The Mouse H-2A Region Influences the Envelope Gene Structure of Tumor-Associated Murine Leukemia Viruses

Abstract

C57BL/10 (B10) strains congenic at the mouse major histocompatibility locus (H-2) were injected with a modified ecotropic SL3-3 murine leukemia virus (MuLV) to determine the effect of the H-2 genes on the envelope gene structure of recombinant MuLVs. All tested strains rapidly developed T-cell lymphomas, and recombinant proviruses were detected in the tumor DNAs by Southern blot. The B10.D2 (H-2^d), B10.Br (H-2^k), B10.Q (H-2^q), and B10.RIII (H-2^r) strains exhibited a TI phenotype in which almost all tumors contained type I recombinants. These recombinants characteristically acquire envelope gene sequences from the endogenous polytropic viruses but retain the 5′ p15E (TM) gene sequences from the ecotropic virus. The parental B10 (H-2^b) strain, however, had a novel phenotype that was designated NS for nonselective. Only 30% of the B10 tumors had detectable type I recombinants, whereas a proportion of the others appeared to contain type II recombinants that lacked the type I-specific ecotropic p15E gene sequences. Studies of other B10 congenic strains with hybrid H-2 loci and selected F₁ animals revealed that the NS phenotype was regulated by a dominant gene(s) that mapped to the A region of H-2^b. These results demonstrate that a host gene within the major histocompatibility complex can influence the genetic evolution of pathogenic retroviruses in vivo.

The consequences of retrovirus infections in mice, cats, sheep, and humans may depend on the genetic evolution of pathogenic retroviruses in vivo (4, 29). Spontaneous mutations and recombination between different viral genomes create genetic diversity in the virus population. This allows for the selection of those variants that replicate more efficiently and are more pathogenic than the original dominant species in the viral population. In the inbred mouse strains AKR, HRS, C58, and CWD, the generation and selection of pathogenic recombinant murine leukemia viruses (MuLVs) is an important step in the development of spontaneous lymphomas (3, 5, 13, 15, 17, 23, 32, 33, 35). Animals of these strains express early in life endogenous ecotropic MuLVs that subsequently acquire pathogenic sequences by recombination with endogenous polytropic and xenotropic viruses. Similarly, the injection of exogenous ecotropic MuLVs, such as SL3-3, into these or other susceptible strains induces leukemias and the formation of envelope gene recombinants (6, 10, 31). The recombinants typically incorporate 5′ envelope gene sequences from one or more of the 20 or so endogenous polytropic viruses. This portion of the gene encodes the receptor binding domain of the major envelope protein, SU or gp70, and thus confers a polytropic host range (2, 3, 5, 13, 15, 17, 23, 27, 32, 33, 35). The polytropic domain within gp70 appears to promote leukemogenesis by enhancing virus replication in lymphoid target cells and through the activation of growth factor signaling pathways (3, 22, 39).

The MuLV envelope gene (env) encodes a common precursor protein, gp85, that is cleaved to yield two products, the mature gp70 and the minor envelope protein, TM or p15E (12, 43). The 5′ portion of the gp70 gene of the recombinant viruses is composed of sequences inherited from the endogenous polytropic viruses, but the 5′ portion of the p15E gene may be derived from either the ecotropic or polytropic virus parent (6, 23, 32, 35, 37). Recombinants in which the 5′ p15E gene contains ecotropic virus sequences are classified as type I recombinants, whereas those that acquire the allelic polytropic virus sequences in this region are designated type II recombinants (Fig. 1; references 6 and 23). We had previously shown that the presence of type I or type II recombinants within lymphoma cells is influenced by a host gene located on mouse chromosome 17. In these experiments, the injection of the ecotropic SL3-3 MuLV induced tumors and type I recombinants in HRS/J mice, whereas tumors from injected CWD mice contained type II recombinants (6, 32, 37). The type I-forming phenotype (TI) of HRS/J mice was dominant with respect to the type II-forming phenotype (TII) of CWD mice and segregated with restriction fragment length polymorphisms on chromosome 17 which are located within the H-2 locus, the mouse major histocompatibility complex.

FIG. 1.

Predicted amino acid sequences of p15E proteins of type I and type II recombinant MuLVs in relationship to known or suspected functional domains. Underlined residues may participate in the formation of a leucine zipper-like region.

