Influenza basic polymerase 2 peptides are recognized by influenza nucleoprotein-specific cytotoxic T lymphocytes

0161-5890/92$5.00 + 0.00 Pergamon Press Ltd

Molecular Immunology, Vol. 29, No. 9, pp. 1089-1096, 1992 Printed in Great Britain.

INFLUENZA BASIC POLYMERASE 2 PEPTIDES ARE RECOGNIZED BY INFLUENZA NUCLEOPROTEIN-SPECIFIC CYTOTOXIC T LYMPHOCYTES ROBERTW. ANDERSON,*~JACKR. BENNINK,~JONATHANW. YEWDELL,?W. LEE MALOY,*$ and JOHN E. COLIGAN*§ *Biological Resources Branch and YLaboratory of Viral Diseases, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD 20892, U.S.A. (First received 20 August 1991; accepted in revised form 3 1 January

1992)

Abstract--Cytotoxic T lymphocytes (CTL) play an important role in limiting viral infections and in eradicating virus from host tissues. Recent progress in understanding the processing and presentation of viral antigens to CTL indicates that the CTL antigen receptor recognizes peptides derived from viral proteins that are bound to an antigen binding groove present in class I major histocompatibility complex (MHC) molecules. In understanding CTL anti-viral responses and in creating vaccines designed to elicit CTL responses, it is critical to identify the portions of viral proteins that bind class I molecules and are recognized by T cell receptors. Previous findings have indicated that a significant portion of the CTL response of H-2d mice to influenza virus is specific for one of the viral polymerases (PB2). To identify the region of PB2 naturally processed and presented by influenza virusinfected mouse cells to CTL, 31 PB2 peptides of 9-16 residues in length were chosen and chemically synthesized. Two peptides, PB2, residues 146-159 and 187-195, were found to sensitize histocompatible target cells for recognition by influenza virus-specific CTL. When CTL were generated to individual viral proteins using influenza-vaccinia recombinant viruses, we found, to our surprise, that PB2-specific CTL failed to recognize cells sensitized with PB2 peptides 146-159 and 187-195. Further analysis showed that these PB2 peptides were, in fact, recognized by nucleoprotein (NP)-specific CTL generated by NP-vat virus priming and influenza A virus stimulation, or NP peptide stimulation in vitro of NP-vat or influenza A-primed CTL. These results demonstrate that while screening peptide libraries one cannot assume that positive peptides necessarily identify the viral protein to which the CTL response is directed.

INTRODUCTION Major histocompatibility complex (MHC)-restricted cytotoxic T lymphocytes (CTL) are a prominent component of immune responses to many viruses. By directly lysing cells, releasing cytokines, or both, CTL function to limit virus dissemination and speed eradication (Yap et al., 1978; Lin and Askonas, 1981; Lukacher et al., 1984). CTL recognize linear sequences of amino acids derived from foreign proteins that are bound to a groove in class I major histocompatibility complex (MHC) class I glycoproteins (Townsend et al., 1986; Bjorkman et al., 1987). Synthetic peptides as short as five residues corresponding to sequences in viral proteins have been used to sensitize cells for recognition by CTL (Reddehase et al., 1989). Much of the progress toward understanding the processing and presentation of viral antigens to CTL has been made studying the mouse CTL response to

influenza A virus (for reviews see Braciale et al., 1987; Yewdell and Hackett, 1989; Townsend and Bodmer, 1989). At least 8 of the 10 viral gene products can be recognized by mouse CTL. The responsiveness to any given protein varies with the class I MHC alleles of the responder animal. CTL from H-2d mice are largely directed to the four viral gene products: nucleoprotein (NP), hemagglutinin (HA), non-structural one (NSl) and basic polymerase two (PB2). HA and NP are recognized in conjunction with Kd, NSl with Ld and PB2 with Dd (Bennink and Yewdell, 1988). Unlike the HA, which exhibits extensive antigenic and sequence variation between different influenza virus isolates, NP, NSl and PB2 are highly conserved among all influenza A viruses. Consequently, almost all of the cross-reactive CTL response in H-2d mice, i.e. the portion of the CTL response able to lyse cells infected with widely divergent influenza virus isolates, is specific for one of these conserved proteins (Bennink et al., 1986).

The antigenic determinants processed from HA and NP for recognition by H-2d-restricted mice have been BPresent address: Laboratory of Viral Diseases, NIAID, NIH, determined. HA determinants are located at positions Bethesda, MD 20892, U.S.A. 202-221 and 523-545 of the HA (Braciale et al., 1989), SPresent address: Magainin Sciences Inc., 5110 Campus Drive, while NP has a single determinant located at residues Plymouth Meeting, PA 19462, U.S.A. $Author to whom correspondence should be addressed: 147-158 (Bodmer et al., 1988); however, the location in the 759 residues John E. Coligan, Biological Resources Branch, NIAID, of the Dd-restricted determinant(s) NIH, Building, 4, Room 413, Bethesda, MD 20892, U.S.A. of PB2 has not been precisely identified, although one 1089

