Analyses of the cytotoxic T lymphocyte responses to glycoprotein and nucleoprotein components of lymphocytic choriomeningitis virus

VIROLOGY

162, 32 l-327

(1988)

Analyses of the Cytotoxic T Lymphocyte Responses to Glycoprotein and Nucleoprotein Components of Lymphocytic Choriomeningitis Virus J. LINDSAY WHITTON,’ Department

of immunology,

PETER J. SOUTHERN, AND MICHAEL B. A. OLDSTONE

Research institute of Scripps Clinic, 10666 N. Torrey Pines Road, La Jolla, California Received August 24, 1987; accepted

92037

October 20, 1987

The outcome of infection by lymphocytic choriomeningitis virus (LCMV) in the natural murine host is determined in large part by the cytotoxic T lymphocyte response (CTL) mounted by the host. In order to define the specificities of CTL induced by LCMV infection, we have cloned and expressed the full-length nucleoprotein (NP) gene and 75% of the glycoprotein (GP) gene of LCMV in vaccinia virus vectors and have used these recombinant viruses to sensitize syngeneic target ceils to lysis by anti-LCMV CTL. We have studied the anti-LCMV CTL responses induced on three different mouse H2 (major histocompatibility complex) backgrounds. First, we find that the relative recognition of the two LCMV proteins differs markedly on different H2 haplotypes; both proteins are seen on the H2bb background, while only NP is recognized on two other haplotypes (H2dd and H29. Second, we show that on the H2bb background the anti-GP CTL response comprises a major component of the overall CTL response, in marked contrast to several other viruses, e.g., influenza virus, vesicular stomatitis virus, and respiratory syncytial virus where anti-GP responses, if present, comprise only a minor portion of the whole. Third, LCMV GP can be a major target antigen for CTL induced by a serotypically distinct strain of LCMV, again in contrast to the above virus systems in which CTL cross-reactivity 0 1988 Academic Press, Inc. among different serotypes is dependent largely on the recognition of “internal” proteins.

tionally to yield two mature products, GP-1 (residues l-262) and GP-2 (residues 263-498) (Buchmeier ef al., 1987). Using segmental reassortants between two different strains of LCMV, both induction of and target cell recognition/lysis by MHC class l-restricted CTL have been mapped to the virus S segment (Riviere et a/., 1986). In order to estimate the relative contributions made by GP and NP, we needed to express each protein independently. The expression vector we selected, vaccinia virus was chosen primarily because, like LCMV, it has a cytoplasmic replication cycle with no known nuclear phase, and we wished to avoid any potential problems generated by exposing the LCMV RNA sequences to a nuclear environment, where splicing and/or other post-transcriptional processing might have taken place. We describe here the expression of LCMV GP and NP in recombinant vaccinia viruses and their use in the analyses of the primary CTL responses induced during LCMV infection.

INTRODUCTION Immune responses play an important role in recognition and elimination of foreign materials. A critical facet of the immune response to virus infection, shown first in studies on lymphocytic choriomeningitis virus (LCMV), is the induction of virus-specific major histocompatibility complex (MHC)-restricted cytotoxic T lymphocytes (CTL) (Zinkernagel and Doherty, 1979). These cells are important in effecting virus clearance in acute infection of the natural murine host, while failure to mount a CTL response leads to lifelong virus persistence (reviewed in Buchmeier et a/., 1980; Lehmann-Grube, 1984; Byrne and Oldstone, 1984). We wish to analyze this virus/CTL interaction at the molecular level, and have set out to localize CTL epitopes within the LCMV proteins and to identify the particular MHC regions with which they interact. This report represents the initial step toward this goal. LCMV, the prototype of the family Arenaviridae, has a genome comprising two single-stranded RNA segments, long (L) and short (S). The S segment of LCMV strain Armstrong (ARM) has been cloned and sequenced and encodes two polypeptides, nucleoprotein (NP; 558 amino acids) and glycoprotein (GP) (Southern et a/., 1987). The latter molecule is synthesized as a 498 amino acid precursor, GP-C (Buchmeier and Oldstone, 1979) which is cleaved post-transla-

