Peptide motif of a cattle MHC class I molecule

Immunology Letters, 45 (1995) 129-136 0165-2478/95/$09.50 0 1995 Elsevier Science B.V. All rights reserved IMLET 2326

Peptide motif of a cattle MHC class I molecule A.I. Bamford

‘**, A. Douglas

‘I Veterinary

Sciences

‘, T. Friede

b, S. Stevanovic

DilGsion, Belfast BT4 3SD, Northern D-69120

(Received 23 November

Ireland;

Heidelberg

A consensus motif for a bovine major histocompatibility complex (MHC) class I molecule, A20, was derived from parainfluenza type-3 (PI-3) virus-infected muscle-derived fibroblast cells and peripheral blood leukocytes by extraction of the naturally processed peptides from MHC class I molecules by treatment with TFA and peptide sequencing of the complex mixture. The results showed that the majority of peptides were 9 amino acids long with position 2 occupied by lysine and position 9 occupied by arginine. The arginine at position 9 suggests that cattle, like humans, but unlike the mouse have permissive TAP transporter molecules accepting peptides with positively charged amino acids at their C-terminus. This is the first report of a MHC ligand motif in cattle.

2. Introduction Cytotoxic T lymphocytes (CTL) represent a branch of the immune system concerned with recognition of host cells that express new antigens as a result of infection or transformation. It has been known for nearly 20 years that CTL recognise antigen in association with cell-surface class I molecules of the major histocompatibility complex (MHC). However, it has only been appreciated since 1986 that these new antigens are comprised of short peptides, generally produced from internally synthesised proteins, which are directly bound to the class I MHC molecules [l]. The 3-dimensional structures of 4 different class I MHC

SSDI 016%2478(94)00244-4

Sciences

Department

1994; accepted 29 November

1. Summary

Veterinary

h DKFZ.

Di-

b and B.M. Adair

a

of Tumour Virus Immunology,

1. Germany

Key words: Ligand motif; MHC class I; Bovine; Cytotoxic

* Corresponding author: Dr. A.I. Bamford, vision, Belfast BT4 3SD, Northern Ireland.

b, H.G. Rammensee

T lymphocyte;

1994)

Parainfluenza

type-3 virus

molecules have all shown a deep cleft formed by the distal domains of the MHC a-subunit into which the peptides bind [2-61. One of the hallmarks of class I MHC molecules is their extraordinary allelic polymorphism. In mouse (H-2) and man (HLA), class I antigens are coded by several hundred kilobases (kb) of DNA [7,8]. The number of classical class I loci is 6 for the H-2 complex (H-2K1, K, Dl, D2, D3 and L) of the BALB/c mouse and 3 for the HLA complex (HLA-A, B, and 0. Four previous international bovine lymphocyte antigen (BoLA) workshops have dealt with the serological definition of BoLA MHC class I polymorphism [9-121. These workshops defined 50 class I specificities, 25 of which were assigned to the BoLA-A locus (prefix ‘A’) while the other 25 remained classified as workshop (prefix ‘w’) specificities [12]. At the 5th International Bovine Lymphocyte Antigen BoLA workshop in 1992, 2 specificities w16 and w32 were upgraded from workshop specificities to full-status BoLA-A specificities and 3 new specificities ~51 (w28), ~52 and ~53 (~28) were defined

b31. There is growing but limited evidence for more than 1 locus [11,14-201 and recent biochemical studies on BoLA molecules have suggested that 2 loci are predominantly expressed at the protein level [21]. However, based on the number of amino acids in the transmembrane region, the published class I cDNA sequences suggest the presence of 3 loci [15,18,19,22,23]. SDSPAGE, lD-IEF and peptide mapping analysis of the charge heterogeneity of bovine class I molecules have confirmed that at least 3 BoLA class I loci are expressed at the protein level [24]. The expression of multiple, different MHC molecules, as well as the structural variety available within a species, is of obvious benefit, since the variability increases the chances 129