1090

R. W. ANDERSON et al.

or more antigenic regions of influenza A/NT/60/68 were mapped to residues l-402 (Reay et al., 1988). Consequently, the objective of the present study was to identify the exact antigenic epitopes recognized by PB2specific CTL by using synthetic peptides. Since it was impractical to synthesize by automated means overlapping peptides that encompassed the 759 residues of PB2, two algorithms (DeLisi and Berzofsky, 1985; Rothbard and Taylor, 1988) historically used for predicting T cell epitopes were utilized to select 3 1 candidate peptides. Two of these peptides were found to sensitize uninfected cells for lysis by influenza-specific CTL; however, much to our surprise, the peptides were not found to be recognized by PB2-specific CTL, but instead by NP-specific CTL. The recognition of processed PB2 protein is discussed. MATERIALS

AND METHODS

Mice Four- to five-week-old female BALB/c (H-2d) mice were obtained from Jackson Laboratory (Bar Harbor, ME). Mice were maintained under virus-free conditions under the supervision of a certified veterinarian. Viruses Influenza A viruses PR8 (A/Puerto Rico/g/34 [HlNl]) and NT60 (A/Northern Territory/60/68 [H3N2]) were grown in the allantoic cavities of lo-day-old embryonated chicken eggs (specific pathogen free eggs from SPAFAS, Inc., Norwich, CT) for approximately 40 hr at 35°C and 55% humidity and stored at - 70°C as infectious stocks. Hemagglutinating activity was determined by chicken red cell agglutination as described (Fazekas de St Groth and Webster, 1966). Vat recombinants were kindly provided by Dr B. Moss (NIAID, Bethesda, MD) and Dr G. Smith (Cambridge University, Cambridge, England) (Smith et al., 1987) and contained cDNA corresponding to influenza A/PR/8/34 virus segment one (basic polymerase two [PB2]) and segment five (nucleoprotein [NP]). The vat recombinants (~256, ~402, ~618) expressing truncated PB2 molecules (NT60), were kindly provided by Dr G. G. Brownlee (University of Oxford, Oxford, England). The recombinants express residues l-256 (~256) I-402 (~402) and 1-618 (~618) of PB2. Cells P815 cells are a DBA/2 (H-2d) mastocytoma cell line. This cell line was maintained in Dulbecco’s modified Eagle’s medium (DMEM) (GIBCO Laboratories, Grand Island, NY) supplemented with 10% heat-inactivated fetal calf serum (FCS) (Hazelton Biologics, Inc., Lenexa, KS), 100 units/ml penicillin and 100 pgg/ml streptomycin, 2 mM glutamine, 50 pgg/ml gentamicin and 25 mM HEPES at 37°C. In vitro cytotoxicity

assay

Virus-specific CTL were generated and assayed essentially as described (Bennink et al., 1987). Mice were immunized intraperitoneally with 100-300 hemagglutin-

ating units of influenza A virus diluted in phosphatebuffered saline (PBS). Three or more weeks later, immune splenocytes (5-7 x 10’) were stimulated in z,itro with influenza A-infected autologous splenocytes (2.5-3.5 x 10’) in 40 ml of Iscove’s medium (GIBCO Laboratories) containing 10% heat-inactivated FCS (HyClone Laboratories, Inc., Logan, UT), 100 units/ml penicillin and 100 pg/ml streptomycin, 2 mM glutamine, 50 pg/ml gentamicin and 25 mM HEPES for six or seven days at 37°C and 9% CO,. Influenza A virus- (2 x 10e5 HAU/ cell) and Vat recombinant virus-infected target cells (25100 pfu/cell) were prepared by infecting 2 x 1O6 cells in 200 ~1 of autoclavable DMEM (GIBCO Laboratories) containing 25 mM HEPES and 0.1% bovine serum albumin (BSA), pH 6.8, for 1 hr. After 1 hr, the target cells were diluted to 3 x 106/ml with complete DMEM and the infection continued for 6 more hr. After 6 hr, the target cells were labeled for 1 hr with 100 PCi 5’Cr at 37°C washed twice and resuspended to 1 x lO’/ml. Peptide-pulsed target cells were prepared in a similar fashion. Briefly, 200 ~1 of autoclavable DMEM with 25 mM HEPES and 0.1% BSA, pH 6.8, containing 2 x lo6 cells was made 50-200 PM for peptide by the addition of 20 ~1 of sensitizing peptide and incubated at 37°C for 1 hr. After this time, the peptide-sensitized target cells were treated as described above for the infected target cells. One hundred microliters (1 x lo4 target cells) were mixed with 100 ~1 of effector cells at various concns. Experimental and control groups were performed in triplicate. The assay was incubated at 37°C 9% CO, for 4 hr, after which time 100 ~1 of the cell-free supernatant was harvested and the free ” Cr counted in a Micromedic 4/600 ME Plus gamma counter (ICN Micromedic Systems, Huntsville, AL). The percent specific release was calculated as follows [(experimental release-spontaneous release)/(total release-spontaneous release)] x 100. The cross-reactivity of NP and PB2 peptides described below was observed in assays performed by different investigators. Peptide synthesis Peptides were synthesized on either the MilliGen 9050 Pepsynthesizer (MilliGen/Biosearch, Burlington, MA) or the ABI model 430 peptide synthesizer (ABI, Foster City, CA) using the manufacturer’s reagents and recommended chemistry cycles. Peptides were cleaved from the resin, precipitated and extracted several times with anhydrous ether and fractionated by gel filtration and/or HPLC. Amino acid composition, amino acid sequence analysis and reverse phase HPLC were performed to confirm peptide sequence and purity. Peptide purity was greater than 95% as determined by HPLC analysis. As the possibility of peptide cross-contamination was of concern, the results described in this manuscript were reproduced with at least three different peptide preparations for each of the PB2 peptides, residues 146-159 and 187-195. Similar results were obtained in experiments performed by different investigators. Further, irrelevant peptide-treated and mock-treated target cells did not give evidence of cross-contamination. It should be

