T-cell receptor (TCR).
The physiological changes observed in septic shock have been associated with elevated levels of TNF-cc and IL-6, both of which are produced upon T-cell stimlllqtion (Calandra et al., 1990, J. Infect. Dis., 161:982-987).

15 A comprehensive survey of the literature reveals that neither adoptive transfer of tumor-specific CTLs nor specific active immnnothPrapy with whole tumor cells or cell-derived p,~.alalions leads to eradication of tumors or long term survival in more than a minority of patients. It has been demonstrated in vitro that peptides have sslcceed~d in priming T-cells where cell-derived preparations have failed (Cox et al., 1994, supra).
20 Peptides that are expressed by the tumors of many individuals may be useful for imsnunotherapy, but the most generally applicable would be those that also are recognized by Iymphocytes obtained from a large number of different cancer patients. Epitopes recognized by multiple CTL lines would be promising cqn~ t~s for use in peptide-based anti-tumor vaccines. In the absence of a reliable iterative method to identify TSAs, there is 25 no way of ~Pc~ g the limits of cross-reactivity.

There are several obstacles which collllil)uL~ to the difficulty of analyzing MHC-associated peptides by classical means. Current protocols involve isolating and assaying extremely pure MHC molecules from antigen ~ s~.lLillg cells. Prior to peptide extraction, all co..l ~",j"stj..g proteinaceous material must be removed (this includes low molecular weight 30 contqminqnt~ that normally escape de~ection by routine methods used to analyze protein purity such as SDS-PAGE) (Chicz and Urban, 1994, Immllnol. Today, 15:155-160).

CA 022~1781 1998-10-14 WO 97/3503~ PCT/US97/04479 Briefly, imml-noaffinity pu,irlcation yields approximately 0.5-1 mg of HLA molecules per gram (11 culture) of B-cell Iymphocytes (yields from B-cells are signifir~ntly higher than those obtained from primary explant tissues). Since the bound peptide is only 8-10 amino acids long, l mg of MHC contains 16 pmol of extractable peptide. Furthermore, the 5 efficiency of peptide extraction is typically 75-80%. Thus, lmg of MHC usually yields 13 pmol of isolated peptide for analysis. The population of bound peptide is estim~t~d to have a complexity > 2000, the majority of which are believed to be self-peptides. Therefore, the average molar amount of each individual peptide present after purification is 13 pmol divided by the population complexity. The utility of a large pool of purified peptides in 10 which each individual species is present in e~cee~ingly minute qu~ntities is limited. At this point, the purifled peptides can be fractionated by HPLC and the fractions assayed for reactivity with cloned CTLs. Tandem mass spectrometry can be used to sequence reactive fractions. However, the complexity of peptides in each fraction often exceeds the number of peptides that can be sequenced with the available material. Thus, although this method 15 has been used successfully, the lack of data in the litelalul~ gleaned from this approach is testimony to the difficulty of its successful execution.

Knowledge of the primary seq~Pnre of MHC, or of known T cell epitopes, has not yielded a key to immunogenicity of such epitopes. I(len~ifir~tion and screening of epitopes has also not been further facilitated by the d~ ion of structural features of the MHC, e.g., 20 using X-ray crystallography. These techni~ es, which in other systems provide for the rational design or i~lrntifir~tir,n of receptor agonists and antagonists, have not proven useful for idenrific~tir,n of T cell epitopes.

Reco.,.bi.larll bacteriophage have been used to produce large libraries. Using the "phage method" (Scott and Smith, 1990, Science 249:386-390; Cwirla, et al., 1990, Proc. Natl.
25 Acad. Sci., 87:6378-6382; Devlin et al., 1990, Science, 249:404-406), very large libraries can be constructed (106-108 rhrmir~l entities). However, in these libraries it is difficult to dissociate a ~ n~lse due to the recombinant fusion protein from one due only to the peptide. Another approach uses primarily chPmir~l m~thorl~, of which the Geysen method (Geysen et al., 1986, Molecular lmmllnr,logy 23:709-715; Geysen et al. 1987, J.
30 Tmmlmnlogic Method 102:259-274) and the recent method of Fodor et al. (1991, Science 251, 767-773) are examples. Furka et al. (1988, 14th International Congress of Biochrmictty, Volume 5, Abstract FR:013; Furka, 1991, Int. J. Peptide Protein Res.

. .

CA 022~1781 1998-10-14 W O 97/35035 PCTrUS97/04479 37:487493), Houghton (U.S. Patent No. 4,631,211, issued Dece~.lb~. 1986) and Rutter et al. (U.S. Patent No. 5,010,175, issued April 23, 1991) describe methods to produce a mixture of peptides that can be tested as agonists or antagonists. However, these methods are deficient as they provide either limited numbers of pred~sign~d peptides, which require 5 some advance predictions about the desired sequence, or in a large (and in the case of Rutter, chaotic and ill~lisc~ a~e) mixture of peptides, leaving one no better off than with naturally purified MHC containing peptide epitopes.

A major advance in scree~ g occurred with the development of synthetic libraries (Needels et al., 1993, "Generation and scree.ling of an oligonucleotide encoded synthetic peptide 10 library," Proc. Natl. Acad. Sci. USA 90:10700~; Ohlmeyer et al., 1993, Proc. Natl.
Acad. Sci. USA 90: 10922-10926; Lam et al., International Patent Publication No. WO
92100252, each of which is incol~olaLed herein by reference in its entirety), and the like can be used to screen for lect;~or ligands.

The synthesis of indexed co..~hlatorial peptide displays has been dcscribed (Ohlmeyer et 15 al., 1993, Proc. Natl. Acad. Sci. USA, 90:10922-26). ~t is possible to synthesize e~ o~c-lcngth peptides on Merrifield resin beads while cG~ ;7ed inert molecular tags allow rapid and ef~lcient decoding of the synthesis history of any unique bead via gas-phase cl~ollla~ography (Ohlmeyer et al., supra) The efficiency of decoding is about 90 percent utilizing a single bead. ~;urthermore, it is not necessary to restrict assays to solid-phase 20 interactions since photocleavable linkages allow controlled release of required amounts of peptide for solution-phase assays. This is irnportant because it may not be possible for peptides to bind directly to surface-localized MHC class I molecules directly (in general, loading APCs with antigen occurs by internalizing the peptide and colul)inillg it with the MHC molecule as it assembles), and even if the APCs can bind directly to the beads, tight 25 packing of the peptide on the surface of the beads may cause enough stearic hindrance so as not to allow access of the MHC/peptide complex to the CTL T-cell receptors. Note that MHC/peptide complexes have a remarkable stability. A feature of most MHC/peptidecomplexes is their llnll~llAlly slow dissociation kinetics, with a half-life in the range of several days. Most peptides ( > 90 % of characterized human HLA-A epitopes) will bind 30 with Afflniti~s o~ 2-50 nM.

CA 022~1781 1998-10-14 W 0 97t35035 PCT~US97/04479 Van der Zee 1989, Eur. J. Imm-lnc)l, 19:43-47) and cowolhel~ have developed a powerful but limited strategy for identifying T-cell epitopes. Briefly, utilizing the "pepscan"
technique, they were able to simnlt~nPously synthesize several dozens of peptides on polyethylene rods arrayed in a 96-well microtiter plate pattern. This is similar to an 5 indexed library in Ihat the position of each pin defines the synthesis history on it. Peptides were then chrmir~lly cleaved from the solid support and supplied to irradiated syngenic thymocytes for antigen p,~st~ ion. The cloned CTL line was then tested for reactivity in a proliferation assay monitored by 3H-thymidine incorporation. This type of analysis particularly suits a CTL stimulation assay since it can be automated using a microtiter plate 10 reader and employs relatively low levels of radiation. The procedure successfully i~lentifi~d a reactive epitope in a deflned region of a 65 kDa mycobacterial heat shock protein with es.~Pnti~lly no background. A second screen where the synthP~i7Pd peptides had one alanine insertion per peptide at each position of the naturally occurring epitope identified an additional seven peptides with di...;l~ Pd yet detectable reactivity, underscoring the 15 tolerances to snbstit~tions in this assay. Ad~ition~lly, screening peptides having a single deletion per peptide (derived from the natural epitope) yielded no reactive peptides, underscoring the specificity endowed by the presence of the nine residues in the naturally occurring epitope.

Notwith~t~n-ling the efforts made to date to identify T cell epitopes, the inventor herein has 20 recognized a clear need in the art for a rapid method to obtain saturating profiles of epitopes which elicit CTL cytolytic activity directed against apl.iol,liale APCs. In several cases, derivatized natural epitopes are more effective than the natural epitope itself, accordingly, there is a need to identify such derivativized natural epitopes. Additionally, ntifir~tion of epitopes from a wide range of independently derived CTLs will allow the 25 design of powerful vaccines which are cross-reactive against dirrel~llt diseases and thus serve a greater cross-section of the population.

~ SUMMARY OF T~:IF INVENTION

The present invention is directed to the design of degenerate, oligopeptide libraries cc,."~,isi"g MHC allele-specific agretopes displaying diverse T cell receptor (TCR) 30 epitopes, and a method of use of these libraries to screen for TCR epitopes.

., ...... ~ . ., ~, CA 022~1781 1998-10-14 In particular, the present invention provides a method for designing oligopeptide libraries such that all species contain high affinity MHC allele-specific binding sites and a rtl,elloile of variable dom~in.c that will interact with complementary TCRs. MHC allele-specific anchor residues have been determined for the common human MHC haplotypes.

5 For example, in a specific embodiment~ a completely degenerate octamer library would have a complexity of 2.56 x 10'~. Two fixed anchor positions reduces the complexity to 6.4 x 10'. The present invention inco~yulates knowledge of TCR plOllliSc,~ily with respect to tolerance to cons~ ti~e substitutions in naturally derived epitopes, combined with knowledge of MHC agretopes, to identify reactive epitopes and all reactive epitope 10 derivatives. Thus, the present invention enables production of a library that can be practicably screened and whose signal to noise ratio is expelh,ltlll~lly tolerable.

The present invention further provides a method for screenhlg an oligopeptide library for bioactive CTL epitopes such that pooled aliquots of solid phase SLlyl)oll~ can be ~imlllt~nPoncly assayed with multiple CTL clones of the same MHC restriction class.

15 In a broad aspect, the present invention is directed to a method for identifying a cytotoxic T
cell epitope. According to the invention, the method col,,yli~es the steps in order of con~ ting a population of at least two cytotoxic T cells ~CTLs) having the same MHC-haplotype restriction with a library of molecules attached to solid phase supports by a releasable linker~ wherein each solid phase support is attached to a single species of 20 molecule, and wherein the structure of the molecule can be dPterrninPd. The library of molecules contains a conserved structural motif corresponding to a structural motif characteristic of peptides that associate with the MHC-haplotype to which the cytotoxic T
cells are res~ricted; this motif is referred to herein and in the art as an "agretope." The library is contacted or exposed to antigen presentation means prior to or ~imlllt~n~ously 25 with the CTLs, which antigen yl~se~ tion means correspond to the MHC-haplotype to which the cytotoxic T cells are restricted. The solid phase ~.lyyoll~ of the library are in separate fractions, so as to facilitate i-lPntifc~tion of molecules that prove to be epitopes recognized by the CTLs. At least a portion of the releasable lirtker is cleaved so as to release at least a portion of the molecule, and the cytotoxic T cells are evaluated as to 30 whether they recognize a molecule present in one or more of the fractions of the library of molecules. Upon observation of such recognition, the method then involves isolating one CA 022~1781 1998-10-14 or more solid phase support supports from the fractions and deL~ -g the structure of a molecule on a solid phase support isolated from the fraction. In a specific embodiment, such molecules are peptides. However, the term peptide is construed herein to encompass peptidomimetics, and non-naturally occurring peptide analogs, as these have been5 developed in the art. It should be further recognized by those of skill in the an that molecules can be designPd by empirical t~r~ni-lu~ c, such as the co~,bi..ato.idl libraries described herein. that topologically and functionally perform as a peptide epitope, but which bear no structural resemblance to peptides (such as morphine activates opioid receptors but has a vastly different structure -- except at the receptor-binding surface -- than 10 ~-endorphin).