The earlier genetic studies of the TI gene of HRS/J and CBA/J mice and the observation that all H-2^k strains exhibited the TI phenotype suggested that the TI gene(s) may be located within or near the H-2 locus (6, 23, 32, 37). To test if the H-2 genes regulate the TI or TII phenotype, we injected a modified ecotropic SL3-3 MuLV into C57BL/10 (B10) strains that were congenic at H-2. As reported here, we discovered that all strains tested except the parental B10 were TI strains. The latter strain exhibited a new phenotype, referred to as NS for nonselective, in which one-half or less of the tumors contained detectable type I recombinant proviruses. The B10 NS phenotype was distinct from and dominant with respect to TI and TII phenotypes of other strains and was regulated by a gene(s) that is located in or near H-2A^b.

MATERIALS AND METHODS

Mice.

CWD breeding stock (cw/+ and cw/cw) was obtained from the Jackson Laboratory, Bar Harbor, Maine, and was maintained at the University of Virginia vivarium (cw/+ to cw/cw). A breeding colony of CWD mice has been continuously maintained by brother-sister matings for 8 years. H-2 congenic and intra-H-2 recombinant congenic mice were obtained from the Jackson Laboratory.

Leukemogenesis.

To induce leukemia, 0.05 to 0.1 ml of reverse transcriptase-positive supernatants from SL3-3NB-infected NIH 3T3 fibroblasts was injected into the peritoneal cavity of animals less than 48 h old (6). The animals were observed regularly for signs of disease and sacrificed by metafane inhalation when moribund. Animals that died in their cages were refrigerated until necropsy. Spleen, thymus, and any other affected tissues were cut into aliquots and immediately stored in liquid nitrogen for use in DNA extraction at a later date.

DNA isolation.

DNA was extracted from tumor tissue aliquots that had been stored in liquid nitrogen by one of two methods. The organic extraction method has been described previously (6). The second was a nonorganic extraction method in which tissue aliquots were added to 5 ml of STE buffer (150 mM NaCl, Tris-Cl [pH 7.4], 20 mM EDTA) and homogenized in 15-ml glass grinders. Homogenate was transferred to fresh tubes, and 5 ml of STE buffer and 1 ml of 10% sodium dodecyl sulfate were added. Homogenate was digested with proteinase K (200 μg/ml) at 50°C for 4 to 5 h. Protein was precipitated by the addition of 3 ml of saturated NaCl solution and vigorous agitation. Following centrifugation, supernatant was transferred to a fresh 50-ml tube, and 2 volumes of room temperature 100% ethanol was added. The DNA precipitate was pelleted and washed with room temperature 70% ethanol. DNA was then lyophilized and resuspended in Tris-EDTA (10:1) at a concentration of between 0.150 and 2.0 mg/ml.

Southern blotting, hybridization probes, and labeling.

Five micrograms of DNA was digested with the appropriate restriction enzyme(s) and blotted by using techniques described previously (6). The pAKV5 probe, a subcloned fragment of the 5′ portion of the AKV ecotropic p15E TM gene, was the gift of Winship Herr (18). The TCR_beta probe hybridizes to both C1 and C2 regions of the beta chain of the T-cell receptor and was a gift of Tak Mak. The probes for the immunoglobulin heavy-chain joining region (J_H) and the light-chain joining region (J_kappa) were gifts of Roger Perlmutter.

All probes were excised from the plasmids by digestion with the appropriate restriction enzymes and purified from low-melting-point agarose gels after electrophoresis. The fragments were labeled with ³²P by the random primer-extension method, using a kit from Boehringer Mannheim (Indianapolis, Ind.).

Amplification of viral sequences by PCR.

PCR was used to selectively amplify envelope and long terminal repeat (LTR) sequences of recombinant viruses. A master mix of buffer, MgCl₂, and Amplitaq (Perkin-Elmer Cetus, Norwalk, Conn.) was added to 500 ng of each DNA sample. Tubes were heated to 84°C in a DNA Thermal Cycler (Perkin-Elmer Cetus), at which point a second mix containing primers and nucleotides was added. The reaction mix was then topped off with a single drop of mineral oil. A 2-min incubation at 94°C was followed by 30 cycles of 1 min at 94°C, 1 min at 49°C, and 30 s at 72°C. Cycles were followed by a single-step incubation at 72°C for 2 min that was immediately followed by 4°C incubation until samples could be analyzed. The presence of PCR product was confirmed by running 2.5 μl of reaction mix on a 1.5% agarose gel in using Tris-agarose-EDTA buffer.

