Specificity of a CTL epitope
Eur. J. Imrnunol. 1992. 22: 191-195
Scott R. Burrows, Stuart J. Rodda., Andreas Suhrbier, H. Mario Geysen. and Denis J. Moss Queensland Institute of Medical Research, The Bancroft Centre, Brisbane and Chiron Mimotopeso, Clayton
The specificity of recognition of a cytotoxic T lymphocyte epitope" An Epstein-Barr virus (EBV)-specific CD8+ cytotoxic T lymphocyte (CTL) clone (LC13) was shown to recognize the minimal peptide determinant FLRGRAYGL from the EBNA3 antigen of the BL74 strain of EBV. The equivalent epitope from the B95-8 strain (FLRGRAYGI) is not recognized when endogenously presented and the peptide is 15-fold less active than FLRGRAYGL. A replacement set of peptides was synthesized in which each residue within FLRGRAYGL was sequentially replaced with all other genetically coded amino acids. These peptides were tested for their ability to sensitize target cells to lysis by LC13. Of the 171 single-amino acid replacement peptides only 1.5 were more active than the peptide FLRGRAYGI. Five peptides had significantly greater activity than FLRGRAYGL and a peptide incorporating the most active of these single-amino acid substitutions (HIRGRAYSL) induced lysis at concentrations approximately 30-fold less thanFLRGRATGL. Simplified theoretical calculations based on this study suggest that CTL LC13 has a specificity for its target epitope of 1 in 4.7 x 10'". This represents the first complete analysis of the role of single amino acids within a minimum epitope on the specificity of CTL recognition.
1 Introduction Tcell epitopes can be defined by the use of synthetic peptides, which when added to APC efficiently mimic naturally processed antigens. Free peptide is thought to bind to cell surface MHC glycoproteins and the TcR then recognizes the resulting complex [ 11. Using synthetic peptides we have previously localized the target determinant of an HLA-B8 restricted CD8+ EBV-specific CTL clone (LC13) to a 15-amino acid sequence (AWNAGFLRGRAYGLD) from the EBV nuclear antigen, EBNA-3 [2, 31. This region of EBNA-3 has been shown to differ in three EBV strains. EBV derived from the BL74 cell line encodes the above sequence and the autologous lymphoblastoid cell line (LCL) transformed with this strain (referred to as LC/BL74 LCL) is recognized by CTL LC13. EBV from the B95-8 and L4 cell lines encode the sequences AWNAGFLRGRAYGID and AWNAGLLRGRAYGQD, respectively (changes from the BL74 sequence underlined) and autologous LCL transformed with these strains (i. e. LC/B95-8LCL and LC/L4 LCL) are not recognized by CTL LC13 [3]. Synthetic peptide AWNAGLLRGRAYGQD is inactive as a target determinant for t h e clone w k r e a s peptide AWNAGFLRGRAYGID can induce target cell recognition if added exogenously at concentrations above S p~ but is 15-fold less active than peptide AWNAGFLR-
[I 96671
*
1Yl
We acknowledge the support of the National Health and Medical Research Council, Canberra; The National Cancer Institute, USA (NIH Grant no. CA 52250.01); and the Wellcorne Trust, GB.
Correspondence: Denis J. Moss, Queensland Institute of Medical Research,The Bancroft Centre, 300 Herston Road, Brisbane 4029. Australia Abbreviation: LCL: Lymphoblastoid cell line
0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1992
GRAYGLD [3]. Previous results have suggested that the B9S-8 epitope is efficiently processed and presented by LC/B95-8 LCL [3]. This is supported by the observation that LC/L4 LCL infected with an EBNA-3 vaccinia construct encoding the B9S-8 sequence, induces high levels of antigen expression and is recognized by LC13 (unpublished results and [4]).These studies suggest that CTL recognition can be lost when only a minor change occurs in the epitope sequence (leucine to isoleucine) and that limiting concentrations of endogenously processed epitope may play a major role in maintaining this fine specificity [ S ] . The dependence of critical amino acid residues in peptide epitopes on specific CTL recognition has been demonstrated in other viruses [6-81 and recent evidence suggests that sequence variations within regions encoding CTL epitopes, which prevent recognition, can result from selective pressures by the immune system [9].The present study was undertaken to fully assess the degree of specificity of a CTL clone for its target epitope by measuring the CTL activity of replacement sets of peptides, in which each amino acid has been sequentially replaced with each of the other genetically coded amino acids. By comparing the levels of lysis of LC/B95-8 LCL and LC/BL74 LCL with the activity of the synthetic peptides encoded by the B95-8 and BL74 EBV strains, the concentration at which the peptides were tested was adjusted to approximate to the level of endogenously presented epitope.
2 Materials and methods 2.1 Cell lines CTL clone LC13 was isolated by co-cultivating fresh lymphocytes from donor LC with gamma-irradiated autologous LCL (transformed with EBVderived from the BL74 cell line) at a responder to stimulator ratio of 100 : 1[2].The clone was maintained with the use of highly purified human rIL2 from E . coli [lo, 111. LC13 was not unique for its 0014-2980/Y2/0101-0191$3.50+ .25/0
192
Eur. J. Immunol. 1992. 22: 191-195
S. R. Burrows, S. J. Rodda, A. Suhrbier et al.
inability to recognize the autologous B95-8 transformant. The majority (18/20) of AWNAGFLRGRAYGLD-specific CTL clones raised from donor LC and another donor [12], after stimulation with the autologous BL74 transformant, failed to lyse the autologous B95-8 transformant (unpublished result). LCL used in this study were generated by transforming lymphocytes from donor LC with EBV derived from the cell lines L4, BL74 and B95-8.
