INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Cytomegaloviruses (CMVs) constitute prototype members of the  subgroup of the herpesvirus family. In the immunocompetent host, primary CMV infection is efficiently controlled by immune functions and does not cause major illness (5). Thereafter, a lifelong infection is established which is characterized by alternate stages of virus productivity and latency. In the immunodeficient host, recurrent CMV infection can result in a wide spectrum of disease manifestations, ranging from life-threatening to asymptomatic (5). Human CMV (HCMV) and mouse CMV (MCMV) show a multitude of similarities in biology and pathogenesis. Infection of the mouse with MCMV is an extensively used model of CMV infection. In vivo, CMV replication occurs in multiple cell types and tissues, including epithelial cells, endothelial cells, fibroblasts, smooth muscle cells, and macrophages (47, 48, 52). CMV gene expression is regulated in a cascade fashion characteristic of herpesviruses and can be subdivided into immediate-early (IE), early (E), and late (L) phases.

CD8⁺ major histocompatibility complex (MHC) class I-restricted T lymphocytes play a crucial role in host defense against CMVs (42, 45). This has been experimentally demonstrated in a murine adoptive transfer model in which MCMV-specific CD8⁺ T cells limit virus dissemination, prevent tissue destruction, and protect against fatal MCMV disease (39). A dominant fraction of the CD8⁺ T-cell response in BALB/c (H-2^d) mice is directed towards the nonstructural IE protein pp89 of MCMV (40, 41). During the nonproductive state of infection, CD8⁺ T cells play a prominent role to preclude recurrent MCMV replication (38).

Remarkably, CMVs have evolved gene functions which prevent CD8⁺ T-cell recognition of infected cells by downregulating MHC class I surface expression (reviewed in reference 20). In the MHC class I pathway of antigen processing and presentation, endogenously synthesized proteins are degraded in the cytosol and peptides are translocated into the endoplasmic reticulum (ER), where they assemble with MHC class I heavy chain and ₂-microglobulin (₂m) to form the trimolecular MHC I complex (18). This complex is exported to the plasma membrane to be recognized by peptide-specific CD8⁺ T lymphocytes. We have identified two MCMV E glycoproteins which prevent the transport of MHC I to the cell surface and CD8⁺ T-cell recognition. Specifically, the m152-encoded glycoprotein gp37/40 retains MHC I complexes in the ER-Golgi intermediate compartment (ERGIC)/cis-Golgi compartment (58). The MCMV gp48 molecule encoded by the gene m06 binds to ₂m-associated MHC class I molecules in the ER. Subsequently, MHC I-gp48 complexes leave the ER, pass the Golgi, and reach the endolysosome, where they become rapidly degraded (44).

In this study we addressed the apparent contradiction between the efficient generation of protective CD8⁺ T lymphocytes and the simultaneous inhibition of peptide presentation in the MHC class I pathway. We reasoned that CD8⁺ T-cell control is not compatible with a general inhibition of antigen presentation in all virus-infected cells. To this end we searched for a cell type differing from the prototypic MHC I-downregulated phenotype of CMV-infected fibroblasts. Macrophages represent professional antigen-presenting cells which are infected in vivo. Here we report that MCMV-infected primary bone marrow-derived macrophages (BMM) and the macrophage cell line J774 constitutively present pp89 peptides. Two features characterize antigen presentation of CMV peptides in infected macrophages. First, transport of MHC class I complexes to the plasma membrane remains intact, although m152/gp40 and m06/gp48 are synthesized. Second, quantitative assessment of pp89 peptides revealed a significantly higher antigen-processing efficiency in macrophages than in fibroblasts. Moreover, the yield of peptide determined in MCMV-infected organs of L^d+  L^d and L^d  L^d+ bone marrow (BM) chimeras indicated that bone marrow-derived cells represent a major source of processed pp89 peptides in vivo. The data suggest that macrophages act as potent inducers and amplifyers of the pp89-specific CD8⁺ T-cell response in vivo.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Mice. BALB/c mice (H-2^d) were purchased from IFFA-CREDO (Lyon, France), BALB/c-H-2^dm2 mice of the H-2^dm2 haplotype (H-2K^dD^dL) were provided by R. Reifenberg (University of Ulm, Ulm, Germany).