MATERIALS

AND METHODS

Cell lines, viruses, and mouse strains HeLa, BALB/Cl7 (H2dd), and SWR/J (H2qq) cells were maintained in 59/oCO* using MEM supplemented with 10% heat-inactivated fetal calf serum, 1 mM L-glutamine, with penicillin, streptomycin, and fungizone. 143 TK- cells, used in the selection of recombinant vaccinia viruses, were maintained in the above medium in

’ To whom requests for reprints should be addressed. 321

0042-6822/88 Copyright All rights

$3.00

0 1988 by Academic Press, Inc. of reproduction I” any form resewed.

322

WHITTON.

SOUTHERN,

the presence of 25 pg/ml BUdR. MC57 (H2bb) cells were grown in RPM1 supplemented as above. Virus strains used [LCMV strains Armstrong (LCMV ARM) and Pasteur (LCMV PAST)] were triple plaque purified; their origins have been described elsewhere (Dutko and Oldstone, 1983). Mice [C57BL/6 (H2bb), BALB/ WEHI (H2dd), and SWR/J (H2qq)] were obtained from the breeding colony at the Research Institute of Scripps Clinic. Construction

of recombinant

vaccinia

viruses

This was achieved by standard techniques (Mackett et al., 1984). The plasmid vector used to introduce the LCMV genes into vaccinia virus by homologous recombination was pSC,, (Chakrabarti er a/., 1985), which contains the gene encoding the enzyme fl-galactosidase; this allows the identification of recombinant plaques by including the histochemical stain X-Gal in the agarose overlay, resulting in the rapid development of blue coloration over a recombinant plaque. Such plaques were picked and plaque-purified four times. Virus stocks used in the experiments were produced in 143 TK- cells, and harvested by subjecting the collected cells to three cycles of -70”/37” freeze/thawing, followed by treatment with 0.13% trypsin for 15 min at 37” to disaggregate virus particles. Northern blot analyses of LCMV-specific recombinant vaccinia

RNAs in

Confluent monolayers of 143 TK- cells (5 X 1O6 cells/lOO-mm plate) were infected at m.o.i. of 3 with total VVGP, WP I or WC~ 1 . Five hours postinfection cytoplasmic RNA was harvested as follows. Cells were rinsed twice with 10 ml ice-cold PBS, and 1 ml ice-cold isotonic lysis buffer (150 mn/l NaCI, 1.5 mNI MgCI,, 10 mhll Tris, pH 7.8, 0.65% (v/v) NP-40) was added. The cells were collected by scraping and retained on ice for 10 min; following centrifugation (Eppendorf centrifuge, 3500 rpm, 3 min) the supernatant was retained and an equal volume of phenol extraction-buffer (7 M urea, 350 mM NaCI, 10 rnn/l EDTA, 10 ml\/lTrispH 7.9, 1% (w/v) SDS) was added. Following one extraction with an equal volume of 1 :l phenol chloroform, the aqueous layer was retained and 3 vol of ethanol was added to precipitate the RNA. The RNAs were analyzed on a 1O/oagarose gel containing 69/o formaldehyde and, prior to transfer to a nitrocellulose filter, were visualized by ethidium bromide/uv fluorescence to confirm integrity and equivalence of amounts loaded in each track. The blot was baked and LCMVspecific cDNA probes to either GP or NP (Southern et a/., 1987) were used to identify appropriate messages expressed from recombinant vaccinia.