In a preferred aspect, the cytotoxic T cells are polyclonal T cells isolated from a site of cytotoxic T cell infiltration from an individual. Alternatively, such cells may be isolated from a site of cytotoxic T cell infiltration from two or more individuals. which two or more individuals share an MHC haplotype. In another embodiment, the CTLs may be two or 15 more cytotoxic T cell lines. In yet another embodiment, the CTLs may be any colllbi~lion of the foregoing.

In a further preferred aspect, the site of cytotoxic T cell infiltration is a tumor. The tumors from which cells or cell lines are obtained can be the same type of tumor in dirrel.,.,l individuals with a shared MHC haplotype, e.g., a melanoma from Mr. A and a melanoma 20 from Ms. B, or different types of tumors from different individuals who share an MHC
haplotype, e.g., a melanoma from Mr. A and a breast cancer from Ms. B.

Alternatively, CTL infiltrates can be from sites of viral infection, au~oi,lL...une inflqmmqtion (such as demyelinated nerve tissue or ce.eblu~ al fluid in MS, infl~mpd joints in arthritis, etc.), transplantation rejection, and like sites of infl~mm~tion or Iymphocyte/leukocyte 25 infiltration.

As noted above, the peptide identified accordhlg to the invention may comprise subunits selected from the group con~istin~ of glycine, L-amino acids, D-amino acids, non-cl~c.cic~l amino acids, and peptirlf)mimt tirS.

CA 022~1781 1998-10-14 W O 97/35035 PCT~US97/04479 Various solid phase supports (also termed herein "beads") can be used in the practice of the invention. Such solid phase supports must be compatible with the biological assay to be performed. and must be inert to the synthesis of the molecule, e.g., peptide. and if present, a coding molecule. Examples of solid phase supports include polystyrene resin, 5 poly(dimethylacryl)amide-grafted styrene-co-divinylbenzene resin, polyamide resin, polystyrene resin grafted with polyethylene glycol, and polydil"~ ylacrylamide resin.

The releasable linker may release upon exposure to an acid, a base, a nucleophile, an electrophile, light, an oxidizing ageM, a reducing agent, or an enzyme.

The invention provides specific ~ilu~;luldl motifs (agretopes) for use in libraries of the 10 invention, inrlt1-iing, but not limited to, LXXXXXXV (SEQ ID NO~ RXXXXXX+
(SEQ ID NO:2); X(D,E) XXXXXX(F,K,Y) (SEQ ID NO:3); RXXXXXXL (SEQ ID
NO:4);X(K,~)XXXXX(L,I) (SEQ ID NO:5); ~,L)XXXXXXK (SEQ ID NO:6);
EXXXXXX(Y,F) (SEQ ID NO:7);XPXXXXX(F,H,W,Y)(SEQID NO:~);
(L,I)XXXXX(H,K) (SEQ ID NO:(9); wherein X intljr~t~s any amino acid residue, and +
15 in~lir~tPs a positively charged amino acid residue.

In a ~.lefell~d aspect, the invention greatly simplifies a primary search for an epitope by incol~ol~.lillg a limited number of r~ es~.llative amino acid residues in the peptides of the library. For example, positively charged amino acid residues may be substituted with an amino acid selected from the group consisting of lysine. arginine, and hicti-linP; negatively 20 charged amino acid residues may be s~ Pd with an amino acid selected from the group consisting of aspartic acid and glllt~mi~. acid; neutral, polar amino acid residues may be ~ul)~ uled with an amino acid selected from the group consi~Lillg of asparagine, ~lul~lllille, serine, lhl~olli,le, tyrosine, and cysteine; nonpolar amino acid residues may be sul,~ ..t. d with an amino acid selected from the group consislhlg of alanine, valine, leucine, 25 isoleucine, proline, phenylalanine, lly~,tol.han, and methionine. In a further embodiment, the nonpolar, aromatic amino acid residues are substituted with an amino acid selected from the group con~i;,ling of tyrosine, threonine, and tryptophan; and the nonpolar aliphatic amino acid residues are substin~d with an amino acid selected from the group collsi~lillg of alanine, valine, leucine, isoleucine, proline, and methionine.

CA 022~1781 1998-10-14 W O 97/35035 PCT~US97/04479 The ~LI-lcl~llc of the molecule (or peptide) dctell~lhled in a simplifiPd primary screen can be refined in a secondary screen. This secondary screening constrains the structure of the library of molecules to have the required agretope, and to have a sequence of chl~rnir~lly similar residues as defined in the primary screen. However, in the secondary screen, all 5 possible amino acids corresponding to a particular residue type are tested to find an epitope with the greatest stim~ tory activity. Thus, the foregoing method may further cunlyli~e the steps of cnnt~~ting at least one of the CTLs in the population of at least two cytotoxic T
cells having the same MHC-haplotype restriction with a library of molecules as set forth above, which library of molecules contains a conserved structural motif collc."~olldillg to a 10 structural motif characteristic of peptides that associate with the MHC-haplotype to which the cytotoxic T cells are restricted, and wherein every amino acid corresponding to the eple~ ative residue is utilized at the position identified for the colle~yonding,epleselltative residue. The library is contacted or exposed to antigen p~esell~dlion means prior to or ~imlllt~n~ollcly with the CTLs, which antigen yles~ n means correspond to 15 the MHC-haplotype to which the cytotoxic T cells are restricted; at least a portion of the releasable linker is cleaved so as to release at least a portion of the molecule; the cytotoxic T cells are evaluated for recognition of a molecule present in one or more of the fractions of the library of molecules; one or more solid phase support supports from the fractions is isolated; and the structure of a molecule on a solid phase support isolated from the fraction 20 is deltllllhled.

In a specific embodiment. the invention provides a method for identifying a high affinity cytotoxic T cell epitope Colll~,lisillg contacting a population of cytotoxic T cells having an MHC-haplotype restriction with a library of molecules ~tt~ched to solid phase supports by a releasable linker, wherein each solid phase support is ~tt~.hl~d to a single species of 2~ molecule, and wherein the structure of the molecule can be determined, which library of molecules contains a conserved structural motif coll~,yollding to a structural motif chaldcte~ ic of peptides that associate with the MHC-haplotype to which the cytotoxic T
cells are restricted, and wherein every amino acid coll~,~yolldillg to a l~plesellld~ e residue ~etermjn~ as set forth above is utilized at the position identified for the corresponding 30 r~,l)lc~ellLdli~te residue; and antigen pl~,c~ ion means, which antigen plescllldlion means correspond to the MHC-haplotype to which the cytotoxic T cells are restricted; wherein the solid phase supports of the library are in separate fractions; cleaving at least a portion of the releasable linker so as to release at least a portion of the molecule; evaluating whether the CA 022~1781 1998-10-14 W O 97/35035 PCT~US97/04479 cytotoxic T cells recognize a molecule present in one or more of the fractions of the library of molecules; isolating one or more solid phase support supports from the fractions; and determining the structure of a molecu}e on a solid phase support isolated from the fraction.

In a further preferred aspect of the invention. a coding molecule is ~tt~rllP~1 to each solid 5 phase support of the library, which coding molecule defines the structure of the molecule ~tt~ Pd to the solid phase support by the releasable linker. In specific embo~limPntc, the coding molecule is a peptide, an oligonucleotide, or, preferably, an inert molecular tag that can be decoded by gas-phase chromatography. An example of the latter is a halogen snbstitll~ed benzene.

10 Thus. in one aspect, the structure of the molecule is determined by analyzing a portion of the molecule rem~ining on the solid phase support. For example, a sequence of the peptide may be cle~e~Tnin~d by sequencing a portion of the peptide remqining on the solid phase support, e.g., using Edmond degradation and lllicluse~ çrlring ter~lniflues. Alternatively, using the coding molecule technology set forth above, the structure of the molecule is lS determined by analyzing the structure of the coding molecule. In a further embodiment, the structure of the molecule is determined after isolating more than one r~n~ te solid phase support; repeating the scleenillg procedure with the isolated c~n~ te supports, however, testing each support in a separate assay, isolating one such solid phase support that demon~Ll~tes CTL activation, and deLellllinillg the structure of a molecule on that solid 20 phase support.

According to the invention, the antigen presentation means may be a purified MHC class I
molecule complexed to ,B2-microglobulin; an intact antigen pl~se.l~ g cell; or a foster antigen pll,senlillg cell. Preferably, the antigen p~es~ ion means is a foster antigen presçnting cell. More preferably, the foster antigen pl~,srllling cell lacks antigen processing 25 activity, whereby it e~resses MHC molecules free of bound peptides. In a specific embodiment, the foster antigen pl.,~e.llillg cell is the human B and T Iymphoblast hybrid cell line 174xCFM.T2, deposited as a publicly available cell line with the American Type Culture Collection (ATCC), 12301 Parklawn Drive, Rockville, Maryland 20852, and acsi~n~d accession number CRL-1992.

CA 022~1781 1998-10-14 The recognition of a molecule present in one or more of the fractions of the library of molecules by the cytotoxic T cells can be evaluated by detecting cytotoxic T cell activation.
For example, but not by way of limitation, cytotoxic T cell activation can be detected by a method selected from the group comi~ting of ~H-thymidine incorporation; metabolic activity 5 detected by conversion of MTT to formazan blue; increased cytokine mRNA expression;
increased cytokine protein production; increased protein synthesis; and chromium release by target cells.

In another aspect, the invention provides a method of identifying a protein antigen comprising identifying the cytotoxic T cell epitope of the protein determined in a secondary 10 screen (or ~eri--r--~enl screen~, co~,ua~ g a sequenre of the T cell epitope it1enrified in step (a) with known sequences of proteins; and determining a protein having a sequence corresponding to the seqUçnre of the T cell epitope. It should be recognized that although the secondary screen of the invention may provide a highly active epitope that does not correspond to the natural epitope, and thus may not provide sequçnre identity, in all 15 likelihood the sequence of the natural epitope will correspond to a portion of the sequçn~e of the antigen, or be so similar as to leave little doubt about the antigen (Blake et al., 1996, J. Exp. Med. 184:121-30).

It should also be recognized that an epitope derived according to the present invention may correspond to an as yet nnill~ntified protein. Thus, the invention provides for identifir~tion 20 of novel protein antigens. Furthermore, the invention provides a method for cloning the cDNA and genomic DNA encoding such protein by generating degenerate oligonucleotide probes or primers based on the sequence of the epitope.

The present invention further provides therapeutic and di~gnostic agents co~ ri~i.,g oligopeptide sequences determined according to the foregoing methods.

25 Diagnostic agents may also include oligonucleotides corresponding to the identified epitope or a region of geno"lic DNA surrounding the epitope locus. For example, in the case of the CDK4 epitope ~i~cu~.sed infra, PCR was used to type individuals from the patient's pedigree for the presence of the CDK4 mutation, thus identifying individuals at risk for developing this rn~ nom~

CA 022~1781 1998-10-14 Accordingly, it is a primary object of the present invention to provide a method to rapidly and efficiently identify CTL epitopes.

It is a particular object to identify such CTL epitopes of tumor antigens. and more particularly ubiquitous epitopes found on a wide variety of tumors.

5 It is another object to identify CTL epitopes of other disease-associated antigens, such as but not limited to viral antigens, auloh~,.uile antigens, and the like.

Still another object of the invention is to prepare a vaccine colllyli~ing an epitope or epitopes i~entified according to the invention for protection from tumors or other ~ ACeS.
in~ iing viral infections. autoimmune disease, and the like.