of binding to peptides derived from a wide spectrum of infectious agents. However, as each individual expresses only a small number of different class I molecules, each must be capable of binding to a broad spectrum of structurally distinct peptides. The peptide binding cleft is lined with many of the amino acid residues that vary among alleles, and which thus control peptide binding specificity. Indeed, the polymorphic amino acid residues determine the location of indentations (pockets) in the groove that can be occupied by amino acid side-chains of bound peptides. The availability of the pockets and their exact location and charge environment are correlated with specific patterns of amino acid residues within bound peptides that can be isolated from certain MHC class I molecules. Extraction of naturally processed peptides from MHC class I molecules by treatment with trifluoroacetic acid (TFA) and analysis of the complex mixture by peptide sequencing has shed considerable light on the nature of the peptide mixture present in the binding groove of MHC class I molecules [25]. At certain cycles of Edman degradation only one or a few amino acids predominated. These were designated ‘anchor’ positions in the eluted peptides and were thought to represent conserved features required for strong association with a particular MHC class I molecule [25]. In addition, pool sequencing indicated a uniform length, e.g., 9 amino acids of most ligands. These parameters - length and the position of the anchor residues - defined a consensus motif for the natural peptides eluted from each MHC class I molecule. Therefore the characterisation of peptides intracellularly bound to MHC class I molecules and the observation that these peptides contain motifs of amino acid residues specific for the MHC molecule to which they bind makes it possible to predict antigenic peptides within proteins of pathogens with known sequences. This paper describes the extraction and characterisation of naturally processed peptides from bovine MHC class I molecules to determine a consensus motif which was used to screen the published sequence of bovine parainfluenza type-3 (bPI-3) virus [26,27] for possible antigenic peptides.

3. Materials

and Methods

3.1. CultiLlation of the B-cell hybridoma

clone W6 /32

The mouse hybridoma W6/32 secretes monoclonal antibody (mAb) specific for monomorphic determinants on HLA-A, B, C and p-2m [28,29] and cross-reacts with bovine class I molecules [30,31]. W6/32 was cultured in RPMI-1640 medium (Gibco, Life Technolo130

gies, Paisley, Scotland) containing 10% HIFCS, 2% Hepes buffer, 1% L-glutamine and 50 pg/ ml of gentamycin. Vented 75 cm’ tissue culture flasks (Costar, Northumberland, UK) were seeded with 1 X lo5 cells/ ml and 20 ml was added to each flask and incubated at 37°C in an atmosphere of 5% CO,. For large-scale production of this suspension cell line, spinner culture flasks (Gibco, Life Technologies, Paisley, Scotland) were used at 37°C. The spent supernatant from all cultures was stored at -70°C prior to purification of the anti-MHC class I antibody. w6/32 was precipitated using ammonium sulphate and purified using a protein-A Sepharose affinity column (Biorad). The purity of the monoclonal was checked by polyacrylamide gel electrophoresis (PAGE). 3.2. Lysis of a muscle-derived persistently infected with bPI-3

fibroblastoid

cell line

A persistently infected muscle-derived cell line from animal Cl0 was grown in plastic flasks (175 cm21 (Nunclon, Intermed, Denmark). This animal was typed serologically and had the BoLA-A locus haplotype A20/ -( - = blank). The cells were scraped into the medium using a cell scraper (Costar, Northumberland, UK), and recovered by centrifugation at 1500 rpm for 7 min. Approximately 5 X lo9 cells were centrifuged to form a tight pellet which were either lysed immediately, or frozen at -20°C until required. The pellet of cells was lysed for 1.5 h in a flask on ice with constant slow stirring, using a magnetic stirrer. A 5-fold lysis was carried out, i.e., volume of lysis buffer = 5 X volume of pellet. The lysis buffer consisted of 0.01 M PBS, pH 7.2, with the following additions: 1% (v/v) NP-40 (Boehringer Mannheim, Lewes, East Sussex, UK), 0.02% (w/v) Pepstatin (Sigma Chemical, Poole, Dorset, UK), 0.02% (w/v) Leupeptin (Sigma), 1 mM PMSF (Sigma), 0.02% Aprotinin (Sigma). After lysis, the cells were centrifuged at 1500 rpm for 7 min and the pellet was checked microscopically using nigrosin stain to confirm that all the cells had been lysed. The supernatant was ultracentrifuged (Beckman L7 ultracentrifuge, Beckman Instruments, High Wycombe, UK) for 1 h 5 min at 41k rpm (TST 41.14) at + 4°C. The supernatant was collected using a Pasteur pipette without disturbing the lipid layer and filtered through a 0.22 pm or 0.45 pm filter (Millipore UK, Watford, Herts, UK). 3.3. Lysis of peripheral