AND OLDSTONE

Preparation

of effector cells (splenic lymphocytes)

Mice were inoculated intraperitoneally with LCMV (1 O6PFU) or vaccinia virus (2 X 1O6 PFU). Spleens were collected after 6 days (vaccinia virus), 7 days (LCMV strain ARM), or 9 days (LCMV strain PAST), cleared of erythrocytes and used in cytotoxicity assays as described. Cytotoxicity

assays

Target cells were mock-infected, or infected with LCMV (48 hr prior to the assay, m.o.i. = 1) or with vaccinia virus (6 hr before assay, m.o.i. = 3) incubated for 1 hr with 5’Cr, and then incubated for 5 hr in triplicate samples with effector cells at effector:target ratios of 5O:l and 25:1, unless otherwise indicated. Supernatants were collected, and percentage specific release of !j’Cr was calculated by 100 X [(cpm sample release - cpm spontaneous release)/(cpm total release - cpm spontaneous release)]. Standard deviation among triplicate samples was less than 5%, and results presented have been confirmed in at least three separate experiments. RESULTS Construction of recombinant LCMV GP and NP

vaccinia

expressing

Figure 1A shows the manipulations involved in cloning the LCMV GP and NP genes into vaccinia virus. Full-length NP was excised, cloned into the Smal site of pSC,, , and recombined into vaccinia virus in the standard manner. A full-length GP clone was not available at the commencement of these procedures, so the longest clone available, defined by a BarnHI-Bglll fragment containing the translational initiation codon and encoding 363 amino acids (about 75% of the total GP), was used. This was cloned into a PUTT plasmid (I. Lindsay Whitton, unpublished observations) which contains translational termination codons in all three reading frames downstream of a multiple cloning site. The resultant reading frame, confirmed by DNA sequencing prior to introduction into vaccinia virus, encodes six non-LCMV amino acids before the stop codon, as shown in Fig. 1A. The entire reading frame was excised on a BarnHI-HindIll fragment, cloned into PSCII I and introduced into vaccinia virus. Recombinant viruses were selected by formation of blue plaques in the presence of BUdR (see Materials and Methods) and further assessed for expression of LCMV sequences (see below). A third virus recombinant was made by recombining the transfer plasmid pSC,, (in the absence of any LCMV sequences) into vaccinia virus. The resultant virus, VVsc,, , is TK- and p-Gal+, and is used as a control in cytotoxicity assays.

CTL RESPONSES TO LCMV INFECTION

323

““SC,, ““GP ““NP

““SC,,

““GP

““NP

,28S-

18S--. -.__ hGATCctctagagtcgcttaattaat!.. I I I I I I I BarnHI

A_

u SW

5:

i

GP Probe

NP Probe

FIG. 1A. Cloning of the LCMV GP and NP genes into VV. The LCMV S segment is shown with its encoded proteins. The cloning procedures are described under Results. The filled box represents a universal translation terminator, and the DNA sequence across the cloning junction is shown. Uppercase letters are GP-derived bases, and lowercase are plasmid-derived nucleotides. Note the addition of six non-LCMV amino acids to the C-terminus of the truncated GP. FIG. 16. Northern blot analysis of total cytoplasmic RNA from cells infected with W sc,, (a control which contains no LCMV sequences), Wep, or WNp, and hybridized with either the GP-specific probe (left-hand panel) or the NP-specific probe (right-hand panel). The blot was exposed at -70” against Kodak XAR5 film with a Cronex lightning plus intensifying screen.

We looked first for expression by the recombinant viruses of RNAs containing the LCMV sequences. Because of the nature of the pSC,, plasmid (Chakrabarti et al., 1985) the size of such RNAs should be a combination of three features: the size of the DNA insert (1200 for GP, 1700 for NP), an additional 350 bases encoded by pSC,, sequences, and a poly(A) tail of indeterminate length (often 200-400 bases). Northern blot analysis of total cytoplasmic RNA prepared from recombinant vaccinia virus-infected cells was carried out (Fig. 1 B). For both NP and GP recombinants, RNAs of appropriate sizes were readily detectable. (The faint second band visible with the NP probe may represent low-level premature transcription termination by VV RNA polymerase at a specific point in the LCMV NP sequences.) To confirm that the LCMV-specific RNAs directed production of the encoded proteins, we used fluorescent antibodies specific for NP or GP moieties to probe for protein expression and could readily demonstrate intense cytoplasmic fluorescence (data not shown). Repeated attempts to display surface expression of GP determinants by immunofluorescence gave weakly positive results. Nevertheless the recombinant viruses clearly are expressing the LCMV sequences at both RNA and protein levels.

