Common Antiviral Cytotoxic T-Lymphocyte Epitope for Diverse Arenaviruses

Abstract

Members of the Arenaviridae family have been isolated from mammalian hosts in disparate geographic locations, leading to their grouping as Old World types (i.e., lymphocytic choriomeningitis virus [LCMV], Lassa fever virus [LFV], Mopeia virus, and Mobala virus) and New World types (i.e., Junin, Machupo, Tacaribe, and Sabia viruses) (C. J. Peters, M. J. Buchmeier, P. E. Rollin, and T. G. Ksiazek, p. 1521–1551, in B. N. Fields, D. M. Knipe, and P. M. Howley [ed.], Fields virology, 3rd ed., 1996; P. J. Southern, p. 1505–1519, in B. N. Fields, D. M. Knipe, and P. M. Howley [ed.], Fields virology, 3rd ed., 1996). Several types in both groups—LFV, Junin, Machupo, and Sabia viruses—cause severe and often lethal human diseases. By sequence comparison, we noted that eight Old World and New World arenaviruses share several amino acids with the nucleoprotein (NP) that consists of amino acids (aa) 118 to 126 (NP 118–126) (RPQASGVYM) of LCMV that comprise the immunodominant cytotoxic T-lymphocyte (CTL) epitope for H-2^d mice (32). This L^d-restricted epitope constituted >97% of the total bulk CTLs produced in the specific antiviral or clonal responses of H-2^d BALB mice. NP 118–126 of the Old World arenaviruses LFV, Mopeia virus, and LCMV and the New World arenavirus Sabia virus bound at high affinity to L^d. The primary H-2^d CTL anti-LCMV response as well as that of a CTL clone responsive to LCMV NP 118–126 recognized target cells coated with NP 118–126 peptides derived from LCMV, LFV, and Mopeia virus but not Sabia virus, indicating that a common functional NP epitope exists among Old World arenaviruses. Use of site-specific amino acid exchanges in the NP CTL epitope among these arenaviruses identified amino acids involved in major histocompatibility complex binding and CTL recognition.

The humoral (antibody) and cellular (T-lymphocyte) responses represent two distinct pathways by which the immune system recognizes and combats viral infections (reviewed in reference 30). Antibodies primarily recognize viruses or viral antigens circulating in the blood and other fluids, whereas T lymphocytes interact with viral antigens in the form of processed peptides bound to host cells and presented in a groove between the two α-helices of the major histocompatibility complex (MHC) glycoproteins (3). Cytotoxic T lymphocytes (CTLs), most of which bear the CD8 molecule, recognize and interact with infected cells that express viral antigen (peptide) presented by MHC class I molecules. Observations of humans and mice with genetic or acquired deficiencies in antibody or T-cell production and experimentally manipulated animals clearly show that control of most acute viral infections is critically dependent on MHC class I-restricted CD8 CTLs (reviewed in reference 30). However, immune clearance of persistent viral infections, in addition to dependence on CD8 CTLs, also requires the participation of CD4 T cells (2, 8, 12, 16, 29).

CTL responses to infection with the arenaviruses lymphocytic choriomeningitis virus (LCMV) and, likely, Lassa fever virus (LFV) essentially rely on cell-mediated CD8 CTL responses (reviewed in references 6, 29, 30, and 34). Features of interactions between arenavirus peptides and MHC class I molecules have been delineated molecularly and structurally by biochemical characterization of the naturally presented peptides (10, 11, 13, 19, 32, 33) and analysis of CTLs as viral escape variants (9, 14, 15, 22). Studies with these and other viruses whose peptides were isolated by elution from MHC class I molecules led to the identification of allele-specific anchor residues within the peptide sequence (reviewed in reference 23). Similarly, the impact of structural factors at nonanchoring residues in peptide-MHC interactions has been identified by using single amino acid mutations of viral gene products (9, 10).