(Emof 5 : 1). The autologous LCL transformed with EBV derived from the L4 cell line (LC/L4 LCL), which is not recognized by the clone LC13, was used as target in these assays. The percent specific lysis results are given as means from tests of duplicate peptide preparations. Peptides were also tested for toxicity by incubation with 51Cr-labeled targets without effector addition. None were found to be toxic.
2.2 Synthesis of peptides
3 Results
Peptides were synthesized in duplicate on polyethylene pins and cleaved from the pins directly into 150 pl per pin of sterile 0.1 M sodium phosphate solution as described [13]. The cleavage chemistry results in the formation of a beta-alanine-diketopiperazinegroup at the C-terminal end of each peptide. This C-terminal structure appears to exert a minimal influence on the function of peptides in Tcell activation assays [14]. The N termini of the peptides were unblocked. The yield of peptide was 11k 4 nmole of the nine-amino acid peptides per pin. Purity of the nine-amino acid peptides on a weight basis was 3 4 k 9 % ( n = 10). Impurities consisted primarily of a mixture of truncated peptides (data not shown). Peptides HIRGRAYSL and FLRGRAYGL, used in Fig. 4b, were synthesized according to the method of Houghten [15].
3.1 Determination of the mimimal epitope To define the minimal active epitope, overlapping peptides of 9, 10, 11 and 12 amino acids in length were synthesized to span the 18mer peptide, NAGFLRGRAYGIDLLRTE. Peptides were screened at 21 p~ for their ability to induce specific lysis of L C / U LCL targets by the autologous EBV-specific CTL clone LC13. This revealed a sharply defined sequence of nine residues, FLRGRAYGI, which contained the minimum sequence for full CTL activity (Fig. 1).
2.3 Cytotoxicity assay
Only the sequence of the B95-8 strain of EBVwas known at the beginning of this study.The nine-residue sequence from the BL74 strain, FLRGRAYGL, was also shown to be the minimal sequence required for CTL activity (see Sect. 3.3; amino acid-deletion peptides for positions 1 and 9, shown as * in Fig. 2, were essentially unrecognized).
Each peptide was incubated (at 21 pM and 0.7 pM) with lo4 51Cr-labeled target cells in 35 pl for 1 h in U-shaped microtiter wells. LC13 effector cells (5 X lo4) were then added in 165 pl for a standard 5-h 51Cr-release assay [16]
An effect observed with longer peptides is a depression in cytotoxicity when the homologous glycine residue is added at the amino-terminal end of active peptides containing the minimal determinant. Titration showed that peptides
NAGFLRGRAYGIDLLRTE
NAGFLRGRAYGIDLLRTE
c GFLRGRAYG FLRGRAYGI GRAYGlDLL RAYGIDLLR AYGIDLLRT
GFLRGRAYGID FLRGRAYGIDL GRAYGIDLLRT RAYGIDLLRTE AYGIDLLRTEG
I -D
L
,
0 1 0 2 0 3 0 4 0 5 0 6 0
FLRGRAYGID LRGRAYGIDL RGRAYGIDLL
FLRGRAYGIDLL LRGRAYGIDLLR RGRAYGIDLLRT GRAYGIDLLRTE RAYGIDLLRTEG
I
YGIDLLRTEG 0
~_-_~ -
, , ___1 0 2 0 3 0 4 0 5 0
SPECIFIC LYSlS (%)
60
YGIDLLRTEGGG
I I I ~D ,
,
,
10
20
30
4
i
n
SPECIFIC LYSlS (%)
Figure 1. Cytotoxicity of CTL clone LC13 for LC/L4 LCL targets primed with 21 pM of peptides of lengths from 9 to 12 residues, homologous with sequence NAGFLRGRAYGIDLLRTE. Nonhomologous glycine residues were added to the carboxy-terminus of some peptides to maintain consistent peptide length.
193
Specificity of a CTL epitope
Eur. J. Immunol. 1992. 22: 191-195
El
1
80 70 60 50 40 30 20 10
m
100
n
80
70.
Ijl
MNPQRSTVWY'
lEFGH
ACDEFGH I KLMNPQRSTI
605040,
10
ACDEFGH I KLMNPQRSTVWY '
m
10090-
100
40~ 3020-
--7-4T+-uy
KLMNPQRSTVWY
10 0
100~
100
90 80-
90 80
70
70
60~ 50-
60
d r ACDEFGH I
---
i KLMNPQRSTVWY*
50
a-
40
30~ 20-
UK
10 0'-, , ACDEFGHIKLMNPQRSTVH ~
I
.
,
,
'
,
'
"
"
"
'
'
r
!L
MNPQRSTVWY'
REPLACEMENT AMINO ACID RESIDUE Figure 2. Effect of single-residue substitutions on cytotoxicity. Effectors and targets were as for Fig. 1. Every one of the 20 genetically coded amino acids, as well as a deletion (*),was tested in each of the nine locations within the "parent" sequence FLRGRAYGL at two concentrations of peptide. The boxed letter represents the parent residue being replaced; the horizontal axis lists the residue replacing the parent residue. Open bars represent the % lysis observed at the higher concentration of peptide (21 p ~ and ) the overlaid solid bars represent % lysis observed with 1/30 of this peptide level (0.7 pM).
GFLRGRAYGUL are both approximately twofold less active than FLRGRAYGUL, respectively (data not shown). 3.2 Correlation of CTL activity obtained with exogenous peptide and endogenous peptide We have previously demonstrated a 15-fold difference in activity between the peptides AWNAGFLRGRAYGLD and AWNAGFLRGRAYGID, which are derived from the EBNA-3 sequence of BL74 and B95-8, respectively (see Fig. 2 in [3]).The minimal epitope peptides FLRGRAYGL and FLRGRAYGI were titrated for CTL activity and an identical 15-fold difference in activity was observed. In the same experiment, levels of lysis were 40% and 3% for the autologous LCL transformed with the BL74 and B95-8 strains of EBV, respectively. These levels of lysis corresponded to the level measured when LC/L4 LCL was incubated with 0.7 pM of peptides FLRGRAYGL and FLRGRAYGI, respectively (equivalent to 0.6 pg/ml of the 15mer in Fig. 3in [3]).The replacement set of peptides was wbsequently tested at 0.7 y~ as this concentration
appeared to mimic the concentration of endogenously presented epitope.