Both GP and NP moieties of LCMV can induce a marked CTL response Next we analyzed the ability of CTL to detect LCMV determinants on the surface of recombinant infected cells. As is shown in Table 1, H2bb target cells (C57!3L/6) infected with either VVGP or VVNP were efficiently lysed by syngeneic spleen cells from LCMV ARM-infected mice, in a manner both Ml-E-restricted (infected H2dd targets are not lysed) and LCMV-specific [VV wild-type (w,,)-infected cells are not lysed, although these cells are productively infected and can be lysed by anti-VV CTL]. These results show that LCMV GP and NP both can provide determinants for presentation by the H2bb class I MHC molecule(s). Induction of an anti-GP response mouse strain

is dependent

on

This group has previously reported that CTL recognition of several LCMV strains varies in a host-dependent manner, and that the variability maps to the MHC locus (Ahmed et a/., 1984). Therefore we wished to see if, by switching between MHC haplotypes, we could identify changes in the pattern of CTL recognition of LCMV NP and GP. Table 1 shows that a change

324

WHITTON,

SOUTHERN,

TABLE 1 THE PATERN OF RECOGNITIONOF WGp AND WNp IS DEPENDENT ON MHC BACKGROUNDUSED

TARGET CELLS

r

SVNGENEIC

I UN

WI

‘WJq

I

lt

vvwt

lm

C 25:l

[ 504 2x1

:C$V

VVw,

0 1 16 0 0 12 II0 0

115

2

VVop

VVw

16

ND

II

1ID

11

I

ALL0 GENEIC

WI

LCMV ARM

15

Note. Effector cells were induced on three MHC backgrounds (H2bb, H2dd, and H2qq); on each background effecters were induced by infection with LCMV strain Armstrong (anti-LCMV ARM), wildtype vaccinia virus (anti W,,,), or by mock-infection (UN). Each effector population was assayed against both syngeneic and allogeneic target cells (indicated in the Table); target cells were uninfected (UN) or infected with LCMV strain Armstrong (LCMV ARM), wild-type vaccinia virus (WA, or a vaccinia recombinant expressing either LCMV-GP (VVep) or LCMV-NP (W,,). Cytotoxicity assays were as described under Materials and Methods. Percentage %r release considered significantly above background is boxed.

of MHC background does indeed result in a change in the pattern of recognition of LCMV proteins. On both the H2dd and the H2qq haplotypes, cells infected with the VVNprecombinant were efficiently lysed; this “internal” protein is, therefore, a target antigen on all MHC haplotypes we have tested, and its recognition by CTL presumably relies upon the surface exposure of an as yet unidentified epitope. In contrast WGp-infected targets, although productively infected (they are lysed by anti-VV CTL; Table l), exhibit little or no lysis when challenged with anti-LCMV CTL, suggesting