10 These and other objects of the present invention will be more completely understood by ,efelellce to the following detailed description of the invention.

DETA~T Fl) DESCRIPTION OF T~ INVENTION

As tliccu.csed above, the present invention is directed to a method for identifying cytotoxic T
cell epitopes. According to the method of the invention, such epitopes are id~ntifi~d by 15 screening a library of molecules, such as peptides~ in a cytotoxic T cell assay.
Acco~.lillgly, the invention is directed to use of solid phase cul"~i"ito,ial libraries of molecules, particularly peptides; cytotoxic T cells (CTLs), such as CTL lines, CTL clones, and CTLs from tumor, illllA.l.,ilA~ory~ or other infiltrates; various antigen pres. .l~ ion means co"~ i,.g MHC-molecules compl~m-on~ry to the CTL lines used to screen the 20 various libraries, inr~ ing autologous MHC-positive cells, MHC-positive cell lines, MHC-transfected cell lines, and free MHC-~2-microglobulin complexes; assays for activation of CTLs by the molecules; and methods to use the molecules for various thc.a~culic and diagnostic indications. Each of these elements is explored in greater detail in the following sections.

CA 022~1781 1998-10-14 Wo 97/35035 PCT/US97/04479 At the outset, an explanation of a strategy of the invention that simplifies s~ ,e~ g for CTL
epitopes will facilitate an unde~ n~ling of the invention, as well as the best mode contemplated by the inventors for the practice of the invention.

Library Comp~exi~es. To sy-.Llle~ize a peptide library of completely dege~ aI~ 8-mers 5 would result in an intractable complexity on the order of 10'~. r;o.lu.lat~ly, several invariant amino acids have been identifPd which serve as anchor residues on the peptide for binding to the MHC molecule. These anchor positions are dirr~lt-l~ for each subclass of MHC class I molecules, but in each case there are two or three dominant anchor positions (Falk et al., 1991, Nature, 351:290-296). I~ ,e~Li-lgly, a single MHC molecule is capable 10 of binding many different peptides as long as the anchoring amino acids are present or at least conserved.

Inclusion of two invariant positions in each peptide reduces the complexity by a factor of 400, resulting in a complexity of 206 or 6.4 x 107 for an octamer library.

This complexity can be further reduced by making some c~lcul~tPd assumptions. For 15 epitopes with high MHC binding affinity (i.e., optimal anchor residues), conservative substhlltions at non-anchor positions do not interfere with recognition by CTL TCRs. That is to say, these conservatively substituted peptides are seen qualitatively as a single entity to the a~plopliate TCR rather than as individual entities. For example, a human HLA-B27 restricted CTL clone specific for the HIV gag p24 protein (residues 263-271) could not 20 distinguish between valine and isoleucine at position five, or methionine and leucine at position six. However a nonconservative s~lhstinltion of glllt~mic acid for glycine at position seven was no longer leco~ ed (Johnson et al., J . Tmmlln- l., 147: 1512- 1521).
The limits of TCR tolerance to conservative substitutions are not known and cannot be accurately acc~ssed until satllld~ g profiles of reactive epitopes are empirically cletprmin~d 25 for a st::lti.ctir~lly signific~nt number of CTLs. SynthPci7ing a library in which one or two r~rese.lIaIi~/es of each class of amino acids are coupled at each position (for example leucine and mPthionin~ but not isoleucine and valine in the case of hydrophobic amino acids) can reduce the library to a manageable complexity. Arguably, a given tumor will elicit CTL clones specific for more than one individual tumor-specific epitope. Screening a 30 less deg~,n~lale library using the CTL clones (in batch) in the proliferation assay can m~int~in st~ticti~i confi~en~e in success even though the complexity is reduced. The CA 022~1781 1998-10-14 trade-off is that a saturating profile of reactive epitopes may not be achieved. However, sc~ g a library in which all non-anchor positions are completely deg~n~al~7 though enormously complex, results in the same signal to noise ratio as an inco"l~,leLe}y deg.,,l~.ale library. The "signal" refers to beads that will correctly register as reactive in the assay and 5 the "noise" is the sum total of all non-specific reactive species. In theory this larger library could provide a saturating profile of reactive peptides~ however, it will not because it cannot be practicably screened in its entirety. The advantage of using the more complex library is that all possible epitopes will be represented. What is not readily obvious is that the chances of identifying a reactive epitope from the more complex library is no greater 10 than those of the less complex library when the same number of beads are assayed. That is to say, the signal to noise ratio, and thus the probability of successful screening, is the same for both libraries.

The cu,l,l,osilion of first (primary) screen MHC allele-specific libraries must (l) have a high affinity for the particular MHC allele (~.e., an effective agretope) and (2) provide a 15 sufficiently diverse repertoire of motifs to interact with TCRs (i.e., the epitopes) in order that the natural epitope and/or its reactive derivatives to be ~ esented to the extent that at least one positive can be detected by sclc~.,i"g a manageable number of beads. Until the limits of TCR promiscuity are known, dt Ir~ ion of the number of beads that must be screened cannot be accurately calculated. However, the few reports in the literature that 20 address this issue are encouraging. If this small sample size is a le~fese,.L~Live cross section of general TCR behavior, then it is predicted that the beads necessary to generate 105 dirrele"~ peptides (e.g., a number c~lc~ ted from Poisson di~Llib,llion sr~ti.cti~s) will suffice to at least identify a reactive derivative. Once irl~ntifi~d, even if it is only weakly reactive, the secondary screen will provide for enumeration of the complete spectrum of reactive 25 derivatives. In most cases the secondary screen will involve far less than 105 different sequen~es.

Thus, a particular advantage of the invention is that it greatly reduces the complexity of the library sclee~ g procedures.

To a first approximation it can be ~$~m~d that collselvaLi~/e substitutions will be tolerated 30 and will perhaps, more often than not, result in partial loss of activity. (Hobohm and Meyerhans, l993, Eur. J. Tmmlln~l., 23:1271-1276), have developed an algorithm to detect CA 022~1781 1998-10-14 MHC binding motifs which correlates well with known, ~I..pilically dele...li.,ed, MHC
binding motifs. The best results were obtained using a similarity matrix based on the physicor~lrmiri~l properties of the amino acids (Taylor, 1986, J. Theor. Biol., 119:205) rather than the Dayhoff matrix PAM250 (Dayhoff et al., 1978, Atlas of protein sequen~.e 5 and structure, Vol. 5, Suppl. 3:345), the three inside matrices 'inside alpha', inside beta' and 'inside other' (Luethy et al., 1991, Proteins, 10:229) or the sllbsti-~ltion probability table for in~--cessible residues (Overington et al., 1992, Protein Sci., 1:216). Thus, the physicoch~mic~l properties matrix seems to embody the criteria important to MHC-peptide-TCR interactions. The table below illustrates these properties for the 10 standard amino acids:
TABLE 1. Pll~ c ' -.mical E~ s Matlix Amino Acid Hydro- Positive Negative ~I~p)~- - Aromatic phobic A l 0 0 0 0 1 1 0 . 0 1 0 S O O O O O

" 1 " in-lica~es the amino acid possesses the property; "0" in~ s it does not.

35 Using these data, a theoretical reactive peptide substitution profile can be generated. For example, if a residue in an epitope is negatively charged, assume that D or E will work to some degree: if a residue is positively charged, H, K or R will work. So if the epitope is YLKDQQLL (anchor residues shown in bold), there are 70,304 conservatively s~bsl;l..led ., CA 022~1781 1998-10-14 W O 97/35035 PCT~US97/04479 derivatives. Therefore~ for a library with a complexity of 206, less than 1000 beads need to be screened in order to identify a derivative. This calculation assumes that all possible combinations of conservative sl~hstinltions retain some activity. A more likely scenario is that only a fraction of these will be detectably active. Screening 106 beads (i.e., ten 96-well S plates at 1000 beads per well) would be successful if only 0.09% of these derivatives demonstrate detectable activity. Additionally, the .~imnlr~n~ous screening of multiple CTLs contributes greatly to the chances of success. It will be possible to further refine these c~lrul~tions when a st~ti~tir~lly .signific~n~ number of complete spectrums of TCR
tolerances are ~ hically determined, e.g., using the methods of the present invention.

10 Multiple s~ l ions within an epitope have been observed to be fully functional by way of a redistribution of the relative corL~ lions of each residue to the trimolecular complex, thus challenging the static definition of agretope and epitope as MHC and TC~ binding sites (Red~l~h~ce and Koszinowski, 1991, Eur. J. Tmmllnol., 21:1697-1701; Boyer et al., Eur. J.
Imm~ln-)l., 20:2145-2148). Also, a mouse CTL epitope derived from pigeon cytochrome C
15 was shown to have complete tolerance to at least 7 s~b~tih~tions (A,V,L,N,Q,K,M) at an internal residue presumed to be part of the epitope rather than the agretope (Ogasawal~ et al., 1990, Int. Immunol., 2:219-224). The sensitivity of the primary screen can be tuned by adjusting the number of beads and/or the number of CTL lines screened. This analysis demon~ tcs the teasibility of utilizing libraries with complexities on the order of 107.

CA 022~1781 1998-10-14 (: o,.~e,osiLions of HLA Allele-Specific Oligopeptide I .ihr~rie$
Table 2 gives the optimal library co~ Gailions for various MHC alleles based on the li~ ulc and the above considerations:

5 MHCA~ele SEQID C~ in~ RLr~lc-~ce' NO:
HLAA2.1~ 1 LXXXXXXV Hendersonetal., 1992 HLA B273 2 RXXXXXX+ Jardetzky etal., 1991 HLA A1 3 X(D,E)XXXXXX(F,K,Y) Hobohn etal., 1993 HLAB14 4 RXXXXXXL Hobohmetal., 1993 HLA B8 5 X(K,R)XXXXX(L,I) Hobohm etal., 1993 HLA A3 6 (M,L)XXXXXXK DiBrino et al., 1993 HLA B44 7 EXXXXXX(Y,F) DiBrino eta/., 1995 HLA B7-l-Sm4 8 XPXXXXX(F.H,W,Y) Sidney etal., 1995 HLA A11 9 (L,I)XXXXX(H,K) Hobohmetal., 1993 15 Anchor residues are shown in bold. "X" in(~ir~S a completely degenerate position. " + "
in(iir~eS a positively charged residue (i.e., H,K,R).
DiBrino, et al., 1993, J. Immunol., 151: 5930-5935; DiBrino et al., 1995, 34:10130-10138; Fruci et al., 1983, ~um. Immunol. 38:187: Henderson et al., 1992, Science, 255:1264; Hobolimetal., 1993, Eur. J. Immunol. 23:1271; Jardetzky, 1991, Nature, 353:326; Sidney et al., 1995, J. Immllnol. 154:247.
2 Crossreactive with HLA A24,A26,A28,A29 (Fruci et al., 1993, Hum. lmmllm~l.
38: 187-192) 3 Crossreactive with HLA A31 (Fruci et al., 1993, sllpra) 4 HLA B7-like supermotif, crossreactive with HLA- B*0701, -B*0801, -B*2705, -B*3501 -03, -B*5401, -Cw*0401, -Cw*0602, -Cw*0702 Solid Phase Combinatorial Libraries Any one of the many COIllbillal~Jl ial library technologies described to date can be employed in the practice of the present invention, inrh~ ng but not limited to synthetic culllbinalolial peptide or molecule libraries (Needels et al., 1993, "Generation and screening of an 30 oligonucleotide encoded synthetic peptide library," Proc. Natl. Acad. Sci. USA 90: 10700-4;
Ohlmeyer et al., 1993. Proc. Natl. Acad. Sci. USA 90-10922-10926; Lam et al.. International Patent Publication No. WO 92/00252; Kocis et al., International Patent Publication No. WO
94/28028, published Dect~-.bcr 8, 1994, each of which is incorporated herein by r~lcnce in its entirety), 35 and the like can be used to screen for CTL epitopes according to the present invention.