blood cells

Five hundred millilitres of blood was collected from animal Cl0 into an equal volume of Alsever’s solution

and separated by centrifugation through Ficoll-Hypaque (Pharmacia, Biosystems, Milton Keynes, UK) at 2500 rpm for 35 min at room temperature. The mononuclear cells were collected from the Ficoll-plasma interface and washed twice with PBS by centrifugation at 2000 rpm for 10 min. The erythrocytes were lysed with 2 volumes of distilled water for 30 s followed by 1 volume of 2.7% saline to restore isotonicity. The polymorphs were recovered by centrifuging at 2000 rpm for 10 min and then pooled together with the mononuclear cells and counted using an improved Neubauer counting chamber. Approximately 1 X lOlo cells were sonicated as described above.

3.4. Isolation of the MHC class I molecules The lysates were applied on a Sepharose CL-4B column (Pharmacia, Milton Keynes, UK) to which glycine was coupled, followed by passage through a Sepharose CL-4B immunoadsorbent column coupled with the MHC class I-specific, cross-reactive mAb w6/32. The columns were connected to a peristaltic pump and a flow rate of 30-50 ml/h was applied. The gels were pre-equilibrated with PBS containing 1% NP-40 prior to application of the lysate which was

allowed to circulate overnight at + 4°C. The depleted lysate was retained and frozen at - 20°C. The columns were separated, and washed with 25 ml each of PBS containing 0.5% NP-40 followed by PBS/O.l% NP-40 and then PBS. All liquid was removed from each column prior to elution. The MHC class I antigen and peptides were eluted off the MHC column by the addition of 1.5 ml of 0.1% TFA (protein sequencing grade, Sigma Chemical, Poole, Dorset, UK). The glycine column was treated in a similar manner. The pooled supematants from each gel were size fractionated using 10 kDa ultrafiltration devices (Amicon, Beverley, MA). The filtrate was collected from each gel, pooled, and dried in vacua, dissolved in 0.1% TFA and subjected to reverse-phase high-performance liquid chromatography (HPLC) on a Superpac Pep-S column (C2/C18, 5 ,um particle size, 4 X 250 nm; Phannacia) and Waters Cl8 300_& Delta Pak column (Millipore) with a flow rate of 1 ml/min. The buffers used for the separation were water containing 0.1% TFA (buffer A) and 80% acetonitrile in water containing 0.081% TFA (buffer B). The peptides were eluted by applying a gradient of lo-75% buffer B over a period of 75 min. The eluant was monitored at 214 nm and 280 nm and fractions (0.5 ml) were collected over the entire gradient.

Au

o.,,

0.10

0.00

0.00

0.0

00.0

40.0

Fig. 1. HPLC separation of peptides eluted from MHC class I molecules extracted from PI-3 virus-infected Absorbances at 214 nm (upper line) and 280 nm (lower line) are shown along with the gradient profile.

*in

00.0

muscle-derived

fibroblastoid

cells.

131

0.m

0.00

Fig. 2. HPLC separation of non-specifically bound peptides from the glycine column (PI-3 virus-infected muscle-derived Absorbances at 214 nm (upper line) and 280 nm (lower line) are shown along with the gradient profile.

fibroblastoid

cell extract).

4. Results

3.5. Peptide sequencing After separation of the peptides by RP-HPLC, several distinct peaks were identified for individual sequencing. In addition, any fractions containing peptide material were pooled and sequenced. Sequencing was performed by Edman degradation in a pulsed-liquid protein sequencer (model 476A) equipped with an online phenylthiohydantoin-derivatized amino acid analyser (model 120A; Applied Biosystems).