Cells. The simian virus 40-transformed BALB.SV (H-2^d) fibroblast cell line has been described previously (11). Primary BALB/c mouse embryo fibroblasts (MEF) were used after three in vitro passages for MCMV infection and extraction of naturally processed peptides. BMM were established as described (14). Granulocyte/macrophage colony-stimulating factor-producing L929 cells were cultured for 6 to 7 days in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal calf serum (FCS), 2 mM L-glutamine, and 1 mM sodium pyruvate. Bone marrow cells were collected from femurs of BALB/c mice and seeded in Iscove's modified Dulbecco's medium containing 30% of the L 929 cell supernatant, 10% FCS, 5% horse serum, 2 mM L-glutamine, and 1 mM sodium pyruvate at 37°C and 10% CO₂-humidified air. After 9 days of culture in Teflon film bags, BMM were harvested and plated in six-well dishes overnight before being infected with MCMV for cytotoxic T lymphocyte (CTL) assays. The BALB/c-derived macrophage cell line J774.A.1 (American Type Culture Collection ATCC TIB-67) was kept in 10% DMEM. P815 mastocytoma cells (ATCC TIB-64) were used as targets in CTL assays.

Virus, infection conditions, and virus titration. The Smith strain of MCMV (ATCC VR-194) was propagated in third-passage MEF and purified by pelleting through a sucrose cushion. More pathogenic salivary gland virus (SGV) of MCMV was prepared as described (28). In some in vitro experiments, the Fc receptor deletion mutant MS94.4 (9) and the m152 deletion mutant MC95.21 (31) were used. Subconfluent layers of cells were infected with MCMV at a multiplicity of infection (MOI) of 5 by centrifugation at 800 × g for 30 min and harvested as indicated. Selective expression of MCMV IE gene products was achieved by infection of cells with MCMV in the presence of cycloheximide (CH; 50 µg/ml), which was replaced 3 h later by actinomycin D (actD; 5 µg/ml). Controlled transition to the E phase was achieved by incubating cells in the absence of drugs for 90 min after removal of CH. Phosphonoacetic acid (PAA; 250 µg/ml) was used to arrest MCMV-infected cells in the E phase and to prevent L-phase gene expression. Animals were infected by intraperitoneal (i.p.) injection of 10⁶ PFU of SGV. Virus titers of virus stock preparations were determined by an in vitro plaque assay. To determine the growth kinetics of MCMV in permissively infected MEF, J774 cells, and BMM, cells were infected with MCMV at an MOI of 5. Supernatant and cells were harvested at different time points postinfection (p.i.) and counted on MEF in triplicate.

BM chimeras. For the generation of BM chimeras, female mice of the strain BALB/c and of the L^d gene deletion mutant strain BALB/c-H-2^dm2 were used at 8 to 10 weeks of age as BM cell donors and recipients. For the hematoablative conditioning, recipient mice were total-body irradiated with a single dose of 9.5 Gy from a ¹³⁷Cs  source (Buchler, Braunschweig, Germany), delivering a dose rate of 1.5 Gy min¹. Donor femoral BM cells were isolated as described (36). For prevention of graft-versus-host disease, BM cells were depleted of T cells by incubation with paramagnetic beads coated with the anti-Thy-1.2 monoclonal antibody (MAb) 30-H12 (Milteny Biotech, Bergisch Gladbach, Germany) and by passage through a Minimacs column in a strong magnetic field (Milteny Biotech). A total of 5 × 10⁶ BM cells were infused intravenously into the tail vein of recipients. The chimeras were used for the experiments 3 to 5 months after transplantation. The chimeric state was controlled by two-color cytofluorometric analysis of spleen cells and found to be consistent in all animals tested. Splenocytes (5 × 10⁶) were incubated with either purified biotinylated anti-Thy-1.2 MAb (Pharmingen, Hamburg, Germany), anti-L^d MAb 28-14-8s (Dianova, Hamburg, Germany), or anti-K^d MAb SF1.1.1 (Dianova) in combination with fluorescein isothiocyanate (FITC)-coupled rat anti-mouse CD4 MAb RM4-5 (Dianova), rat anti-mouse CD8a MAb 53-6.7 (Dianova), rat anti-mouse CD11b MAb M1/70 (Dianova), or rat anti-mouse CD45R MAb RA3-6B2 (Dianova) to distinguish cells of donor and recipient origin. Biotin-conjugated antibodies were stained with phycoerythrin-avidin (Dianova). Nonspecific FcR-mediated binding of murine antibodies was prevented by blocking with MAb 2.4G2 (ATCC HB197), specific for the murine Fc-II receptor.