The final volume for all PCRs was 25 μl. With the exception of primers, all reagents were provided by Perkin-Elmer Cetus. Final concentrations of reagents were 200 mM each of the four deoxynucleoside triphosphates, 1× PCR buffer II, 2.5 mM MgCl₂, 0.02 U of Amplitaq per ml, 20 ng of the experimental DNA template per ml, and 0.5 mM each primer. Primers were obtained from the University of Virginia sequencing facility. The 5′ primer was GCGAATTCTCTATAGTCCCTGAGACTG, which is specific for polytropic gp70 sequences, and the 3′ primer was TTCCCGGGTCTCTTGAAACTGTTGTTG, which is specific for ecotropic sequences in the LTR.

Cloning and DNA analysis.

PCR products were cloned by using the TA cloning system (Invitrogen, San Diego, Calif.) according to the manufacturer’s directions. In brief, PCR mixtures were diluted 1:100, and 1 μl was added to 6 μl of sterile water, 1 μl of 10× ligation buffer, 2 μl of pCR vector (25 ng/μl), and 1 μl of T4 DNA ligase and incubated at 9°C overnight. One microliter of this TA ligation reaction was transformed into 50 μl of competent Escherichia coli INVaF cells to which was added 450 μl of SOC medium for a total volume of 500 μl; 25 μl was plated on LB agar plates containing kanamycin (50 mg/ml) and 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal; 50 ng/ml). White colonies were screened by using selective restriction enzyme digests and sequenced by using a dideoxy-based Sequenase kit from U.S. Biochemical (Cleveland, Ohio). The primers used for PCR were also used for DNA sequence analysis. Internal primers which anneal to both polytropic and ecotropic p15E sequences, CCGCCCATAGTAAGTCCTCC and CATTCTTCTTTTAGGGCAGCAC, were also obtained from the University of Virginia sequencing facility. [³⁵S]dATP-labeled reaction products were run on 8% urea-acrylamide gels with TBA (40 mM Tris, 20 mM boric acid, 2 mM EDTA) buffer.

Statistics.

The statistical significance of differences in disease latencies and lymphoma incidence were determined by Student’s t test. The assignment of phenotypes of the B10 strains was confirmed statistically by the Wilcoxon rank sum test, a nonparametric substitution for the t test. Analyses were performed with Medlog version 92.4a (Information Analysis Corporation) as previously described (23). The Wilcoxon rank sum test showed that the ratio of type I provirus-containing lymphomas to total lymphomas for the B10 strain (NS) was statistically different from this ratio for the congenic strains which expressed the TI phenotype. P values were 0.001 for comparison of B10.Br to B10, 0.001 for comparison of B10.D2 to B10, and 0.003 for comparison of B10.Q to B10.

RESULTS

The SL3-3NB virus induces T-cell lymphomas in B10 and B10 H-2 congenic mice.

To determine the recombinant virus-forming phenotype of B10 mice and four related H-2 congenic strains, neonatal animals were injected with the SL3-3NB ecotropic MuLV. SL3-3NB is derived from a modified AKR SL3-3 MuLV provirus in which 200 bp of the gag gene is derived from the Moloney MuLV (36). This confers NB tropism and negates the suppressive effects of the B10 Fv-1^b allele on viral replication and leukemogenecity (20, 36). SL3-3NB rapidly induced lymphoma in each of the H-2 congenic strains, and the majority of animals developed both thymic and splenic tumors. As shown in Table 1, the incidence of lymphoma varied from 80 to 100% and latencies ranged from 4.6 to 6.4 months. Analysis of selected tumors by histopathology revealed lymphoblastic lymphomas, which is compatible with thymic or nonthymic T-cell lymphomas. Southern blot analysis confirmed that the tumors were of T-cell origin, as all tested tumors contained rearrangements of the T-cell receptor beta chain, with or without rearranged immunoglobulin heavy-chain genes (data not shown).

