The Immunoevasive Function Encoded by the Mouse Cytomegalovirus Gene m152 Protects the Virus against T Cell Control in Vivo

Mouse NIH 3T3 cells (American Type Culture Collection [ATCC] CRL1658) were grown in DME supplemented with 10% newborn calf serum. Primary mouse embryonic fibroblasts (MEFs) prepared from BALB/cJ mice and B12 cells 24 were grown in MEM with 10% FCS. The Smith strain of MCMV (VR-194; ATCC) and the recombinant viruses were propagated on third-passage MEFs and purified by sucrose gradient centrifugation. Tissue culture–grown virus preparations were used throughout.

Plasmid constructions were performed by standard methods 25. Plasmid p152KO used for generating m152⁻ recombinant viruses was constructed by ligation of a 5-kb NotI–BamHI fragment comprising a loxP-flanked lacZ cassette 26 into the XhoI/NheI-digested plasmid pEcoOΔMB (all sites were blunt-ended by treatment with Klenow DNA polymerase). Plasmid pEcoOΔMB contains a 5.0-kb EcoRI–MluI fragment of the MCMV genome (MCMV nucleotides 209,756–214,714) encompassing the m152 gene 27. To generate recombination plasmid pm152gpt, the Escherichia coli gpt gene was flanked with loxP sites and inserted into an XhoI site of plasmid pEcoOΔMB at the 3′ end of the m152 gene.

Recombinant viruses were generated by homologous recombination in NIH 3T3 as described previously 26. LacZ⁺ recombinants were identified by 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-gal) staining and isolated by at least five rounds of limiting dilution passage. Recombinant viruses carrying the gpt marker were first enriched by selection in medium that was supplemented with 12.5 μg/ml mycophenolic acid (GIBCO BRL) and 100 μg/ml xanthine (Sigma Chemical Co.) 28 and further purified by limiting dilution. LacZ⁻ and gpt⁻ mutants were generated by a single passage through the recombinase Cre⁺ cell line, N2 26. LacZ⁻ recombinants were identified as white plaques after X-gal color screening and purified by limiting dilution. Gpt⁻ mutants were selected on STO cells (ATCC CRL-1503) in medium containing 20 μg/ml 6-thioguanine (Sigma Chemical Co.) as described previously 29. To characterize the recombinant virus genomes, viral DNA was isolated from infected cells and analyzed by Southern blot analysis 26.

B12 cells were infected with wild-type MCMV or m152 recombinant viruses. Cells were pulse labeled at 37°C for 60 min with 500 μCi/ml [³⁵S]methionine (1,200 Ci/mmol; Amersham) in methionine-free MEM supplemented with 5% dialysed FCS and chased in the presence of 10 mM nonlabeled methionine for 2 h. Labeled cells were washed in ice cold PBS and disrupted in lysis buffer (140 mM NaCl, 20 mM Tris/HCl, pH 7.6, 5 mM MgCl₂, 1% NP-40, and 1 mM PMSF). Cytoplasmic extracts were precleared by incubation with normal mouse serum, antiactin mAb (Boehringer Mannheim), and protein A–coupled Sepharose (Pharmacia). Immunoprecipitations were performed with anti-K^d mAb MA-215 ascitic fluid, and immune complexes were retrieved using protein A–coupled Sepharose. Endoglycosidase H (Endo H; Boehringer Mannheim) digestion and SDS-PAGE were performed as described previously 23.

