Abstract
Murine CMV (MCMV), a β-herpesvirus, infects dendritic cells (DC) and impairs their function. The underlying events are poorly described. In this study, we identify MCMV m138 as the viral gene responsible for promoting the rapid disappearance of the costimulatory molecule B7-1 (CD80) from the cell surface of DC. This was unexpected, as m138 was previously identified as fcr-1, a putative virus-encoded FcR. m138 impaired the ability of DC to activate CD8+ T cells. Biochemical analysis and immunocytochemistry showed that m138 targets B7-1 in the secretory pathway and reroutes it to lysosomal associated membrane glycoprotein-1+ compartments. These results show a novel function for m138 in MCMV infection and identify the first viral protein to target B7-1.
Herpesviruses are large DNA viruses that establish latent infection in the host by actively limiting the immune response (reviewed in Ref. 1). To disable host cell immune strategies, murine CMV (MCMV)3 possesses a dsDNA genome of ∼230 kb that potentially encodes 170 predicted genes (2). The central core of the genome is conserved among all β-herpesviruses and contains genes essential for virus assembly and replication (3). In contrast, the genome termini contain >100 predicted genes that are dispensable for viral replication in vitro (4), but are considered critical for infection and immune modulation in vivo (5). Although for a small number of these genes, immune modulatory activities have been described, the majority possess unknown potential for novel immune evasion strategies.
CD8+ CTLs serve as the major immune effectors that control CMV infection in both humans (6, 7) and mice (8, 9). CMV-infected cells are targeted by TCR recognition of host cell MHC class I molecules displaying virus-derived peptide(s). Given the critical nature of the TCR:MHC class I-peptide interaction, it is attacked by CMV at multiple levels. Immune evasion proteins impair the MHC class I molecule by promoting its retention (10, 11) or degradation (12, 13, 14), by blocking the peptide-loading complex (15), or by preventing TCR recognition at the cell surface (16). Disruption of MHC class I Ag presentation has mostly been studied in non-APC, thereby mimicking the effector stage of recognition, in which an infected cell in the tissue must avoid an already primed host cell immune response. For the virus, it is equally important to target the initiation of events. This occurs in the secondary lymphoid organs in which circulating naive T cells are primed by dendritic cells (DC) expressing virus-derived peptide in the context of MHC class I. In addition to MHC class I-peptide complexes, DC display multiple costimulatory molecules that enhance TCR-mediated signaling and enable optimal T cell activation (17). Costimulatory molecule-deficient mice display impaired viral clearance in a number of viral infection models, illustrating the critical role for costimulation in eliciting antiviral immunity (reviewed in Ref. 18).
The B7 family of costimulatory molecules is the best described to date (reviewed in Ref. 19) and currently includes seven known members: B7-1 (CD80), B7-2 (CD86), inducible costimulatory molecule ligand (ICOSL), programmed death ligand (PD-L)1, PD-L2, B7H3, and B7H4, all of which are expressed by APC. Through interactions with their receptors on T cells, the B7 family members modulate TCR signaling. Impaired expression of B7 costimulatory molecules occurs upon infection with numerous viruses, including Kaposi’s sarcoma-associated herpesvirus (KSHV) (20), lymphocytic choriomeningitis virus (21), varicella-zoster virus (22), vaccinia (23), and HIV type 1 (HIV-1) (24). In most cases, neither the viral genes responsible, nor the mechanism of modulation, are known. Two exceptions are the targeting of B7-2 by K5, an E3 ubiquitin ligase expressed by KSHV (20), and modB7-2 encoded by MCMV (25). Consequently, the removal of specific costimulatory molecules from the DC surface affords a particularly potent strategy of viral-mediated immune evasion. In this study, we investigated the ability of MCMV to modulate B7 costimulatory molecule expression in DC. We have identified the first virus protein to target B7-1 and have identified a novel function for the previously described MCMV fcr-1 gene, m138.
Materials and Methods
Cells
The DC line D2SC/1 (provided by P. Ricciardi-Castagnoli (University of Milano-Bicocca, Italy) and referred to as DC2 throughout this work) (26), DC2.4 (27, 28), M210B4 (29), Chinese hamster ovary (CHO) cells (American Type Culture Collection), CHO-B7-1 (provided by A. Sharpe, Harvard University, Boston, MA) (30), NIH 3T3 (American Type Culture Collection), B3Z (provided by N. Shastri, University of California, Berkeley, CA) (31), and BALB/c murine embryonic fibroblasts were grown, as previously described. Primary bone marrow-derived DC (BMDC) were generated from bone marrow harvested from C57BL/6 mice (The Jackson Laboratory). Mice were maintained at the Whitehead Institute for Biomedical Research Animal Facility, according to institutional guidelines. Bone marrow was cultured in the presence of 200 μg/ml GM-CSF (PeproTech) and 20 μg/ml IL-4 (PeproTech) for 4–5 days.