CA 022~1781 1998-10-14 W O 97~5035 PCTAUS97/04479 In a pr~ ,d aspect, the invention employs the solid phase library technirlue described by Ohlmeyer et al. (1993, Proc. Natl. Acad. Sci. USA 90:10922-26; this r~f~ nce is incorporated herein by reference in its entirety). Halogen ~lb~ lJlrd benzenes linked to tag-liner tert-butyl esters co~ ule the inert molecular tags that encode the sequence of the 5 unique peptide co-synthP~i7~d on any given bead in the library. A brief description of the tag synthesis, peptide synthesis, and encoding/decoding strategy is p~se"led below.

The molecular tags used as encoding molecules are precipitated from di~ hyl~orrn~
(DMF) containing 8-bromo-1-octanol and 2,4,6-tricholorophenol by the addition of ceslum carbonate. The solution is then heated to 80~C for 2 hours, washed with .SM NaOH, IM
10 HCI, and finally H2O, at which point the organic phase is evaporated. The resulting tag alcohol is a colorless oil.
The tag alcohol is then added to a 2 M solution of phosgene (in toluene) to produce a crude chloroformate. After evaporation of the solvent, the compound is dissolved in CH~Cl2 and pyridine and incubated with tert-Butyl4-(hydroxymethyl)-3-nillobenzoate. The resulting 15 tag-linker tert-butyl ester is isolated from the organic phase and purified by chrollldlography with the product being a clear oil. To generate unique tags, halogen-~ub~Lilul~d benzene compounds are reacted with the elec~luyholic tag. Each derivative has a different gas chromatography retention time. This property confers the ability to encode the unique synthesis history of individual beads. Electron capture capillary gas chromatography can 20 selectively detect the tags at levels less than 1 pmol.

The peptides are synthesi7ed on Merrified resin beads (or other suitable resin) such that the peptides are linked by photocleavable crosslinkers by a typical split-synthesis method. As each amino acid is added, a corresponding mixture of acyl carbonate-activated linker tag acids is co-ligated, but with a linker which is not photocleavable. This allows release of the 25 peptide with retention of the coding molecules during the screening procedure. The combination of tag molecules added at each step corresponds to the specific amino acid residue added in that step, thus serving as a record of the synthetic history of any given bead.

When a bead of interest is identified, the seqll~nre of the peptide that was synthesized on it 30 can be deduced by the following method. The bead is loaded into a Pyrex capillary tube and washed with DMF. It is then suspended in 1 ~l DMF and sealed in the capillary tube CA 022~1781 1998-10-14 W O 97t35035 PCTrUS97/04479 and irradiated to release the tag alcohols. The capillary tube is then opened and the tag alcohols are trim~Athl~ilyted with bis(trimethylsilyl) aret:lmi~e. The solution above the bead is then injected into an electron capture, capillary gas chromatograph for analysis. The resulting profile of tag elution on the gas chromatogram allows the amino acid seql-~An~e of S the co-synth~ci7~d peptide to be directly tletPnnin~d~

The term "peptide" is used in its broadest sense to refer to a compound of two or more subunit amino acids, amino acid analogs, or peptidomimP.tics. The subunits may be linked by peptide bonds. In another embodiment, the subunit may be linked by other the bonds, e.g., ester, ether~ etc. As used herein the term "amino acid" refers to either natural and/or 10 unnatural or synthetic amino acids, inrlu~1ing glycine and both the D or L optical isomers, and amino acid analogs and peptidomim~tics. A peptide of three or more amino acids is commonly called an oligopeptide if the peptide chain is short. If the peptide chain is long, the peptide is commonly called a polypeptide or a protein.

Peptide libraries can include u~ .Luldl amino acids. Thus, peptides of the invention may 15 comprise D-amino acids, a co.l.binalion of D- and L-amino acids, and various "design~r"
amino acids (e.g., ~-methyl amino acids, Ca-methyl amino acids, and Na-methyl amino acids, etc.) to convey special properties to peptides in the library. Ad~litionqlly, by qcSigning specific amino acids at specific coupling steps, peptide libraries with a-helices, ~
turns, ~ sheets, y-turns, and cyclic peptides can be generated. Generally, it is believed that 20 o~-helical secondary structure or random secondary structure is preferred.

In a specific aspect of the invention, the peptides of a library may co,ll~lise a special amino acid at the C-tellllil-us which incul~oldtes either a CO2H or CONH2 side chain to ~imlll~A a free glycine or a glycine-amide group. Another way to consider this special residue would be as a D or L amino acid analog with a side chain consi~li,1g of the linker or bound to the 25 bead. In one embodiment, the pseudo-free C-terminal residue may be of the D or the L
optical configuration; in another embodiment, a racemic mixture of D and L-isomers may be used.

In an n lriiri~mql embodiment, pyrogll~t~nqte may be inrllld~Ad as the N-terminal residue of the peptides of the library. Although pyro~lntqmqt~ is not amenable to seqnenre by Edman 30 degradation, id~Antifi~qtion of the peptide se~u~Anre can be accomplished by a coded library ~ . .. .. . . ... . . . ..

CA 022~1781 1998-10-14 W 097/3S035 PCT~US97/04479 strategy, or by limiting substitution to only 50% of the peptides on a given bead with N-termin~i pyroglllt~m~te, thus leaving enough non-pyrogl~ peptide on the bead for direct seq~lenring. One of ordinary skill would readily recognize that this te~nique could be used for sequenring of any peptide that incorporates a residue resistant to Edman S degradation at the N-terminus. Specific activity of a peptide that cu~ es a blocked N-terminal group, e.g., pyroglutamate, when the particular N-terminal group is present in 50% of the peptides, would readily be demonstrated by colll~aling activity of a completely (100%) blocked peptide with a non-blocked (0%) peptide.

In a further embodiment, s~ulliLs of peptides that confer useful chPmir~l and structural 10 properties will be chosen. For example, peptides COlll~ illg D-amino acids will be lesi~ to L-amino acid-specific proteases in vivo. In addition, the present invention envisions preparing libraries of peptides that have more well defined ~llu-;lulal properties, and the use of pepti-lomim~tirs, and pepti~1omimPtir bonds, such as ester bonds, to prepare libraries with novel properties. In another elllbodillltlll, a peptide library may be g~lle,~led 15 that incorporates a reduced peptide bond, i.e., R,-CH2-NH-R2, where R, and R2 are amino acid residues or seql~enres. A reduced peptide bond may be introduced as a dipeptide subunit. Such a molecule would be l~ L~ to peptide bond hydrolysis, e.g., prolease activity. Such libraries would provide ligands with unique function and activity, such as ext~n~ed half-lives in vivo due to reCi~t~nre to metabolic breakdown. or protease activity.
20 Furthermore, it is well known that in certain systems con~tlàilled peptides show enhqnred functional activity (Hruby, 1982. Life Sciences 31:189-199; Hruby et al., 1990, Biochem J.
268:249-262); the present invention provides a method to produce a constrained peptide that incoll,old~s random sequences at all other positions.

Non-classical amino acids that indllce co,.J~,".anonal Cu~.~lruin~5. The following non-25 cl~scir~l amino acids may be illcol~ul~t~d in the peptide library in order to introduce particular conformational motifs: 1,2,3,4-tetrahydroisoquinoline-3-carboxylate (Kazmierski et al., 1991, J. Am. Chem. Soc. 113:2275-2283); (2S,3S)-methyl-phenylalanine, (2S,3R)-methyl-phenylalanine, (2R,3S)-methyl-phenylalanine and (2R,3R)-methyl-phenylalanine (K~7mierski and Hruby, 1991, Tetrahedron l,ett.); 2-alllhlo~ dhydlùllaL,lllllalene-2-30 carboxylic acid (Landis, 1989, Ph.D. Thesis, University of Arizona); hydroxy-1,2,3,4-tetrahydroi.coquin-line-3-carboxylate (Miyake et al., 1989, J. Takeda Res. Labs. 43:53-76);
~-carboline (D and L) (K~7mi~rski, 1988, Ph.D. Thesis, University of Arizona); HIC

CA 022~1781 1998-10-14 (hi~ti-linP isoquinoline carboxylic acid) (Zechel et al., 1991, Int. J. Pep. Protein Res. 43);
and HIC (hi.~ti-linP cyclic urea) (Dharanipragada).

The following amino acid analogs and peptidomimPtics may be incorporated into a library to induce or favor specific secondary structures: LL-Acp (LL-3-amino-2-propenidone-6-5 carboxylic acid), a ~-turn inducing dipeptide analog (Kemp et al., 1985, J. Org. Chem.
50:5834-5838); ~-sheet inducing analogs (Kemp et al., 1988, Tetrahedron Lett. 29:5081-5082); ~-tum inducing analogs (Kemp et al., 1988, Tetrahedron Lett. 29:5057-S060);
~-helix inducing analogs (Kemp et al., 1988, Tetrahedron Lett. 29:4935-4938); y-turn inducing analogs (Kemp et al., 1989, J. Org. Chem. 54: 109: 115); and analogs provided by 10 the following ,erel.,.,ces: Nagai and Sato, 1985, Tetrahedron Lett. 26:647-650; DiMaio et al.. 1989, J. Chem. Soc. Perkin Trans. p. 1687; also a Gly-Ala turn analog (Kahn et al., 1989, Tetrahedron Lett. 30:2317); amide bond isostere (Jones et al., 1988, Tetrahedron Lett. 29:3853-3856); tretrazol (Zabrocki et al., 1988, J. Am. Chem. Soc. 110:5875-5880);
DTC (S~m~nPn et al., 1990, Int. J. Protein Pep. Res. 35:501 :509); and analogs taught in 15 Olson et al., 1990, J . Am. Chem. Sci. 112:323-333 and Garvey et al., 1990, J . Org.
Chem. 56:436. Collrol,l,ationally restricted mimPti~s of beta turns and beta bulges, and peptides containing them, are described in U.S. Patent No. 5,440,013, issued August 8, 1995 to Kahn.

Determination of the seqllence of peptides that incorporate such non-classical amino acids is 20 readily accomplished by the use of a coded library. Alternatively, a combination of initial Edman degradation followed by amino acid analysis of the residual chain can be used to the structure of a peptide with desired activity. Mass spectral analysis may be employed.

Solid phase supports and linkers. A solid phase support for use in the present invention 25 will be inert to the reaction conditions for synthesis. A solid phase support for use in the present invention must have reactive groups in order to attach a monomer subunit, or for ~tt~ hing a linker or handle which can serve as the initial binding point for a monomer subunit. In one embodiment, the solid phase support may be suitable for in vivo use, i.e., it may serve as a carrier for or support for direct applications of the library (e.g., TentaGel, 30 Rapp Polymere, Tubingen, Germany).

CA 022~1781 1998-10-14 W O 97~5035 PCT~US97/04479 As used herein, solid phase support is not limited to a specific type of support. ~ather a large number of supports are available and are known tO one of ordinary skill in the art.
Solid phase supports include silica gels, resins. derivatized plastic films, glass beads, cotton, plastic beads, alumina gels. A suitable solid phase support may be selected on the 5 basis of desired end use and suitability for various synthetic protocols. For example, for peptide synthesis. solid phase support may refer to resins such as polystyrene (e.g., PAM-resin obtained from Bachem Inc., Pel-in~ Laboldtolics, etc.), POLYHIPE0 resin (obtained from Aminotech, Canada), polyamide resin (obtained from Peninsula Laboratories), poly~ly,cne resin graRed with polyethylene glycol (TentaGel~, Rapp 10 Polymere, Tubingen, Germany) or polydimelllylacrylarnide resin (obtained fromMilligen/Biosearch, California). In a pleft.l.,d embodiment for peptide synthesis, solid phase support refers to polydimethylacrylamide resin.