3.3 Screening single-residue replacement peptides for CTL activity To assess the role of individual residues in the minimal active epitope, FLRGRAYGI,, sets of peptides in which each residue had been sequentially replaced with each of the other 19 genetically coded amino acids were tested for their ability to sensitize targets to CTL lysis. Peptides in which an amino acid had been deleted altogether were also synthesized and are represented with the symbol "*". The substituted peptides were screened at two concentrations (21 y ~ open ; bars and 0.7 y ~ filled ; bars; Fig. 2). The patterns of reactivity reveal that the two residues at the carboxy and amino termini of the peptide were much more replaceable than those in the central core. Substitution of any amino acid within the central RGRAY sequence of the epitope gave peptides with activity less than that of the FLRGRAYGI- peptide. Overall, 15/171 single-amino acid
Eur. J. Immunol. 1992. 22: 191-195
S. R. Burrows, S. J. Rodda, A. Suhrbier et al.
194
replacement peptides displayed activity significantly (k2 SD) higher than FLRGRAYGI, suggesting that these sequences would have some activjty at levels naturally presented on LCL.While 7 of these 1.5were less active than FLRGRAYGL, five substituted peptides (ELRGRAYGL, HLRGRAYGL, YLRGRAYGL, FIRGRAYGL and FLRGRAYSL) had significantly higher activity than FLRGRAYGL (Fig. 2).
-
3.4 Effect of peptide extension on residue replaceability
To determine whether the relatively high replaceability of the carboxy- and amino-terminal residues of the minimal epitope, was maintained when peptides were extended beyond the minimal length, longer replacement peptides were synthesized and tested for activity (at 21 pM). The replacement sets were as follows: substitution of F in the peptides GFLRGRAYGL and AGFLRGRAYGL; and substitution of L in the peptide FLRGRAYGLD. No significant difference in the pattern of replaceability was observed for peptide replacement sets of either F or L as a result of using sequences extended beyond the minimal epitope. The results for the set of peptides with replacements for F in the sequence AGFLRGRAYGL is shown in Fig. 3. The significant specific lysis observed with the deletion peptide (i.e. the 10mer peptide AG*LRGRAYGL) is consistent with the result obtained for GLKGRAYGL in Fig. 2 when used at 21 pM.
0.001
0.1
0.01
The activity of the five substituted peptides, which showed higher activity than FLRGRAYGL, were compared with FLRGRAYGL by titration. FLRGRAYSL was S f o l d more active, FIRGRAYGL, ELRGRAYGL and HLRGRAYGL were approximately Pfold more active, while YLRGRAYGL was 2-3-fold more active than F L R G n Y G L (Fig. 4a). A peptide incorporating three of these active amino acid replacements, HIRGRAYSL, was
RGRAYSL
100
10
Peptide concentration (pM)
loo
.-aa
i? -*
60
.-co
40
-
1
FLRGRAYGL
(b)
"IRGRAYSL
o 0
g
20 0 0.001
0.01 Peptide
3.5 Titration of peptides with higher activity than FLRGRAYGL
1
FLRGRAYGL ELRGRAYGL HLRGRAYGL MRGRAYGL FIRGRAYGL
0.1
1
10
100
concentration (pM)
Figure 4. (a) Titration of peptides FLRGRAYGL, ELRGRAYGL, HLRGRAYGL, YLRGRAYGL, FIRGRAYGLT and FLRGRAYS using effectors and targets for Fig. 1. (b) Titration of peptides FLRGRAYGL and BRGRAYSL using effectors and targets as for Fig. 1.
as
found to be approximately 30-fold more active than FLRGRAYGL (Fig. 4b; note both peptides in this figure were prepared using the method of Houghten [lS]).
4 Discussion
AGIFILRGRAYGL U
100
90
'
-80 70
$60 A50
Y
!km
0
E30
a20 10 0
A
REPLACEMENT AMINO ACID RESIDUE Figure 3. Effect of single-residue substitutions for phenylalanine on cytotoxic activity of the peptide AGFLRGRAYGL. Effectors and targets were as for Fig. 1. Each peptide was tested at 21 FM.
The minimal epitope recognized by CTL line LC13 was the nonapeptide FLRGRAYGL (Figs. 1 and 2).This epitope is encoded by the BL74 virus strain and is 15-fold more efficient as an epitope for CTL LC13 than the analogous sequence encoded by the B9S-8 strain (this report and [3]). In addition, a concentratin of 0.7 p~ of peptides FLRGRAYGL and FLRGRAYGI, when added exogenously to L C / L ~ gave , similar values of lysis as autologous B cells infected with BL74 and B9S-8 viral strains, respectively. Taken together, these data indicate that peptides, which give equal or lower lysis than the B95-8 sequence (FLRGRAYGI) when added at a concentration of 0.7 PM to LC/L4, are-unlikely to be active if naturally processed and presented by an LCL. The following discussion is, therefore, based on the replacement set data obtained at 0.7 p~ (solid bars in Fig. 2).0nly 15 of the 171single-amino acid substituted peptides had significantly (+ 2 SD) higher activity than FLRGRAYGI. These 15 substititons were all located in the two amino acids at the amino- and carboxyterminal ends of the epitope.The central residues, RGRAY, did not tolerate any substitutions.