Isolation of endogenously processed peptides, cytolytic assay, and quantification of peptides in infected cells and organs. Peptides were extracted as described previously (15). In brief, cells were infected with MCMV at an MOI of 3 for 16 h in the presence of PAA, treated with trypsin, resuspended in medium, washed, and counted before peptides were extracted. In the case of organs, organs (three per group) were collected 48 h p.i. and passed through a steel mesh, and 5% trifluoroacetic acid (TFA) was added to a cell lysate to achieve a pH of 2.0. After homogenization, extracts were sonicated, left for 30 min on ice, and ultracentrifuged for 45 min at 100,000 × g. Supernatants were removed and passed through a Sephadex G25 column with an isocratic flow of 5 ml of 0.1% TFA per min. Low-molecular-weight fractions were collected and passed through a SepPack C18 reversed-phase unit (Waters, Eschborn, Germany). Bound material was eluted with 1.5 ml of 50% acetonitrile (AcN) and 1.5 ml of 100% AcN. Eluates were pooled, concentrated to a final volume of 2 ml, and further purified on a PepS (Amersham Pharmacia, Freiburg, Germany) reversed-phase high-pressure liquid chromatography (HPLC) column. Extract (1 ml) was loaded and eluted with a flow rate of 0.8 ml/min on a linear AcN gradient with the following protocol: solution A (0.1% TFA); solution B (70% AcN, 0.1% TFA), 0 to 4 min at 25% B, 4 to 17 min of linear increase to 90% B; 17 to 21 min at 90% B; 21 to 23 min of linear decrease to 25% B; and 23 to 28 min at 25% B. Fractions were collected every 1 min and stored at 70°C. H-2L^d-restricted polyclonal CTL lines specific for pp89 peptide were cultivated as described (11). For the quantification of antigenic peptides derived from MCMV-infected cells, HPLC fractions 10 to 16 were adjusted to the cell count before being tested in fivefold dilution steps. Freeze-dried HPLC fractions were reconstituted with culture medium and incubated with 10^{3 51}Cr-labeled P815 target cells. After peptide binding for 60 min at 37°C, 10⁴ CTL were added, and the samples were tested in a 4-h cytolytic assay. The spontaneous release of target cells was between 5 and 15%.

Metabolic labeling, immunoprecipitation, and endo H treatment. Cells were labeled with [³⁵S]methionine (1,200 Ci/mmol; Amersham, Braunschweig, Germany) at a concentration of 500 µCi/ml for 60 min as described previously (11). After being washed with phosphate-buffered saline, cells were lysed in 1 ml of lysis buffer (1% [vol/vol] IGEPAL [Sigma, Munich, Germany], 5 mM MgCl₂, 140 mM NaCl, 20 mM Tris [pH 7.6], 0.2 mM phenylmethylsulfonyl fluoride). Lysates were centrifuged at 13,000 × g for 30 min and precleared by adding the appropriate preimmune serum and protein A-Sepharose (Pharmacia). ³⁵S incorporation into proteins was quantitated in all experiments by liquid scintillation counting of an aliquot of the lysate. Lysates were adjusted before immunoprecipitations were performed with ascites fluid or antiserum, as indicated. Immune complexes were mock treated or digested with 2 mU of endoglycosidase H (endo H; Boehringer, Mannheim, Germany) overnight at 37°C, eluted with sample buffer, and analyzed by 10 to 13.5% polyacrylamide gradient gel electrophoresis. The gels were dried and exposed to BioMaxMR films (Kodak) at 70°C for 1 to 7 days.

Antibodies. For detection of MHC class I molecules by immunoprecipitation, the 3-domain-specific MAb 28-14-8s (ATCC HB 27), recognizing free and ₂m-associated L^d, was used. MAb 20/234/28, recognizing MCMV e1 proteins (6), and the rabbit antiserum b5/1 against pp89 (11), gp37/40 (58), and gp48 (44) have been described. For flow cytometry, supernatant from the mouse Fc- receptor (FcRII)-specific rat hybridoma 2.4G2 (ATCC HB 197) was used to block Fc-mediated binding of antibodies to macrophages. Hybridoma supernatant from 30-5-7s (ATCC HB31) was used to detect surface H-2L^d molecules. MAb HB157 (ATCC) specific for HLA-B7 was used as an isotype control. Rat anti-mouse CD29 MAb was purchased from Pharmingen (Heidelberg, Germany).