Cross-reactive

noted that these peptides were prepared on two different synthesizers using different chemistries and purified on two different HPLC systems located in different laboratories. RESULTS Algorithmic

prediction

of antigenic

synthetic

1091

CTL epitopes of influenza A virus

peptides

To identify short synthetic peptides derived from the PB2 protein that can sensitize cells to recognition by influenza A virus-specific CTL, overlapping peptides from the entire PB2 protein could, in theory, be synthesized and screened. Since PB2 is 759 amino acids long, however, this would entail synthesizing more than 70 15mers overlapping by five amino acids at each end. As an alternative, two predictive algorithms, chosen for historic reasons, were used to select regions of PB2 likely to represent CTL recognition sites. One scheme is based on the amphipathic properties of a peptide, which postulates that peptides able to form helices with polar and apolar faces are more likely to associate with the antigen binding groove of MHC molecules and be recognized by T cells (DeLisi and Berzofsky, 1985). The Rothbard and Taylor algorithm (1988) was established empirically and is predicated on the assertion that T cell epitopes are more likely to contain a glycine or a charged amino acid residue, followed by two to three hydrophobic residues and ending with a glycine or a polar amino acid residue (Rothbard and Taylor, 1988). It should be noted that these two schemes are not mutually exclusive; however, recent evidence does not support the structural requirements these schemes proposed (Bjorkman et al., 1987; Falk et al., 1990; Rotzsche et al., 1990; van Bleek and Nathenson, 1990). The amphipathic algorithm predicted seven regions in PB2 with amphipathicity indices ranging from 25.1 to 103.8 for blocks of 11 amino acids in an cc-helix. As these are fairly large regions, they were synthesized as 15 peptides ranging from 12 to 16 residues. The Rothbard and Taylor algorithm predicted 34 regions of antigenicity. Thirteen of these epitopes overlap with amphipathic regions of the molecule and were contained within 11 of the peptides synthesized as amphipathic peptides. Many of the Rothbard and Taylor epitopes are quite close; therefore, the remaining 21 regions were synthesized in the form of 16 additional peptides ranging from 9 to 16 residues. Thus, a total of 31 peptides representing approximately 55% of PB2 were synthesized (Table 1). The peptides were assayed for their ability to sensitize target cells for recognition by influenza A/PR8 virus primed and A/NT60 virus stimulated CTL. Two of the 3 1 peptides, corresponding to PB2 residues 146-l 59 and 187-195, consistently sensitized target cells (representative data are shown in Table 2). In these experiments, P8 15 cells infected with influenza A/PR8, the full length PB2vat recombinant, NP-vat recombinant, the truncated PB2-vat recombinants, ~402 and ~618, are recognized by the influenza-specific effecters. Similar data were obtained with AjPR8 primed and stimulated effecters (data not shown). P815 cells pulsed with the positive

Table 1. PB2 peptides analyzed in this study Amphipathic peptides TVHYPKIYKTYFER-14’ PB2,Si-100 VERLKHGTFGPVHFR-15 PB2, 101-l 15 PB2,247-261 PB2,270-283 PB2, 354-368 PB2,404&419 PB2,420-435 PB2,436-450 PB2,470-484 PB2,485-499 PB2,555-570 PB2,57ll582 PB2, 583-596 PB2,68 l-694 PB2,6955706

RRAAVSADmLLE15” GIRMVDILKQNPTE-14 RKATRRLIQLIVSGR-15” NQRLNPMHQLLRHFQK-16 DAKVLFQNWGVEPIDN16” VMGMIGILPDMTPSI15” TERVVVSIDRFLRVR15“ DQRGNVLLSPEEY SE- 15 EF-=SLVPKAIRGQ16“ YSGFVRTLFQQM-12” RDVLGTFDTAQIIK-14” RRYGPALSINELSN-14” LAKGEKANVLIG12

Peptides with Rothbard and Taylor’s sequences (underlined) KTTVDHMAHKKYTS-14 PB2, I-15 KYPITADKmMIPE-16 PB2, 33-48 PB2,62-76 DAGSDRVMVSPLAVT15 PB2, 136-150 KEAQDVIMEVVFPNE15 PB2, 146-159 VFPNEVGARILTSE-14 PB2, 187-195 ERELVRKTR-09 PB2,207-22 1 YIEVLHLTQGTCWEQ-15 QAVGTCKAAMGLR-13 PB2,285-297 PB2, 3033313 SFGGFTFKRTS-1 1 PB2, 319-330 KREEEVLTGNLQT13 PB2, 384-398 SQEDCMIKAVRGDLN15 PB2,534-546 RNWETVKIQWSQN13 PB2,620-632 RGSGMmRGNS13 PB2,636-645 NYNKATKRLT-IO PB2,653-667 KDAGTLTEDPDEGTAG15 PB2,725-738 DSQTATKRIRMAIN13 “These peptides also contain Rothbard Taylor’s sequences (underlined).