Eur. J. Immunol. 1992. 22: 191-195
To test the possibility that the higher replaceability of residues at the amino- and carboxy-terminal ends of the minimal sequence was an artefact due to the use of a minimal determinant, the replaceability of the amino- and carboxy-terminal residues, F and L , were tested with peptides extended at either end by one or two naturally occurring amino acids. The replaceability patterns appeared independent of extensions outside the minimal epitope. It is unclear what the length of the naturally processed epitope is in this case, although it has been suggested that many CTL epitopes are nine amino acids long [17, 181.
Specificity of a CTL epitope
195
5 References
1 Townsend, A . and Bodmer, H., Annu. Rev. Immunol. 1989. 7: 601. 2 Burrows, S. R., Sculley, T. B., Misko, I. S., Schmidt, C. and Moss, D. J., J. Exp. Med. 1990. 171: 345. 3 Apolloni, A , , Moss, D. J., Stumm. R., Burrows. S. R.. Suhrbier, A., Misko, I . S.. Schmidt. C. and Sculley,T. B., Eur. J. Imrnunol. 1992. 22:183. 4 Murray, R. J., Kurilla, M. G . , Griffin, H. M., Brooks. M.. Arrand. J. R., Rowe, M., Burrows, S. R.. Moss, D. J.. Kieff, E. and Rickinson, A . B., Proc. Natl. Acad. Sci. USA 1990. 87: 2906. 5 Milligan. G . N., Morrison, L. A., Gorka. J., Braciale,V. L. and Braciale, T. J., J. Immunol. 1990. 145: 3188. The present study and a previous report [19] have revealed 6 Bastin. J., Rothbard, J., Davey. J., Jones, I. and Townsend. A , , substituted analogues which had higher activity than the J. Exp. Med. 1987. 165: 1508. natural sequence; FLRGRAYSL was 15-fold and 7 Joly, E., Salvato. M..Whitton, J. L. and Oldstone. M. B. A . , J. HLRGRAYGL and FIRGRAYGcwere 9-fold more active Virol. 1989. 63: 1845. than FLRGRAYGL at half maximal kiling. When these 8 Takahashi, H., Merli, S., Putney, S. D., Houghten, R.. Moss. replacements were combined, the resulting peptide, HIRB., Germain, R. N . and Berzofsky, J. A., Science 1989. 246; GRAYSL, triggered target cell lysis at a concentrzon 118. approximately 30-fold less than FLRGRAYGL. This data 9 Pircher, H., Moskophidis, D.. Rohrer, U., Burki, K., Hengartner, H. and Zinkernagel, R.M., Nature 1990. 346: 629. suggests that the effect on activity of multiple amino acid replacements is roughly equal to the additive effects of the 10 Wang, A . , Lu. S. D. and Mark, D. F., Science 1984. 224: 1431. single-residue replacements. 11 Rosenberg, S. A , , Grimm, E. A , . McGrogan, M., Doyle. M., Kawasaki, E., Koths, K.. Mark, D. F., Science 1984. 223: Assuming that the effect of all multiple substitutions result 1412. in the simple sum of the effect of independent substitutions, 12 Misko, I. S.. Schmidt, C., Mcm, D. J., Burrows, S. R. and this replacement set data suggests that a theoretical 240 Scu1ley.T. B., J. Gen. Virol. 1991. 72: 405. epitopes (10 substitutions for F, multiplied by 3 for L, 4 for 13 Maeji, J. N.. Bray. A. B. and Geysen. H. M.. .I. Immunol. G and 2 for L) might all be effective targets for LC13. The Meth0d.s 1990. 134: 23. value of 240 may, however, be an underestimate since the 14 Mutch, D. A ., Rodda, S. J.. Benstead. M.,Valerio, R. M., Geysen, H . M., Peptide Res. 1991. 4: 132. presence of substitutions, which enhance peptide activity, may allow some substitutions of the irreplaceable central 15 Houghten, R. A., Proc. Nutl. Acud. Sci. USA 1985. 82: 5131. residues. Making some simplifying assumptions there are I6 Misko, I. S., Pope, J. H., Hutter, R . , Soszynski,T. D. and Kane. 209 possible 9mers and since LC13 only recognizes 240 of R. G . , Int. J. Cancer 1984. 3.7: 239. them, this clone has a specificity of 1in 4.7 x 101(’(240/209)), 17 Falk, K., Rotzschke, O., StevanoviC, S.. Jung, G . and Rama level which exceeds class I1 T cell specificity by two orders mensee, H., Nature 1991. 351: 290. of magnitude [20] and antibody specificity by five orders of 18 Schumacher,T. N.M.,DeBruijn, M. L. H.,Vernie,L. N.. Kast, magnitude 1211. If this level of specificity is generally true W. M., Melief, C. J. M., Neefjes, J. J. and Ploegh, H. L., Nature 1991. 350: 703. for CTL recognition it would have implications for autoimmunity and vaccine design because (a) cross-reaction I9 Bodmer, H. C., Pemberton, R. M., Rothbard, J. B. and Askonas. B. A , , Cell 1988. 52: 253. between closely related sequences would be unlikely and (b) the majority (at least 91% in the case of FLRGRAYGL) 20 Suhrbier, A., Rodda, S. J., Ho. P. C.. Csurhcs, P.. Dunkley, H.. Saul, A . , Geyscn. H. M. and Rzepczyk, C., .I. Imrnunol.1991. of amino acid mutations would result in loss of recognition 147: 2507. by the CTL. 21 Geysen, H. M., Mason,T. J. and Rodda, S. J., J. Mol. Recog. 1988. 1: 32. r l L 2 was kindly provided by the Cetus Corporution. We t h u d Richard Luuricellu and his team for synthesizing the peptides.
Received June 19, 1991; in revised form September 19, 1991.