TABLE 1.

Analysis of envelope gene recombinant viruses in SL3-3NB-injected H-2 congenic mice

Strain	H-2 allele	Latency (mo)	Lymphomaa		Type I virus		Pheno- type
			%	No. with lymphoma/ no. tested	%b	nc
B10	b	6.1	89	17/19	29	7/24	NS
B10.Br	k	6.4	100	11/11	91	10/11	TI
B10.D2	d	6.3	86	18/21	94	17/18	TI
B10.Q	q	4.6	100	5/5	100	5/5	TI
B10.RIII	r	5.0	80	4/5	100	4/4	TI

^aIncidence of lymphoma in animals that survived 2 months or more. 

^bPercentage of animals with type I virus versus all animals with lymphoma. 

^cNumber of animals with type I viruses/all animals with lymphoma. 

To determine if recombinant viruses with type I envelope genes were present in lymphomatous tissues, the tumor DNA was cleaved with EcoRI and PstI and subjected to Southern blot analysis. The blots were hybridized to the pAKV5 probe, which is specific for the ecotropic 5′ p15E gene sequences that are retained in the type I but not type II recombinants (18). This probe will detect the 1.4-kb EcoRI-PstI proviral fragment from the 3′ end of type I proviruses as well as the 0.6-kb PstI-PstI fragment seen in some type I variants (Fig. 2). The SL3-3NB and endogenous ecotropic proviruses lack an EcoRI site within env and thus generate an 8.2-kb PstI-PstI fragment that also hybridizes to the probe. The allelic region of the type II proviruses produces a 0.6-kb PstI-PstI fragment that does not hybridize to this probe (6).

FIG. 2.

Strategy for Southern blot detection of type I viruses. The relevant restriction enzyme sites and relative position of the AKV5 probe within the different types of proviruses are shown. The AKV5 probe hybridizes to the ecotropic p15E gene sequences but not to the polytropic p15E gene sequences found in the type II recombinant and endogenous polytropic proviruses.

As shown in Fig. 3A, the 8.2-kb band of the endogenous or SL3-3NB ecotropic proviruses and the 1.4-kb band of type I recombinant proviruses were seen in almost all tumors from B10.D2 mice. Similarly, as summarized in Table 1, most tumors from the B10.Br, B10.RIII, and B10.Q strains also contained detectable type I recombinants by this assay. In contrast, type I proviruses (including one with a 0.6-kb band [data not shown]) were found in only 30% of lymphomas from B10 mice (H-2^b), a result that is not compatible with the TI phenotype (Fig. 3B). The presence of type I recombinants among the individual B10 tumors did not correlate with latency or pattern of disease or with the type of gene rearrangements. The proportion of tumors with type I recombinants in the B10 strain was statistically different from the frequency of these recombinants in tumors from the other strains (see Materials and Methods). The B10 phenotype was also distinct from the TII phenotype of CWD mice whose tumors lack detectable type I proviruses (32). We therefore assigned B10 mice the NS phenotype, which was conferred by a gene or genes within H-2^b.

FIG. 3.

Detection of type I recombinants in lymphomas from B10.D2 and B10 mice. Shown are autoradiographs of Southern blots of tumor DNAs from control and SL3-3NB-injected mice. Approximate band sizes in kilobases are shown at the left. (A) Lanes: +, positive control DNA from SL3-3NB-injected B10.Br mouse; −, negative control DNA from uninjected B10.D2 mouse; A to R, lymphoma DNAs from SL3-3NB-injected B10.D2 mice. (B) Lanes: +, positive control DNA from SL3-3NB-injected B10.Br mouse; −, negative control DNA from uninjected B10 mouse; A to R, lymphoma DNAs from SL3-3NB-injected B10 mice.

One explanation for the absence of type I recombinants in some B10 tumors would be the presence of type II recombinants as seen in SL3-3-induced tumors of CWD mice. However, additional Southern blot assays to detect type II proviruses with a nonecotropic p15E gene probe (6) were inconclusive due to the background from the B10 endogenous viruses (data not shown). However, type II-like recombinants were detected in 4 of the 12 B10 lymphomas, using other Southern blot and PCR amplification techniques that can identify subsets of recombinant MuLVs (data not shown). It is unclear if the remaining eight tumors lacked recombinant viruses or contained atypical forms that escaped detection by these assays.