and

peptide, NP, residues 147-158, are recognized, while an irrelevant peptide, SV40 T antigen, residues 225-239, was not recognized. Untreated (mock) and vat-infected cells were not recognized. Other H-2d target cells pulsed with PB2, 1466159 and 1877195 were recognized, while H-2 incompatible PB2 peptide pulsed cells were not recognized (data not shown). The PB2 peptides recognized by the influenza-specific CTL were predicted by the Rothbard and Taylor algorithm. Recognition

of PB2 peptides

by PB2-speciJic

CTL

The CTL used to test the peptide library were derived by priming and restimulating splenocytes with PR8 and, therefore, consisted of CTL specific for HA, NP and NSI, in addition to PB2 (Bennink and Yewdell, 1988). Although we expected that the peptides would be recognized by PBZspecific CTL only, it was essential to determine this experimentally. To this end, PB2-specific CTL were generated by priming mice with PB2-Vat and stimulating splenocytes in vitro with PR8. It has been shown (Bennink and Yewdell, 1988) that CTL generated in this manner recognize PB2, and not other influenza virus gene products.

1092

R.W. ANDERSON et al. Table 2. Recognition of PB2 peptides by BALB/ cByJ CTL % Specific release” Primed Stimulated

PR8h NT60 30 : 1

15:l

8:1

PR8 Mock

90 13

91 9

77 7

PBZVac ~256~ ~402~ v618d NP-Vat Vat

59 15 61 32 90 13

44 12 48 22 91 8

24 6 27 11 100 6

NP, 147-158 PB2, 146-I 59 PB2, 187-195 SV40 T, 225-239

93 87 78 17

88 78 74 13

65 69 56 11

Effector: target’ ratio

“CTL cultures were assayed in a 5iCr release assay as described in Materials and Methods. ‘CTL primed with A/PR8 and stimulated with A/NT60 recognize the influenza A virus NP, PB2 and NSl (Bennick et al., 1986). ‘Infected or peptide sensitized P815 cells (H-2d), a mastocytoma cell line, were used as the target cells. dPB2-Vac recombinant viruses expressing residues l-256 (v256), l-402 (~402) and l-618 (~618).

To our surprise, PB2 peptides were not recognized by PB2-specific CTL (Table 3). This is not due to a general failure of the peptides to sensitize target cells in this experiment, since sensitized cells were well recognized by polyclonal anti-influenza CTL populations (Table 3) nor is it attributable to a poor response of the PB2specific CTL, since these cells efficiently recognized either PR8 or PB2-Vat infected cells. The specificity of this response is shown by the failure of these effector cells to recognize uninfected cells, or cells infected with Vat, or NP-Vat. In the same experiment, we characterized the specificity of PB2-specific CTL using three Vat recombinants expressing truncated forms of PB2 (Table 3). These are the same recombinants used by Reay et al. (1988) to map the CTL response to PB2. Like Reay et al., we found that cells expressing the first 402 or 618 residues of PB2 were recognized by PB2-specific CTL and, to a lesser extent, by polyclonal anti-influenza CTL. We failed, however, to detect specific recognition of cells expressing the first 256 residues of PB2 relative to cells infected with NP-Vat or Vat, despite the fact that these cells were infected equally as well as cells expressing the longer truncated gene products, as judged by their lysis by Vat-specific CTL. Further, the PB2 expression of ~256 is equivalent to ~402 and ~618 as determined by immunofluorescence (data not shown). As the PB2 peptides able to sensitize cells are contained within the 256 residue fragment, this finding provides additional evidence that

Table 3. Recognition of PB2 peptides by PB2-specific effecters % Specific release0 Primed Stimulated

PR8b NT60

PB2-Vat’ PR8

PB2-Vacd Vat

Effector: target’ ratio 15 : 1

8 :1

15 : 1

8:1

15:1

8:1

PR8 Mock

91 9

77 7

77 16

54 8

9 11

6 9

PB2-Vat ~256’ ~402~ v618/ NP-Vat Vat

44 12 48 22 91 8

24 6 27 11 100 6

85 15 65 33 19 18

69 8 46 19 13 10

86 90 89 89 88 77

81 77 86 78 89 70

NP, 147-158 PB2, 146-159 PB2, 187-195 SV40 T, 225-239

88 78 74 13

65 69 56 11

12 14 12 18

7 8 7 9

9 10 11 11

8 7 9 9

“CTL cultures were assayed in a j’ Cr release assay as described in Materials and Methods. hCTL primed with A/PR8 and stimulated with A/NT60 recognize the influenza A virus NP, PB2 and NSl (Bennick et al., 1986). ‘CTL primed with PB2-Vat and stimulated with AjPR8 recognize the influenza A PB2 (Bennick and Yewdell, 1988). “CTL primed with PB2-Vat and stimulated with Vat recognize Vat proteins (Bennick and Yewdell, 1988). ‘Infected or peptide sensitized P815 cells (H-2d), a mastocytoma cell line, were used as the target cells. “PB2-Vat recombinant viruses expressing residues l-256 (v256), l-402 (~402) and l-618 (~618).