AND OLDSTONE

that CTL directed against GP residues l-363 are at best a minor component of the H2ddand H2qqsplenocyte populations. Anti-GP CTL comprise a significant the CTL response on H2bb

component

of

While CTL responses to virus glycoproteins have been previously described (Rosenthal et a/., 1987; Yewdell et a/., 1986; Finberg era/., 1982) results from several systems have indicated that they are a minor portion of the overall anti-viral CTL response, and in several instances studies similar in outline to those presented here have failed to detect any anti-GP CTL [e.g., RSV (Bangham et al., 1986) VSV (Puddington et a/., 1986)]. Our results on the H2bb background have shown that anti-GP lysis is quantitatively comparable to anti-NP lysis, and to confirm that the H2bb anti-GP CTLs comprise a significant proportion of the overall reactivity, we assessed the lysis of LCMV-infected and VVGp-infected cells at different effector:target (E:T) ratios (Fig. 2). The E:T ratio at which 50% Cr release is obtained is only threefold greater for VVGp-infected targets than for LCMV-infected targets, suggesting that approximately one-third of the CTL response is directed against LCMV GP. Cross-reactive

CTL recognize

LCMV GP

In other virus systems, any anti-GP response detected has almost always been strain-specific; that is, anti-GP CTL induced by one serotype of virus are unable to cross-react with serologically distinct strains. The great majority of cross-reactive CTL characterized have been directed against “internal” virus proteins (Kees and Krammer, 1984; Townsend and Skehel, 1984; Yewdell et a/., 1985). In order to establish whether LCMV GP could be the target for cross-reactive CTL, LCMV strain Pasteur (LCMV-PAST), a biolog-

ID0

6.25

12.5 E:T

25 50 RSNO

100

200

FIG. 2. Figure 2 shows the percentage specific 6’Cr release from three populations of H2bb targets, infected with Wsc,, , WGp, or LCMV strain Armstrong (LCMV ARM) and incubated at various E:T ratios as shown. Dotted line shows the E:T ratio required to achieve 50% lysis under these circumstances.

CTL RESPONSES TO LCMV INFECTION

ically and serotypically distinct strain of LCMV, was used to induce CTL on the H2bb background. These anti-PAST CTL (Table 2) are able to lyse syngeneic targets infected with the WGP recombinant, indicating that PAST also induces anti-GP CTL on the H2bb background, and that these cells are able to recognize ARM GP, i.e., are cross-reactive. We next tested the reactivity of anti-PAST CTL against VVGp on the H2dd and H2qq backgrounds (where anti-PAST CTL efficiently lyse ARM-infected cells; Ahmed et al., 1984), and we see no target cell lysis (Table 2) suggesting that these two LCMV strains are similar in the ability of their GP moieties to induce a CTL response on the H2bb haplotype alone. DISCUSSION With the advent of relatively straightforward methods by which isolated virus components can be expressed in vitro, it has become possible to study the interaction between the host cellular immune system and individual virus proteins. The data we present here, from studies of a primary CTL response to natural infection, bear upon several aspects of CTL recognition of viral antigens in association with class I MHC molecules. First, we have shown that both GP and NP moieties of LCMV can be efficiently used as target antigens for CTL recognition and lysis. Thus the GP, which contains the major antibody neutralization site of the virus (Parekh and Buchmeier, 1986) also contains at least one major CTL epitope on the H2bb background. During acute LCMV infection, both GP-1 and GP-2 moi-

TABLE 2 THE LMIC ACTIVITIES OF CTL INDUCED BY INFECTION WITH LCMV STRAIN PASTEUR(ANTI-LCMV PAST CTL) AGAINSTSYNGENEICTARGETS INFECTEDWITH LCMV STRAIN ARMSTRONG (LCMV ARM), WILD-TYPE VACCINIAVIRUS (W,), ORAVACCINIA RECOMBINANTEXPRESSINGTHE GP OF LCMV ARM (VV,,) UN