The nucleoprotein (NP) sequence of LCMV consisting of amino acids (aa) 118 to 126 (NP 118–126) (RPQASGVYM) was found to be L^d restricted and to constitute >97% of the total clonal or total bulk primary CTL responses (32). A single injection of recombinant vaccinia virus (VV)-LCMV NP vaccine incorporating this L^d epitope provided complete protection for H-2^d mice later challenged intracerebrally with an ordinarily lethal dose of virus (18, 19). Further, study of this CTL epitope indicated that one-third of mice from nine established haplotypes (H-2^d, H-2^u, and H-2^q) possessed MHC class I glycoproteins capable of presenting the LCMV NP 118–126 peptide for recognition and lysis by virus-specific CTLs (19). This concept was further strengthened when incorporation of this virus-specific NP CTL epitope into a recombinant VV vaccine and administration of a single dose protected mice of these three haplotypes from an ordinarily lethal challenge of virus. This outcome indicated the presence of a common epitope among disparate MHC class I types and suggested the possibility of developing an effective CTL vaccine for outbred populations, such as humans (31). To further explore this possibility, we studied the NP CTL epitope (117/118)–126 for its representation among other arenaviruses. We noted that both the Old World (LVF, Mopeia) and New World (Machupo, Sabia, Junin, Pichinde, Oliveros, and Tacaribe) viruses had 22 to 78% homology with the LCMV NP 118–126 epitope. Sequence analysis of the NP (117/118)–126 among the arenaviruses combined with site-specific amino acid exchanges revealed the requirement for peptide binding to MHC class I L^d molecules and the requirement for subsequent cross-recognition of these MHC-restricted peptides by LCMV-specific L^d-restricted CD8⁺ CTLs. Of the nine arenaviruses studied, the NP 118–126 peptide from four, LCMV, LFV, Mopeia virus, and Sabia virus, bound at high affinity to L^d MHC class I molecules. Binding of the Sabia virus peptide NP 118–126 to L^d indicated that a Ser (Sabia virus aa 119) would substitute for the known position 2 (P2) MHC anchor residue Pro (aa 119: LCMV, Mopeia virus, LFV) and that Ala in P3 might serve as an ancillary MHC anchor. The three Old World arenaviruses, LCMV, LFV, and Mopeia virus, but not the New World virus Sabia virus, showed functional CTL cross-reactivity when their peptides were used to coat MHC-matched target cells. Single-amino-acid exchanges between LCMV and Sabia virus indicated that an amino acid change of Gly at P5 (aa 122) for a Ser or Ala aborted cross-recognition by anti-LCMV CD8⁺ CTLs. Hence, a recombinant, plasmid or subunit viral vaccine including a common NP CTL epitope could offer immune protection from diseases caused by Old World but not New World arenaviruses.

MATERIALS AND METHODS

Mouse strains, cell lines, and virus stocks.

BALB/WEHI or BALB/cByJ (H-2^d) and C57BL/6 (H-2^b) mice were obtained from the breeding colony at The Scripps Research Institute and used when 6 to 10 weeks of age. BALB/cl 7 (H-2^d), MC57 (H-2^b), and human T2 cells transfected with H-2 L^d (T2-L^d) were used in CTL and binding experiments. Cells were grown in RPMI 1640 (BALB/cl 7, MC57) or Iscove's modified Dulbecco's medium (T2-L^d) containing 8% bovine serum and l-glutamine as described previously (10, 11, 13–15, 20, 32, 33). Gentamicin (400 μg/ml) was added to Iscove's modified Dulbecco's medium to maintain selection of T2-L^d cells. The LCMV Armstrong clone 53b (LCMV ARM) was used at an MOI of 1 to infect cells. The origin, cloning, sequencing, plaque purification, and quantification of LCMV on Vero cells, expression of viral NP or glycoprotein (GP), and their identification with NP- or GP-specific monoclonal antibodies have been described previously (10, 14, 15, 32). Details describing the construction and use of recombinant VV expressing full-length LCMV ARM GP, NP, or NP epitope aa 118 to 127 have been published (14, 15, 31, 32).

Peptides.

Peptides were synthesized by the solid-phase method using the N-9-fluorenyl(methylcarbonyl) (Fmoc) chemistry, purified by high-pressure liquid chromatography on a RP300-C8 reverse-phase column (Brownlee lab), and identified by fast atom bombardment or electrospray mass spectrometry (10, 20).

MHC binding studies.