Eur. J. Imrnunol. 1992. 22: 191-195
Scott R. Burrows, Stuart J. Rodda., Andreas Suhrbier, H. Mario Geysen. and Denis J. Moss Queensland Institute of Medical Research, The Bancroft Centre, Brisbane and Chiron Mimotopeso, Clayton
The specificity of recognition of a cytotoxic T lymphocyte epitope" An Epstein-Barr virus (EBV)-specific CD8+ cytotoxic T lymphocyte (CTL) clone (LC13) was shown to recognize the minimal peptide determinant FLRGRAYGL from the EBNA3 antigen of the BL74 strain of EBV. The equivalent epitope from the B95-8 strain (FLRGRAYGI) is not recognized when endogenously presented and the peptide is 15-fold less active than FLRGRAYGL. A replacement set of peptides was synthesized in which each residue within FLRGRAYGL was sequentially replaced with all other genetically coded amino acids. These peptides were tested for their ability to sensitize target cells to lysis by LC13. Of the 171 single-amino acid replacement peptides only 1.5 were more active than the peptide FLRGRAYGI. Five peptides had significantly greater activity than FLRGRAYGL and a peptide incorporating the most active of these single-amino acid substitutions (HIRGRAYSL) induced lysis at concentrations approximately 30-fold less thanFLRGRATGL. Simplified theoretical calculations based on this study suggest that CTL LC13 has a specificity for its target epitope of 1 in 4.7 x 10'". This represents the first complete analysis of the role of single amino acids within a minimum epitope on the specificity of CTL recognition.
1 Introduction Tcell epitopes can be defined by the use of synthetic peptides, which when added to APC efficiently mimic naturally processed antigens. Free peptide is thought to bind to cell surface MHC glycoproteins and the TcR then recognizes the resulting complex [ 11. Using synthetic peptides we have previously localized the target determinant of an HLA-B8 restricted CD8+ EBV-specific CTL clone (LC13) to a 15-amino acid sequence (AWNAGFLRGRAYGLD) from the EBV nuclear antigen, EBNA-3 [2, 31. This region of EBNA-3 has been shown to differ in three EBV strains. EBV derived from the BL74 cell line encodes the above sequence and the autologous lymphoblastoid cell line (LCL) transformed with this strain (referred to as LC/BL74 LCL) is recognized by CTL LC13. EBV from the B95-8 and L4 cell lines encode the sequences AWNAGFLRGRAYGID and AWNAGLLRGRAYGQD, respectively (changes from the BL74 sequence underlined) and autologous LCL transformed with these strains (i. e. LC/B95-8LCL and LC/L4 LCL) are not recognized by CTL LC13 [3]. Synthetic peptide AWNAGLLRGRAYGQD is inactive as a target determinant for t h e clone w k r e a s peptide AWNAGFLRGRAYGID can induce target cell recognition if added exogenously at concentrations above S p~ but is 15-fold less active than peptide AWNAGFLR-
[I 96671
*
1Yl
We acknowledge the support of the National Health and Medical Research Council, Canberra; The National Cancer Institute, USA (NIH Grant no. CA 52250.01); and the Wellcorne Trust, GB.
Correspondence: Denis J. Moss, Queensland Institute of Medical Research,The Bancroft Centre, 300 Herston Road, Brisbane 4029. Australia Abbreviation: LCL: Lymphoblastoid cell line
0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1992
GRAYGLD [3]. Previous results have suggested that the B9S-8 epitope is efficiently processed and presented by LC/B95-8 LCL [3]. This is supported by the observation that LC/L4 LCL infected with an EBNA-3 vaccinia construct encoding the B9S-8 sequence, induces high levels of antigen expression and is recognized by LC13 (unpublished results and [4]).These studies suggest that CTL recognition can be lost when only a minor change occurs in the epitope sequence (leucine to isoleucine) and that limiting concentrations of endogenously processed epitope may play a major role in maintaining this fine specificity [ S ] . The dependence of critical amino acid residues in peptide epitopes on specific CTL recognition has been demonstrated in other viruses [6-81 and recent evidence suggests that sequence variations within regions encoding CTL epitopes, which prevent recognition, can result from selective pressures by the immune system [9].The present study was undertaken to fully assess the degree of specificity of a CTL clone for its target epitope by measuring the CTL activity of replacement sets of peptides, in which each amino acid has been sequentially replaced with each of the other genetically coded amino acids. By comparing the levels of lysis of LC/B95-8 LCL and LC/BL74 LCL with the activity of the synthetic peptides encoded by the B95-8 and BL74 EBV strains, the concentration at which the peptides were tested was adjusted to approximate to the level of endogenously presented epitope.
2 Materials and methods 2.1 Cell lines CTL clone LC13 was isolated by co-cultivating fresh lymphocytes from donor LC with gamma-irradiated autologous LCL (transformed with EBVderived from the BL74 cell line) at a responder to stimulator ratio of 100 : 1[2].The clone was maintained with the use of highly purified human rIL2 from E . coli [lo, 111. LC13 was not unique for its 0014-2980/Y2/0101-0191$3.50+ .25/0
192
Eur. J. Immunol. 1992. 22: 191-195
S. R. Burrows, S. J. Rodda, A. Suhrbier et al.
inability to recognize the autologous B95-8 transformant. The majority (18/20) of AWNAGFLRGRAYGLD-specific CTL clones raised from donor LC and another donor [12], after stimulation with the autologous BL74 transformant, failed to lyse the autologous B95-8 transformant (unpublished result). LCL used in this study were generated by transforming lymphocytes from donor LC with EBV derived from the cell lines L4, BL74 and B95-8.