B10 recombinant virus contains endogenous polytropic envelope gene sequences.

Given the novel phenotype of the B10 mice, it was important to confirm that the structure of the envelope genes of the B10 recombinants was similar to that found in recombinants recovered from other mouse strains. A portion of the envelope gene of a type I-like recombinant virus from B10 tumor E9 was amplified by PCR and cloned into a plasmid vector for subsequent DNA sequence analysis. In Fig. 4, the sequence of the E9-TA9 PCR fragment is compared to sequences of envelope genes of endogenous and other recombinant MuLVs. The nonecotropic sequences found in the envelope gene of the B10 E9-TA9 virus were highly homologous to those found in the HRS/J endogenous polytropic virus and recombinant viruses from other strains. This observation argues that the SL3-3-induced recombinant viruses in B10 mice acquire endogenous polytropic virus sequences by a process that is identical or highly similar to that utilized by recombinants from the other strains. Thus, it is unlikely that a unique recombination process or envelope gene donor in B10 mice confers the NS phenotype.

FIG. 4.

Comparison of partial envelope gene sequence of the B10 recombinant provirus E9-TA9 with sequences of other endogenous and recombinant proviruses. The sequences are compared to those of the endogenous polytropic virus MX27 (POLY). The first base corresponds to position 863 of the MX27 sequence (28). ECO is the endogenous ecotropic virus AKV623, CWNT25 is a type II recombinant from a CWD mouse, and PTV-1 is a type I recombinant from an HRS/J mouse (34). Dots in the sequence indicate homology with the corresponding nucleotide of the POLY sequence; nucleotide substitutions are shown by placement of the appropriate symbol. The sequence of E9-TA9 beyond position 710 was not determined. The most 3′ extents of substitution by endogenous polytropic sequences in PTV-1 and E9-TA9 are shown. The overlined sequences indicate the relative position of the AKV5 probe (ECO sequences) that distinguishes type I from type II proviruses. Restriction enzyme sites for PstI (polytropic only), XbaI (ecotropic only), and BglII are underlined.

An interesting feature of the E9-TA9 virus was that the entire gp70 gene was probably derived from an endogenous polytropic virus. The switch from polytropic to ecotropic virus sequences occurred just 3′ of the polytropic virus-specific PstI site and immediately 5′ of the ecotropic-specific p15E gene sequences. Type I recombinants with a similar envelope gene structure have been detected by Southern blot assays in tumors from HRS/J mice, which led to the hypothesis that the type I env virus phenotype is determined by ecotropic p15E gene sequences located 3′ of the PstI site (6). However, additional experiments are required to confirm that the E9 env sequences confer the type I virus phenotype.

The NS phenotype of B10 mice is linked to H-2A^b.

To map the region within H-2 that regulates the NS or TI phenotype, we studied SL3-3NB-induced tumors recovered from B10 H-2 congenic strains that carry hybrid H-2 regions and tumors from selected F₁ crosses. The results of the Southern blot analysis are summarized in Table 2. The B10A(4R) strain exhibited a TI phenotype, which indicated that the NS phenotype was not conferred by the H-2E^b or H-2D^b locus or adjacent genes. A role for H-2K^b was also excluded since B10.MBR mice had a TI phenotype. However, B10.A(5R) mice appeared to be an NS strain since only 3 of 10 tumors contained type I recombinants. This result strongly suggests that the NS phenotype is regulated by H-2A^b or an adjacent sequence, such as an LMP or TAP gene.

TABLE 2.

Comparison of H-2 genes and phenotypes of SL3-3NB-injected mice

Strain	Haplotype of corresponding H-2 gene					Type I virus		Phenotype
	K	LMPa	A	E	D	%b	n
Intra-H-2-recombinant congenic strains
B10.A(4R)	k	k	k	b	b	89	16/18	TI
B10.MBR	b	k	k	k	q	95	19/20	TI
B10.A(5R)	b	b	b	k	d	30	3/10	NS
NS × TI crosses
B10 × B10.Br	b/k	b/k	b/k	b/k	b/k	33	6/18	NS
B10 × B10.MBR	b/b	b/k	b/k	b/k	b/q	37	7/19	NS
B10.Br × B10.A(5R)	b/k	b/k	b/k	k/k	d/k	40	2/5	NS
B10.MBR × B10.A(5R)	b/b	b/k	b/k	k/k	d/q	47	16/34	NS
NS × TII cross
B10 × CWD	b/b	b/kc	b/d	b/?	b/?	36	4/11	NS

^aRepresents both LMP and TAP loci. 