The solid phase ~uyyOrlS of the invention also comprise a cleavable linker. As used herein, a cleavable linker refers to any molecule that provides spatial distance between the support 15 and the peptide to be synthesi7~ and which can be cleaved to provide for release of the peptide from the support into solution. Linkers can be covalently ~ttArhPd on the solid phase support prior to coupling with a Na-Boc or N~-Fmoc or otherwise ayylopliately protected amino acids. Various linkers can be used to attach the oligomer to solid phase support.

20 Exarnples of spacer linkers include aminobutyric acid, aminocaproic acid, 7-aminoheptanoic acid, and 8-aminocaprylic acid. Fmoc-Ami.-rlcAr~loic acid is co.l.l..clcially available from Bachem Biochem, and is the preferred embodiment. In a further c~bodi~cnt, linkers can additionally col~.ylise one or more ~-alanines as spacers.

In addition, the solid-support could be mctlifiPd to meet specific re4ui-l,l--cl-t~ for the 25 particular purpose of bioassay or detection. MoAifirAtion of solid phase support may be made by incorporation of a specific linker. For example, modified solid phase support could be made acid-sensitive, base-sensitive, nucleophilic-sensitive, electrophilic sensitive, pholosr...~;l;ve, oxidation sensitive or reduction sensitive.

In ~iflitirJn to the linkers described above, selectively cleavable linkers may be employed.
30 For example, an ultraviolet light sensitive linker, ONb. can be used (see Barany and CA 022~1781 1998-10-14 W 097/35035 PCT~US97/04479 Albericio, 1985, J. Am. Chem. Soc. 107:4936-4942). Other cleavable linkers require hydrogenolysis or photolysis. Examples of photosensitive (photocleavable) linkers are found in Wang (1976, J.Org. Chem. 41 :32-58), Hammer et al. (1990, Int. J. Pept. Protein Res. 36:31-45), and Kreib-Cordonier et al. (1990, in Peptides - Ch.,.~ y, Structure and 5 Biology, Rivier and Marshall, eds., pp. 895-897). Landen (1977, Methods Enzym.47:145-149) used aqueous formic acid to cleave Asp-Pro bonds; this d~ul,loach has been used to characterize T-cell delel~ ,a~ in conjunction with the Geysen pin synthesis method (Van der Zee et al., 1989, Eur.J.Immunol. 191:43~7). Other potential linker groups cleavable under basic conditions include those based on p-(hydroxylmethyl) benzoic 10 acid (Atherton et al., 1981, J. Chem. Soc. Perkin I:538-546) and hydroxyacetic acid (Baleaux et al., 1986, Int. J. Pept. Protein Res. 28:22-28). Geysen et al. (1990, J.
lmm-lnol. Methods 134:23-33) reported peptide cleavage by a dike~opi~,.,.d~.i"e mPe~
An enzyme may specifically cleave a linker that cO"l~ises a sequence that is sensitive or a substrate for enzyme cleavage, e.g., protease cleavage of a peptide 15 In certain i.,~l~n-~es, one may derivatize 10-90% of the resin by s~lhstin~tion with the cleavable linker, and the rem~ining 90-10% sllbstituted with a noncleavable linker to ensure that after cleavage of linker enough peptide will remain for seq~-erl-~in~. Preferably, however, a cleavable linker is used in co,l,bi,la~ion with a coded library strategy.
Co",bi"alions of cleavable linkers can also be used to allow sequP.r~ti~l cleaving from a 20 single bead.

Anti~en ~sel~lh-~ Means FosterAntigen Presennng Cells. The human cell line 174xCEM.T2, referred to as T2, contains a mutation in its antigen p,.,ce~i"g pathway that restricts the association of endogenous peptides with cell surface MHC class I molecules (Zweerink et al., 1993, J.
25 Immunol., 150:1763-1771). This is due to a large homozygous deletion in the MHC class II region e.lc.)".l.~cci"~ the genes TAPl, TAP2, LMPl, and LMP2 which are required for antigen p,ese,.l~lion to MHC class I-restricted CD8+ CTLs. In effect, only "empty" MHC
class I molecules are pl~s~ ed on the surface of these cells. Exogenous peptide added to the culture medium binds to these MHC molecules provided that the peptide contains the 30 allele-specific binding motif. These T2 cells are what will be referred to as "foster" APCs.

CA 022~1781 1998-10-14 W O 97/35035 PCT~US97/04479 Retroviral infection or transfection of T2 cells with specific reco..~i"alll MHC alleles allows for redirection of the MHC restriction profile. Libraries tailored to the reco.,~
allele will be prefel~ ially presellled by them because the anchor residues will prevent efficient binding to the endogenous allele. In at least one case, the cell line 174 x CEM.T2 5 was transfected with a mouse H-2Ld MHC allele which rendered the cells sensitive to an H-2Ld restricted CTL clone (Crump~ker et al., 1992, J. Immunol., 148:3004). ThistPchniq~le allows the generation of recombinant foster APCs specific for any MHCrestricted CTL for which the variable chain of the MHC allele is cloned.

It has been de~.~ol~Ll~ted in several cases that transfection of non-p~of~,i,siol1al APCs with 10 allogenic MHC alleles aids greatly in the immunogenicity of the recombinant cell line (Leong et al., 1994. Int. J. Cancer, 59:212-216; Os~rand-Rosenberg et al., 1991, Int. J.
Cancer Suppl., 6:61-68). That is to say, imm--nosensitivity is proportional to the level of expression of the MHC proteins. Thus, recombinant T2 cells should be ideal APCs.
High level expression of MHC molecules makes the APC "more visible" to the CTLs.15 E~,u-esshlg the MHC allele of interest in T2 cells using a pow~lrul lldns,,lil~tional promotor (e.g., the CMV plolll-~tor) results in a more reactive APC (most likely due to a higher concentration of reactive MHC-peptide complexes on the cell surface). Note that since only one type of MHC allele will be able to interact with a given library, the presence of or expression level of the endogenous allele will not colllpr~ ulise specificity if the library is 20 design.od to bind to the newly transfected allele.

Alternative to the generation of suitable APCs are also possible as described:

Cell free antigen presentation. Recently activation of CTLs has been achieved byinrub~ting the antigenic peptide with purified MHC class I molecules complexed to B2-microglobulin (Huang et al., 1994, T~ ily, 1:607-613). In this cell-free MHC/peptide 25 binding assay, it was shown that the Kr~ and Kd approached physiologic levels, reaching equilibrium in 1-2 minutes. This elimin~t~s the need for intact antigen presenting cells and may prove to be more efficient. Since there are numerous prece-1entc in the lilelalule utilizing intact, irradiated APCs to assay peptides in solution, all further discussion will be restricted to this method of antigen prcse.llalion. However, preliminary exp~,li"le,lls will 30 address the c~ln~aldli~/e efficacy of the cell-free approach.

CA 022~1781 1998-10-14 W O 97/3~035 PCT~US97/04479 Antigen Painting. It has been demonstrated that glycosyl-phosphotidylinosi~ol (GPI)-modified proteins possess the ability to reincorporate themselves back into cell ,l,c:",blalles aRer purification. (Medof et al.. J. ~xp. Med., 160:1558-1578). Huang et al.
(I"....l--~ily, 1:607-613) have exploited this property in order to create APCs of specific S composition for the presemation of antigen to CTLs. They devised expression vectors for ,B2-microglobulin and the HLA-A2.1 allele. l'he proteins were expressed in Sç~m.oi-lrr S2 Drasophila melanogaster cells, known to support GPI-modification. After purification, the proteins could be inr-~hat~d together with a purified antigenic peptide which resulted in a trirnolecular complex capable of efficiently inserting itself into the membranes of autologous 10 cells. in essence~ these protein ~ lules were used to "paint" the APC surface, coll~tlling the ability to stim~Jl~t~ a CTL clone that was specific for the antigenic peptide. Cell coating was shown to occur rapidly and to be protein concentration dependent. This method of generating APCs bypasses the need for gene transfer into the APC and permits control of antigenic peptide den~ities at the cell surfaces. It is possible that this approach would allow 15 the scree.lil,g of a greater number of beads/well, since the problem of saturating the MHC
binding sites can be managed by "painting" the APC at a higher MHC/peptide density.

Expression of Signal Accessory Molecules. Research ~rcum~ t~d over the past several years has de~l,o~ ldLed convincingly that resting T cells require at least two signals for ~ induction of cytokine gene expression and proliferation (Schwartz, R.H., 1990, Science, 20 248: 1349-1356; Jenlcins, M.K., 1992, Immunol. Today, 13:69-73). One signal, the one that confers specificity, can be produced by interaclion of the TCR/CD3 complex with an a~ opriate MHC/peptide complex. The second signal is not antigen specific and is termed the "co-stimlll~tnry" signal. This signal was originally defined as an activity provided by bone-marrow-derived accessory cells such as macrophages and dendritic cells, the so called 25 "professional" APCs. Several molecules have been shown to enhance co-stim~ tory activity. These are heat stable antigen (HSA) (Liu, Y., et al., 1992, J. Exp. Med., 175:437-445), chondroitin sulfate-mo-lifiPd MHC invariant chain (Ii-CS) (Naujokas, M.F., et al., 1993, Cell, 74:257-268), intr~(~ell~ r adhesion molecule 1 (ICAM-l) (Van Seventer, G.A., 1990, J. lmmnnol., 144:4579-4586), B7-1, and B7-2/B70 (Schwartz, R.H., 1992, 30 Cell, 71: 1065-1068). These molecules each appear to assist co-stimnlAtion by interacting with their cognate ligands on the T cells. The stim~ tory ability of APCs prepared by either method described above may be enh~nred by the introduction of genes that have been shown to provide co-stim--l~tory signals. The benefits of such enh~nrem~nt may include CA 022~1781 1998-10-14 the need for lower peptide cullcelll,dtions and an improved signal-to-noise ratio in large-scale screens.

Method of Detection and I~lçntifit~tion of Reactive Oligopeptides In general, a screening assay of the invention will involve the steps of contacting antigen 5 presentation means, e.g., antigen plesenlillg cells. with a limited number of individual beads from a library (either a primary library with the greatest degeneracy of structure types, or a secondary library with degenclaLe residues of limited chPmil~l types), cleaving peptides from the library to bind to MHC molecules of the antigen l,lesentation means, and ~imlllt~n-oously or subsequently contacting the MHC-peptide complexes to the CTL(s) of 10 interest. Various methods are known to evaluate CTL activation, including but not limited to tritiated thymidine incorporation (indicative of DNA synthesis), and examination of the population for growth or proliferation, e.g., by identifir~tion of colonies. In another embodiment, the tetrazolium salt MTT (3-~4,5-dimethyl-thazol-2-yl)-2,5-diphenyl tetrazolium bromide) may be added (Mossman, 1983, l. lmmllnol Methods 65:55-63; Niks 15 and Otto, 1990, J. lmmlmol. Methods 130:140-151). Succinate dehydrogenase, found in mitochondria of viable cells, converts the MTT to formazan blue. Thus, concentrated blue color would indicate metabolically active cells. In yet another embodiment, incorporation of radiolabel, e.g., tritiated thymidine, may be assayed to indicate proliferation of cells.
Similarly, protein synthesis may be shown by incorporation of 35S-methionine. In still 20 another embodiment, cytotoxicity and cell killing assays, such as the classical chlullliu release assay, may be employed to evaluate epitope-specific CTL activation.