An MHC stabilization assay was performed. RMAS cells (D^b K^b) transfected with L^d molecules and P815 (H-2^d) were used to measure L^d stabilization in the presence of increasing peptide concentrations, using the monoclonal antibody HB27, 28-14-8S (anti-D^b, anti-L^d), Y3 (anti-K^b) or SF1-1.1.1 (anti-K^d) staining followed by flow cytometry analysis, as previously described (10, 20; M. B. A. Oldstone, J. E. Gairin, H. Mazarguil, F. Laval, V. C. Asensio, I. L. Campbell, S. J. DeArmond, B. Coon, and D. Homann, unpublished data.). The peptides were considered MHC binders when displaying a peptide concentration giving half of the maximal stabilization effect (SC₅₀) affinity values of 50 μM or lower.

Computer analysis.

Molecular modeling of the interaction between H-2 L^d and viral peptides was performed using Insite (Biosym Technology, San Diego, Calif.). The calculated binding motif for L^d and crystallographic studies on this molecule are published (1, 26).

Generation and detection of CTLs.

Primary LCMV-specific CTLs were generated by priming mice intraperitoneally with 10⁵ PFU of LCMV ARM. Spleens from such mice were removed 7 days after inoculation, and single-cell suspensions of spleens (free from erythrocytes) were prepared in complete RPMI 1640 media as described previously (14, 15, 29, 31, 32). These cells were then tested for their ability to lyse virus-infected or peptide-coated targets as reported previously (14, 15, 24, 29, 31, 32). The L^d-restricted CTL clone HD-8 has been described, and its specificities have been recorded (32).

A standard 5- to 6-h ⁵¹Cr-release assay (14, 15, 24, 31, 33) was used to measure CTL activity. Target cells were infected with LCMV ARM at an MOI of 1, 48 h before the assay, or with recombinant VV at an MOI of 3, 16 h before the assay. Peptides at various concentrations were used to coat uninfected target cells. The percent specific lysis was calculated as 100[(CPM release by CTL − CPM of spontaneous release)/(CPM of total release − CPM of spontaneous release)].

Intracellular cytokine assays were done on BALB spleen cells obtained 7 days after infection with LCMV ARM. Lymphocytes were harvested as described above, restimulated for 5 h in the presence of the indicated arenavirus peptides (0.1 μg/ml) plus recombinant human interleukin 2 (50 U/ml) and brefeldin A (1 μg/ml), and subsequently stained for CD8, gamma interferon (IFN-γ), and tumor necrosis factor alpha (TNF-α) as described previously (2, 24).

RESULTS AND DISCUSSION

The L^d MHC CD8⁺ CTL-restricted peptide sequence was remarkably similar among diverse arenaviruses.