these peptides are not recognized by CTL that recognize determinants processed from PB2 synthesized by target cells. Consistent with this conclusion, in a series of experiments in which we confirmed that recognition of PB2 is solely Dd-restricted, we failed to observe Dd(or Kd- or Ld-) restricted recognition of PB2, 146-159 or PB2, 187-195 (data not shown). Recognition

of PB2 peptides

by NP-spec$c

CTL

Given the failure of PB2-specific CTL to recognize the PB2 peptides, it seemed likely that the peptides were recognized by CTL specific for another viral protein. The fact that the peptides were recognized by crossreactive CTL, induced by priming and restimulation with heterologous influenza viruses, indicated they were not recognized by HA-specific CTL, which are not present in these CTL populations (Bennink et al., 1986). Conversely, since a substantial fraction of cross-reactive CTL recognize NP (Wysocka and Bennink, 1988; Yewdell et al., 1985), NP-specific CTL were a prime candidate for recognizing the PB2 peptides. To examine this possibility, splenocytes from NP-Vat recombinant virus primed mice were stimulated in uitro with influenza A virus or with the NP 147-158 peptide (Table 4). Polyclonal influenza-specific CTL were included in the experiment as positive controls. Both populations of NP-specific CTL recognized the PB2

Cross-reactive

CTL epitopes

Table 4. Recognition

of influenza

of PB2 peptides

1093

A virus

by NP-specific

CTL

% Specific release” PR8b PR8

Primed Stimulated

PR8 NP, 147-158

NP-Vacd PR8

NP-Vat’ NP, 147-158

20 : 1

1O:l

2O:l

10: 1

20 : 1

1O:l

2O:l

1O:l

PR8 Mock

75 1

62 1

71 7

71 3

64 3

52 6

41 3

24 2

PB2-Vat NP-Vat Vat

36 52 3

37 39 3

14 83 8

11 77 6

11 74 6

14 57 7

14 42 9

12 28 7

NP, 147-158 PB2, 146158 PB2, 187-195 SV40 T, 225-239

33 38 37 4

20 27 31 4

82 68 66 8

57 57 54 5

56 49 46 5

35 36 36 6

26 28 24 5

13 16 17 4

Effector:

target’ ratio

“CTL cultures were assayed in a “Cr release assay as described in Materials and Methods. bCTL primed and stimulated with AjPR8 recognize the influenza A virus HA, NP, PB2 and NSl (Bennick and Yewdell, 1988). ‘CTL primed with AjPR8 and stimulated with NP, 147-158 recognize the influenza A virus NP, 147-158. dCTL primed with NP-VAc and stimulated with AjPR8 recognize the influenza A NP (Bennick and Yewdsell, 1988). ‘CTL primed with NP-Vat and stimulated with NP, 147-158 recognize influenza A virus NP, 147-158. ‘Infected or peptide sensitized P815 cells (H-2d), a mastocytoma cell line, were used as the target cells.

peptides. The extent of recognition of PB2 peptides by the various CTL populations paralleled their recognition of NP, as measured by their lysis of NP-Vat infected cells. The specificity of the anti-NP CTL populations for NP is shown by their failure to lyse PB2-Vat infected cells significantly above control values. These findings demonstrate that the recognition of PB2 peptides 146-159 and 187-195 is based on a cross-reaction with the NP 147-158 peptide. To determine the extent of cross-reactivity, virusspecific and NP-specific CTL were used to determine the dose response curves for the NP, 147-158, PB2, 146-159 and 187-195 peptides. Representative results from virus-specific effecters are shown in Fig. 1. The PB2 peptides are at least three log,, less effective at sensitizing target cells. DISCUSSION The objective of the present study was to precisely define the region or regions recognized by PBZspecific CTL using synthetic peptides. As PB2 is rather large, two algorithms that identify motifs statistically favored to represent T cell determinants were used to identify potential epitopes for historic reasons (DeLisi and Berzofsky, 1985; Rothbard and Taylor, 1988). Thirtyone peptides were selected that were either amphipathic, contained Rothbard and Taylor sequences, or both. Out of the 31 synthetic peptides, two were found to be recognized by influenza A virus-specific CTL.

Surprisingly, the effecters recognizing these PB2 peptides in polyclonal anti-influenza populations were apparently induced by the NP 147-158 determinant, and not by determinants derived from PB2. Thus, while the predictive algorithms may help in identifying antigenic regions of proteins, they cannot guarantee the specificity and restriction of the effecters that recognize these epitopes. For large proteins such as PB2, the best strategy for locating antigenic determinants is to first localize the determinant to segments of 50 to 100 residues using Vat recombinants expressing genetically truncated versions of the protein, and then precisely identify the determinant(s) using synthetic peptides. The determinant(s) normally processed from PB2 for CTL recognition remain to be defined. Our finding that PB2-specific CTL recognize cells expressing PB2 residues l-402 but not residues l-256 suggests that a CTL determinant is located between residues 256 and 402. It is also possible, though less likely based on published reports of presentation of truncated proteins to CTL (Tevethia er al., 1983; Townsend et al., 1985), that the loss of residues 257-402 affects the processing or presentation of a site located in l-256 fragment. Comparison of the PB2, 146-159 and 187-195 peptides with the NP, 147-158, peptide reveals a fair degree of homology around the middle of the 147-158 sequence (Table 5). This is particularly striking with the 187-195 peptide, which is identical at four of five residues in this region. Neither of the PB2 peptides, however, possesses a tyrosine corresponding to position 148 of the NP