DNA constructs
The N-terminal hemagglutinin (HA) construct (provided by B. Lilley, Harvard University, Boston, MA) contained the H-2Kb signal sequence, followed by the HA tag in the pcDNA 3.1 expression vector (Invitrogen Life Technologies). The m138 gene was introduced into the EcoRI/NotI site using the following primer sequences: 5′-CGCAATTGCATCAATTACCTGCGTGC-3′ and 5′-AAGAAGCGGCCGCTTAGGCGTAGTCGGGGAC-3′. GFP-tagged B7-1 was generated using B7-1 cDNA (provided by G. Freeman). The B7-1 gene was cloned into the XhoI/SalI site of the pEGFP-N1 expression vector (BD Clontech) by PCR. The primer sequences used were as follows: 5′-CCGCTCGAGACCACCATGGCTTGCAATTGTCAG-3′ and 5′-CGCGTCGACGCAAGGAAGACGGTCTGTTC-3′. m138 truncation proteins were generated by the following primer combinations: tailm138, 5′-ACGGCCTTCCGCGCTGACGCG-3′; CD4TMtailm138, 5′-CCAATGGCCCTGATTGTGCTG-3′ and 5′-CTGGCGGCCGCTCAAATGGGGCTACATGTC-3′. The human CD4 transmembrane and tail domain (aa 396–458) was generated by PCR using the primers 5′-CCAATGGCCCTGATTGTGCTG-3′ and 5′-CTGGCGGCCGCTCAAATGGGGCTACATGTC-3′. Cells were transiently transfected using Fugene-6 (Roche).
Viruses
MCMV were propagated in M210B4 cells. MCMV infections were performed by centrifugation at 2000 rpm for 30 min at 25°C. Supernatant was harvested from DC2 infected with wild-type MCMV 16 h postinfection and filtered through a 0.45-μm syringe filter (Corning Glass). The pMIG retrovirus, containing an internal ribosomal entry site followed by GFP, was used to generate stable cell lines. The m138 gene was cloned into the BglI/EcoRI site by PCR using the primer sequences 5′-GAGGATCCACCACCATGGCGCCTTCGACGCTG-3′ and 5′-CCCGCAATTGTTACGT GTGACGTACGC-3′.
Mutagenesis of MCMV genome
Mutagenesis was performed by homologous recombination between a linear DNA PCR fragment and the MCMV-BAC pSM3fr in Escherichia coli, as previously described (4). To delete MCMV genes, linear DNA fragments containing the kanamycin resistance gene flanked by regions of homology to the MCMV genome were generated. The specific primers used to generate Δm138MCMV were as follows: 5′-GGGTCAGTCATTAGTAAGTGTTAGTAGTCGATGACTTGAGCGTCGTGGAATGCCTTCGAATTC-3′ and 5′-CTCAAGTCGCCATCATCTCTCGGTCGGCAGAGCCGAGGCGACAAGGACGACGACGACAAGTAAG-3′. The loss of m138 was confirmed by restriction enzyme pattern analysis of the rMCMV genome with HindIII and by immunoblotting for m138 protein in infected cell lysates.
Flow cytometry
Cells were harvested and stained with the following Abs: anti-B7-1 PE (16-10A1; BD Biosciences), anti-B7-2 PE (GL1; BD Biosciences), anti-ICOSL PE (HK5.3; eBioscience), anti-PD-L1 (MIH5; eBioscience), anti-PD-L2 (TY25; eBioscience), anti-CD11c (HL3; BD Biosciences), and anti-mouse IgG2,κ-PE (BD Biosciences). When DC were analyzed, cells were incubated with purified anti-mouse CD16/CD32 (BD Pharmingen) before cell surface staining. To detect MCMV infection, cells were fixed with 0.5% paraformaldehyde (EM grade; Electron Microscopy Sciences) and permeabilized with 0.5% saponin (Sigma-Aldrich). Cells were stained with mouse anti-pp89 (provided by S. Jonjic, University of Rijeka, Croatia), followed by goat anti-mouse IgG (H + L)-Alexa 660 (Molecular Probes). Detection of Ig binding at the cell surface was undertaken by incubation of virus-infected cells with mouse IgG (Sigma-Aldrich) for 30 min, followed by staining with anti-mouse IgG2,κ-PE (BD Biosciences). Flow cytometry analysis was performed using a FACSCalibur flow cytometer (BD Biosciences) and analyzed with CellQuest software (BD Biosciences).