Flow cytometry. Trypsinized cells were preincubated in 5% goat serum. In the case of macrophages, Fc receptors were blocked by incubation with 2.4G2 supernatant before cells were stained with MAbs. Bound antibodies were visualized by addition of fluoresceinated goat anti-mouse immunoglobulin G2a (IgG2a) isotype-specific antibodies (Medac, Hamburg, Germany). As a negative control, cells were incubated with the second antibody alone. As an isotype control, hybridoma supernatant of MAb B27MI (ATCC HB 157) recognizing HLA-B27 was used. A total of 10⁴ cells were analyzed for each fluorescent profile on a FACScan IV (Becton Dickinson & Co., San Jose, Calif.).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Constitutive presentation of pp89 during productive MCMV infection of primary and J774 macrophages. After infection of BALB/c MEF with MCMV, H-2 L^d-restricted presentation of the pp89-derived peptide YPHFMPTNL is detectable only when viral gene expression is restricted to genes of the IE phase (Fig. 1). A 30-min period of MCMV E gene transcription in infected MEF and subsequent translation of E gene mRNA are sufficient to prevent presentation of pp89 peptides to CD8⁺ CTL despite the sustained synthesis and stability of pp89 during the E phase (11). Accordingly, pp89 presentation is virtually absent during all phases of permissive infection of fibroblasts (22). Macrophages are professional antigen-presenting cells which have been shown to support CMV replication and to be infected in vivo (48, 52). As expected, MCMV-infected primary BMM were lysed by pp89-specific CTL under selective and enhanced expression of IE genes (Fig. 1). Remarkably, pp89 presentation was also observed with undiminished efficiency under conditions allowing expression of E and L genes. The same type of result was found when investigating the macrophage cell line J774. Macrophages infected for 6 h before addition of actD to terminate E gene expression and macrophages preactivated with lipopolysaccharide exhibited the same phenotype of pp89 presentation (data not shown). To exclude the possibility that recognition of pp89 in MCMV-infected BMM and J774 cells was due to an abortive infection restricting viral gene expression to IE genes, the growth kinetics of MCMV in macrophages and MEF were studied. MCMV replication was highly productive in MEF, reaching titers higher than 10⁸ PFU/ml at 48 h p.i. (Fig. 2). MCMV replication in BMM and J774 yielded titers of between 10⁶ and 10⁷ PFU/ml after 72 h p.i., confirming a productive infection. Altogether, infected macrophages, unlike fibroblasts, efficiently present pp89 peptides to CD8⁺ T cells throughout the replication cycle.

View larger version (14K): [in this window] [in a new window]   FIG. 1.   pp89 presentation in MCMV-infected MEF, BMM, and J774 macrophages. Presentation of pp89 peptides was tested by pp89-specific BALB/c CTL in a standard 4-h 51Cr release assay at the indicated effector-to-target cell ratio (E/T). Target cells were infected with MCMV as indicated or not infected (inf.). CH/actD, restriction of MCMV gene expression to the IE phase by infecting cells in the presence of CH, which was replaced after 3 h by actD to achieve selective and enhanced IE protein synthesis. CH/  /actD, conditions of controlled early gene expression by infecting cells in the presence of CH, which was washed out after 3 h, allowing IE protein synthesis followed by subsequent E gene transcription and translation. The transcription of MCMV genes was terminated after 90 min by the addition of actD. PAA was used to arrest gene expression in the E phase.

View larger version (19K): [in this window] [in a new window]   FIG. 2.   Growth kinetics of MCMV in MEF, BMM, and J774 macrophages. Subconfluently growing cells were infected with MCMV at an MOI of 5. Duplicate samples were collected at various time points after infection, and cell-associated virus was released by two freeze-thaw cycles. The titer of each sample was determined by a standard plaque assay on MEF.