R. W. ANDERSONet al.

1094

+

PB2 146159

i_

PB2 185-197

Concentration (mM) Fig. 1. Titration of cross-reactive influenza A virus CTL epitope. P81.5 target cells were sensitized with lo-fold serial dilutions of the NP, 147-158, PB2, 146159 and PB2, 185-197 peptides. They were assayed in a “Cr release assay as described in Materials and Methods with CTL primed and stimulated with AIPR8 at an effector-to-target ratio of 25: 1.

sequence. This tyrosine is present in al1 of the previously described peptides known to be naturally processed from proteins for H-2Kd restricted CTL recognition (Maryanski et al., 1990). It has been proposed that it contacts the antigen binding groove of Kd, and is believed to confer specificity on peptide-MHC interactions. Analysis of other peptides recognized in the context of H-2Kd, epitopes of the influenza HA (Braciale et al., 1989; Kuwano et al., 1988), plasmodium circumsporozoite (Romero et al., 1989) and class I molecules (Maryanski et al., 1986, 1988) reveals that they also have corresponding tyrosines (Table 5). It is worth considering why the PB2 determinants we have identified as cross-reacting with NP-specific CTL are not normally processed for CTL recognition. First, it is possible that the relevant portion of PB2 never has the opportunity to associate with Kd, either because the region containing the determinants is rapidly degraded, is not transported into the exocytic compartment where antigen association occurs, or is not provided to Kd in a length capable of binding. Second, in the conditions under which antigen association normally occurs in the Table 5. Comparison Peptide PB2, 146-159 PB2, 187-195 NP, 147-158 HA (Jap), 523-545 HA (PRO), 512-534 HA (Jap), 202-221 HLA-A24, 170-182 H-2Kd, 17&182 PolyPro analog P. berghei CS, 249-260

cell, the PB2 determinants might have a lower affinity for class I than they do for class I expressed at the cell surface, where exogenous peptides associate with class I molecules (Yewdell and Bennink, 1989; Nuchtern et al., 1989; Hosken et al., 1989). Perhaps the conserved tyrosine present in Kd-restricted determinants is essential for binding to class I intracellularly. Third. and most likely, the concentration of the PB2 dete~inants might not reach sufficient levels in the cells to produce the number of determinants required for T cell recognition. Whereas the NP, 147-158 peptide is able to sensitize cells at concns as low as lo-l3 M, sensitization with the PB2 peptides require concns at least three log,, higher. We feel that this relatively low binding aflinity offers the most likely explanation for our failure to demonstrate that the cross-reactive PB2 peptides are H-2Kd restricted when H-2Kd transfected L cells and congenic cell lines were utilized as target cells. In this case, it seems possible that the PB2 peptides only weakly compete with endogenously processed peptides present in the H-2Kd peptide binding groove. Nonetheless, the fact that NP, 147-158 primed effector cells, which are known to be H-2Kd

of cross-reactive PB2 epitopes with sequences of H-2Kd-restricted Sequence VFPNEVGAR I LTSE ERE LVRKTR TYQRTRALVRTG VYQILAIYATVAGSLSLAIMMAG IYQILAIYSTVASSLVLLVSLGA RTLYQNVGTYVSVGTSTLNK RYLENGKETLQRA RYLKNGNETLLRT AYPPPPPTLA NDDSYIPSAEKI

epitopes

Reference This study This study Braciale et al. (1989) Bennick et al. (1986) Kuwano et al. ( 1988) Bennick et al. (1986) Maryanski et al. (1986) Maryanski et al. (1988) Maryanski et al. (1990) Romero et al. (1989)

Cross-reactive

restricted, recognized the PI32 peptide sensitized target cells offers indirect support that the PB2 peptides are recognized in the context of H-2Kd. Shimojo et al. (1989) described cross-recognition of the human rotavirus VP4, residues 86-100 by an HLA-A2.1 restricted human CTL clone induced by the influenza A matrix peptide residues 57-68. Recently, Kuwano et al. (1991) demonstrated recognition of influenza HA and NSl peptides by an H-2Kd restricted influenza-specific CTL clone. Similar observations have been reported with the CTL response to lymphocytic choriomeningitis virus (LCMV) (Joly et al., 1989). An LCMV- (Armstrong strain) specific, H-2Db-restricted CTL clone, which mapped to Armstrong glycoprotein residues 272-286, lysed cells pulsed with Pasteur strain-derived peptide, residues 272-293, while failing to lyse Pasteur straininfected cells. Others have made analogous observations. Influenza-infected HLA-Aw69 expressing cells were not recognized by influenza matrix protein (residues 58-69)specific, HLA-A2 restricted CTL, while HLA-Aw69 cells pulsed with the matrix peptide were recognized by these CTL (Bodmer et al., 1989). SV40 large T antigen peptide (residues 205-219 or 489-511) pulsed H-2bm” and H-2bm14cells are recognized by SV40-specific CTL clones, but the endogenous T antigen synthesized in these SV40 transformed cells fails to sensitize these cells for recognition (Tevethia et al., 1990). In these studies, the naturally processed antigen failed to associate with the appropriate MHC molecules in sufficient concentration to sensitize cells for recognition by the CTL. Finally, it should be appreciated that there is no fundamental reason why naturally processed determinants from different proteins would be prevented from cross-reacting. Indeed, this must occur at some finite probability. To cite one possible example of crossreactivity between dete~inants derived from distinct proteins, numerous virus-specific clones have been identified that also recognize noninfected cells expressing different alleles of class I molecules. For example, HSV(Jennings, 1985) and Moloney-specific (Stukart et al., 1984) H-2Kb-restricted CTL recognize uninfected cells expressing Kbmmolecules. It has been proposed that many of these cells recognize determinants derived from self proteins in association with the allelic class I molecule. Given the number of proteins produced by large DNA viruses such as Vat or herpes viruses, it would not be too shocking to discover T cell clones that recognized distinct peptides in a cross-reactive

manner.