Immunoblotting
Cells were harvested and lysed in 1% SDS. Protein quantity was determined by a bicinchonic acid protein assay (Pierce). Immunoblotting was performed by standard techniques. Anti-m138 rabbit polyclonal serum was generated by immunization with a mixture of three peptides (amino acid residues 115–127, 213–225, and 499–513) (Cocalico). Proteins were detected using rabbit anti-B7-1 (Abcam), mouse anti-β-actin (Sigma- Aldrich), anti-mouse IgG-HRP (Southern Biotechnology Associates), and anti-rabbit IgG-HRP (Southern Biotechnology Associates).
Radio or biotin labeling and immunoprecipitation
For radiolabeling, cells were starved in complete medium lacking cysteine and methionine, supplemented with 0.5 μCi/ml [35S]methionine/cysteine. Chase periods were performed with medium containing 2.5 mM methionine and 0.5 mM cysteine. For surface biotin labeling, cells were washed with PBS and labeled with 0.2 mg/ml sulfo-NHS- LC-biotin (Pierce) in PBS. Following labeling, cells were washed with PBS/10 mM glycine. Labeled cells were lysed in 1% detergent (Nonidet P-40 or digitonin). Proteins were immunoprecipitated with specific Abs and protein A-agarose (RepliGen). Immunoprecipitations were performed from equivalent amounts of radioactive protein following the pulse label and preclearing of cell lysates. Radioactive counts were determined by trichloroacetic acid precipitation of 10 μl of total lysate. Endoglycosidase H (Endo H; New England Biolabs) digestion was performed by incubation at 37°C for 1 h. Anti-GFP rabbit polyclonal serum (Abcam), anti-HA rabbit polyclonal serum (12CA5), and mouse Ig (Sigma-Aldrich) were used for immunoprecipitation. Radioactive polypeptides were visualized by fluorography and exposure to Kodak X-OMAT films.
Immunostaining and fluorescent confocal microscopy
Cells were grown on coverslips and fixed with 4% paraformaldehyde (EM grade; Electron Microscopy Sciences) in PBS. Cells were permeabilized with 1% Triton X-100. The following Abs were used: anti-HA (3F10; Roche), anti-B7-1 (16-10A1; BD Biosciences), anti-lysosomal-associated membrane glycoprotein-1 (LAMP-1) (Abcam), anti-early endosomal Ag-1 (EEA-1) (Abcam), anti-protein disulfide isomerase (PDI; Abcam), anti-hamster IgG-Cy3 (Jackson ImmunoResearch Laboratories), and anti-rabbit Ig-Alexa 488, 647 (Molecular Probes). To label transferrin (Tfn)-containing compartments, cells were serum starved and incubated with Tfn-Alexa 594 (Molecular Probes). Cells were imaged with a spinning disk confocal microscope.
RT-PCR
RNA extraction was performed (Qiagen) and cDNA generated (Invitrogen Life Technologies). RT-PCR was performed using B7-1 primers (32).
Ag presentation assay
DC2.4 were pulsed with 1.1-μm-diameter latex beads (Sigma-Aldrich) or 0.89-μm diameter flash red fluorescent beads (Bangs Laboratories) previously adsorbed with OVA (10 mg/ml) for 6 h. Cells were fixed with 0.5% paraformaldehyde (EM grade; Electron Microscopy Sciences) and washed with PBS. Indicated cultures were incubated with 25 μg/ml anti-B7-1 (BD Pharmingen). A total of 1 × 106 B3Z T cells was added to each culture and harvested 24 h later. B3Z T cells are CD8+ OVA-specific T cells that express the lacZ gene under the transcriptional control of NF-AT (31). Expression of lacZ was assessed by the chromogenic β-galactosidase enzyme assay (Promega).