LCMV ARM

VVW1

H2bb-

VVGP

325

eties are expressed on the cell membrane and are detectable by fluorescent antibody analyses. As infection persists, however, expression of GP decreases dramatically both in vitro and in vivo (Welsh and Buchmeier, 1979; Oldstone and Buchmeier, 1982). It is not known if this decreased GP expression has any part to play in the establishment and/or maintenance of persistence, but the mapping to GP of sites critical to both humoral and cellular immune response may be of interest in this regard. Second, we find that the virus protein used as a source of CTL determinants varies, dependent on the mouse strain used. Previous work using congenic mouse strains differing only at the H2 locus has shown that patterns of CTL cross-reactivity between different LCMV strains is dependent on the class I restricting molecules (Ahmed et al., 1984) and the observations we present here confirm that CTL responses to a particular virus protein (in this case LCMV GP) vary greatly in mice with differing MHC backgrounds. These results complement findings recently reported in the influenza system (Pala and Askonas, 1986; McMichael et a/., 1986; Townsend et al., 1986a), and have important implications for vaccine development. We would suggest that vaccination with the VVGp recombinant may induce a CTL response only in H2bb mice, while HZdd and H2qq recipients would remain unstimulated, and therefore susceptible to LCMV challenge. We are currently investigating this possibility. Our results reinforce the concern that subunit vaccines may be effective in generating a CTL response only on certain MHC backgrounds; and moreover, suggest that to maximize the chance of a vaccine being universally effective in a population, a variety of antigenic determinants may have to be provided. Third, we show that a major part of the anti-LCMV response on the H2bb background is directed against the GP molecule. Analyses of the CTL responses to influenza virus, VSV, and RSV on a variety of MHC backgrounds have suggested an emerging pattern wherein the great majority of the response is directed against “internal” viral proteins; anti-glycoprotein responses have been said to be either undetectable (e.g., against RSV G protein) or a minor component. However, other studies have suggested that a response to virus glycoproteins can be detected depending on the strain of virus and/or host MHC used (Yewdell et al., 1986; Braciale et a/. (1986). It is clear from our studies that, in a manner defined by MHC, the LCMV glycoprotein can induce a very brisk CTL response. Furthermore, most reports from other virus systems agree that anti-glycoprotein CTL are almost always strain-specific; CTL which react across serotypic boundaries are invariably directed against the

326

WHIl-fON,

SOUTHERN,

“internal” virus components. In contrast, we demonstrate that a cross-reactive CTL response can be mounted to LCMV ARM GP, as CTL from a serologitally distinct PAST strain efficiently lyse targets expressing the ARM GP. Finally, our results address to some extent the nature of the structure recognized by the T cell receptor on CTL. Do the anti-GP CTL see native GP or degraded GP in the context of the MHC? The fact that the GP molecule encoded in VVGp is recognized despite its being truncated indicates that full-length GP-C is not required to allow the successful presentation of the CTL epitope(s) contained therein. WGp does, however, encode full-length GP-1, and one could argue that this protein is seen in intact form by the CTL. Using fluorescent antibody techniques, however, we could detect only very weak cell-surface expression of GP. Our findings, that neither intact GP-C nor efficient GP-1 surface expression is necessary for CTL recognition, suggest that CTL recognition of this virus glycoprotein may not require cell-surface expression of the intact molecule. Experiments are in progress to clarify this point. This interpretation is consistent with observations in the influenza system where truncated antigens, not expressed in quantity on the cell membrane, are successfully presented by class I MHC molecules (Townsend et a/., 1986a, b). CTL recognition of the LCMV NP is not unexpected in the light of the results recently reported for several other “internal” virus proteins (Bennink et a/., 1987; Yewdell et al., 1985, 1986; Bangham et a/., 1986). Cell-surface detection by antibody has been reported for the LCMV NP (Zeller et a/., 1986), but the relationship between regions detected by antibody and those seen by CTL is uncertain. Thus our studies underline the variability of viral CTL epitopes and their dependence on the precise nature of the restricting element. More precise mapping of epitopes of GP and NP using truncated cDNAs will establish whether there is any commonality in the viral epitopes selected on the different MHC backgrounds and will allow more detailed analyses of the exact requirements for induction of, and target cell recognition by, CTL. ACKNOWLEDGMENTS This is Publication Number 4668-IMM from the Department of Immunology, Research Institute of Scripps Clinic. We thank A. Tishon and J. R. Gebhard for excellent technical assistance, and are grateful to Mrs. G. Schilling for expert secretarial help. This work was supported in part by USPHS Grants Al-09484, NS-12428, and AG-04342.