Highly efficient peptide processing in permissively infected J774 macrophages. Intact pp89 presentation by infected macrophages could be due to the absence of MCMV control of the MHC class I presentation pathway. Alternatively, a quantitative difference in pp89 processing, i.e., the generation of a higher yield of pp89 peptides, could provide a larger number of YPHFMPTNL-loaded L^d molecules, some of which could escape from viral inhibition and reach the cell surface. To distinguish between these possibilities, we studied the relationship between pp89 synthesis and the yield of finally trimmed antigenic peptides in J774 macrophages and MEF. As demonstrated in Fig. 3, pp89 biosynthesis and subsequent processing to pp84 was much higher in MEF than in J774, in which pp89 was abundantly expressed not earlier than 36 h p.i. Pulse-chase experiments revealed no major differences in pp89 and pp84 stability between MEF and macrophages within 270 min of chase (data not shown). Relevant for CTL recognition is the processing of the polypeptide to the antigenic epitope rather than the synthesis of the protein. To test the capacity of pp89 peptide processing in both cell types, extraction of peptides was performed. The individual fractions were assayed with pp89-specific CTL for antigenic peptide activity, which showed no qualitative differences (Fig. 4). Consistently, the antigenic activity was detected in fractions 12 to 16 of our HPLC separation and peaked in fraction 14, coeluting with synthetic YPHFMPTNL. Serial dilution of fractions 10 to 16 revealed a similar amount of biological activity in MEF and J774. The peptide content per cell was calculated from the titration of a synthetic YPHFMPTNL standard. Accordingly, approximately 10⁴ peptide copies per cell could be recovered from MEF as well as from J774 macrophages. Thus, the efficiency with which YPHFMPTNL is processed differs between cell types. Macrophages process pp89 much more efficiently, and it is not the amount of the viral protein which represents the limiting factor for antigen presentation.

View larger version (18K): [in this window] [in a new window]   FIG. 3.   Biosynthesis of pp89 in MCMV-infected MEF and J774 macrophages. Subconfluently growing cells were infected with MCMV at an MOI of 3 for various times as indicated. Cells were metabolically pulse labeled with [35S]methionine for 1 h. pp89 molecules and cleavage products pp84 and pp76 were immunoprecipitated with pp89-specific antibodies.

View larger version (18K): [in this window] [in a new window]   FIG. 4.   Detection and quantification of antigenic pp89 peptides in MCMV-infected MEF and J774 macrophages. BALB/c MEF and J774 cells were infected with MCMV at an MOI of 3 for 16 h in the presence of PAA to prevent late gene expression. Cells were harvested and counted before acid-soluble molecules were extracted from identical cell numbers and separated by size exclusion chromatography, followed by reverse-phase HPLC using the indicated AcN gradient for elution. HPLC fractions were analyzed in triplicate with pp89-specific CTL for their content of antigenic peptides. The quantitative assessment of the antigenic activity in fractions 10 to 16 was performed by serial dilution from undiluted to 1:25.

Expression of MHC I-subversive MCMV glycoproteins in macrophages. The genes m152 and m06 both encode type I transmembrane glycoproteins which interfere with the MHC class I pathway of antigen presentation. While transcription of m152 is initiated in MCMV-infected fibroblasts at 2 h p.i. and declines after 12 h p.i. (58), m06 belongs to a later set of E genes, reaching maximum expression 3 to 6 h p.i., and is produced at high levels throughout the replication cycle (44). As a marker for E gene expression, the e1-encoded phosphoproteins of 36 and 38 kDa (6) which are expressed simultaneously with m152/gp40, were precipitated from lysates of metabolically labeled J774 macrophages and MEF at 12, 24, and 36 h p.i. (Fig. 5). This transcription unit is under control of the same IE genes as m152. In J774 cells, E1 products were detected in both fibroblasts and macrophages, and at later times in the replication cycle E1 expression in J774 surpassed that seen in MEF (Fig. 5). To compare synthesis of m152/gp40 and m06/gp48 during the course of infection, lysates were subjected to immunoprecipitation with antibodies specific for gp40 and gp48, respectively. Figure 5 shows that m152/gp40 was synthesized in J774 macrophages with a temporal kinetics and in a similar quantity as in fibroblasts. Similarly, expression levels of m06/gp48 in J774 compared to MEF were indistinguishable (Fig. 5), and in both cell types gp48-MHC I complexes were formed. The data suggested that expression of m152 and m06 in macrophages is as abundant as in MEF and that a lack of synthesis of the inhibitory proteins cannot explain antigen presentation in macrophages.

View larger version (53K): [in this window] [in a new window]   FIG. 5.   Biosynthesis of MHC I-subversive MCMV glycoproteins in infected MEF and J774 macrophages. Cells were mock infected or infected with MCMV MS94.4 at an MOI of 3 for various times as indicated and pulse labeled with [35S]methionine for 1 h. Cell lysates were split and immunoprecipitated with (A) MAb 20/234/28, recognizing the 36- and 38-kDa E1 proteins, (B) rabbit antibodies recognizing gp37/40 molecules, or (C) rabbit antibodies specific for gp48. Immunoprecipitates were separated by SDS-10% PAGE.