Results
MCMV modulates cell surface expression of B7 costimulatory molecules
To study MCMV infection of DC, we used D2SC/1 (DC2) (26), a DC line that can readily be infected with MCMV. The impact of MCMV infection on the cell surface expression of B7 costimulatory molecules was examined by flow cytometry (Fig. 1⇓A). Uninfected DC2 expressed high levels of B7-1 and B7-2 and low levels of PD-L1, PD-L2, and ICOSL (data not shown). The extent of MCMV infection of DC2 was determined by intracellular staining for pp89, an immediate early MCMV protein. In all cases, MCMV infection at a multiplicity of infection (MOI) = 10 resulted in >50% of DC2 expressing the pp89 protein (data not shown). Upon MCMV infection, the expression of both PD-L1 and PD-L2 was up-regulated in MCMV-infected DC2, most notably PD-L1, which was expressed at 2-fold higher levels compared with uninfected cells. The low expression of ICOSL remained unaltered in MCMV-infected DC2. The most striking effect of MCMV infection was the specific targeting of the B7-1 and B7-2 molecules, which could no longer be detected at the cell surface 16 h following infection. We focused on MCMV interference with B7-1, given that there are no known viral proteins that interfere with its expression. Examination of uninfected cells and MCMV-infected DC2 at various time points following infection showed that cell surface expression of B7-1 is rapidly lost. B7-1 was completely absent from the cell surface as early as 6 h following infection (Fig. 1⇓B).
MCMV modulates cell surface expression of B7 costimulatory molecules. A, DC2 were infected with MCMV (MOI = 10) or left uninfected and examined 16 h postinfection. The graph displays the mean percentage of the mean fluorescence intensity (MFI) of infected (pp89+) cells relative to uninfected cells ± SEM. B, DC2 were infected with MCMV (MOI = 10). The graph displays the mean percentage MFI of B7-1 expressed by pp89+ MCMV-infected cells or uninfected cells relative to that at 0 h postinfection. Background fluorescence in the absence of Ab (dashed line). C, DC2 were infected with wild-type MCMV (MOI = 10), UV-inactivated MCMV, or left uninfected and examined 16 h postinfection. For wild-type MCMV-infected cells, the histograms are gated on pp89+ DC2. D, DC2 were infected with wild-type MCMV (MOI = 10), treated with supernatant harvested from MCMV-infected DC2, or left uninfected and examined 16 h postinfection. For wild-type MCMV-infected cells, the histograms are gated on pp89+ DC2.
Because viral infection can alter the cell surface expression of proteins via secondary effects, we determined whether a specific MCMV gene was responsible for B7-1 down-modulation. DC2 were exposed to either UV-inactivated MCMV or supernatant harvested from live MCMV-infected DC2. In both cases, the exclusion of active virus was confirmed by the complete absence of pp89-expressing cells (data not shown). Although up-regulation of PD-L1 was promoted by exposure to both UV-inactivated MCMV and MCMV-infected DC2 supernatant, the loss of B7-1 from the cell surface required direct infection with live MCMV (Fig. 1⇑, C and D). Therefore, the up-regulation of PD-L1 promoted by MCMV does not require live virus and might be attributable to cytokines produced by DC2 in response to MCMV exposure. In contrast, active virus-mediated mechanisms are responsible for down-modulation of B7-1.
MCMV m138 is responsible for down-modulation of B7-1
To determine the identity of the MCMV gene responsible for modulation of B7-1, we used six MCMV deletion mutants (Δm01–22MCMV, Δm32–36MCMV, Δm37–43MCMV, Δm128- 133MCMV, Δm128–139MCMV, and Δm01–17 + m144–158 + m159–170MCMV) that target regions that are nonessential for viral replication in vitro (4). In total, a combined 73 individual MCMV genes were examined for their ability to manipulate B7-1 expression (data not shown). Using MCMV deletion mutants spanning progressively smaller genomic regions (Δm134–136MCMV, Δm137–139MCMV, Δm137MCMV, Δm138MCMV, Δm139MCMV) and by overexpression of the individual genes (data not shown), the open reading frame responsible for B7-1 modulation was identified as m138 or fcr-1, a previously characterized MCMV-encoded FcR (33). Introduction of the full-length m138 gene into DC2, in the absence of any other MCMV gene, caused the down-modulation of B7-1 from the cell surface, but not that of B7-2 (Fig. 2⇓A). Expression of m138 promoted loss of B7-1 from the cell surface in all cell lines examined. To demonstrate that m138 specifically down-regulated B7-1 during MCMV infection, a Δm138 MCMV deletion mutant virus was generated. Immunoblotting with m138 antiserum confirmed the absence of the m138 protein in Δm138 MCMV-infected cell lysates (Fig. 2⇓B). Infection of DC2 with Δm138 MCMV at MOI = 10 generated an equivalent number of pp89-expressing cells, as did wild-type MCMV (data not shown). In contrast to wild-type virus, Δm138 MCMV was incapable of down-regulating B7-1 from the cell surface, whereas B7-2 expression was abrogated similar to infection with wild-type virus (Fig. 2⇓C).