(Emof 5 : 1). The autologous LCL transformed with EBV derived from the L4 cell line (LC/L4 LCL), which is not recognized by the clone LC13, was used as target in these assays. The percent specific lysis results are given as means from tests of duplicate peptide preparations. Peptides were also tested for toxicity by incubation with 51Cr-labeled targets without effector addition. None were found to be toxic.
2.2 Synthesis of peptides
3 Results
Peptides were synthesized in duplicate on polyethylene pins and cleaved from the pins directly into 150 pl per pin of sterile 0.1 M sodium phosphate solution as described [13]. The cleavage chemistry results in the formation of a beta-alanine-diketopiperazinegroup at the C-terminal end of each peptide. This C-terminal structure appears to exert a minimal influence on the function of peptides in Tcell activation assays [14]. The N termini of the peptides were unblocked. The yield of peptide was 11k 4 nmole of the nine-amino acid peptides per pin. Purity of the nine-amino acid peptides on a weight basis was 3 4 k 9 % ( n = 10). Impurities consisted primarily of a mixture of truncated peptides (data not shown). Peptides HIRGRAYSL and FLRGRAYGL, used in Fig. 4b, were synthesized according to the method of Houghten [15].
3.1 Determination of the mimimal epitope To define the minimal active epitope, overlapping peptides of 9, 10, 11 and 12 amino acids in length were synthesized to span the 18mer peptide, NAGFLRGRAYGIDLLRTE. Peptides were screened at 21 p~ for their ability to induce specific lysis of L C / U LCL targets by the autologous EBV-specific CTL clone LC13. This revealed a sharply defined sequence of nine residues, FLRGRAYGI, which contained the minimum sequence for full CTL activity (Fig. 1).
2.3 Cytotoxicity assay
Only the sequence of the B95-8 strain of EBVwas known at the beginning of this study.The nine-residue sequence from the BL74 strain, FLRGRAYGL, was also shown to be the minimal sequence required for CTL activity (see Sect. 3.3; amino acid-deletion peptides for positions 1 and 9, shown as * in Fig. 2, were essentially unrecognized).
Each peptide was incubated (at 21 pM and 0.7 pM) with lo4 51Cr-labeled target cells in 35 pl for 1 h in U-shaped microtiter wells. LC13 effector cells (5 X lo4) were then added in 165 pl for a standard 5-h 51Cr-release assay [16]
An effect observed with longer peptides is a depression in cytotoxicity when the homologous glycine residue is added at the amino-terminal end of active peptides containing the minimal determinant. Titration showed that peptides
NAGFLRGRAYGIDLLRTE
NAGFLRGRAYGIDLLRTE
c GFLRGRAYG FLRGRAYGI GRAYGlDLL RAYGIDLLR AYGIDLLRT
GFLRGRAYGID FLRGRAYGIDL GRAYGIDLLRT RAYGIDLLRTE AYGIDLLRTEG
I -D
L
,
0 1 0 2 0 3 0 4 0 5 0 6 0
FLRGRAYGID LRGRAYGIDL RGRAYGIDLL
FLRGRAYGIDLL LRGRAYGIDLLR RGRAYGIDLLRT GRAYGIDLLRTE RAYGIDLLRTEG
I
YGIDLLRTEG 0
~_-_~ -
, , ___1 0 2 0 3 0 4 0 5 0
SPECIFIC LYSlS (%)
60
YGIDLLRTEGGG
I I I ~D ,
,
,
10
20
30
4
i
n
SPECIFIC LYSlS (%)
Figure 1. Cytotoxicity of CTL clone LC13 for LC/L4 LCL targets primed with 21 pM of peptides of lengths from 9 to 12 residues, homologous with sequence NAGFLRGRAYGIDLLRTE. Nonhomologous glycine residues were added to the carboxy-terminus of some peptides to maintain consistent peptide length.
193
Specificity of a CTL epitope
Eur. J. Immunol. 1992. 22: 191-195
El
1
80 70 60 50 40 30 20 10
m
100
n
80
70.
Ijl
MNPQRSTVWY'
lEFGH
ACDEFGH I KLMNPQRSTI
605040,
10
ACDEFGH I KLMNPQRSTVWY '
m
10090-
100
40~ 3020-
--7-4T+-uy
KLMNPQRSTVWY
10 0
100~
100
90 80-
90 80
70
70
60~ 50-
60
d r ACDEFGH I
---
i KLMNPQRSTVWY*
50
a-
40
30~ 20-
UK
10 0'-, , ACDEFGHIKLMNPQRSTVH ~
I
.
,
,
'
,
'
"
"
"
'
'
r
!L
MNPQRSTVWY'
REPLACEMENT AMINO ACID RESIDUE Figure 2. Effect of single-residue substitutions on cytotoxicity. Effectors and targets were as for Fig. 1. Every one of the 20 genetically coded amino acids, as well as a deletion (*),was tested in each of the nine locations within the "parent" sequence FLRGRAYGL at two concentrations of peptide. The boxed letter represents the parent residue being replaced; the horizontal axis lists the residue replacing the parent residue. Open bars represent the % lysis observed at the higher concentration of peptide (21 p ~ and ) the overlaid solid bars represent % lysis observed with 1/30 of this peptide level (0.7 pM).
GFLRGRAYGUL are both approximately twofold less active than FLRGRAYGUL, respectively (data not shown). 3.2 Correlation of CTL activity obtained with exogenous peptide and endogenous peptide We have previously demonstrated a 15-fold difference in activity between the peptides AWNAGFLRGRAYGLD and AWNAGFLRGRAYGID, which are derived from the EBNA-3 sequence of BL74 and B95-8, respectively (see Fig. 2 in [3]).The minimal epitope peptides FLRGRAYGL and FLRGRAYGI were titrated for CTL activity and an identical 15-fold difference in activity was observed. In the same experiment, levels of lysis were 40% and 3% for the autologous LCL transformed with the BL74 and B95-8 strains of EBV, respectively. These levels of lysis corresponded to the level measured when LC/L4 LCL was incubated with 0.7 pM of peptides FLRGRAYGL and FLRGRAYGI, respectively (equivalent to 0.6 pg/ml of the 15mer in Fig. 3in [3]).The replacement set of peptides was wbsequently tested at 0.7 y~ as this concentration
appeared to mimic the concentration of endogenously presented epitope.