MHC class I complexes reach the cell surface in MCMV-infected macrophages. At the onset of E gene expression, newly assembled MHC I complexes fail to leave the ERGIC of MCMV-infected fibroblasts (11). This block to MHC I transport is mediated by m152/gp40 (58). MHC I molecule maturation can be monitored by MHC I oligosaccharide chain processing from endo H-sensitive to endo H-resistant forms which are generated in the medial-Golgi compartment. Accordingly, impaired MHC I transport is reflected by the reduction in endo H-resistant MHC I molecules. We therefore determined synthesis of H-2L^d molecules and the appearance of endo H-resistant forms at 16 h p.i. In mock-infected cells, L^d molecules slowly matured to endo H-resistant forms within a chase period of more than 6 h (Fig. 6). In MCMV-infected MEF and J774 macrophages, acquisition of endo H resistance by L^d molecules was significantly delayed compared to mock controls (Fig. 6). Nevertheless, the relative amount of endo H-resistant forms after the chase period in infected cells was comparable to that determined in mock-infected cells (Fig. 6), suggesting a significant leakage of the transport block. After endo H digestion, coprecipitating proteins of about 34 and 32 kDa, corresponding to the endo H-sensitive forms of m06/gp48 and m04/gp34, respectively (29, 44), were observed, suggesting complex formation with L^d in infected MEF and J774. The data indicated that L^d complexes are eventually released from the transport block and reach the medial-Golgi compartment in MCMV-infected MEF as well as in macrophages.

View larger version (60K): [in this window] [in a new window]   FIG. 6.   MHC class I molecule maturation in MCMV-infected MEF and J774 macrophages. Cells were mock infected or infected with MCMV MS94.4 at an MOI of 3 for 16 h before being pulse labeled with [35S]methionine for 1 h and chased for the indicated times. Cell lysates were immunoprecipitated with MAb 28-14-8s, recognizing H-2Ld molecules. Half of the samples were treated with endo H prior to separation by 11 to 14% gradient SDS-PAGE. Ldr, endo H-resistant Ld molecules; Lds, endo H-sensitive Ld molecules; open arrowhead, endo H-sensitive MCMV gp48 band; solid arrowhead, endo H-sensitive MCMV gp34 band.

View larger version (29K): [in this window] [in a new window]   FIG. 7.   Detection of surface-expressed MHC I molecules on MCMV-infected macrophages and fibroblasts. J774 (left panel) and BALB.SV (right panel) fibroblasts were mock infected (dotted line) or infected with wildtype MCMV at an MOI of 5 (bold line) or MCMV m152 deletion mutant MC95.21 at an MOI of 5 (broken line) for 18 h. Before staining with MAb 30-5-7s, recognizing H-2 Ld, HB157 (isotype control), and anti-CD29, respectively, cells were incubated with MAb 2.4G2 to block Fc receptors. Bound antibodies were visualized by FITC-coupled goat anti-mouse IgG2a antibodies. As a negative control, cells were incubated with the second antibody alone (thin line).

BM-dependent and -independent processing of pp89 peptides in infected organs. The results described above predict that pp89 peptides are generated with high efficiency in MCMV-infected myelomonocytic cells such as macrophages. In contrast, peptides processed in infected fibroblasts and perhaps other stromal cells remain intracellular and are prevented from presentation on the cell surface. Accordingly, MCMV replication occurs primarily in stromal and parenchymal tissue cells (37). The detection of pp89 peptides in infected tissues is H-2L^d dependent and differs widely between organs (15). To address the relative significance of these compartments for the supply of antigenic viral peptides in vivo, BALB/c H-2^d(L^d+)  BALB/c H-2^dm2 (L^d) and BALB/c H-2^dm2 (L^d)  BALB/c H-2^d (L^d+) BM chimeras were constructed. The chimeric state was controlled by two-color cytofluorometric analysis of spleen cells (Table 1). At 48 h after MCMV infection, the spleen, lungs, and liver were subjected to peptide extraction. As expected, organs of MCMV-infected BALB/c H-2^dm2 (L^d) mice did not contain antigenic material (data not shown). In both chimeric settings, extracts of the spleen, lungs, and liver contained antigenic activity eluting in HPLC fraction 14 (Fig. 8). Serial dilution of fraction 14 reproduced a higher peptide yield in the spleen and the liver than in the lungs (15) but roughly similar amounts of the naturally processed YPHFMPTNL in both experimental settings. Due to residual host-derived L^d+ BM-derived cells in BALB/c H-2^dm2 (L^d)  BALB/c H-2^d (L^d+) chimeras (see Table 1), the yield of pp89 peptides cannot be attributed entirely to BM-independent stromal and parenchymal cells. In BALB/c H-2^d (L^d+)  BALB/c H-2^dm2 (L^d) chimeras, however, all pp89 peptides must originate from donor-derived hematopoietic cells. Since the peptide concentration in the organs of both experimental groups is comparable, this cellular compartment that includes macrophages must constitute the major source of antigenic viral peptides. Remarkably, this result was true not only for lymphoid tissue like the spleen, but also for solid organs like liver and lungs (Fig. 8). From these data, we infer that BM-dependent cells contribute significantly to the processing of viral peptides in parenchymal organs in vivo.