Target cells were labeled for 90 min with Na₂⁵¹CrO₄, and a 4-h standard release assay was performed with 10³ target cells and a graded number of effector cells in fivefold dilution steps as described 21,30. In short, for selective and enhanced expression of immediate-early (IE) genes, MEFs were infected with 0.5 PFU of recombinant viruses or wild-type MCMV per cell by centrifugation (800 g, 30 min). Infection was performed in the presence of cycloheximide (50 μg/ml), which was removed 3 h later by washing with medium containing actinomycin D (5 μg/ml). Limited E gene expression after CH treatment was achieved by removal of cycloheximide using inhibitor-free medium and by adding actinomycin D to the final concentration of 5 μg/ml after 1.5 h. To generate pp89-specific polyclonal CTLs, MCMV-primed spleen cells were restimulated with pp89-derived antigenic peptide 21, and recombinant IL-2 (100 U/ml) was added 5 d later. Cultures were restimulated with gamma-irradiated syngeneic MEFs pulsed with antigenic peptide at a concentration of 10⁻⁸ M. Data represent the mean percentage of specific lysis from three replicate cultures (see Fig. 2 B).

BALB/c (H-2^d haplotype) and C57BL/6 mice (H-2^b haplotype) were bred at the Central Animal Facilities at the Medical Faculty, University of Rijeka. Mice homozygous for the μ chain mutation (C57BL/6 background; reference 31) were provided by Dr. Klaus Rajewsky (Institute for Genetics, Cologne, Germany) and were backcrossed on the BALB/c background for 10 generations. Mice heterozygous (μMT^−/+) and homozygous (μMT^−/−) for the μ chain mutation were distinguished by ELISA for the presence or absence of IgM in mouse sera, as described previously 32. Mice homozygous for the β2 microglobulin mutation (β₂m^−/−; supplied by Dr. Rudolf Jaehnisch, Whitehead Institute of Biomedical Research, Cambridge, MA) fail to express ternary MHC class I complexes and are devoid of CD8⁺ T lymphocytes 33. Mice homozygous for the deletion of the gene encoding the CD8 molecule (CD8^−/−) were obtained from the Centre de Developpement des Techniques Avancées pour l'Expérimentation Animale, Institut de Transgenose, Orleans, France. The absence of CD8⁺ T lymphocytes in β₂m^−/− and CD8^−/− mice was verified by flow cytometry as described previously 34. Neonatal mice, 24 h and 4 d postpartum, were injected intraperitoneally with recombinant viruses or wild-type MCMV. 6–8-wk-old mice were injected either in the posterior footpad or i.p. with 2 × 10⁵ PFU of virus in a volume of 50 and 500 μl of diluent, respectively, as described 35.

Plaque assays were performed in MEF as described previously 36,37. Statistical significance of differences between the experimental groups was determined by the Mann-Whitney exact rank sum test. Virus titers (x and y) were considered significantly different for P (x versus y) < alpha = 0.05 (one sided), where P is the observed probability value and alpha is a selected significance level.

In vivo depletion of CD4⁺ and CD8⁺ T lymphocyte subsets was performed by intraperitoneal injection of mAbs (rat anti–mouse) to CD4 (YTS 191.1) and/or CD8 (YTS 169.4) molecules 38. Adult and newborn mice received 1 mg and 250 μg of antilymphocyte antibodies, respectively, at the time of injection and every fifth day throughout the experiment. The efficacy of T lymphocyte depletion was >95%, as assessed by cytofluorometric analysis of spleen cells using FITC- or PE-conjugated antibodies directed against mouse CD4 and CD8 molecules (Becton Dickinson; nos. 1333 and 1447).

Donor T lymphocytes were harvested from spleens of uninfected (nonprimed) or latently infected (MCMV-primed) mice. Recipient mice were injected with 2 × 10⁵ PFU of virus in a rear footpad 12 h after gamma irradiation (6.5 Gy). Immediately after infection, 2 × 10⁵ nylon wool–purified cells were injected intravenously into recipient mice. Mice that did not receive cell transfer were used as negative controls. Mice were killed on day 14 after infection, and tissues were harvested for virus titer determinations.