A brief overview of a specific embodiment of the assay is as follows: roughly speaking, ten 96-well plates with 1000 beads per well will accommodate lo6 beads; ten 96-well plates with 100 beads per well will accommodate 105 beads. In order to minimi7e both the 25 number of CTL cells required per screen and the amount of manual manipulations, the eluted peptides can be further pooled to yield wells with any desired complexity. For example, based on expelilll~llL~ with soluble libraries, it should be possible to screen 107 peptides in 96-well plates (10,000 peptides per well) with as few as 2 X lo6 CTL cells.
After cleaving a pe~ ge of the peptides from the beads, incubating them with 30 garnma-irradiated foster APCs and the cloned CTL line(s), positive wells determined by 3H-thymidine incorporatiOn will be further ex~min~d. Alternatively, as pointed out above, cytokine production or cytolytic 5'Cr-release assays may be used (Coulie et al., 1992, Int~ J.

CA 022~1781 1998-10-14 W 0 97~5035 PCTAJS97/04479 Cancer, 50:289-291). Beads from each positive well will be separated and assayedindividually as before, utilizing an adtlition~l pe.c~ ge of the peptide from each bead.
Positive individual beads will then be decoded, identifying the reactive amino acid sequence. Analysis of all positives will give a partial profile of conservatively substituted 5 epitopes which stim~ te the CTL clone tested. At this point, the peptide can be resynrh~si7ed and retested. Also, a second library (of minimal complexity) can be syntht~si7ed with replese~ ons of all conservative substitutions in order to enumerate the complete spectrum of derivatives tolerated by a particular CTL. By screening multiple CTLs (of the same MHC restriction) sim~llt~nPously~ the search for c,u~sleaclhlg epitopes is 10 greatly facilitated.

The specific details of the preferred assay are as follows: the library being screened is selected to match (in terms of anchor residues) the MHC allele expressed by the tumor.
The library is plated out in 96-well flat bolLo,.led plates at a density of approximately 100 to about 5000, preferably 1000-5000, beads per well (i.e., 20-1000, p,~r, ldbly 20-100, plates 15 for a library of 107 peptides) . Each bead, on average, contains 200 pmol of peptide, so release of 50% of the product yields 100 pmol of solubilized peptide per well. The final assay volume can be 100-200,u1, so 100 pmol results in a final conce~ Lion of 0.5-1 ~LM.
At this conce.ltlalion, the peptide will spontaneously bind to the MHC molecules on the surface of the foster APCs. For the purposes of conserving peptide, the reaction may be 20 supple~..f .l~d with free ~2-microglobulin, allowing the peptides to bind at concentrations of .01-0.1~LM (Rock et al., 1992, Analysis of the association of peptides of optimal length to class I molecules on the surface of cells., Proc. Natl. Acad. Sci. USA, 89:8918-8922).
Note that an extremely small number of MHC molecules need to be occupied with the peptide in order to elicit a re~,uonse because a single complex can serially engage and 25 trigger up to approximately 200 TCRs (Valitut~i et al., 1995, Serial triggering of many T
cell receptors by a few peptide-MHC complexes., Nature, 375:148-151). The method used for the partial release depends on the type of cleavable linker used. At this point, in order to be conservative with CTLs, wells are further pooled so that each well contains ~ 10,000 peptides. In this way, 10' beads can be screened in ten 96-well plates. The master plates 30 cont~ining the beads and 1 ~ ~ ght~r plates containing unused peptide can be stored at -70~C for reuse. The 2~ d~ hter plates containing the pooled peptides are ready for SC,~.,lli~lg. These plates may be prepared in advance and stored under the a~lu~Jlia conditions so that the library is ready whenever the CTLs are at peak activity.

CA 022~1781 1998-10-14 Strategy 1. The supernatant from each well is distributed to replica plates and 1-2x1~3 irradiated (1500 rads) foster APCs (expressing the proper MHC allele) are added to each well. Next, the cloned CTLs are added to a total of 103-104 cells le~le~cllli,.g equal al~uunLs of 10-20 different clones of the same MHC restriction such that the total final 5 volume per well is 200 ~1 and the plates are incubated in a hllmirlifi~d CO, incubator for 4 days at 37~C. Each well is then pulsed with 18.5 k~q of ~3H] dThd to measure CTLproliferation. 16 hours later, the radioactivity incorporated into the DNA of mitotically active CTLs is assayed by scintillation cûunting (Estaquier et al., 1994, The mixotope: a co",l,i~.atorial peptide library as a T cell and B cell immunogen., Eur. J. Immunol., 10 24:2789-2795). The m~nit~lde of the proliferative response may serve as a preli,llh~a,y screen for crossreacting epitopes. The greater the ,esl)onse the more likely it is that more than one CTL clone was stiml-l~tçd. While all reactive peptides are of interest. the most efficacious vaccine c~n~ t-os will be those that clOSSL. act with CTLs derived from independent donors and which are restricted by the most common MHC alleles. Note that 15 it~ ir~ -l ion of epitopes cont~ining the HLA B7-like supermotif (see TABLE 2) would be of great value as vaccine c~ntli-~t~os since it will bind to many HLA B alleles which are r~ es~"led in over 40% of individuals from all major ethnic groups (Sidney et al., 1995, Several HLA alleles share overlapping peptide specificities., J. Immunol., 154:247-259).

Strategy 2. Alternatively, the first step is to a~ 5'Cr-labeled T2 cells to the wells of 20 the 2~ daughter plates, followed by the addition of the CTLs. After 4 hours the released 5'Cr is measured in the standard manner. When a positive well is identified, the 10 wells from the 1~ ti~lleht~r plate that correspond to that well are similarly assayed. At this point, the epitope search is narrowed down to the beads in a single well on one of the master plates.
Wells that register positive will be further analyzed as follows: The beads that correspond to the positive well are manually disl~ d (1 per well) to new plates and the rem~ining peptide is released from each. These plates are assayed as before, and in this way the reactive bead(s) are unambiguously isolated. The positive bead(s) can be rapidly and 30 efficiently decoded since the molecular tags that encode the bead's synthesis history has rem~ined on the bead (coupled with a non-photocleavable cros.clinker). For example, analysis of the bead(s) by electron capture capillary gas chromatography immptli~tely reveals the peptide sequence that was synthtoci7f cl on that bead (Ohlmeyer et al., 1993, CA 022~1781 1998-10-14 supra). Thus the unambiguous i.1entifir~tion of an epitope can be achieved in approximately ten days using the 3H-thymidine incorporation assay and in as few as two days if a 5'Cr-release assay is used.

In another embodirnent~ application of the library beads to the surface of freshly poured top 5 agar in a standard tissue culture plate, followed by release of a portion of the peptide, will result in a three dimensional concentration gradient of eluted peptide around each bead.
Antigen l,lese..ii-lg cells could be present in the top agar or applied to the surface after peptide release. Next, the CTL(s) of interest are plated over the top agar/peptide/APCs, followed by inrllh~tion at 37~C for 4-12 hours. Reactive beads may be detected by the 10 formation of plaques, where the size of the plaque in~ir~tPs the m~gninlde of the lesl)onse.
Positive beads can then be taken from the plate, washed, and sequenced. This assay requires very little manual manipulation of the beads and the entire library can be scleened sim-llt~nPoucly (in one step) in as little as four hours. Furtherrnore, the beads can be recovered, washed in 6M gu~n~ m, and reused.

15 In another embodiment, the described method for the identification of CD8+ MHC Class I-restricted CTL epitopes can be applied to the i-lPn-ific~ion of CD4+ MHC Class Il-restricted helper T-cell (Th) c~ o~cs. In this case, MHC Class II allele-specific libraries are synrhpci7pd such that haplotype-specific anchor residues are r~ e~ led at the a~plop.iale positions. MHC Class Il agretopic motifs have been identified for the comrnon 20 alleles (R~mmPncee, 1995, Curr. Opin. Imrnunol. 7:85-96; Altuvia et al., 1994, Mol.
Imm-lnol. 24:375-379; Reay et al., 1994, J. Immunol. 152:3946-3957; Verreck et al., 1994, Eur. J. Irnmnol. 24:375-379; Sinig~Eli~ and Hammer, 1994, Curr. Opin. Irnmunol.

6:52-56; Rotzschke and Fallc, 1994, Curr. Opin. Tmmllnol. 6:45-51). The overall length of the peptides will be 12-20 amino acid residues, and previously described mPtho-lc may be 25 employed to limit library complexity. The screening process is identic~l to that described for MHC Class l-associated epitopes except that B Iymphoblastoid cell lines (B-LCL) are used for antigen presentation rather than T2 cells. In a plere,.~d aspect, previously characterized B-LCLs that are defective in antigen processing (Mellins et al., 1991, J. Exp.
Med. 174:1607-1615); thus allowing specific p.~s~;..l~lion of exogenously added antigen, 30 are employed. The libraries are screened for reactivity with isolated CD4+ MHC Class 11 allele-specific Th cells. Reactivity may be measured by 3H-thymidine incorporation . .

CA 022~1781 1998-10-14 W 0 97~5035 PCT~US97/04479 according to the method of Mellins et al. (sl~pra), or by any of the mf~thor~s previously described for MHC Class I-associated epitope sc~ee,li-.g.

The final step is to synthesize a library (of minim~l complexity~ which lC~ Sc~
conservatively s-lhsrirll~ed derivatives of the identified epitope in order to isolate the most 5 efficient cytolytic stim~ tor of the CTL clone(s). This second library can be used to sort out which of the CTLs originally assayed are responding to the peptide as well as identify the most efficacious peptide derivative. Note that the naturally occurring epitope may not be the most efficieM stimlll~tor. If one organizes the amino acids into ch~lir~lly related groups, col.lposilion and complexity of the derivative libraries can be readily c~rul~t.od.
10 The amino acid groupings chosen for the design of secondary screen libraries is shown in TABLE 3. This table provides for the design of derivative libraries that are diverse yet easily manageable in terms of size. The amino acids can be loosely grouped according to their physic~rh~mir~l properties as follows:

Amino Acid Group Class A,V,L,I,P,F,W,M NONPOLAR SIDE CHAINS
G,S,T,C,Y,N,Q UNCHARGED POLAR SIDE CHAINS
D,E NEGATIVELY CHARGED SIDE CHAINS
K,R,H POSITIVELY CHARGED SIDE CHAINS

20 For example, a derivative library for the sequence YLKDQQLL (SEQ ID NO: 10), actually a known HLA-B8 epitope, would look like this:
X,X2K X3X4X5X6L (SEQ ID NO: 11) Xl = G,S,T,C,Y,N,Q
X2 = A,V,L,I,P,F,W,M
25 X3 = D,E, X4 = G,S,T,C,Y,N,Q, X5 = G,S,T,C,Y,N,Q
X6 = A,V,L,I,P,F,W,M
where the HLA-B8 anchor residues are shown in bold type. This library would have a 30 complexity of 43,904. De~ hlillg the complete spectrum of reactive derivatives will provide inforrnation as to the extent and limits of TCR promiscuity and allow the design of better primary screen libraries.

CA 022~1781 1998-10-14 Therapeutics With Identified Peptide Epitopes Cancer treatment andprevennon. Cancer cells contain many new antigens potentially recognizable by the immune system. Given the speed with which epitopes can be i-ientifi~ custom anti-cancer vaccines can be generated for affected individuals by isolating 5 TILs from patients with solid tumors, determining their MHC restriction, and assaying these CTLs against the applo~liale library for reactive epitopes. The short time frame heralds a new therapeutic clinical tl~dtlllell~ modality for cancer patients. These vaccines will be both tlt~ for affected individuals as well as preventive therapy against recL~ ce (or establi~hm~nt of the disease in patients which present with a familial genetic 10 predisposition to it). Inoculation of individuals who have never had the cancer is expected to be quite successful as preventive therapy, even though a tumor antigen-specific CTL
response has not yet been elicited, because in most cases high affinity peptides seem to be immunogenic suggesting that holes in the functional T cell repertoire, if they exist, may be relatively rare (Sette et al., 1994, The relationship between class I binding affinity and 15 imrnunogenici~y of potential cytotoxic T cell epitopes., J. Immunol., 153:5586-5592). In mice, vaccination with a~ ol),iate epitopes not only elimin~tPs established tumors but also protects against tumor re-establichm~nt after inocnl~tion with otherwise lethal doses of tumor cells (Bystryn et al., 1993, supra).