^bPercentage of animals with tumors with type I viruses as calculated from n (see Table 1, footnote c). 

^cCWD was assigned H-2^k alleles of TAP-1 and LMP-7 and H-2A^d based on genetic markers as determined by Southern blot analysis (26a). The origins of TAP-2, LMP-2, and other genes remain undetermined. 

To determine if the NS phenotype of B10 and B10A(5R) was dominant or recessive to the TI phenotype of B10.Br and B10.MBR, we tested F₁ crosses of these strains. As summarized in Table 2, less than 50% of the SL3-3NB-induced tumors from the F₁ animals contained type I proviruses. This result indicated that the NS phenotype of the B10 and B10A(5R) mice was dominant compared to the TI phenotype of the other two strains. This was somewhat surprising given that the TI phenotype of HRS/J mice is dominant with respect to the TII phenotype of CWD mice. As might be predicted from these results, however, the B10 NS was also dominant over the CWD TII phenotype, as 4 of the 11 tumors from the (B10 × CWD)F₁ animals contained type I proviruses. The results from both sets of F₁ crosses indicate that a single copy of the H-2A^b region was sufficient to confer the NS phenotype.

DISCUSSION

The major finding of this study is that a host gene(s) that maps to the H-2A^b region influenced the envelope gene structure of tumor-associated recombinant MuLVs. Also, the H-2^b strain B10 exhibited a novel phenotype in which about 30% of the SL3-3NB-induced tumors contained type I recombinant MuLVs. At least a portion of the other tumors appeared to contain type II recombinants. The B10 phenotype, designated NS for nonselective, was distinct from the TI phenotype of other mouse strains in which type I recombinants are found in almost all tumors and the TII phenotype in which type II recombinants can be detected but type I recombinants are absent (6, 32). Based on the analysis of the H-2 recombinant strains, B10.A(4R), B10A.(5R), and B10.MBR, the sequences that confer the NS phenotype were located within or near H-2A^b. The NS phenotype of both B10 and B10.A(5R) mice proved to be dominant in crosses with the TI strains, B10.Br and B10.MBR, and the TII strain, CWD.

Whether the NS gene is allelic to the TI and TII genes is not clear. As discussed earlier, there are data to suggest that the TI gene is also located within H-2. For instance, the TI phenotypes of HRS/J and CBA/J mice are linked to polymorphisms located within the H-2A region (6). Also, all H-2^k strains that have been tested, including HRS/J, CBA/J, and B10.Br, exhibit the TI phenotype, whereas B10 mice and another H-2^b strain, C57L/J, appear to share the NS phenotype (5a). On the other hand, the H-2A^d region alone does not appear to confer the TI phenotype to the CWD strain. As determined by genetic markers, CWD mice appear to carry H-2A^d sequences yet have a TII phenotype (26a).

How did the H-2A^b genes confer the NS phenotype to B10 and B10.A(5R) mice? We propose that the NS gene, like the TI gene, acts at a step that is subsequent to the generation of the recombinant viruses (34). This hypothesis is supported by the analysis of the gp70 and p15E gene sequences of the B10 recombinant E9. The results suggest that the process of recombination between the parental SL3-3NB and endogenous polytropic viruses in B10 mice is similar or identical to that seen in other strains that generate either type I or type II recombinants. Moreover, the known candidates for the NS gene, H-2A, LMP-2, LMP-7, TAP-1, and TAP-2, all encode products that are involved in antigen processing, transport, or presentation (1, 25, 38). This implies that the presence or absence of type-specific antiviral immune responses is responsible for the different phenotypes. The H-2 region is known to influence the oncogenicity and disease latency of exogenous or endogenous MuLVs in other systems. In some cases, the effects of the H-2 genes may be mediated by antiviral immune responses directed against the envelope proteins (16, 19, 26, 30, 40–42, 46, 47). T-cell epitopes have been found in the p15E proteins of the Friend and AKV MuLVs, although these are found outside the type-specific domain (6, 7, 24, 34). Whether any of these reported immune responses to the envelope proteins would be capable of mediating the selection of specific types of recombinant viruses is not known.