4. I. HPLC profiles of peptides eluted from MHC class I molecules isolated from bovine muscle-derived fibroblastoid cells persistently infected with bPI-3 HPLC separation of peptides eluted from MHC class I molecules from muscle-derived fibroblastoid cells are shown in Fig. 1. Samples were separated by reverse-

TABLE 1 CONSENSUS MOTIF DERIVED FROM THE SEQUENCE DATA OF PEPTIDES ELUTED FROM MHC MOLECULES PI-3 VIRUS-INFECTED MUSCLE-DERIVED FIBROBLASTOID CELLS AND PERIPHERAL BLOOD LEUKOCYTES

EXTRACTED

Amino acid position 1 Anchor residues Strong

Weak

132

2

3

4

5

6

7

8

A I

K

D P

D P

K I P V D

I V L

R

V L S E F D

D A F Y L R

E K V A

A V E K L

L R E A Q

D A K E

L K D V I

K

Y R

Q

FROM

8

?

d

8

Fig. 3. HPLC separation of peptides eluted from MHC class I molecules line) and gradient profile (dotted line) are shown.

extracted

phase chromatography on a C,/C,, column eluted with a TFA/water/ acetonitrile gradient. Fig. 2 shows the HPLC separation of non-specifically bound peptides from the glycine column. Fractions 48, 57, 59 and 78 were sequenced and were found to be either the Nterminus of bovine µglobulin or a2 macroglobulin. For the pool sequencing, 50 ,ul was taken from each fraction and the results showed that the majority of the peptides eluted from the MHC class I molecules were 9 amino acids long with the following characteristics: position 2 was exclusively occupied by lysine and TABLE 2 SELECTED A20

REGIONS

FROM THE PROTEINS

9. 8

?

w

OF bPI-3 VIRUS WHICH

from peripheral

blood leukocytes.

Absorbance

at 214 nm (solid

position 9 was occupied by arginine. The amino acids between position 3 and 8 were mainly hydrophobic in character. 4.2. HPLC profiles of peptides eluted from MHC class I molecules isolated from bovine peripheral blood leukocytes HPLC separation of peptides eluted from MHC class I molecules from peripheral blood leukocytes is shown in Fig. 3 and Fig. 4 shows the HPLC separation of

CONFORM

TO THE PREDICTED

MHC CLASS

I MOTIF FOR

Amino acid position 1 HN protein NP protein C protein M protein P protein

62 467 5 111 255 430 504 542

2

3

4

5

6

7

8

9

I I K R M H Y M

R R R R R R R R

70 475 13 119 263 438 512 550

133

8 d

4 8

8 i

Fig. 4. HPLC separation of non-specifically bound peptides from the glycine column (peripheral (solid line) and gradient profile (dotted line) are shown.

non-specifically bound peptides from the glycine column. Few peaks were obtained from the MHC class I column, however 500 ~1 was taken from each fraction, pooled and the volume reduced by evaporation for sequencing. The pool sequencing results were not as strong as seen with the fibroblast cell line due to a poorer yield of extracted MHC class I-associated peptides; however, the pattern of anchor residues was identical to that obtained using muscle-derived fibroblastoid cells. 4.3. Screening of the published sequence of bovine parainfluenza type-3 using the consensus motif The shown lished protein

sequence results defined a consensus motif as in Table 1 which was used to screen the pubsequence of bPI-3. Selected regions from each conforming to this motif are shown in Table 2.

5. Discussion This paper describes for the first time a consensus motif for a bovine MHC class I molecule (A201, which 134