To investigate the significance of the m152 gene product in the course of infection, a targeted deletion of the m152 gene and subsequent reintroduction of this gene into the MCMV genome was performed (Fig. 1 A). The recombinant virus ΔMC95.21 was generated by homologous recombination between the wild-type MCMV genome and the recombination plasmid p152KO. In this plasmid, a 1.2-kb XhoI–NheI fragment containing the m152 gene was replaced by a loxP-flanked E. coli lacZ gene. The lacZ marker was excised by passaging the ΔMC95.21 recombinant through the recombinase Cre⁺ cell line N2 26 to create the m152⁻lacZ⁻ deletion mutant ΔMC95.24. To generate a revertant virus, the m152 gene, together with the loxP-flanked gpt gene, was reinserted by homologous recombination into the ΔMC95.21 genome. After positive selection 28 of the m152⁺gpt⁺ virus rMC95.26, the gpt marker gene was again removed by passaging the virus through the recombinase Cre⁺ cell line N2 to generate the m152⁺gpt⁻ revertant virus (designated rMC96.27).

Southern blot analysis of the recombinant virus genomes confirmed the recombination events at the expected positions (Fig. 1 B). In the mutant virus genomes, the original 23.3-kb HindIII E fragment and the 5.9-kb EcoRI O fragment were replaced by expected new HindIII and EcoRI fragments. HindIII fragments of 20.0 and 6.5 kb and EcoRI fragments of 4.8 and 3.5 kb are evident in ΔMC95.21, whereas in the m152⁻lacZ⁻ deletion mutant, ΔMC95.24, a 22-kb HindIII fragment and a 4.7-kb EcoRI fragment were found. In the genomes of the rMC95.26 and rMC96.27 revertant viruses, HindIII fragments of 21.2 and 2.9, and 23.2 kb, and EcoRI fragments of 6.4 and 0.4 kb, and 5.5 and 0.5 kb, were observed, respectively. Note that the size of the HindIII E fragment in the rMC96.27 genome is identical to that in the wild-type MCMV genome, whereas a new EcoRI site purposely introduced outside of the m152 open reading frame enabled us to discriminate between the constructed revertant and wild-type MCMV. Comparison of the HindIII, EcoRI, and XbaI digestion patterns of the recombinant genomes with those of the wild-type MCMV genome confirmed that the recombinant viruses were free of detectable deletions or insertions in any other region of the viral genome (data not shown).

The altered glycosylation pattern of newly synthesized molecules can be used to locate the export block of nascent MHC class I molecules in MCMV-infected cells 1,22,23. Correctly assembled MHC class I complexes retained by the m152 gene product in the ERGIC/cis-Golgi compartment of MCMV-infected cells are not processed by medial-Golgi enzymes to complex glycans. Therefore, the majority of MHC class I molecules from cells infected with wild-type MCMV exhibit high mannose N-linked glycans typical for this compartment that are sensitive to Endo H and migrate faster in gels after digestion with Endo H (Fig. 2 A). In contrast, MHC class I complexes in cells infected with the m152 deletion mutant ΔMC95.21 as well as in uninfected cells acquire Endo H–resistant glycans, indicating the normal egress from the ERGIC/cis-Golgi compartment. As expected, the MHC class I transport was affected again in cells infected with the revertant virus rMC96.27, demonstrated by the reappearance of molecules sensitive to Endo H digestion.

The transport arrest of MHC class I molecules by the MCMV m152 gene product at early (E) times of virus replication prevents surface expression of these molecules and thus the recognition and lysis of infected cells by specific CTLs 1. A deletion of this gene should restore the recognition of infected cells by CTLs under the experimental conditions. To test this, MEFs were infected with the m152 deletion mutant ΔMC95.21, the revertant virus rMC96.27, or wild-type MCMV. Infected cells were arrested in the IE or E phase of the MCMV replication cycle and used in a CTL assay with MHC class I–restricted CTLs specific for the MCMV antigen pp89 21,30,36. As expected, recognition and cytolysis were equivalent for cells infected with all three viruses during the IE phase of the viral replication cycle, a time at which the m152 gene product is not yet expressed (Fig. 2 B). However, recognition was impaired during the E phase when cells infected with wild-type or revertant virus were used as targets. In contrast, efficient recognition of ΔMC95.21-infected cells was seen, confirming that retention of MHC class I molecules and the associated block in antigen presentation is mediated under these conditions exclusively by the m152 gene.