View this table: [in this window] [in a new window]   TABLE 1.   Establishment of chimerism in BALB/c H-2d (Ld+)  BALB/c H-2dm2 (Ld) and BALB/c H-2dm2 (Ldm)  BALB/c H-2d (Ld+) BM chimeras

View larger version (22K): [in this window] [in a new window]   FIG. 8.   BM-dependent and -independent processing of pp89 peptides in spleen, lungs, and liver of MCMV-infected BM chimeras. BM chimeras BALB/c  BALB/c-H-2dm2 and BALB/c-H-2dm2  BALB/c were infected with 106 PFU of SGV MCMV intraperitoneally. Spleen, lungs, and liver were removed 2 days p.i. and subjected to peptide extraction. HPLC fractions (, undiluted; , diluted 1:5; , diluted 1:25) were tested with pp89-specific CTL in a standard 51chromium release assay. The AcN gradient is indicated in the upper right panel.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Both MCMV and HCMV are endowed with a multitude of gene functions designed to downregulate the MHC class I-dependent antigen presentation pathway in infected cells (1, 2, 21, 26, 27, 35, 44, 55, 56, 58). In this report we document that this hallmark of CMV biology has a remarkable exception. Antigen presentation of pp89 peptides by macrophages occurs despite a significantly lower synthesis rate of the antigenic protein, which is compensated for by a higher processing rate of the antigenic peptide. Second, macrophages are exempt from effective MHC I downregulation. Peptide presentation is manifest despite the synthesis of the MHC I-subversive MCMV proteins. Macrophage-derived viral peptides dominate the immune response in vivo, since large amounts of peptide processing occur in BM-derived cells in CMV-infected parenchymal organs.

Our findings address several aspects of CMV biology. Previous studies have suggested a pivotal role for cells of the monocyte-macrophage lineage in acute and particularly in latent CMV infection. The permissiveness of macrophages for CMV replication in vitro (13, 17, 25) was confirmed by the detection of viral DNA and RNA in blood mononuclear phagocytes in vivo (8, 52). HCMV proteins representing all stages of permissive infection have been demonstrated in tissue macrophages in congenitally infected fetuses and immunodeficient adults (47, 48). Monocyte-derived macrophages infected in vitro exhibit a nonlytic type of infection and accumulate infectious virus in cytoplasmic vacuoles. The poor virus release from macrophages was interpreted as a mechanism for viral persistence (12). In vivo, MCMV exhibits a strong tropism to mononuclear phagocytes which may be used as vehicles mediating virus dissemination (8, 52). On the other hand, in macrophage-depleted mice, MCMV replication is strongly enhanced, suggesting that macrophages exert a protective effect, presumably by the induction of early cytokine responses of lymphocytes (17). Our findings explain these results in that macrophages rapidly take up MCMV virions in infected tissues and constitutively present viral peptides to activate cytotoxic and secretory antiviral CD8⁺ T-cell responses long before CMV virions are formed.