8

si

blood leukocyte

extract). Absorbance

at 214 nm

has been used to screen the published protein sequences of the respiratory pathogen, bPI-3, for possible antigenie sequences. The animal used in this study was serologically typed and presented with a BoLA-A locus haplotype of w20/ -. This animal was not a true homozygote; however, the MHC class I-associated peptides which were extracted from w6/32_precipitated MHC class I molecules gave strong signals following sequencing, suggesting that the peptides were extracted mainly from 1 MHC class I molecule. Although a contribution of other MHC molecules is possible, from the results presented in this study it would seem that w6/32 does not precipitate the untypeable BoLA product as well as typeable BoLA-A molecules, since such a clear motif was found. Other workers have shown that w6/32 precipitates [ 35S]methionine-labelled typed BoLA-A products in large amounts [16] and it has been suggested that the other untypeable products are either relatively poorly labelled with [ 35S]methionine or w6/32 does not strongly precipitate them [16]. Alternatively, these untypeable MHC class I products may represent differentially spliced products of the previ-

ously typed bovine MHC class I locus [32]. The analyses showed that the majority of peptides were 9 amino acids long with position 2 occupied by lysine and position 9 occupied by arginine. This motif is remarkably similar to that of HLA-B27, which also has positively charged anchors at positions 2 and 9. The arginine at position 9 suggests that cattle, like humans can transport with their TAP molecules, peptides into the endoplasmic reticulum or early Golgi with positively charged amino acids at their C-terminus [33]. Thus, the presence of this positively charged amino acid suggests that the bovine TAP system is functionally more similar to humans than to the mouse. The use of the consensus motif to predict antigenic regions of PI-3 proteins yielded relatively few sequence matches. This was to be expected because of the imposition of the sequence restraint by the consensus motif. Only approximately 1% of the potential nonapeptides from any given protein should match the required anchor amino acids of a given consensus motif [34] and of these sequences, only a fraction will contain the necessary auxilliary positions. Therefore, the actual frequency of matched sequences should be quite low. The proteins of the envelope of the virus, HN and F, contained only 1 possible region each, however better fits were found in the internal proteins of the virus. Only a total of 8 regions, 1 in the HN protein, 1 in the NP protein, 3 in the P protein, 1 in the C protein and 2 in the M protein, conformed to the consensus motif. Two regions stood out above the rest as being extremely good fits: the P protein residues 430-438 (LKKMDESHR) and the M protein residues 11 l-l 19 (TKLDIEVRR). Both have the anchor positions of K at position 2 and R at position 9; however, on more close inspection the P protein region also has K at 3 and D at 5 and the M protein region has D at 4, V at 7 and R at 8. These other residues between positions 2 and 9 are amino acids which gave strong signals at these intermediary positions. A total of 4 amino acids in the P protein sequence and 5 amino acids in the M protein sequence match the determined consensus motif and it would seem likely that one or indeed both of these regions are T-cell epitopes presented by BoLA A20 MHC class I molecules. In order to confirm that these regions are indeed T-cell epitopes of this virus, it would be necessary to synthesise these peptides and test them in cytotoxicity assays using targets expressing the MHC class I allele A20 and T cells which are specific for bPI-3 and BoLA-A20-restricted. Bevan and co-workers [35] showed that using the consensus motif for H-2Kd 11 nonamers were selected from the sequence of Listeria monocytogenes listeriolysin a 59 kDa protein; however, only 1 of these peptides was recognised by specific

H-2Kd-restricted CTL. Therefore, testing of predicted peptides using a biological assay is essential. Knowledge of consensus motifs can also pinpoint epitopes within larger active sequences. For instance, the HLAA2 consensus motif was used to predict a nonamer within the sequence of an active 25 mer as the epitope recognised by an HLA-A2-restricted CTL clone [25]. There are numerous examples of antigenic peptides predicted and confirmed using consensus motif and biological assay data; however, it must be realised that conformation to a consensus motif is neither sufficient nor absolutely required for selection of peptides to be presented with MHC class I molecules and it must be logically assumed that further requirements need to be met. For instance, some peptides lacking anchor residues still possess sensitizing activities and therefore must bind possibly with a low affinity to the restricting MHC class I molecule [36]. This can result in some weakly reactive epitopes being overlooked by relying on the motif alone. Conversely, relying solely on data from sensitization of target cells by a synthetic peptide does not imply that the same peptide is present naturally on target cells. In this case one is assuming that the sensitizing activity is due to the synthetic peptide itself, and not to a more potent contaminant or degradation product. Thus, the final proof for a peptide to be a natural MHC ligand is to extract it from MHC molecules. The ability to analyse the antigen-presentation system and to predict T-cell epitopes within known protein sequences of pathogens using allele-specific consensus motif data may lead to the development of synthetic vaccine preparations capable of generating vigorous memory CTL responses without triggering unwanted side effects. However, to be effective and applicable in an outbred population of cattle, peptide-based vaccines would have to contain a cocktail of antigenic peptides that would protect animals with different MHC class I combinations. Some of the most frequent BoLA-A class I alleles are A6, AlO, A14, A20 and A31 [37], and if the consensus motifs were known for these alleles, peptides could be selected from known antigen sequences with a pre-determined spectrum of immune responses. In conclusion, the knowledge of the characteristics of peptides that can bind efficiently to MHC class I molecules will undoubtedly hasten the development of second generation subunit vaccines. However, parameters such as the availability of antigen in vivo (stability, adequate antigen processing) and the induction of second signals (cytokine production, accessory molecule stimulation) still need to be addressed in order to highlight possible dangers or shortcomings when using synthetic peptide-based subunit vaccines. 135