MCMV m138 is responsible for down-modulation of B7-1. A, DC2 were transduced with empty vector-pMIG retrovirus or pMIG retrovirus-expressing m138 and stained with anti-B7-1, anti-B7-2, or an isotype control (IgG2, κ) Ab. The histograms are gated for GFP+ cells representing cells transduced with the pMIG retrovirus. The data are representative of five independent experiments. B, 3T3 fibroblasts were infected with wild-type MCMV (MOI = 1), Δm138 MCMV (MOI = 1), or left uninfected. A total of 10 μg of 1% SDS lysates was examined by immunoblotting for m138 or β-actin protein using anti-m138 or β-actin antiserum. C, DC2 were infected with wild-type MCMV (MOI = 10) or Δm138 MCMV (MOI = 10) or left uninfected. Sixteen hours postinfection, B7-1 or B7-2 cell surface expression was examined by flow cytometry. The histograms are gated on pp89+, MCMV-infected DC2. The data are representative of three independent experiments. D, Primary BMDC (day 4) were infected with wild-type MCMV (MOI = 10) or Δm138 MCMV (MOI = 10) or left uninfected and examined 16 h postinfection. The histogram is gated on CD11c+ cells (for uninfected cells) or pp89+, MCMV-infected CD11c+ cells (for wild-type and Δm138 MCMV-infected cells). The graph summarizes the percentage of mean fluorescence intensity (MFI) of B7-1 expression relative to uninfected BMDC (mean ± SEM). The data are representative of three individual experiments.
To examine the impact of m138 on the expression of B7-1 in primary DC, BMDC were infected with wild-type MCMV (MOI = 10) or Δm138 MCMV (MOI = 10) (Fig. 2⇑D). MCMV infection of BMDC at day 4 did not alter the number of CD11c+ cells generated, and equivalent levels of infection, as assessed by staining for pp89, were observed with both viruses (data not shown). Infection with wild-type MCMV caused down-regulation of B7-1 in CD11c+, pp89+ cells to an extent similar to that observed for the cell lines examined. Infection with Δm138 MCMV did not attenuate B7-1 expression, confirming that m138 is the MCMV gene responsible for B7-1 modulation. Δm138 MCMV promoted the generation of a B7-1high population, as the BMDC were activated in response to virus exposure. Therefore, m138 was identified as the MCMV gene that specifically and independently interferes with the cell surface expression of B7-1 and does so in primary BMDC.
MCMV m138 impairs the ability of DC to activate T cells
The costimulatory signal provided by B7-1 engagement of CD28 is required for optimal T cell responses (34). Therefore, the impact of m138 expression on the ability of DC to stimulate T cell responses was examined. To do so, an assay was designed in which m138 could be examined in the absence of the numerous immune evasion genes present in the MCMV genome. This excluded the direct infection of DC with virus. Instead, m138 was introduced into the DC line DC2.4 (H-2b) using the pMIG retrovirus. Stable cell lines expressing either empty vector or m138 were pulsed with OVA-coated beads, and their ability to promote stimulation of OVA-specific B3Z CD8+ T cells was examined. This is an in vitro assay of cross-presentation and was determined to be dose dependent, Ag specific, and proteasome dependent (data not shown). Cells loaded with a high dose of the OVA-derived class I peptide, SIINFEKL, served as positive controls for B3Z T cell activation. The expression of m138 by DC2.4 did not alter the H-2Kb levels at the cell surface (Fig. 3⇓A), the phagocytosis of OVA beads (Fig. 3⇓B), or the loading of H-2Kb with the SIINFEKL peptide, as determined by staining with the anti-H-2Kb-SIINFEKL Ab 25D1.16 (Fig. 3⇓C). In the presence of m138, the response to SIINFEKL-loaded DC2.4 was not altered due to the strong signal provided by high dose of peptide that overcomes any requirement for B7-1-mediated costimulation. In contrast, m138 significantly impaired the ability of DC2.4 to promote the activation of B3Z T cells in response to OVA beads (Fig. 3⇓D). This response was greater than the suppression observed when the assay was performed in the presence of anti-B7-1 blocking Ab. In this case, B3Z T cell activation was 62 ± 3% (mean ± SEM) of activation observed in the absence of Ab. Therefore, the expression of m138 and the consequent loss of B7-1 from the cell surface impact the ability of DC to promote optimal T cell activation.