3.3 Screening single-residue replacement peptides for CTL activity To assess the role of individual residues in the minimal active epitope, FLRGRAYGI,, sets of peptides in which each residue had been sequentially replaced with each of the other 19 genetically coded amino acids were tested for their ability to sensitize targets to CTL lysis. Peptides in which an amino acid had been deleted altogether were also synthesized and are represented with the symbol "*". The substituted peptides were screened at two concentrations (21 y ~ open ; bars and 0.7 y ~ filled ; bars; Fig. 2). The patterns of reactivity reveal that the two residues at the carboxy and amino termini of the peptide were much more replaceable than those in the central core. Substitution of any amino acid within the central RGRAY sequence of the epitope gave peptides with activity less than that of the FLRGRAYGI- peptide. Overall, 15/171 single-amino acid
Eur. J. Immunol. 1992. 22: 191-195
S. R. Burrows, S. J. Rodda, A. Suhrbier et al.
194
replacement peptides displayed activity significantly (k2 SD) higher than FLRGRAYGI, suggesting that these sequences would have some activjty at levels naturally presented on LCL.While 7 of these 1.5were less active than FLRGRAYGL, five substituted peptides (ELRGRAYGL, HLRGRAYGL, YLRGRAYGL, FIRGRAYGL and FLRGRAYSL) had significantly higher activity than FLRGRAYGL (Fig. 2).
-
3.4 Effect of peptide extension on residue replaceability
To determine whether the relatively high replaceability of the carboxy- and amino-terminal residues of the minimal epitope, was maintained when peptides were extended beyond the minimal length, longer replacement peptides were synthesized and tested for activity (at 21 pM). The replacement sets were as follows: substitution of F in the peptides GFLRGRAYGL and AGFLRGRAYGL; and substitution of L in the peptide FLRGRAYGLD. No significant difference in the pattern of replaceability was observed for peptide replacement sets of either F or L as a result of using sequences extended beyond the minimal epitope. The results for the set of peptides with replacements for F in the sequence AGFLRGRAYGL is shown in Fig. 3. The significant specific lysis observed with the deletion peptide (i.e. the 10mer peptide AG*LRGRAYGL) is consistent with the result obtained for GLKGRAYGL in Fig. 2 when used at 21 pM.
0.001
0.1
0.01
The activity of the five substituted peptides, which showed higher activity than FLRGRAYGL, were compared with FLRGRAYGL by titration. FLRGRAYSL was S f o l d more active, FIRGRAYGL, ELRGRAYGL and HLRGRAYGL were approximately Pfold more active, while YLRGRAYGL was 2-3-fold more active than F L R G n Y G L (Fig. 4a). A peptide incorporating three of these active amino acid replacements, HIRGRAYSL, was
RGRAYSL
100
10
Peptide concentration (pM)
loo
.-aa
i? -*
60
.-co
40
-
1
FLRGRAYGL
(b)
"IRGRAYSL
o 0
g
20 0 0.001
0.01 Peptide
3.5 Titration of peptides with higher activity than FLRGRAYGL
1
FLRGRAYGL ELRGRAYGL HLRGRAYGL MRGRAYGL FIRGRAYGL
0.1
1
10
100
concentration (pM)
Figure 4. (a) Titration of peptides FLRGRAYGL, ELRGRAYGL, HLRGRAYGL, YLRGRAYGL, FIRGRAYGLT and FLRGRAYS using effectors and targets for Fig. 1. (b) Titration of peptides FLRGRAYGL and BRGRAYSL using effectors and targets as for Fig. 1.
as
found to be approximately 30-fold more active than FLRGRAYGL (Fig. 4b; note both peptides in this figure were prepared using the method of Houghten [lS]).
4 Discussion
AGIFILRGRAYGL U
100
90
'
-80 70
$60 A50
Y
!km
0
E30
a20 10 0
A
REPLACEMENT AMINO ACID RESIDUE Figure 3. Effect of single-residue substitutions for phenylalanine on cytotoxic activity of the peptide AGFLRGRAYGL. Effectors and targets were as for Fig. 1. Each peptide was tested at 21 FM.
The minimal epitope recognized by CTL line LC13 was the nonapeptide FLRGRAYGL (Figs. 1 and 2).This epitope is encoded by the BL74 virus strain and is 15-fold more efficient as an epitope for CTL LC13 than the analogous sequence encoded by the B9S-8 strain (this report and [3]). In addition, a concentratin of 0.7 p~ of peptides FLRGRAYGL and FLRGRAYGI, when added exogenously to L C / L ~ gave , similar values of lysis as autologous B cells infected with BL74 and B9S-8 viral strains, respectively. Taken together, these data indicate that peptides, which give equal or lower lysis than the B95-8 sequence (FLRGRAYGI) when added at a concentration of 0.7 PM to LC/L4, are-unlikely to be active if naturally processed and presented by an LCL. The following discussion is, therefore, based on the replacement set data obtained at 0.7 p~ (solid bars in Fig. 2).0nly 15 of the 171single-amino acid substituted peptides had significantly (+ 2 SD) higher activity than FLRGRAYGI. These 15 substititons were all located in the two amino acids at the amino- and carboxyterminal ends of the epitope.The central residues, RGRAY, did not tolerate any substitutions.