Acknowledgements The authors would like to thank Dr. E.J. Glass (Roslin Institute, Edinburgh) for the class I typing of the animal used in this study.

[14] [15] [16]

References [l] Townsend, A. and Bodmer, H. (1989) Annu. Rev. Immunol. 7, 601. [2] Bjorkman, P.J., Saper, M.A., Samraoui, B., Bennett, W.S., Strominger, J.L. and Wiley, D.C. (1987) Nature 329, 506. [3] Garrett, T.P.J., Saper, M.A., Bjorkman, P.J., Strominger, J.L. and Wiley, D.C. (1989) Nature 342, 692. [4] Madden, D.R., Gorga, J.C., Strominger, J.L. and Wiley, D.C. (1991) Nature 353, 321. [5] Fremont, D.H., Matsumura, M., Stura, E.A., Peterson, P.A. and Wilson, LA. (1992) Science 257, 919. [6] Zhang, W., Young, A.C.M., Imarai, M., Nathenson, S.G., and Sacchettini, J.C. (1992) Proc. Natl. Acad. Sci. USA 89, 8403. [7] Steinmetz, M., Stephan, D. and Fischer-Lindhal, K. (1986) Cell 44, 895. [8] Carroll, M.C., Katzman, O., Alicot, E.M., Koller, B.H., Geraghty, D.E., Strominger, J.L. and Spies, T. (1987) Proc. Natl. Acad. Sci. USA 84, 8535. [9] Spooner, R.L., Oliver, R.A., Sales, D.I., McCoubrey, C.M., Millar, P., Morgan, A.G., Amorena, B., Bailey, E., Bernoco, D., Brandon, M., Bull, R.W., Caldwell, J., Cwik, S., van Dam, R.H., Dodd, J., Gahne, B., Grosclaude, F., Hall., J.G., Hines, H., Leveziel, H., Newman, M.J., Stear, M.J., Stone, W.H. and Vaiman, M. (1979) Anim. Blood Groups Biochem. Genet. 10, 63. [lo] Anonymous (1982) Anim. Blood Groups Biochem. Genet. 13, 33. [ll] Bull, R.W., Lewin, H.A., Wu, M.C., Peterbaugh, K., Antczak, D., Bernoco, D., Cwik, S., Dam, L., Davies, C., Dawkins, R.L., Dufty, J.H., Gerlach, J., Hines, H.C., Lazary, S., Leibold, W., Leveziel, H., Lie, O., Lindberg, P.G., Meggiolaro, D., Meyer, E., Oliver, R., Ross, M., Simon, M., Spooner, R.L., Stear, M.J., Teale, A.J. and Templeton, J.W. (1989) Anim. Genet. 20, 109. [12] Bernoco, D., Lewin, H.A., Andersson, L., Arriens, M.A., Byrns, G., Cwik, S., Davies, C.J., Hines, H.C., Leibold, W., Lie, O., Meggiolaro, D., Oliver, R.A., Ostergard, H., Spooner, R.L., Stewart-Haynes, J.A., Teale, A.J., Templeton, J.W. and Zanotti, M. (1991) Anim. Genet. 22, 477. [13] Davies, C.J., Joosten, I., Bernoco, D., Arriens, M.A., Bester, J., Ceriotti, G., Ellis, S., Hensen, E.J., Hines, H.C., Horin, P.,