One explanation for the results shown here is that the H-2 genes regulate type-specific immune responses. The p15E proteins of type I and type II recombinants differ from one another by 10 amino acids which are clustered in the amino-terminal portion of the molecule and thus are likely to be immunologically distinct (Fig. 1; references 6 and 34). Also, H-2A^b and LMP-2^b encode proteins that are involved in antigen processing or presentation. These two loci appear to be the best candidates for the NS gene, since the cognate proteins have the greatest degree of nonhomology compared to the analogous non-H2^b proteins (38, 44, 45). It is tempting to speculate that the NS phenotype is related to the failure of H-2A^b of the NS strains to mount antiviral responses directed against type I- or type II-specific p15E antigens, whereas the non-b H-2A molecules from TI strains are capable of presenting type II but not type I-specific peptides leading to the selection of type I recombinants. The problem with this simple model is that it does not explain the fact that NS is dominant with respect to TI. The model would predict that F₁ animals that coexpress b and non-b H-2A molecules will mount anti-type II virus immune responses and thus exhibit a TI phenotype.

Dominance of NS, however, is compatible with a model in which the viruses induce ineffectual immune responses that paradoxically promote leukemogenesis. The ineffective responses could stimulate the proliferation and increase in number of pre-T cells that serve as targets for viral replication and subsequent transformation. This effect would be similar to that proposed to explain how antiviral host responses augment lymphomagenesis by the Moloney MuLV (19). Thus, one possibility is that the H-2^b NS protein of B10 mice induces ineffectual immune responses against both type I and type II viruses, in essence stimulating both types similarly and allowing other subtle factors, such as replicative advantage, to come into play. In contrast, the analogous non-b H-2 proteins expressed in other strains might elicit responses only to type I recombinants overcoming the less significant forces. In the case of CWD mice, a unique lack of response might allow the selection of exclusively type II viruses via another mechanism. Thus, the NS phenotype would be dominant in F₁ crosses with TI animals due to the expression of the H-2^b proteins and stimulation of both viral types. The TI phenotype, in turn, would be dominant to CWD since a response favorable to the selection of type I viruses would be produced.

Conversely, the dominance of the NS genes could be explained by the induction of tolerance to the envelope proteins of both type I and type II viruses. One possibility is that the NS gene product mediates the deletion of T-cell clones that react with type-specific envelope sequences. A similar mechanism is thought to explain why the expression of H-2E enhances rather than suppresses the growth of leukemia cells in Friend MuLV-infected mice (24). Less likely, immune tolerance may result from a specific interaction between the allele-specific domain of the NS gene product and p15E that blocks the processing, transport, or presentation of viral antigens. If this is so, perhaps the functions of the corresponding non-b H-2 protein would be blocked by the p15E proteins of type I but not type II viruses, resulting in the selection of the former. This postulated effect of p15E on antigen presentation is analogous to that reported for the ICP47 protein of herpes simplex virus (14). However, to explain the dominance of NS, the NS protein-p15E interaction must act as a dominant negative in tissues of (NS × TI)F₁ mice, which seems unlikely.

Finally, our data do not exclude the possibility that the NS gene functions by a nonimmune mechanism. The efficiency of virus replication may be increased or inhibited by a direct interaction between the allele-specific region of the NS gene product with the type-specific domain of pI5E. This could influence maturation or processing of viral envelope proteins, virion assembly or release, or the early stages of entry of viruses into cells. The effect may be similar to that of variants of the human protein CKR5 which interact with the human immunodeficiency virus envelope protein and influence the efficiency of viral replication in vitro and in vivo (8, 9, 11). Certainly, a first step in elucidating how the H-2A^b gene(s) confers the NS phenotype in B10 mice will be to determine if the mechanism is immune or nonimmune. Regardless, the generation and selection of recombinant MuLVs in H-2 congenic strains of B10 mice provides a new model system to study how the major histocompatibility complex influences the genetic evolution of pathogenic retroviruses in vivo.