The L^d MHC molecule bound viral peptides via an MHC anchor of proline (P) at P2 and a methionine (M), phenylalanine (F), leucine (L), or isoleucine (I) at P9 or P10 according to elution, alignment, and immunochemical studies (23). Others (9) employing limited single-amino-acid substitution analysis reported that replacing the authentic P at P2 with an alanine (A) or arginine (R), S at P5 with asparagine (N), or M at P9 with a lysine (K) or L diminished or abolished the capacity of NP peptide to increase cell surface L^d expression and to induce L^d stabilization in L cells (H-2^k) manipulated to express L^d molecules. In addition, crystallographic analysis of the L^d MHC molecule suggested that P3 may also act as an auxiliary MHC anchor with P4 and P8 forming contacts for the CD8⁺-T-cell receptor (1, 26). With the L^d-restricted LCMV ARM immunodominant CTL epitope RPQASGVYM (32) used as a guide, the corresponding NP sequences of several Old World and New World arenaviruses were obtained. As shown in Table 1, NP 118–126 of the Old World LFVs Nigeria (G) and Josiah (J) as well as Mopeia virus shared from 6 to 7 aa with the LCMV ARM peptide. These viruses bore homology to the MHC binding anchor of a P at P2 and an M at P9 but not the suggested P3 auxiliary MHC anchor LCMV ARM (Q), LFV (G and J), and Mopeia (L) residues. As anticipated, L^d-expressing cells bound tightly at concentrations of 4 μM for Mopeia virus, 20 μM for LFV (G), and 5 μM for LFV (J) (Table 1; Fig. 1). In contrast, the corresponding aligned NP sequence from the New World arenaviruses had less homology to the LCMV prototype, i.e., 2 to 4 aa were shared with the 9 aa residues of LCMV ARM. Further, as shown in Table 1, none of the New World arenaviruses, Junin, Machupo, Sabia, Pichinde, Oliveros, or Tacaribe, had a P in P2 or a Q or L in P3, although Junin, Machupo, Sabia, and Tacaribe viruses had the correct amino acids to serve as an MHC anchor in P9. Therefore, it was surprising that Sabia bound well to the L^d molecules (30 μM [Table 1 and Fig. 1], and in a separate experiment, 10 μM [Table 2]). Sequence comparison showed that Machupo virus, a nonbinder (affinity of good binders is less than 50 μM), and Sabia virus, a good binder, shared serine (S) in P2 but differed with respect to amino acids in P3; Sabia virus had an alanine (A), and Machupo virus had a glycine (G). Further, Sabia virus, like the Old World viruses, had an arginine (R) in P1, whereas Machupo virus had a glutamic acid (E) in that position. To evaluate the role of P at P2 and E at P1 in binding to L^d, we substituted E for R in LCMV ARM at P1 and R for E in Machupo virus at the same amino acid residue. Table 2 shows that these substitutions did not change the high binding efficiency of wild-type (wt) LCMV ARM NP 118–126 (parental sequence) or mutant LCMV ARM NP 118–126 (E 118) to L^d or reverse the low or absent binding of wt Machupo NP 117–125 or mutant Machupo NP 117–125 (R 117) to L^d. From these results, we conclude that amino acid E in P1 does not interfere with binding to the L^d molecule if P is at P2. The binding of Sabia to L^d suggests that an S in P2 with the appropriate flanking amino acid could serve as an ancillary MHC anchor for P2.

TABLE 1.

LCMV NP 118–126 sequence homology among arenaviruses and their binding affinities to H-2L^d arenavirus peptides^a

Virus strain	Peptide identification	Peptide sequenceb	Homology (%)c	Ld binding affinityd (SC50 [μM])
LCMV ARM	NP 118–126	R P Q A S G V Y M	9/9 (100)	0.5 ± 0.4
Mopeia	NP 118–126	R P L A A G V Y M	7/9 (78)	4 ± 1
Lassa (G)	NP 118–126	R P L S S G V Y M	7/9 (78)	20 ± 8
Lassa (J)	NP 118–126	R P L S A G V Y M	6/9 (67)	5 ± 3
Junin	Ncap 117–125	E T G S Q G V Y M	4/9 (44)	>100
Machupo	Ncap 117–125	E S G P Q G L Y M	3/9 (33)	>100
Sabia	Ncap 118–126	R S A S G G Y Y L	3/9 (33)	30 ± 20
Pichinde (E124)	Ncap 116–124	S L S Q P G V Y E	3/9 (33)	No binding
Pichinde (G124)	Ncap 116–124	S L S Q P G V Y G	3/9 (33)	No binding
Oliveros	Ncap 117–125	E R S T P G V Y Q	3/9 (33)	No binding
Tacaribe	Ncap 117–125	E S N G T N A Y M	2/9 (22)	No binding
Influenza	PB1 591–599 (Ld-irrelevant peptide)	V S D G G P N L Y	0/9 (0)	No binding

^aBinding affinity was determined by measuring peptide-induced stabilization of H-2 L^d molecules at the surfaces of P815 cells. 

^bBoldface type indicates homology. 

^cNumber of homologous amino acids/number in sequence. 

^d“No binding” indicates that no stabilization effect was observed even at the highest peptide concentration tested (10⁻⁴ M). 

FIG. 1.

Data for the MHC stabilization assay using target cells expressing L^d molecules. Tenfold dilutions of various arenavirus peptides, monoclonal antibody to L^d, and fluorescence-activated cell sorting were used. LCMV, LFV, Mopeia virus, and Sabia virus NP peptides aa 117/118 to 126 bind to L^d. Data were confirmed in two repeated studies. See Materials and Methods and Table 1 for details.