Multistep growth curves of recombinant and wild-type viruses served to assess whether the deletion of the m152 gene affects virus growth in cell culture. After infection of NIH 3T3 fibroblasts at a multiplicity of infection of 0.1 PFU per cell, replication of the m152 deletion mutant and revertant were indistinguishable from that of MCMV wild-type virus (Fig. 3). Identical results were obtained by comparing the replication capacity of the m152 deletion mutants, the revertant virus, and MCMV wild type on primary MEFs (not shown), indicating that the m152 gene product is completely dispensable for virus growth in fibroblasts.

Considering the fact that three different MCMV genes affect nascent MHC molecules and that m152 merely represents the gene that is expressed first, it was not clear whether or not the deletion of this gene would have any detectable impact on the susceptibility of the virus to immune control in vivo. Whereas adult mice control the infection with tissue culture–derived wild-type MCMV effectively, young mice allow virus replication to high titers 39,40. To detect even minor differences in virulence due to deletion of the single m152 gene, we assayed virus replication in neonatal mice. To avoid the potential influence of marker gene products on the biological properties of mutant viruses, the in vivo experiments were performed mainly with the m152 deletion mutant ΔMC95.24 and the revertant virus rMC96.27, although the other mutants gave comparable results (data not shown). Neonatal mice were injected with 100 PFU of the m152 deletion mutant, the revertant virus, or wild-type MCMV and monitored for 30 d. After infection with wild-type MCMV or revertant virus, 53 and 75%, respectively, of animals succumbed to infection (Fig. 4 A). In contrast, infection with the m152 deletion mutant was survived by the majority of mice (25% mortality). With respect to clinical signs, all three groups of mice exhibited during the first week of infection significant runting and a general failure to thrive compared with mock-infected controls. By 14–20 d after infection, however, most animals that survived the infection with the m152 deletion mutant had recovered. In contrast, clinical signs persisted throughout the course of observation for wild-type MCMV and revertant virus–infected mice. The different disease courses correlated with the body weights of infected mice. On day 26 after infection, the average body weight of mice that survived infection with ΔMC95.24 was comparable to that of the control group (9.79 ± 1.86 and 10.9 ± 1.16 g, respectively), whereas mice infected with the revertant virus still appeared runted (7.04 ± 1.70 g; data not shown).

To assess whether the differences were due to an altered tissue tropism associated with the m152 deletion, virus titers were determined for lungs, spleen (Fig. 4 B), and salivary glands (data not shown). The mutant ΔMC95.24 yielded lower titers in the spleen and lungs as compared with wild-type MCMV and the revertant virus. Although the differences in virus titers in tissues of neonatal mice did not exceed 1–2 log₁₀ steps, this finding was reproducible both in MCMV-sensitive (BALB/c) as well as MCMV-resistant (C57BL/6) mouse strains. In the salivary glands, this observation could not be made. In this organ, the virus titer yielded by the m152 deletion mutant was indistinguishable from that of the wild-type and revertant virus. In this context, it is of interest to note that we have demonstrated earlier that the salivary gland represents the only organ in which MCMV replication is exempt from CD8⁺ T cell control 41. Altogether, the lack of the m152 gene results in an attenuated course of infection and in restricted virus growth.