MCMV m138 impairs the ability of DC to activate T cells. DC2.4 transduced with empty vector pMIG (DC2.4) or pMIG-m138 (DC2.4 + m138) were examined for the following: A, cell surface levels of B7-1 or H-2Kb; B, uptake of OVA-fluorescent beads; or C, cell surface levels of H-2Kb-SIINFEKL in response to loading with 1 μM SIINFEKL, pulsing with OVA beads, or in the absence of Ag. D, B3Z activation by Ag-expressing DC2.4 was assessed upon expression of m138 or in the presence of an anti-B7-1 Ab. Activation was assessed using a chromogenic assay for lacZ expression. The data are presented as the percentage of B3Z activation observed in response to Ag presented by DC2.4 transduced with empty vector (mean ± SEM). The graphs summarize four independent experiments.
Characterization of MCMV m138 protein
m138 encodes a 569-aa type I glycoprotein (2). To investigate the expression kinetics of m138 protein during MCMV infection, DC2 were infected with wild-type MCMV (MOI = 10) and harvested at various time points. m138, detected by immunoblotting as an 80-kDa polypeptide, was observed as early as 2 h postinfection and persisted throughout the infectious cycle (Fig. 4⇓A). To characterize the biosynthesis of m138, an N-terminal HA-tagged m138 protein was generated for pulse-chase analysis of transiently transfected cells. HA-m138 was capable of B7-1 down-modulation similar to untagged and C-terminal HA-tagged m138 protein (data not shown). m138 was immunoprecipitated from radiolabeled CHO lysates and digested with Endo H to determine the maturation of its associated glycans. The m138 protein was expressed as a 75- to 80-kDa polypeptide that matured to a slower migrating polypeptide of ∼85 kDa (Fig. 4⇓B). The maturation pattern was consistent with m138 possessing N-linked glycans that remained Endo H sensitive. Even following up to 4 h of chase, m138 remained Endo H sensitive (data not shown). The increase in m.w., together with the observed heterogeneity, is consistent with O-linked glycosylation, predicted to occur in the proline-glutamate-serine-threonine domain of m138.
Characterization of the MCMV m138 protein. A, DC2 were infected with MCMV (MOI = 10), and the expression of m138 in 60 μg of 1% SDS lysates was examined by immunoblotting with anti-m138 antiserum. Detection of β-actin served as a loading control. B, CHO-expressing HA-m138 were radiolabeled for 15 min and chased for 45 and 90 min. m138 was recovered from 1% Nonidet P-40 lysates by immunoprecipitation (anti-HA) and digested with Endo H, where indicated. m138-associated glycans are designated Endo H sensitive (S) or O linked (O). C, CHO-expressing HA-m138 were radiolabeled for 15 min and subject to 1% Nonidet P-40 lysis. m138 was recovered by anti-HA immunoprecipitation or using 10 μg of mouse Ig. D, CHO-expressing B7-1-GFP, B7-1-GFP, plus HA-m138 or HA-m138 were surface biotinylated for 30 min. B7-1 and m138 were recovered from 1% Nonidet P-40 lysates by immunoprecipitation with anti-GFP and anti-HA Abs, respectively. Biotinylated proteins were detected by blotting with streptavidin-HRP. The membrane was stripped and reblotted using anti-m138 antiserum. E, DC2 cells were not infected or infected with wild-type MCMV (MOI = 10) or Δm138 MCMV (MOI = 10) and incubated with mouse Ig. Binding of Ig at the cell surface was detected by anti-mouse Ig Ab. The histograms are gated on pp89+ cells. The data are representative of two independent experiments.