TABLE 2.

Association of P2 and P3 MHC anchor residues for binding of arenaviruses to L^d class I MHC molecule

Virus strain	Peptide identification	Peptide sequencea	H-2 Ld binding affinity (μM)b
LCMV ARM	NP 118–126 (wt)	R P Q A S G V Y M	<1
LCMV ARM	NP 118–126 (E118)	E P Q A S G V Y M	2
Machupo	NP 117–125 (wt)	E S G P Q G L Y M	>100
Machupo	NP 117–125 (R117)	R S G P Q G L Y M	>100
Sabia	NP 118–126 (wt)	R S A S G G Y Y L	10

^aBoldface type indicates amino acid under study. 

^bBinding affinity was determined by measuring peptide-induced stabilization of H-2 L^d molecules at the surface of P815 cells. 

L^d-restricted LCMV ARM CD8⁺ CTLs lyse target cells coated with related NP peptides from diverse Arenaviridae.

For the next series of experiments, we inoculated BALB H-2^d mice with LCMV ARM to generate primary day 7 anti-LCMV ARM CTLs. The lytic capacity of these CTLs was then tested against ⁵¹Cr-labeled target cells coated with various concentrations of related NP peptides from Old World and New World arenaviruses. As expected (Table 3), LCMV ARM day 7 CTLs in a virus-specific MHC-restricted manner killed target cells expressing the relevant Old World peptides from Mopeia, LFV (G), and LFV (J) viruses. For confirmation, a well-characterized L^d-restricted CTL clone (HD-8) specific for LCMV ARM NP 118–126 (32) was used at effector-to-target ratios of 5:1 and 1:1 (5:1 shown in Table 3). Since these CTLs also lysed L^d targets coated with peptides from Old World arenaviruses, CTLs generated specifically to LCMV could cross-react and kill targets of Old World Arenaviridae. In contrast to these findings, Sabia virus peptide sequence NP 118–126, which bound robustly to L^d, was not recognized by LCMV ARM CTLs (Table 3), indicating a lack of cross-reactivity by LCMV to the CTL peptide epitope of the New World Sabia virus.

TABLE 3.

Lysis of H-2^d targets coated with arenaviral NP peptides by BALB day 7 primary LCMV ARM bulk CTLs or by L^d restricted CTL clone HD-8^a

Virus strain	Peptide identification	Peptide sequenceb	Specific 51Cr release by:
			H-2d (BALB/cl 7) targets coated with:						H-2b (MC57) targets (10−5 M)	H-2d targets infected with:
			d7 P° bulk CTLs			Ld CTL LCMV Cl HD-8				LCMV ARM	VV-ARM NP	VV-ARM GP
			10−5 M	10−7 M	10−9 M	10−5 M	10−7 M	10−9 M
LCMV ARM	NP 118–126	R P Q A S G V Y M	49 ± 3	52 ± 4	0	55 ± 6	48 ± 4	6 ± 2	0	56 ± 3	34 ± 4	2 ± 1
Mopeia	NP 118–126	R P L A A G V Y M	26 ± 5	28 ± 4	0	72 ± 6	52 ± 4	23 ± 3	ND	ND	ND	ND
Lassa (G)	NP 118–126	R P L S S G V Y M	28 ± 4	21 ± 5	3 ± 2	38 ± 6	29 ± 3	0	ND	ND	ND	ND
Lassa (J)	NP 118–126	R P L S A G V Y M	22 ± 3	14 ± 2	0	41 ± 4	21 ± 3	0	ND	ND	ND	ND
Junin	NP 117–125	E T G S Q G V Y M	0	0	0	0	0	0	ND	ND	ND	ND
Machupo	NP 117–125	E S G P Q G L Y M	0	0	0	2 ± 1	0	NDc	ND	ND	ND	ND
Sabia	NP 118–126	R S A S G G Y Y L	0	0	0	0	0	0	ND	ND	ND	ND
Pichinde	NP 116–124	S L S Q P G V Y G	0	0	0	4 ± 2	0	ND	ND	ND	ND	ND
Oliveros	NP 117–125	E R S T P G V Y Q	0	0	0	2 ± 1	0	ND	ND	ND	ND	ND
Tacaribe	NP 117–125	E S N G T N A Y M	0	0	0	5 ± 2	0	ND	ND	ND	ND	ND
LCMV ARM	NP 396–405	F Q P Q N G Q F I	2 ± 2	0	0	6 ± 4	0	ND	32 ± 3d	2 ± 1d	1 ± 1d	0d

^aSee Materials and Methods for details. d7, day 7. 