Recent advances in vaccine adjuvants provide effective means of .~ .hlg peptides so 20 that they impact maximally on the immune system (Del-Giudice~ 1994. Hsp70: a carrier molecule with built-in adjuvanticity., Experientia, 50:1061-1066). These peptide vaccines will be of great value in treating mf~t~t~ tumors that are generally ulllts~uonsive to conventional therapies. Note that tumors arising from the homozygous deletion of recessive oncogenes are not likely to betray themselves to a humoral (antibody) response and would 25 thus be treated more effectively by eliciting a cellular, CTL response. The ability to catalog large numbers of CTL epitopes with this technology will allow the i~entifi~tion of widely cross-reactive epitopes that are shared between independently derived tumors. In the case of mPl~n-~m~ there is recent evidence that the same T-cell-defined tumor antigens are e~lessed by independent human melanoma and breast cancers suggesting that ~r~n~rolll~ation-associated events may give rise to recurrent expression of the same tumor antigen in different tumors of related tissue and cellular origin (Sahasrabudhe et al., 1993, sup~a).

CA 022=,1781 1998-10-14 WO 97/3503~ PCT/US97/04479 Viral Diseases. Viral infections are ideal c~n~ tç~ for immunotherapy. Immunological responses to viral pathogens are sometimes ineffective as in the case of the lentiviruses such as HIV which causes AIDS. The high rates of spontaneous ml-t~tion make these viruses elusive to the im~nune system. However, a saturating profile of CTL epitopes p~ ,e.lL~d on 5 infected cells will identify shared antigens among dirrelel,~ selc lyl~es in P~senti~l genes that are largely intolerant to mutation which would allow the design of more efr~ ., vaccines.

Autoimmune Diseases. These are diseases in which the body's immune system ~e~,~ollds against self tissues. They include most forms of arthritis, ulcerative colitis, and multiple sclerosis. This technology can identify the endogenous elements that are recognized as 10 foreign - a giant step towards the development of L~ ..lr~ ; using gene therapy or other approaches. One of our interests is the design of synthetic CTL epitopes which can act as "suicide substrates" for CTLs that mediate au~oimmunity. That is to say, peptides which have a high affinity for the MHC allele but fail to activate the TCR could effectively mask the cellular immune l~,.,ponse against cells presc.l~ing the antigen in question. In support of 15 this approach, it is believed that the long latency period of the HIV virus is due to an antiviral immune lc~ollse and a m~ ... by which the virus finally evades the immune system is by ~;e~ illg epitopes that occupy the MHC molecules but do not stim~ e a TCR Iytic response, inducing specific T cell anergy (Klen~ ".~n et al., 1995, The effects of natural altered peptide ligands on the whole blood cytotoxic T Iymphocyte response to 20 human immnnol1ifirien~y virus., Eur. J. Immunol.. 25:1927-1931).

Diagnostic ~eagents. Defined CTL epitopes can be used to clinically characterize tumors and viral pathogens in order to determine, in advance, the predicted efficacy of an in vivo vaccine trial. This can be achieved by a simple proliferation assay of a patient's peripheral blood mono,-uclear cells using defined CTL epitopes as stiml~l~tors. Peptides which elicit a 25 response are viable vaccine c~n~ t~s for that patient. Cataloging large numbers of CTL
epitopes of defined MHC restrictions, as can be achieved with this technology, will make feasible the rapid typing and customized vaccine formulation for affected or gen~ti~lly predisposed individuals.

Identification of Tumor Genes. It is predicted that optimally reactive peptides will, more 30 often than not, reflect the structure of naturally occurring epitopes. This is because, in general, there are fewer ways to generate gain-of-function rather than loss-of-function -CA 022~1781 1998-10-14 W O 97/35035 PCTrUS97/04479 mutations by amino acid substitutions. It is possible to clone genes which contain the defined epitope within their seqn~nre by classical methods (i.e., hybridization of synthetic oligonucleotides to phage libraries, RT-PCR, antibody screening of phage e~ c~sion libraries, etc.). Since many proteins will contain processing sites which will gene.~tt 5 peptides that bind to a variety of MHC alleles, vaccination with the complete protein from which the natural epitope was derived will allow the design of vaccines which can largely overcome the problem of MHC restriction. It may be possible to identify these proteins based on their ~cplese~ tion in currently available sequence ~l~t~b~ces (i.e., genbank, PIR, Swiss-Prot, etc.). Protein seqllenres which contain more than one i(lentified epitope (or 10 derivatives of i~lentifi~d epitopes) would be strong c~n~ t~s for vaccines which may be independent of MHC restriction.

It is expected that the i-1Pntifir~tion of proteins from tumor cells or virally infected cells that are targeted by the jmmune system will identify genes that play a direct role in their pl~,sellled abnormal phenotypes. Recently (Wolfel, et al., 1995, A pl6'NK4a-illse"sili./e 15 CDK4 mutant targeted by cytolytic T Iymphocytes in a human melanoma. Science,269:I281-4), an HLA-A2.1-restricted human anti-melanoma CTL epitope which colle~ol1ds to a UV-specific mutation in the cyclin-dependent kinase 4 gene (CDK4) was identified. This is the first example of the identification of a gene ~ .onsible for tumorogenesis by isolation and analysis of an anti-tumor CTL epitope. The ~ltili7~ti~n of 20 CTLs may be an effective means of pursuing tumor genes, possibly more effective than the conventional te~hni~luec of subtractive hybridization or lc~lesç~ tional difference analysis (Lisitsyn et al., 1993, Cloning the dirr~ .ces between two complex genomes., Science, 259:946-51). Knowledge of these genes will aid in underss~n(ling the molecular mPf h~nicmc underlying tumorogenesis and may suggest other clinical treatment mo-l~litiçs 25 such as gene therapy.

Induction of Active Immunity Through CTL Infusion. Objective ~ntitllmor responses can be observed when TILs are infused with IL-2 in patients with met~ct~tir melanoma (Roscllbelg, et al., 1991, In Biologic Therapy of Cancer, Devita et al., eds. Phili~lrhi~
Lippincott, pp. 214-236; Rosenbelg, N. Engl. 1. Med., 1988; 319:1676-1680). The 30 proliferation of TILs in vitro is dependent on the persi~Lel~t presence of antigen.
Traditionally, the source of antigen is irradiated cells grown from the tumor from which the TILs were isolated. Most of the ~ntitllmnr CTLs obtained thus far have been g~lle~alcd CA 022~1781 1998-10-14 W O 97~5035 PCT~US97/04479 against m,o!~nom~c, because met~ct~tir melanoma cells are rather easy to adapt to culture, providing convenient sources of antigen for CTLs with unknown epitopes (Van Pel, et al., 1995 lmmnnol. Rev. 145:229-250). It is critical to obtain antigen sources in order to stimlll~te CTLs to expand clones for use in CTL infusion therapy. In most cases (e.g., 5 prostate cancer, pancreatic carcinoma, lung tumor~, ovarian cancer), it is difficult to establish continuous cultures from primary explaMs. In these cases, the method described above is unique in that parental tumor cell line is required only for an initial small-scale expansion of the CTLs and to test the efficacy of the identified peptide epitopes.

The present invention is not to be limited in scope by the specific embodiments describe 10 herein. Indeed, various modifir~tions of the invention in addition to those described herein will become appa~elll to those skilled in the art from the foregoing description and the acco...l)~.,ying figures. Such modifications are inten-l~d to fall within the scope of the appended claims.

Various publications are cited herein, the disclosures of which are incorporated by 15 ~ ence intheir entireties.

CA 022~l78l l998-l0-l4 W O 97/35035 37 PCT~US97/04479 ~:Qu~ LISTING

( I ) ~F~F.RAr. INFORMATION:
(i) APPLICANT: Nicolette, Charles A.
(ii) TITLE OF INVENTION: METHOD FOR IDkNll~YlNG CYTOTOXIC T-CELL
EPITOPES
(iii) NUMBER OF S:Q~kN~:S: 11 (iv) CORRESPONDENCE ADDRESS:
(A) ADDRESSEE: David A. Jackson, Esq.
(B) STREET: 411 Hackensack Ave, Continental Plaza, 4th Floor (C) CITY: Hackensack (D) STATE: New Jersey (E) CO~NTRY: USA
(F) ZIP: 07601 (v) CONPUTER READABLE FORM:
(A) MEDIUM TYPE: Floppy disk (B) COMPUTER: IBM PC compatible (C) OPERATING SYSTEM: PC-DOS/MS-DOS
(D) SOFTWARE: PatentIn Release #1.0, Version #1.30 (vi) CURRENT APPLICATION DATA:
(A) APPLICATION NUMBER:
(B) FILING DATE: 20-MAR-1997 (C) CLASSIFICATION:
(viii) ATTORNEY~AGENT INFORMATION:
(A) NAME: Jackson Esq., David A.
(B) REGISTRATION NUMBER: 26,742 (C) REFERENCE/DOCKET NUMBER: 1302-1-001 PCTP
(ix) TELECOMMUNICATION INFORMATION:
(A) TELEPHONE: 201-487-5800 (B) TELEFAX: 201-343-1684 (2) INFORMATION FOR SEQ ID NO:1:
(i) S~QuL~ CHARACTERISTICS:
(A) LENGTH: 8 amino acids (B) TYPE: amino acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: peptide (iii) HYPOTHETICAL: NO

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:1:

CA 022~1781 1998-10-14 W O 97/35035 PCT~US97/04479 Leu Xaa Xaa Xaa Xaa Xaa Xaa Val (2) INFORMATION FOR SEQ ID NO:2:
(i) ~QU~N~: CHARACTERISTICS:
(A) LENGTH: 8 amino acids (B) TYPE: amino acid (C) STR~NnFnN~.SS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: peptide (A) DESCRIPTION: Xaa at Position 8 is a positively charged residue (iii) HYPOTHETICAL: NO

(xi) ~:QU~N~ DESCRIPTION: SEQ ID NO:2:
Arg Xaa Xaa Xaa Xaa Xaa Xaa Xaa (2) INFORMATION FOR SEQ ID NO:3:
Q~N~ CHARACTERISTICS:
(A) LENGTH: 9 amino acids (B) TYPE: amino acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: peptide (A) DESCRIPTION: Xaa at Position 2 is Glu or Asp; Xaa at Position 9 is Phe, Lys, or Tyr (iii) HYPOTHETICAL: NO

(xi) ~QU~C~ DESCRIPTION: SEQ ID NO:3:
Xaa Xaa Xaa Xaa Xaa Xaa Xaa Xaa Xaa (2) INFORMATION FOR SEQ ID NO:4:
(i) ~Qu~ CHARACTERISTICS:
(A) LENGTH: 8 amino acids (B) TYPE: amino acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: peptide CA 022~1781 1998-10-14 W 097/35035 39 PCTrUS97/04479 (iii) ~Y~O~ CAL: NO

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:4:
Arg Xaa Xaa Xaa Xaa Xaa Xaa Leu (2) INFORMATION FOR SEQ ID NO:5:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 8 amino acids (B) TYPE: amino acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: peptide (A) DESCRIPTION: Xaa at Position 2 is Lys or Arg; Xaa at Position 8 is Leu or Ile ~iii) HYPO~ lCAL: NO

(xi) ~Q~NC~: DESCRIPTION: SEQ ID NO:5:
Xaa Xaa Xaa Xaa Xaa Xaa Xaa Xaa (2) INFORMATION FOR SEQ ID NO:6:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 8 amino acids (B) TYPE: amino acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: peptide (A) DESCRIPTION: Xaa at Position 1 is Met or Leu (iii) HYPOTHETICAL: NO