Immunodeficient mice were used to assess whether the attenuated phenotype of the m152 deletion mutant indeed reflected an enhanced sensitivity to T cell control. BALB/c mice were immunodepleted by gamma irradiation and by injection with cytolytic antibodies to T lymphocytes and NK cells. In immunodepleted animals, all three viruses replicated to high titers without significant titer differences (data not shown). This demonstrated already that the attenuated phenotype of the m152 deletion mutant is caused by an increased sensitivity to immune control mechanisms. The m152 deletion mutant replicates to lower virus titers than the revertant virus (Fig. 5 A, left panels; P < 0.005) in undepleted BALB/c as well as C57BL/6 mice (Fig. 5 A, top and bottom panels, respectively). This growth restriction was abrogated after depletion of T lymphocyte subsets (Fig. 5 A, right panels), indicating that the attenuated phenotype of the deletion mutant is caused by an enhanced sensitivity to T cell control.

B cell–deficient (μMT^−/−) mice were employed to identify the relative role of T cell subsets (Fig. 5 B). Due to the lack of specific antibodies, MCMV spreads rapidly in μMT^−/− mice, and detection of infectious virus is facilitated 35. 8-wk-old μMT^−/− mice were depleted of only CD8⁺ T lymphocytes, depleted of both CD8⁺ and CD4⁺ T cell subsets, or left undepleted. Virus titers were determined 10 d after infection. The growth restriction of the m152 deletion mutant was notable in particular in the lungs of nondepleted mice, resulting in titer differences ranging from 2 to 3 log₁₀ (Fig. 5 B, left panels). After depletion of CD8⁺ T lymphocytes or of both T cell subsets, mutant and revertant virus reached comparable virus titers (Fig. 5 B, center and right panels). These data demonstrate that CD8⁺ T cells are the relevant cell subset responsible for the replication inhibition associated with the m152 gene deficiency. Furthermore, the attenuating effect is also seen in adult mice. Although the differences are not significant, in BALB/c μMT^−/− mice, the m152 deletion mutant reached slightly lower titers than the revertant, even after depletion of T cell subsets.

The m152 gene function affects antigen presentation in the MHC class I pathway. Therefore, in mice in which this presentation pathway is defective, the specific defect of the virus should be phenotypically complemented. To test this, we used MHC class I–deficient C57BL/6 β₂m^−/− mice 33 and mice deficient for CD8⁺ T lymphocytes due to the deletion of the CD8 gene (C57BL/6 CD8^−/− mice). 4-d-old mice were infected with 1,000 PFU of either the m152 deletion mutant or the revertant virus. In contrast to the situation in immunocompetent mice, no difference in the titers between the two viruses was found in three replicate experiments performed in β₂m^−/− and CD8^−/− mice. One representative experiment is shown in Fig. 6. Essentially, the same message was obtained in adult CD8^−/− mice infected with the m152 deletion mutant or the revertant virus. However, adult mice of the C57BL/6 strain cleared both viruses so efficiently that the titers in tissues were below the threshold levels when assayed 10 d after infection. Therefore, to enhance the virus replication and to get a measurable virus load in tissues, we had to deplete NK cells in vivo (data not shown). Altogether, these experiments show that attenuation of the m152 deletion mutant is directly linked to functions required for antigen presentation and recognition in the MHC class I pathway.

Adoptive cell transfer into immunodepleted recipients was used to determine the sensitivity of MCMV N2 virus-primed as well as to naive lymphocytes. Lethal MCMV infection in gamma-irradiated BALB/c mice is therapeutically prevented by adoptive transfer of as few as 10⁵ MCMV-primed CD8⁺ T cells, whereas the same number of naive lymphocytes or primed CD4⁺ T cells is ineffective 37,41,42. As the product of the m152 gene downregulates presentation of viral antigens in the MHC class I pathway, we expected to see an increased sensitivity of the mutant to primed T cells and perhaps also a more effective priming of T lymphocytes.

To test this, 2 × 10⁵ lymphocytes derived from BALB/c mice, either MCMV primed or naive, were intravenously transferred into syngeneic gamma-irradiated recipients 12 h after infection with wild-type MCMV, the m152 deletion mutant, or the revertant virus strain. Adoptive T cell control of MCMV is a selective function of CD8 T cells but not of CD4 T cells and is more effective in spleen and liver than in the lungs 37,42. Accordingly, the replication of the m152 deletion mutant is more efficiently controlled in these organs than the revertant virus.