Eur. J. Immunol. 1992. 22: 191-195
To test the possibility that the higher replaceability of residues at the amino- and carboxy-terminal ends of the minimal sequence was an artefact due to the use of a minimal determinant, the replaceability of the amino- and carboxy-terminal residues, F and L , were tested with peptides extended at either end by one or two naturally occurring amino acids. The replaceability patterns appeared independent of extensions outside the minimal epitope. It is unclear what the length of the naturally processed epitope is in this case, although it has been suggested that many CTL epitopes are nine amino acids long [17, 181.
Specificity of a CTL epitope
195
5 References
1 Townsend, A . and Bodmer, H., Annu. Rev. Immunol. 1989. 7: 601. 2 Burrows, S. R., Sculley, T. B., Misko, I. S., Schmidt, C. and Moss, D. J., J. Exp. Med. 1990. 171: 345. 3 Apolloni, A , , Moss, D. J., Stumm. R., Burrows. S. R.. Suhrbier, A., Misko, I . S.. Schmidt. C. and Sculley,T. B., Eur. J. Imrnunol. 1992. 22:183. 4 Murray, R. J., Kurilla, M. G . , Griffin, H. M., Brooks. M.. Arrand. J. R., Rowe, M., Burrows, S. R.. Moss, D. J.. Kieff, E. and Rickinson, A . B., Proc. Natl. Acad. Sci. USA 1990. 87: 2906. 5 Milligan. G . N., Morrison, L. A., Gorka. J., Braciale,V. L. and Braciale, T. J., J. Immunol. 1990. 145: 3188. The present study and a previous report [19] have revealed 6 Bastin. J., Rothbard, J., Davey. J., Jones, I. and Townsend. A , , substituted analogues which had higher activity than the J. Exp. Med. 1987. 165: 1508. natural sequence; FLRGRAYSL was 15-fold and 7 Joly, E., Salvato. M..Whitton, J. L. and Oldstone. M. B. A . , J. HLRGRAYGL and FIRGRAYGcwere 9-fold more active Virol. 1989. 63: 1845. than FLRGRAYGL at half maximal kiling. When these 8 Takahashi, H., Merli, S., Putney, S. D., Houghten, R.. Moss. replacements were combined, the resulting peptide, HIRB., Germain, R. N . and Berzofsky, J. A., Science 1989. 246; GRAYSL, triggered target cell lysis at a concentrzon 118. approximately 30-fold less than FLRGRAYGL. This data 9 Pircher, H., Moskophidis, D.. Rohrer, U., Burki, K., Hengartner, H. and Zinkernagel, R.M., Nature 1990. 346: 629. suggests that the effect on activity of multiple amino acid replacements is roughly equal to the additive effects of the 10 Wang, A . , Lu. S. D. and Mark, D. F., Science 1984. 224: 1431. single-residue replacements. 11 Rosenberg, S. A , , Grimm, E. A , . McGrogan, M., Doyle. M., Kawasaki, E., Koths, K.. Mark, D. F., Science 1984. 223: Assuming that the effect of all multiple substitutions result 1412. in the simple sum of the effect of independent substitutions, 12 Misko, I. S.. Schmidt, C., Mcm, D. J., Burrows, S. R. and this replacement set data suggests that a theoretical 240 Scu1ley.T. B., J. Gen. Virol. 1991. 72: 405. epitopes (10 substitutions for F, multiplied by 3 for L, 4 for 13 Maeji, J. N.. Bray. A. B. and Geysen. H. M.. .I. Immunol. G and 2 for L) might all be effective targets for LC13. The Meth0d.s 1990. 134: 23. value of 240 may, however, be an underestimate since the 14 Mutch, D. A ., Rodda, S. J.. Benstead. M.,Valerio, R. M., Geysen, H . M., Peptide Res. 1991. 4: 132. presence of substitutions, which enhance peptide activity, may allow some substitutions of the irreplaceable central 15 Houghten, R. A., Proc. Nutl. Acud. Sci. USA 1985. 82: 5131. residues. Making some simplifying assumptions there are I6 Misko, I. S., Pope, J. H., Hutter, R . , Soszynski,T. D. and Kane. 209 possible 9mers and since LC13 only recognizes 240 of R. G . , Int. J. Cancer 1984. 3.7: 239. them, this clone has a specificity of 1in 4.7 x 101(’(240/209)), 17 Falk, K., Rotzschke, O., StevanoviC, S.. Jung, G . and Rama level which exceeds class I1 T cell specificity by two orders mensee, H., Nature 1991. 351: 290. of magnitude [20] and antibody specificity by five orders of 18 Schumacher,T. N.M.,DeBruijn, M. L. H.,Vernie,L. N.. Kast, magnitude 1211. If this level of specificity is generally true W. M., Melief, C. J. M., Neefjes, J. J. and Ploegh, H. L., Nature 1991. 350: 703. for CTL recognition it would have implications for autoimmunity and vaccine design because (a) cross-reaction I9 Bodmer, H. C., Pemberton, R. M., Rothbard, J. B. and Askonas. B. A , , Cell 1988. 52: 253. between closely related sequences would be unlikely and (b) the majority (at least 91% in the case of FLRGRAYGL) 20 Suhrbier, A., Rodda, S. J., Ho. P. C.. Csurhcs, P.. Dunkley, H.. Saul, A . , Geyscn. H. M. and Rzepczyk, C., .I. Imrnunol.1991. of amino acid mutations would result in loss of recognition 147: 2507. by the CTL. 21 Geysen, H. M., Mason,T. J. and Rodda, S. J., J. Mol. Recog. 1988. 1: 32. r l L 2 was kindly provided by the Cetus Corporution. We t h u d Richard Luuricellu and his team for synthesizing the peptides.
Received June 19, 1991; in revised form September 19, 1991.