(xi) ~Qu~:N~ DESCRIPTION: SEQ ID NO:6:
Xaa Xaa Xaa Xaa Xaa Xaa Xaa Lys (2) INFORMATION FOR SEQ ID NO:7:
(i) ~U~ CHARACTERISTICS:
(A) LENGTH: 8 amino acids (8) TYPE: amino acid CA 022~1781 1998-10-14 PCTrUS97/04479 (C) STRAND~DNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: peptide (A) DESCRIPTION: Xaa at Position 8 is Phe or Tyr (iii) HYPOTHETICAL: NO

(xi) ~:Q~N~ DESCRIPTION: SEQ ID NO:7:
Glu Xaa Xaa Xaa Xaa Xaa Xaa Xaa (2) INFORMATION FOR SEQ ID NO:8:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 8 amino acids (B) TYPE: amino acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: peptide (A) DESCRIPTION: Xaa at Position 8 is Phe, His, Trp, or Tyr (iii) HYPOTHETICAL: NO

(xi) ~Q~ DESCRIPTION: SEQ ID NO:8:
Xaa Pro Xaa Xaa Xaa Xaa Xaa Xaa (2) INFORMATION FOR SEQ ID NO:9:
(i) ~Q~ CHARACTERISTICS:
(A) LENGTH: 7 amino acids (B) TYPE: amino acid (C) STRAN~:~N~SS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: peptide (A) DESCRIPTION: Xaa at Position 1 is Ile or Leu; Xaa at Position 7 is His or Lys (iii) HYPOTHETICAL: NO

(xi) ~Uu~N~ DESCRIPTION: SEQ ID NO:9:
Xaa Xaa Xaa Xaa Xaa Xaa Xaa CA 022~l78l l998-l0-l4 (2) INFORMATION FOR SEQ ID NO:l0:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 8 amino acids (B) TYPE: amino acid (C) STR~Nn~nN~.~S: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: peptide (iii) HYPOTHETICAL: NO

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:l0:
Tyr Leu Lys Asp Gln Gln Leu Leu l 5 (2) INFORMATION FOR SEQ TD NO:ll:
Uu~N~ CHARACTERISTICS:
(A) LENGTH: 8 amino acids (B) TYPE: amino acid (C) STR~Nn~nNFSS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: peptide (A) DESCRIPTION: Xaa at Position 1 is Gly, Ser, Thr, Cys, Tyr, Asn or Gln; Xaa at Position 2 is Ala, Val, Leu, Ile, Pro, Phe, Trp, or Met;
Xaa at Position 4 is Asp or Glu; Xaa at Position 5 is Gly, Ser, Thr, Cys, Tyr, Asn, or Gln; Xaa at Position 6 is Gly, Ser, Thr, Cys, Tyr, Asn, or Gln; Xaa at Position 7 is Ala, Val, Leu, Ile, Pro, Phe, Trp, Met (iii) HYPOTHETICAL: NO

(xi) ~uU~C~ DESCRIPTION: SEQ ID NO:ll:
Xaa Xaa Lys Xaa Xaa Xaa Xaa Leu l 5

Claims (28)

WHAT IS CLAIMED IS:

1. A method for identifying a cytotoxic T cell epitope comprising the steps in order of:
a) contacting a population of at least two cytotoxic T cells having the same MHC-haplotype restriction with i) a library of molecules attached to solid phase supports by a releasable linker, wherein each solid phase support is attached to a single species of molecule, and wherein the structure of the molecule can be determined, which library of molecules contains a conserved structural motif corresponding to a structural motif characteristic of peptides that associate with the MHC-haplotype to which the cytotoxic T cells are restricted; and ii) antigen presentation means, which antigen presentation means correspond to the MHC-haplotype to which the cytotoxic T cells are restricted;
wherein the solid phase supports of the library are in separate fractions;
b) cleaving at least a portion of the releasable linker so as to release at least a portion of the molecule;
c) evaluating whether the cytotoxic T cells recognize a molecule present in one or more of the fractions of the library of molecules;
d) isolating one or more solid phase support supports from the fractions; and e) determining the structure of a molecule on a solid phase support isolated from the fraction.

2. The method according to claim 1, wherein the cytotoxic T cells are selected from the group consisting of a) polyclonal T cells isolated from a site of cytotoxic T cell infiltration from an individual;
b) cells isolated from a site of cytotoxic T cell infiltration from two or more individuals, which two or more individuals share an MHC haplotype;
c) two or more cytotoxic T cell lines; and d) any combination thereof.

3. The method according to claim 2, wherein the site of cytotoxic T cell infiltration is a tumor.

4. The method according to claim 1, wherein the molecules are peptides.

5. The method according to claim 4, wherein the peptides comprise subunits selected from the group consisting of glycine, L-amino acids, D-amino acids, non-classical amino acids, and peptidomimetics.

6. The method according to claim 1, wherein the solid phase support is selected from the group consisting of polystyrene resin, poly(dimethylacryl)amide-grafted styrene-co-divinylbenzene resin, polyamide resin, polystyrene resin grafted with polyethylene glycol, and polydimethylacrylamide resin.

7. The method according to claim 1, wherein the releasable linker releases upon exposure to an acid, a base, a nucleophile, an electrophile, light, an oxidizing agent, a reducing agent, or an enzyme.

8. The method according to claim 1, wherein the structural motif contained in the library of molecules is selected from the group consisting of LXXXXXXV (SEQ ID NO:1);
RXXXXXX+ (SEQ ID NO:2); X(D,E) XXXXXX(F,K,Y) (SEQ ID NO:3); RXXXXXXL
(SEQ ID NO:4); X(K,R)XXXXX(L,I) (SEQ ID NO:5);(M,L)XXXXXXK (SEQ ID
NO:6); EXXXXXX(Y,F) (SEQ ID NO:7); XPXXXXX(F,H,W,Y) (SEQ ID NO:8);
(L,I)XXXXX(H,K) (SEQ ID NO:(9); wherein X indicated any amino acid residue, and +
indicates a positively charged amino acid residue.

9. The method according to claim 4, wherein a limited number of representative amino acid residues are incorporated in the peptides of the library.

10. The method according to claim 9, wherein positively charged amino acid residues are substituted with an amino acid selected from the group consisting of lysine, arginine, and bistidine; negatively charged amino acid residues are substituted with an amino acid selected from the group consisting of aspartic acid and glutamic acid; neutral, polar amino acid residues are substituted with an amino acid selected from the group consisting of asparagine, glutamine, serine, threonine, tyrosine, and cysteine; nonpolar amino acid residues are substituted with an amino acid selected from the group consisting of alanine, valine, leucine, isoleucine, proline, phenylalanine, tryptophan, and methionine.

11. The method according to claim 10, wherein the nonpolar, aromatic amino acid residues are substituted with an amino acid selected from the group consisting of tyrosine, threonine, and tryptophan; and the nonpolar aliphatic amino acid residues are substituted with an amino acid selected from the group consisting of alanine, valine, leucine, isoleucine, proline, and methionine.

12. The method according to claim 1 further comprising a coding molecule attached to each solid phase support of the library, which coding molecule defines the structure of the molecule attached to the solid phase support by the releasable linker.

13. The method according to claim 12 wherein the coding molecule is selected from the group consisting of a peptide and an oligonucleotide.

14. The method according to claim 10, wherein the coding molecule is an inert molecular tag that can be decoded by gas-phase chromatography.

15. The method according to claim 1, wherein the antigen presentation means is selected from the group consisting of a purified MHC class I molecule complexed to .beta.2-microglobulin; an intact antigen presenting cell; and a foster antigen presenting cell.

16. The method according to claim 1, wherein the antigen presentation means is afoster antigen presenting cell.

17. The method according to claim 16, wherein the foster antigen presenting cell lacks antigen processing activity, whereby it expresses MHC molecules free of bound peptides.

18. The method according to claim 17 wherein the foster antigen presenting cell is cell line 174xCFM.T2.

19. The method according to claim 1, wherein the recognition of a molecule present in one or more of the fractions of the library of molecules by the cytotoxic T cells is evaluated by detecting cytotoxic T cell activation.

20. The method according to claim 19 wherein cytotoxic T cell activation is detected by a method selected from the group consisting of 3H-thymidine incorporation; metabolic activity detected by conversion of MTT to formazan blue; increased cytokine mRNAexpression; increased cytokine protein production; and chromium release by target cells.

21. The method of claim 1 wherein the structure of the molecule is determined byanalyzing a portion of the molecule remaining on the solid phase support.

22. The method according to claim 4, wherein a sequence of the peptide is determined by sequencing a portion of the peptide remaining on the solid phase support.

23. The method according to claim 12, wherein the structure of the molecule is determined by analyzing the structure of the coding molecule.

24. The method according to claim 1, wherein the structure of the molecule is determined after isolating more than one candidate solid phase support; repeating steps a) through c), isolating one solid phase support in step c), and determining the structure of a molecule on the solid phase support isolated in step c).

25. The method according to claim 9, further comprising the steps in order of:

a) contacting the population of at least two cytotoxic T cells having the same MHC-haplotype restriction with i) a library of molecules attached to solid phase supports by a releasable linker, wherein each solid phase support is attached to a single species of molecule, and wherein the structure of the molecule can be determined, which library of molecules contains a conserved structural motif corresponding to a structural motif characteristic of peptides that associate with the MHC-haplotype to which the cytotoxic T cells are restricted, and wherein every amino acid corresponding to the representative residue is utilized at the position identified for the corresponding representative residue; and ii) antigen presentation means, which antigen presentation means correspond to the MHC-haplotype to which the cytotoxic T cells are restricted;
wherein the solid phase supports of the library are in separate fractions;
b) cleaving at least a portion of the releasable linker so as to release at least a portion of the molecule;
c) evaluating whether the cytotoxic T cells recognize a molecule present in one or more of the fractions of the library of molecules;
d) isolating one or more solid phase support supports from the fractions; and e) determining the structure of a molecule on a solid phase support isolated from the fraction.

26. A method for identifying a high affinity cytotoxic T cell epitope comprising:
a) contacting a population of cytotoxic T cells having an MHC-haplotype restriction with i) a library of molecules attached to solid phase supports by a releasable linker, wherein each solid phase support is attached to a single species of molecule, and wherein the structure of the molecule can be determined, which library of molecules contains a conserved structural motif corresponding to a structural motif characteristic of peptides that associate with the MHC-haplotype to which the cytotoxic T cells are restricted, and wherein every amino acid corresponding to a representative residue determined according to the method of claim 9 is utilized at the position identified for the corresponding representative residue; and ii) antigen presentation means, which antigen presentation means correspond to the MHC-haplotype to which the cytotoxic T cells are restricted;
wherein the solid phase supports of the library are in separate fractions;
b) cleaving at least a portion of the releasable linker so as to release at least a portion of the molecule;
c) evaluating whether the cytotoxic T cells recognize a molecule present in one or more of the fractions of the library of molecules;
d) isolating one or more solid phase support supports from the fractions; and e) determining the structure of a molecule on a solid phase support isolated from the fraction.

27. A method of identifying a protein antigen comprising:
a) identifying the cytotoxic T cell epitope of the protein according to the method of claim 25;
b) comparing a sequence of the T cell epitope identified in step (a) with known sequences of proteins; and c) determining a protein having a sequence corresponding to the sequence of the T cell epitope.

28. A method of identifying a protein antigen comprising:
a) identifying the cytotoxic T cell epitope of the protein according to the method of claim 26;
b) comparing a sequence of the T cell epitope identified in step (a) with known sequences of proteins; and c) determining a protein having a sequence corresponding to the sequence of the T cell epitope.