^bBoldface type indicates homology. 

^cND, not determined. 

^dLysis caused by addition of day 7 P° bulk C57BL/6 (H-2^b) CTLs. 

In addition to lysis of infected or peptide-coated target cells, CTLs exhibit effector activity by release of cytokines, such as IFN-γ or TNF-α (reviewed in references 7 and 30). To assay the efficiency of arenavirus peptides in stimulating cytokine production by LCMV ARM-specific CD8⁺ T cells, IFN-γ- or TNF-α-producing cells were quantitated. As Fig. 2 demonstrates, of the viruses tested, only LCMV ARM, Mopeia virus, and LFV (only data for LFV [J] shown) peptides specifically stimulated LCMV-specific CD8⁺ T cells to make the cytokines, but the Sabia virus peptide was unable to. In agreement with the extent of lysis by HD-8 CTLs, CTL accumulation of IFN-γ or TNF-α was enhanced with the Mopeia peptide over that with the corresponding LFV peptide. Sequence comparison suggested that the G in P5 in Sabia virus, compared to S or A at that position among Old World arenaviruses, or the change from V in P7 in Old World arenaviruses to Y in Sabia virus, prevented LCMV ARM CTL recognition. To test for these various possibilities, single-amino-acid mutations were made at positions P4 and P7 and at some other regions in LCMV ARM- and Sabia L^d-restricted peptides. As shown in Table 4, while change of V to Y or vice versa at P7 failed to alter CTL lysis, the change at P5 from S to G or vice versa did allow LCMV ARM CTLs to lyse L^d targets coated with Sabia virus mutant peptide. From these data we conclude that the amino acid S in P5 is a T-cell-receptor contact residue that tolerates an A mutation (see Mopeia virus and LFV sequences, Table 3), while a G mutation prevents CTL-mediated lysis. We then made a mutation in P3 of Sabia changing the A to G, the residue in the non-L^d-binding Machupo virus peptide. The Sabia virus peptide (G in P3 and S in P5) was about half as efficient in preparing the target cell for lysis by LCMV-specific CTLs as Sabia virus peptides with A or Q in P3 and S in P5. These data suggest that ancillary MHC binding occurs via A in P3 of Sabia virus.

FIG. 2.

Intracellular cytokine assay done on lymphocytes obtained from BALB mice 7 days after infection with LCMV. Lymphocytes were incubated with various arenavirus NP peptides (see Table 1), recombinant interleukin-2, and brefeldin A and stained for both surface CD8 molecules and intracellular IFN-γ and TNF-α by FACS. Only LFV and Mopeia virus NP peptides consisting of aa 118 to 126 cross-react at the CD8⁺-T-cell level with LCMV CD8⁺ CTLs. Data displayed are the mean values of triplicate samples, and each bar represents 1 SD. Data are representative of two assays. See Materials and Methods and references 2 and 24 for details.

TABLE 4.

Substitution of S for G in NP aa 122 of Sabia virus allows its recognition by L^d-restricted LCMV-specific day 7 primary CTLs and HD-8 CTL clone