m138 was previously identified in a screen for MCMV proteins that bind Ig (33). We also examined this function for m138. HA-m138 was recovered from radiolabeled CHO lysates either by anti-HA or mouse Ig immunoprecipitation. Consistent with previous findings, we also observed m138 binding to mouse Ig (Fig. 4⇑C). Given this, we examined the putative FcR function of m138, by assessing expression of m138 at the cell surface. In the absence of an anti-m138 Ab for flow cytometry, we investigated this by cell surface biotinylation of transiently transfected CHO cells expressing m138 alone, or cells expressing B7-1 or B7-1 plus m138 (serving as positive and negative controls for cell surface access). The m138 or B7-1 proteins were recovered by immunoprecipitation, and the presence of the biotin tag, acquired only upon cell surface display, was detected by streptavidin blotting. Surface biotinylation was equivalent in all cases, as observed by immunoblotting of total unfractionated lysate (data not shown). As expected, in the absence of m138, B7-1 readily accessed the cell surface, as inferred by the detection of biotinylated B7-1. This was not the case in the presence of m138, consistent with the flow cytometry data. m138 did not gain access to the cell surface at detectable levels (Fig. 4⇑D). Its expression and recovery were verified by immunoblotting with anti-m138 antiserum. Finally, we assessed the impact of m138 expression upon the binding of Ig at the cell surface of MCMV-infected cells. The capacity of MCMV-infected cells to bind cell surface Ig was not impaired upon infection with Δm138 MCMV (Fig. 4⇑E).
Fate of B7-1 in the presence of MCMV m138
The fate of B7-1 upon MCMV infection, and specifically the contribution of m138 to the behavior of B7-1, was investigated. First, we examined whether the failure to detect B7-1 at the cell surface was due to MCMV-mediated inhibition of B7-1 transcription. DC2 were infected with MCMV-GFP, and GFP-positive cells were isolated by flow cytometry. RT-PCR of RNA extracted from CHO, CHO-B7-1, uninfected DC2, or MCMV-infected DC2 showed the presence of a B7-1-specific product amplified from CHO-B7-1, DC2, and MCMV-infected DC2 (Fig. 5⇓A). Therefore, MCMV infection does not shut off B7-1 gene transcription.
The fate of B7-1 in the presence of MCMV m138. A, RT-PCR analysis of B7-1 or β-actin transcription in CHO, CHO-B7-1, DC2, and MCMV-infected DC2. B, CHO-expressing B7-1-GFP in the presence or absence of m138 were radiolabeled for 15 min and chased for 45 and 90 min. B7-1-GFP was recovered from 1% Nonidet P-40 lysates by immunoprecipitation (anti-GFP) and digested with Endo H, where indicated. The B7-1-associated glycans are designated Endo H sensitive (S) or Endo H resistant (R). C, CHO-B7-1 or CHO-B7-1-m138 were subject to 1% SDS lysis, and 10 μg of lysate was examined for the expression of B7-1 or β-actin by immunoblotting.
To assess where in the course of its biosynthetic maturation m138 targets B7-1, pulse-chase analysis, in conjunction with Endo H digestion, was performed. GFP-tagged B7-1 was used to facilitate immunoprecipitation of B7-1 with anti-GFP serum. The GFP tag did not interfere with m138 down-modulation of B7-1 expression (data not shown). CHO cells were used, rather than DC2, given that the level of B7-1 expression by DC2 was insufficient for the methodology undertaken. Analysis was performed using transiently transfected cells. In the absence of m138, B7-1 acquired Endo H-resistant complex oligosaccharides as the protein matured (Fig. 5⇑B, top panel). A shift in molecular mass of ∼24 kDa was observed, consistent with B7-1 containing eight potential N-linked glycans. In contrast, in the presence of m138, B7-1 glycan maturation was not observed (Fig. 5⇑B, bottom panel). This suggests that m138 targets B7-1 early in the secretory pathway, before complex oligosaccharide acquisition that occurs in the trans-Golgi.
The ultimate fate of the B7-1 protein in the presence of m138 was examined by generating a CHO stable cell line that expressed both B7-1 and m138. Immunoblotting of cell lysates for B7-1 expression showed significantly reduced detection of B7-1 in the presence of m138 (Fig. 5⇑C).
MCMV m138 promotes mislocalization of B7-1 to LAMP-1+ compartments
Confocal microscopy analysis was performed to examine the cellular localization of m138, and B7-1 in the presence of m138. In the absence of m138, B7-1 is localized at the cell surface, as expected. In contrast, m138 promoted B7-1 mislocalization to intracellular vesicles (Fig. 6⇓A). To identify the specific cellular compartment to which B7-1 was mislocalized, costaining with markers of the endosomal or lysosomal pathway was performed. The m138-mediated accumulation of B7-1 occurred in organelles that were EEA-1 negative (Fig. 6⇓B, left panel). In addition, the compartments lacked fluorescently labeled Tfn that was added to intact cells to allow visualization of TfnR-positive early endosomes (Fig. 6⇓B, middle panel). Therefore, B7-1 is not mislocalized in early, recycling endosomes. In contrast, vesicles containing B7-1 costained with LAMP-1 (Fig. 6⇓B, right panel).