1095

CTL epitopes of influenza A virus

The positive selection

of lymphocytes that occurs in the thymus must require such cross-reactivity, as the T cell receptor repertoire could never be exposed to all of the (viral) peptides that may be bound by self MHC molecules. The limited number of thymic peptides used to select the repertoire must fairly resemble peptides that mature lymphocytes will encounter in peripheral tissue. Acknowledgements-We thank Drs B, Moss, G. Smith and G. G. Brownlee for recombinant vat viruses and Ms. Natalie Payne for excellent technical assistance. We also thank Drs Beatriz Carreno and William Biddison for their critical review of this manuscript.

REFERENCES Bennink J. R. and Yewdell J. W. (1988) Murine cytotoxic T lymphocyte recognition of individual influenza virus proteins: high frequency of nonresponder MHC class I alleles. f. exp. Med. 168, 1935-1939.

Bennink J. R., Yewdell J. W., Smith G. L. and Moss B. (1986) Recognition of cloned influenza virus hemagglutinin gene products by cytotoxic T lymphocytes. J. Viral. 57, 786-791. Bennink J. R., Yewdell J. W., Smith G. L. and Moss B. (1987) Anti-influenza virus cytotoxic T lymphocytes recognize the three viral polymerases and a non-structural protein: responsiveness to individual viral antigens is major histocompatibility complex controlled. J. Virof. 61, 1098-l 102. Bjorkman P. J., Saper M. A., Samraoui B., Bennet W. S., Strominger J. L. and Wiley D. C. (1987) Structure of the human class I histocompatibility antigen, HLA-AZ. Nature 329, 506-512. Bodmer H. C., Gotch F. M. and McMichael A. J. (1989) Class I cross-restricted T cells reveal low responder allele due to processing of viral antigen. Nature 337, 653--655. Bodmer H. C., Pemberton R., Rothbard J. and Askonas B. (1988) Enhanced recognition of a modified peptide antigen by cytotoxic T cells specific for influenza nucleoprotein. Cell 52, 253-258. Braciale T. J., Morrison L. A., Sweetser M. T., Sambrook J., Gething M. J. and Braciale V. L. (1987) Antigen presentation pathways to class I and class II MHC-restricted T lymphocytes. fmmun. Reu. 98, 95-114. Braciale T. J., Sweetser M. T., Morrison L. A., Kittlesen D. J. and Braciale V. L. (1989) Class I major histocompatibility complex-restricted cytolytic T lymphocytes recognize a limited number of sites on the influenza hemagglutinin. Proc. natn. Acad. Sci. U.S.A. 86, 277-281,

DeLisi C. and Berzofksy J. A. (1985) T-cell antigenic sites tend to be amphipathic structures. Proc. natn. Acad. Sci. U.S.A. 82, 7048-7052.

Falk K., Rotzsche 0. and Rammensee H.-G. (1990) Cellular peptide composition governed by major histocompatibility complex class I molecules. nature 348, 248-251. Fazekas de St Groth S. and Webster R. G. (1966) Disquisition on original antigenic sin. I. Evidence in man. J. exp. Med. 124, 331-345. Hosken N. A., Bevan M. J. and Carbone F. R. (1989) Class I-restricted presentation occurs without internalization or processing of exogenous antigenic peptides. J. Immun. 142, 1079-1083.

Jennings S. R. (1985) Cross-reactive recognition of mouse cells expressing the bm3 and bmll mutations with H-2Kbrestricted herpes simplex virus-specific cytotoxic T lymphocytes. J. Immun. 135, 3530-3536. Joly E., Salvato M., Whitton J. L. and Oldstone M. B. A. (1989) Pol~orphism of cytotoxic T-I~phocyte clones that recognize a defined nine-amino acid immunodominant domain of lymphocytic choriomeningitis virus glycoprotein. J. Viral. 63, 1845-1851. Kuwano K., Braciale T. J. and Ennis E. A. (1988) Localization of a cross-reactive CTL epitope to the transmembrane region on the hemagglutinin of influenza Hl and H2 viruses. FASEB

J. 2, 2221.

Kuwano K., Reyes V. E., Humphreys R. E. and Ennis F. (1991) Recognition of disparate HA and NSl peptides by an H-2Kd-restricted, influenza specific CTL clone. Nolec. immun.

28, l-7.