Virus and amino acid substitutions	Peptide sequencea	Specific 51Cr release byb
		H-2d targets coated with:						H-2b target, H-2d CTLs, LCMV ARM
		d7 P° H-2d CTLs			Ld CTL clone HD-8
		10−5 M	10−7 M	10−9 M	10−5 M	10−7 M	10−9 M
LCMV ARM NP 118–126
wt	R P Q A S G V Y M	41	28	6	56	22	0	0
120A, 121S, 122G, 124Y	R P A S G G Y Y L	0	2	4	3	6	1	4
120A, 121S	R P A S S G V Y L	29	15	3	46	38	9	3
122G	R P Q A G G V Y L	33	30	9	62	65	34
120A	R P A A S G V Y L	22	17	17	13	12	6	5
124Y	R P Q A S G Y Y L	43	33	16	32	24	17	4
122G, 124Y	R P Q A G G Y Y L	5	2	1	6	3	1	0
Sabia NP 118–126
wt	R S A S G G Y Y L	6	0	0	0	0	0	3
122S	R S A S S G Y Y L	35	21	3	35	17	0	0
124V	R S A S G G V Y L	4	3	3	4	0	0	8
120Q, 121A, 122S, 124V	R S Q A S G V Y L	23	28	6	52	36	0	0
120G, 122S	R S G S S G Y Y L	19	11	4	24	16	0	0
Controls
H-2d target cells infected with LCMV ARM		(50:1)c 60			(5:1) 54			(50:1)  5
H-2b target cells infected with LCMV ARM		(50:1)  3			(5:1)  0			(50:1) 46

^aBoldface type shows amino acid substitutions. 

^bResults from a 5-h ⁵¹Cr release assay. Log dilutions of various peptide-coated, ⁵¹Cr-labeled uninfected target cells were made. Spleen CTLs harvested 7 days after LCMV infection (10⁵ PFU intraperitoneally) were added at an effector-to-target ratio of 50:1 or 25:1, with data shown for 50:1. HD-8 CTL clone, L^d restricted and specific for LCMV NP 118–126, was added at a 5:1 effector-to-target ratio. 

^cEffector-to-target ratios are shown in parentheses. 

In conclusion, CTLs generated in H-2^d BALB mice against LCMV ARM cross-react with NP 118–126 sequences of other Old World arenaviruses. These cross-reactive CTLs efficiently lyse both Mopeia virus and LFV peptide-coated target cells presented by L^d MHC molecules and express the antiviral effector cytokines IFN-γ and TNF-α. In contrast, no CTL cross-reactivity was evident between Old World and New World arenaviruses with respect to either lytic or cytokine activity. Cumulatively, our results suggest the possibility of a single vaccine for Old World arenaviruses and argue against attempting to use one vaccine to protect against both Old World and New World arenaviruses.

The initial classification of Arenaviridae into Old World and New World viruses on the basis of complement-fixing antibody differences before the age of molecular biology appears more justified as sequence information emerges. Further, the observation of cross-CTL recognition among the Old World arenaviruses studied here, but not with the New World arenaviruses, supports this classification. Our data stress two points. First, the CTL commonality shown here among Old World arenaviruses may signify that the sharing of their epitopes among several human HLA motifs would permit the design of a universal vaccine for Old World arenaviruses. Studies to test this possibility are currently under way utilizing sequences for various motif patterns and humanized transgenic mice (28). Second, the arenaviruses used here may have originated from geographically separate sources and could reflect differences between their natural rodent hosts on each continent. Alpha-dystroglycan has recently been identified as the receptor for LCMV and LFV (4). Interestingly, recent studies of binding of arenaviruses to this receptor (S. Kunz, C. Spiropoulou, and M. B. A. Oldstone, unpublished data) indicate that the Old World arenaviruses tested (LFV, Mopeia virus, Mobala virus, and LCMV) all bind with high affinity to this receptor, while in contrast the majority of New World arenaviruses bind with an affinity that is 2 to 3 logs lower. The exception is the clan containing the Oliveros and Latino New World arenaviruses, which bind at high affinity to alpha-dystroglycan.

Studies on vaccination to control LFV infection in rodents have shown that GP and NP vaccines are both effective (5, 17). However, in subhuman primates, both components of LFV GP (GP1 plus GP2) were required for a protective vaccine; neither LFV NP, GP1, nor GP2 alone was effective (6). However, new studies in which vaccination successfully protected nonhuman primates from Ebola virus (27) stressed the need of a uniquely designed vaccine protocol to achieve optimal results. Until such approaches are undertaken for LFV and until data concerning LFV-specific T cells becomes available from human or animal models, the roles for various components of the immune response against the viral GP and/or NP remains uncertain.