REFERENCES R., BYRNE,J. A., and OLDSTONE, M. B. A. (1984) Virus specificity of cytotoxic T lymphocytes generated during acute lympho-

AHMED,

AND OLDSTONE cytic choriomeningitis virus infection role of the H-2 region in determining cross reactivity for different lymphocytic choriomeningitis virus strains. J. Viral. 51, 34-41. BANGHAM, C. R. M., OPENSHAW,P. J. M., BALL, L. A., KING, A. M. Q., WERTZ, G. W., and ASKONAS, B. A. (1986). Human and murine cytotoxic T cells specific to respiratory syncytial virus recognize the viral nucleoprotein N but not the major glycoprotein G expressed by vaccinia virus recombinants. J. Immunol. 137, 3973-3977. 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 nonstructural protein responsiveness to individual viral antigens is major histocompatibility complex controlled. J. Viral. 61, 1098-l 102. BRACIALE, T. J., HENKEL, T. J., LUKACHER, A., and BRACIALE, V. L. (1986). Fine specificity and antigenic receptor expression among influenza virus-specific cytotoxic T lymphocyte clones. J. Immunol. 137,995-1002. BUCHMEIER, M. I., and OLDSTONE, M. B. A. (1979). Protein structure of lymphocytic choriomeningitis virus evidence for a cell associated precursor of the virion glycopeptides. Virology 99, 111-120. BUCHMEIER,M. J., SOUTHERN, P. J., PAREKH, B. S., WOODDELL, M. K., and OLDSTONE, M. B. A. (1987). Site-specific antibodies define a cleavage site conserved among arenavirus GP-C glycoproteins. J. Viral. 61, 982-985. BUCHMEIER,M. J., WELSH, R. M., DUTKO, F. J., and OLDSTONE, M. B.A. (1980). The virology and immunobiology of lymphocytic choriomeningitis virus infection. In “Advances in Immunology” (F. 1. Dixon, and H. G. Kunkel, Eds.), Vol. 30, pp. 275-331. Academic Press, New York. BYRNE, J. A., and OLDSTONE, M. B. A. (1984). Biology of cloned cytotoxic T lymphocytes specific for lymphocytic choriomeningitis virus clearance of virus in ho. J. Viral. 51, 682-686. CHAKRABARTI,S., BRECHLING,K., and Moss, B. (1985). Vaccinia virus expression vector: Coexpression of &galactosidase provides visual screening of recombinant virus plaques. Mol. Cell. Biol. 5, 3403-3409. DUTKO, F. J., and OLDSTONE, M. B. A. (1983). Genomic and biological variation among commonly used lymphocytic choriomeningitis virus strains. J. Gen. Viral. 64, 1689-1698. FINBERG,R., SPRIGGS,D. R., and FIELDS, B. N. (1982). Host immune response to reovirus: Cytotoxic T lymphocytes recognize the major neutralization domain of the viral hemagglutinin. J. Immunol. 129,2235-2238. KEES, U., and KRAMMER, P. (1984). Most influenza A virus specific memory cytotoxic T lymphocytes react with antigenic epitopes associated with internal viral determinants. J. fxp. Med. 151, 365-377. LEHMANN-GRUBE, F. (1984). Portraits of viruses: Arenaviruses. Inrefvirology22, 121-145. MACKE~, M., SMITH, G. L., and Moss, B. (1984). General method for production and selection of infectious vaccinia virus recombinants expressing foreign genes. J. Iho/. 49, 857-864. MCMICHAEL, A. J., GOTCH, F. M., and ROTHBARD,J. (1986). HLA B37 determines an influenza A virus nucleoprotein epitope recognized by cytotoxic T lymphocytes. J. fxp. Med. 164, 1397-1406. OLDSTONE, M. B. A., and BUCHMEIER. M. J. (1982). Restricted expression of viral glycoprotein in cells of persistently infected mice. Nature (London) 300, 360-362. PALA, P., and ASKONAS, B. A. (1986). Low responder MHC alleles for Tc recognition of influenza nucleoprotein. Immunogenetics 23, 379-384. PAREKH,B. S., and BUCHMEIER,M. J. (1986). Proteins of lymphocytic