Lin Y.-L. and Askonas B. A. (1981) Biological properties of an influenza A virus specific killer T cell clone. Inhibition

1096

R. W. ANDERSONet al.

of virus replication in vivo and induction of delayed type hypersensitivity reactions. J. exp. Med. 154, 225-234. Lukacher A. E., Braciale V. L. and Braciale T. J. (1984) In vivo effector function of influenza virus-specific cytotoxic T lymphocyte clones is highly specific. J. exp. Med. 160, 814-826. Maryanski J. L., Pala P., Cerrotini J.-C. and Corradin G. (1988) Synthetic peptides as antigens and competitors in recognition by H-2 restricted cytolytic T cells specific for HLA. J. exp. Med. 167, 1391-1405. Maryanski J. L., Pala P., Corradin G., Jordan B. R. and Cerrotini J. C. (1986) H-2 restricted cytolytic T cells specific for HLA can recognize a synthetic HLA peptide. Nature 324, 578-579.

Maryanski J. L., Verdini A. S., Weber P. C., Salemme F. R. and Corradin G. (1990) Competitor analogs for defined T cell antigens: peptides incorporating a putative binding motif and polyproline or polyglycine spacers. Cell 60, 63-72. Nuchtern J. G., Bonifacino J. S., Biddison W. E. and Klausner R. D. (1989) Brefeldin A implicates egress from endoplasmic reticulum in class I restricted antigen presentation. Nature 339, 223-226.

Reay P. A., Jones I. M. and Brownlee G. G. (1988) A region of the influenza A/NT/60/68 PB2 protein containing an antigenie determinant recognized by murine H-2Dd restricted cytotoxic T lymphocytes. Virology 167, 261-268. Reddehase M. J., Rothbard J. B. and Koszinowski U. H. (1989) A pentapeptide as minimal antigenic determinant for MHC class I-restricted T lymphocytes. Nature 337,651-653. Romero P., Maryanski J. L., Corradin G., Nussenzweig R. S., Nussenzweig V. and Zavala F. (1989) Cloned cytotoxic T cells recognize an epitope in the circumsporozoite protein and protect against malaria. Nature 341, 323-326. Rothbard J. B. and Taylor W. R. (1988) A sequence pattern common to T cell epitopes. EMBO J. 7, 93-100. Rotzsche O., Falk K., Deres K., Schild H., Norda M., Metzger J., Jung G. and Rammensee H.-G. (1990) Isolation and analysis of naturally processed viral peptides as recognized by cytotoxic T cells. Nature 348, 252-254. Shimojo N., Maloy W. L., Anderson R. W., Biddison W. E. and Coligan J. E. (1989) Specificity of peptide binding by the HLA-A2.1 molecule. J. Immun. 143, 2939-2947. Smith G. L., Levin J. Z., Palese P. and Moss B. (1987) Synthesis and cellular location of the ten influenza polypeptides individually expressed by recombinant Vat viruses. Virology 160, 336-345.

Stukart M. J., Boes J. and Melief C. J. M. (1984) Recognition of H-2Kb mutant target cells by Moloney virus-specific cytotoxic T lymphocytes from bm13 (H-2Db mutant) mice. II. Relationship of Kbm3and Kbm” in restriction specificities and alloderminants. J. Immun. 133, 28-32. Tevethia S. S., Lewis M., Tanaka Y., Milici J., Knowles B., Maloy W. L. and Anderson R. (1990) Dissection of H-2Dbrestricted cytotoxic T-lymphocyte epitopes on simian virus 40 T antigen by the use of synthetic peptides and H-2Dbm mutants. J. Viral. 64, 1192-1200. Tevethia S. S., Tevethia M. J., Lewis A. J., Reddy V. B. and Weissman S. M. (1983) Biology of simian virus 40 (SV40) transplantation antigen (TrAg). IX. Analysis of TrAg in mouse cells synthesizing truncated SV40 large T antigen. ViroZogy 128, 3 19-330. Townsend A. and Bodmer H. (1989) Antigen recognition by class I-restricted lymphocytes. A. Rev. Immun. 7, 601-624. Townsend A. R., Gotch F. M. and Davey J. (1985) Cytotoxic T cells recognize fragments of the influenza nucleoprotein. Cell 42, 4577467. Townsend A. R. M., Rothbard J., Gotch F. M., Bahadur G., Wraith D. and McMichael A. J. (1986) The epitopes of influenza nucleoprotein recognized by cytotoxic T lymphocytes can be defined with short synthetic peptides. Cell 44, 959-968. Van Bleek G. M. and Nathenson J. G. (1990) Isolation of an endogenously processed immunodominant viral peptide from the class I H-2Kb molecule. Nature 348, 213-216. Wysocka M. and Bennink J. R. (1988) Limiting dilution analysis of memory cytotoxic T lymphocytes specific for individual influenza virus gene products. CeN. Immun. 112, 425-429.

Yap K. L., Ada G. L. and McKenzie I. F. C. (1978) Transfer of specific cytotoxic T lymphocytes protects mice inoculated with influenza virus. Nature 273, 238-239. Yewdell J. W. and Bennink J. R. (1989) Brefeldin A specifically inhibits presentation of protein antigen to cytotoxic T lymphocytes. Science 244, 1072-1075. Yewdell J. W., Bennink J. R., Smith G. L. and Moss B. (1985) Influenza A virus nucleoprotein is a major target antigen for cross reactive anti-influenza A virus cytotoxic T lymphocytes. Proc. natn. Acad. Sci. U.S.A. 82, 1785-1789. Yewdell J. W. and Hackett C. J. (1989) Specificity and function of T lymphocytes induced by influenza A viruses. In The Influenza Viruses (Edited by Krug R.), pp. 361-429. Plenum Publishing, New York.