Infectious HCMV has been recovered in vitro from a CD14⁺ CD64⁺ macrophage expressing dendritic cell markers (51), supporting the notion that monocyte precursors harbor latent HCMV genomes in vivo (16, 30, 53). This implies also that HCMV utilizes a selective monocyte-derived macrophage differentiation pathway for virus production. If antigens are expressed either during random episodes of nonproductive reactivation in latency (23, 30, 33, 34, 57) or after progression to later stages of replication, productively CMV-infected monocytes/macrophages should expose themselves to an immediate cytotoxic T-cell attack much more efficiently than other cells, e.g., fibroblasts. This would purge the majority of virus-producing cells from the macrophage pool and decrease the load of CMV genomes in peripheral blood mononuclear cells, which is very low in healthy seropositive individuals (49, 53). The frequency of HCMV-specific circulating effector CTL in healthy seropositive donors as assessed by MHC tetramers is surprisingly high (P. A. H. Moss, personal communication), suggesting that the CD8⁺ T-cell response is repeatedly boosted. We propose that macrophages play a role in maintaining a high CMV-specific CTL precursor frequency. Accordingly, depletion of CD8⁺ lymphocytes is required for the immediate recurrence of latent MCMV infection in vivo (38).

Gamma interferon induces the formation of HCMV-permissive monocyte-derived macrophages (50). Thus, another scenario is conceivable in which peptide presentation to CD8⁺ T lymphocytes is used by the virus to increase in a paracrine fashion the pool of macrophages for infection. The potential differences between transcriptional programs of CMV during latency versus productive infection is not clear. Latently infected monocyte precursors that do not express pp89 differ from productively infected macrophages with regard to antigen presentation and thus avoid recognition by CD8⁺ T cells. How the pool of latently infected macrophages is maintained despite the activation/replication episodes remains to be investigated.

The MHC I-subversive glycoproteins of MCMV control the MHC I pathway in MEF but fail to downregulate MHC I molecules in infected macrophages. At present, the molecular basis for this difference is not yet clear, but the physiological pathway for MHC class I molecules differs between cell types as well (reviewed in reference 54). Macrophages stably expressing m152/gp40 or m06/gp48 could be instrumental to getting more insight into the MHC class I phenotype of macrophages during infection. On the other hand, CMV still does affect distinct immune functions of macrophages, including the expression of surface receptors like MHC class II and CD14 (12, 19, 24, 43). The resistance of macrophages to CMV-induced downregulation of MHC I contrasts not only with fibroblasts, but also with a variety of epithelial and stromal cell types which exhibit strongly reduced levels of MHC I even after semipermissive infection with HCMV (C. Benz, U. Reusch, W. Muranyi, W. Brune, and H. Hengel, unpublished data) and proves rather the exception than the rule. Macrophages belong to professional antigen-presenting cells, which are indispensable for the induction of antiviral cytotoxic T cells (46). Why does CMV then infect monocytes if these cells increase the virus's immunogenicity, at least after differentiation into macrophages? At first glance, it may appear advantageous to the virus to achieve a maximum of immune avoidance. However, if applied perfectly, this strategy would harm the host and reduce the length of coexistence with the virus.

Despite reduced protein synthesis, production of antigenic peptides is quantitatively similar in macrophages and fibroblasts. The pp89 protein is not part of the virion, and its detection requires de novo viral protein synthesis. This fact, along with the determination of pp89 peptides within 48 h p.i., excludes that cross-presentation (7) has biased our result. In a previous study, we have shown that the yield of pp89 peptides in MCMV-infected organs of BALB/c mice is not linked to the extent of virus replication but governed by gamma interferon (15). Here we extend this finding by demonstrating two principal cell sources for pp89 peptides. One is represented by nonhematopoietic stromal and parenchymal cells, while the other is formed by BM-derived cells, which include macrophages. Although productive MCMV infection occurs primarily in stromal and parenchymal tissue cells (37), both compartments contribute quantitatively to a similar degree to the pp89 peptide pool. However, only the peptide pool processed in macrophages is presented to CTL. The extraction of large amounts of pp89 peptides from BALB/c H-2^d (L^d+)  BALB/c H-2^dm2 (L^d) chimeras reflects efficient processing in cells of the myeloid lineage. Our finding also explains the unexpected high frequency of donor-derived CTL even in BM chimeras lacking the prevailing H-2 L^d molecule in recipient tissues (3). According to current concepts, peptides processed and presented by BM cells have an inherently high immunogenicity for naive T cells, while nonhematopoietic cells are unable to stimulate a primary CTL response (4, 46), but guide immune recognition of primed T-cell responses. In contrast to BM-derived professional antigen-presenting cells like macrophages and dendritic cells, stromal cells lack costimulatory and adhesion molecules required for T-cell induction. In addition, nonhematopoietic cells are unable to migrate to lymphoid organs for the initiation of immune responses (32). Therefore, CMV counteracts the effector rather than the induction phase of the antiviral CD8⁺ T-cell response.