Small numbers (∼10⁵) of naive T lymphocytes fail to protect mice against MCMV infection 41. This was reproduced for mice infected with the revertant virus; however, the number of 2 × 10⁵ naive lymphocytes already decreased the titers of the m152 deletion mutant (Fig. 7). This is a function of T lymphocytes, as depletion of the CD8 T cell subset eliminated this activity (data not shown). Transfer of graded numbers of naive cells into gamma-irradiated mice showed that the number of naive T cells had to be increased by 100-fold to achieve an effect on wild-type MCMV comparable to the effect on revertant MCMV (data not shown). We therefore concluded that deletion of the m152 gene increases the antigenicity of the virus.

Herpesvirus genomes contain several genes coding for potential immunomodulatory functions. Shared between viruses of the α- and the β-herpesvirus family is the expression of gene functions that interfere with peptide presentation in the MHC class I pathway in vitro. Herpesviruses are highly species specific, and so are the functions of the genes that affect this pathway. No cellular homologue has been detected for any of these genes so far 4. For a better understanding of the contribution of each individual gene to the biology of the virus infection, experiments in the natural host are required. Here, we report on the in vivo function of the immunomodulatory protein encoded by gene m152 of MCMV during infection of its natural host.

To prove that a gene has a predicted immunoregulatory function in vivo, three aspects must be addressed. First, the deletion of the gene from the genome should not affect virus growth in cells in the absence of immune control. Second, a phenotype seen in vivo should be lifted by a targeted revertant of the virus. Third, the attenuation due to lack of the immunomodulatory function of the virus should be phenotypically complemented in a host that is genetically or functionally disabled to exert the control that is specifically affected by the deleted viral gene product.

Only the fulfillment of all three requirements confirms the prediction of the in vitro studies. Not all virus genes that have an effect on specific immune effector mechanisms in vitro show this effect in vivo as their main function. One such example is the Fc receptor function encoded by gene m138 of MCMV. The Fc receptor is expressed at the cell surface and selectively binds mouse IgG in vitro. The deletion of m138 results in strong attenuation of the mutant virus in vivo that is lifted by the specific revertant. However, in Ig-deficient mice, the attenuation is still present, proving that attenuation of the virus due to the deletion of the Fc receptor is not linked to Ig control 26,43.

The m152 gene encodes the glycoprotein gp40, which arrests the export of nascent mouse but not human MHC class I molecules 16. If this was the major function of the protein, then the deletion of the gene should be dispensable for virus growth in fibroblasts but should restrict replication in immunocompetent animals. This prediction was fulfilled by the m152 deletion mutant viruses. Virus growth in vivo but not in fibroblasts was affected by the mutation. Furthermore, the MHC class I complex transport and the capacity to present viral peptides to CD8 T lymphocytes was restored.

The revertant virus regained wild-type properties in vivo and fulfilled the second requirement by proving the causal linkage between targeted deletion and biological phenotype. As with HCMV infection in humans, the primary infection of mice even with wild-type MCMV is usually asymptomatic. Newborn mice and mice that are a few days old are much more sensitive than adult mice to tissue culture–grown virus, due to the immaturity of the NK cell response 44. In neonates, the infection with 10² PFU causes a high percentage of mortality and runting in survivors. The attenuating effect of the m152 gene deletion resulted in a higher number of survivors and an earlier cessation of runting.