CTL RESPONSES TO LCMV INFECTION choriomeningitis virus antigenic topography of the viral glycoproteins. virology 153, 168-l 78. PUDDINGTON,L., BEVAN, M. J., ROSE, 1. K., and LEFRANCOIS,L. (1986). N protein is the predominant antigen recognized by vesicular stomatitis virus specific cytotoxic T Cells. J. Viral. 60, 708-717. RIVIERE,Y., SOUTHERN, P. J., AHMED, R., and OLDSTONE, M. B. A. (1986). Biology of cloned cytotoxic T lymphocytes specific for lymphocy-tic choriomeningitis virus. V. Recognition is restricted to gene products encoded by the viral S RNA segment. J. Immunol. 136, 304-307. ROSENTHAL,K. L., SMILEY, J. R., SOUTH, S., and JOHNSON,D. C. (1987). Cells expressing herpes simplex virus glycoprotein gC but not gB, gD, or gE are recognized by murine virus-specific cytotoxic T lymphocytes. 1. Viral. 61, 2438-2447. SOUTHERN, P. J., SINGH, M. K., RIVIERE,Y., JACOBY, D. R., BUCHMEIER, M. J., and OLDSTONE, M. B. A. (1987). Molecular characterization of the genomic S RNA segment from lymphocytic choriomeningitis virus. Virology 157, 145-l 55. TOWNSEND, A. R. M., BASTIN, J., GOULD, K., and BROWNLEE,G. G. (1986a). Cytotoxic T lymphocytes recognize influenza hemagglutinin that lacks a signal sequence. Nature (London) 234, 575. TOWNSEND,A. R. M., ROTHBARD,J., GOTCH, F., BAHADUR, G., WRAITH, D. C., and MCMICHAEL, A. J., (1986b). The epitopes of influenza nucleoprotein recognized by cytotoxic T lymphocytes can be defined with short synthetic peptides. Cell 44, 959-969.

327

TOWNSEND, A. R., and SKEHEL, 1. J. (1984). The influenza virus nucleoprotein gene controls the induction of both subtype specific and cross-reactive T cells. J. hp. Med. 160, 552-563. WELSH, R. M., and BUCHMEIER, M. J. (1979). Protein analysis of defective interfering lymphocytic choriomeningitis virus and persistently infected cells. virology 96, 503-515. YEWDELL,1. W., BENNINK, J. R., MACKE~, M., LEFRANCOIS,L., LYLES, D. S., and Moss, B. (1986). Recognition of cloned vesicular stomatitis virus internal and external gene products by cytotoxic T lymphocytes. J. fxp. Med. 163, 1529-l 538. 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. Nat/. Acad. Sci. USA 82, 1785-l 789. ZELLER,W., BRUNS, M., and LEHMANN-GRUBE, F. (1986). Viral nucleoprotein can be demonstrated on the surface of lymphocytic choriomeningitis virus infected cells. Med. Microbial. Immunol. 175, 89-92. ZINKERNAGEL, R. M., and DOHERTY, P. C. (1979). MHC restricted cytotoxic T cells: Studies on the biological role of polymorphic major transplantation antigens determining T-cell restriction specificity, function, and responsiveness. ln “Advances in Immunology” (F. 1. Dixon, and H. G. Kunkel), Vol. 27, pp. 52-180. Academic Press, New York.