136

[17] [18]

[19] [20] [21] [22] [23] [24] [25] [26] [27] [28]

Kristensen, B., Lcwin, H.A., Meggiolaro, D., Morgan, A.L.G., Morita, M., Nilsson, Ph. R., Oliver, R.A., Orlova, A., Ostergard, H., Park, C.A., Schuberth, H.-J., Simon, M., Spooner, R.L., and Stewart, J.A. (1994) Eur. J. Immunogenetics 21, 239. Stear, M.J., Newman, M.J. and Nicholas, F.W. (1982) Tissue Antigens 20, 289. Ennis, P., Jackson, A.P. and Parham, P. (1988) J. Immunol. 141, 642. Watkins, D.I., Shadduck, J.A., Stone, M.E., Lewin, H.A., and Letvin, N.L. (19891 J. Immunogenetics 16, 233. Toye, P.G., MacHugh, N.D., Bensaid, A., Alberti, S., Teale, A.J. and Morrison, WI. (1990) Immunol. 70, 20. Bensaid, A., Kaushal, A., Baldwin, C.L., Clevers, H., Young, J.R., Kemp, S.J., MacHugh, N.D., Toye, P.G. and Teale, A.J. (1991) Immunogenetics 33, 247. Ellis, S.A., Braem, K.A. and Morrison, W.I. (1993) Immunogenetics 37, 49. Garber, T.L., Hughes, A.L., Letvin, N.L., Templeton, J.W. and Watkins, D.I. (1993) Immunogenetics 38, 11. Joosten, I., Teale, A.J., van der Poel, A. and Hensen, E.J. (1992) Anim. Genet. 23, 113. Brown, P., Spooner, R.L. and Clark, A.J. (1989) Immunogenetics 29, 58. Garber, T.L., Hughes, A.L., Watkins, D.I. and Templeton, J.W. (1994) Immunogenetics 39, 257. Al-Murrani, S.W.K., Glass, E.J., Hopkins, J. (1994) Anim. Genet. 25, 165. Falk, K., Rotzschke, O., Stevanovic, S., Jung, G., Rammensee, H-G. (1991) Nature 351, 290. Sakai, Y., Suzu, S., Shioda, T. and Shibuta, H. (1987) Nucl. Acids Res. 15, 2927. Suzu, S., Sakai, Y., Shioda, T. and Shibuta, H. (1987) Nucl. Acids Res. 15, 2945. Barnstable, C.J., Bodmer, W.F., Brown, G., et al. (1978) Cell 14,

[29] irodsky, F.M. and Parham, P. (1982) J. Immunol. 128, 129. [30] B&sky, F.M., Stone, W.H. and Parham, P. (1981) Hum. Immunol. 3, 143-152 [31] Joosten, I., Oliver, R.A., Spooner, R.L., Williams, J.L., Hepkema, B.G., Sanders, M.F. and Hensen, E. (1988) Anim. Genet. 19, 103. [32] Lalanne, J.-L., Cachet, M., Kummer, A.-M., Gachelain, G., and Kourilsky, P. (1983) Proc. Natl. Acad. Sci. USA. 80, 7561. [33] Rotzschke, 0. and Falk, K. (1991) Immunol. Today 12, 447. [34] Falk, K., Rotzschke, O., Deres, K., Metzger, J., Jung, G. and Rammensee, H.-G. (1991) J. Exp. Med. 174, 425. [35] Pamer, E.G., Harty, J.T. and Bevan, M.J. (1991) Nature 353, 852. [36] Tsomides, T.J., Walker, B.D. and Eisen, H.N. (1991) Proc. Natl. Acad. Sci. USA 88, 11276. [37] Davies, C.J. and Antczak, D.F. (1991) Anim. Genet. 22, 31.