The third requirement was also fulfilled: loss of the phenotypic difference between deletion mutant and revertant virus in the absence of the host immune function affected by the viral gene product. gp40 blocks the export of nascent MHC class I molecules already loaded with viral peptides. The predictable consequence is the inhibition of CD8 T cell priming and CD8 T cell effector function. Loss of the m152 gene should lead to an increased sensitivity of the virus to lymphocytes. Indeed, the virus mutant grew to smaller titers in the various tissues tested. This attenuation did reflect a more stringent control of the deletion mutant by T cell functions, as elimination of T cells resulted in comparable tissue titers of mutant and revertant virus. Furthermore, the attenuating effect of the m152 deletion mutant was absent in C57BL/6 mice that failed to form the functional MHC class I molecules due to the lack of β2-microglobulin expression and also in mice that have a defect in the maturation of MHC class I–restricted CD8⁺ T cells due to the deletion of the CD8 gene. Altogether, this study proves for the first time that in their natural host, herpesviruses benefit from functions that inhibit antigen presentation in the MHC class I pathway in vivo.

It remains open whether the observed function is the only function of the m152 gene product in vivo. MHC class I molecules activate CD8⁺ T cells and, at the same time, inhibit NK cells 45,46. Accordingly, a prediction of the transport block of MHC class I molecules due to m152 gene expression is the susceptibility of MCMV-infected cells for NK cell–mediated destruction in vivo. A deletion of the m152 gene and the restoration of MHC class I molecule transport should result in an enhanced resistance of infected cells to NK cell control in vivo. Our data do not support this assumption. Preliminary studies suggest that the lack of the m152 gene certainly does not make the virus more resistant to control by NK cells (Krmpotic, A., B. Polic, and S. Jonjic, unpublished data). Both HCMV and MCMV genes code for glycoproteins that show homology to MHC class I molecules, UL18 in HCMV 47 and m144 in MCMV 27,48. It has been hypothesized that these viral MHC class I homologues are capable of engaging NK cell inhibitory receptors to protect cells from lysis due to the downregulation of MHC class I expression. Attenuation of MCMV harboring a deletion in the m144 gene has been explained by enhanced control by NK cells in vivo 48. However, a more recent study on UL18 functions failed to confirm the inhibitory function of viral MHC class I homologues on NK cells 49. Therefore, the potential interaction of m152 with m144 needs to be addressed.

Another explanation is that the remaining functions of the genes m04 and m06 fully complement the expected NK cell effect of m152. The genes m04 and m06 have an effect on MHC class I molecules. Both genes are expressed later than m152 during the MCMV replication cycle, and both genes encode glycoproteins that bind tightly to MHC class I molecules. gp34, encoded by the m04 gene, forms a complex with MHC class I molecules that can be detected on the surfaces of infected cells, but the functional consequence is not clear 17. The m06 gene product gp48 binds to MHC class I molecules in the ER and reroutes them to lysosomes for rapid proteolytic degradation 18. This leads to the downregulation of MHC class I surface expression in the late phase of the replication cycle 18,23. Here, we show that the m04 and m06 genes cannot fully compensate all aspects of the loss of the m152 function. Thus, the interaction between the m152 gene product and other viral gene functions is not yet clear and remains to be tested. We are in the process of constructing double and triple deletion mutants to determine the individual contribution of each of the MHC class I–reactive genes and MHC class I homologues in immune evasion. To this end, we have recently pioneered the cloning of infectious herpesvirus genomes and have developed targeted and random mutagenesis techniques 50,51.

Our results show for the first time that genes that inhibit antigen presentation in the MHC class I pathway provide a significant growth advantage for CMV during primary infection. What is the potential benefit for the virus? The conditions of primary infection define the load of latent viral genomes and the risk of recurrence of the CMV infection 39. Accordingly, we predict that the m152 gene allows a higher number of MCMV genomes to establish a latent infection, thereby enhancing the chance for reactivation and transmission to the next host and thus escaping extinction.

This work was supported by grants of the Bundesministerium fuer Bildung und Forschung and the Deutsche Forschungsgemeinschaft to U.H. Koszinowski and by project 006204 from the Croatian Ministry of Science to S. Jonjic.