MCMV m138 promotes mislocalization of B7-1 to LAMP-1+ compartments. Confocal microscopy analysis of CHO expressing: A, B7-1-GFP, in the presence or absence of m138; B, B7-1-GFP and HA-m138, stained with anti-EEA-1, LAMP-1 Abs, or pulsed with Tfn-Alexa 594 for 30 min; C, HA-m138, stained with anti-HA Ab; D, HA-m138, stained with anti-HA and anti-PDI Ab; E, B7-1 and HA-m138, stained with anti-HA Ab; F, B7-1-GFP and HA-m138, stained with anti-HA and LAMP-1 Ab.
The cellular localization of m138 was also examined by confocal microscopy. Analysis of the distribution of m138 by immunofluorescence showed its localization in large punctate vesicles (Fig. 6⇑C), similar to the images presented by Thale et al. (33). m138 was detected in intracellular compartments together with markers of the ER: PDI (Fig. 6⇑D) and calnexin (data not shown), in addition to the lysosomal marker LAMP-1 (Fig. 6⇑F). Given the similar localization pattern of m138, and B7-1 in the presence of m138, we examined a possible colocalization of the two proteins by immunofluorescence microscopy. Colocalization of m138 and B7-1 was indeed observed (Fig. 6⇑E). Colocalization of m138, B7-1, and LAMP-1 was also detected (Fig. 6⇑F).
m138 and B7-1 interact
A potential interaction between the m138 and B7-1 proteins was further investigated. Radiolabeled CHO cells transiently transfected with either B7-1, m138, or B7-1 plus m138 were lysed in 1% digitonin. The mild lysis conditions should favor the preservation of protein coassociation. The m138 protein was recovered by immunoprecipitation (Fig. 7⇓, left panel). The supernatant of the m138 immunoprecipitate (proteins not associated with m138), in addition to the pelleted material (m138 and its associated proteins), was subjected to anti-GFP immunoprecipitation to recover B7-1. In the supernatant, B7-1 was recovered, showing the protein to be present upon expression in the absence of m138 (Fig. 7⇓, middle panel). In the presence of m138, some B7-1 is recovered from the supernatant, demonstrating the presence of a fraction of B7-1 that does not associate with m138. m138 is also detected in the B7-1 immunoprecipitation due to its capacity to bind Ig (in this case, the anti-GFP Ab used to immunoprecipitate B7-1). For the pelleted material, we recovered a protein of the same m.w. as immature B7-1, but only when m138 was present (Fig. 7⇓, right panel). The data therefore indicate that the mechanism of m138-mediated down-modulation of B7-1 occurs via an interaction between the m138 and B7-1 proteins.
m138 and B7-1 interact. CHO expressing B7-1-GFP (1), HA-m138 (2), or B7-1-GFP plus HA-m138 (3) were radiolabeled for 60 min. HA-m138 was immunoprecipitated (IP) from 1% digitonin lysates using anti-HA Ab. Both the recovered immunoprecipitate (pellet) and the supernatant (S/N) were subjected to B7-1 immunoprecipitation (re-IP) using anti-GFP Ab.
The cytoplasmic tail and transmembrane domain of MCMV m138 are not required for modulation of B7-1 expression
Many proteins involved in lysosomal trafficking pathways possess consensus-sorting motifs in their cytoplasmic tails (35). Inspection of the amino acid composition of m138 did not reveal any obvious sorting motifs. Therefore, to examine the role of the cytoplasmic tail (aa 555–569) and/or the transmembrane domain (aa 534–554) of m138, we generated N-terminal HA-tagged m138 truncation mutant proteins (Fig. 8⇓A): 1) Δtailm138, lacking the 15 most C-terminal amino acids, and 2) ΔCD4TMtailm138, in which the m138 transmembrane region and tail were replaced with human CD4 transmembrane and tail. Expression of the proteins was confirmed by immunoblotting for HA (data not shown). Δtailm138 or ΔCD4TMtailm138 (Fig. 8⇓B) promoted the loss of B7-1 from the cell surface of CHO-B7-1 cells, equivalent to full-length m138. The down-modulation was not as extensive as that observed for DC2, presumably due to the high levels of B7-1 protein expressed by CHO-B7-1. Regardless, the mutant m138 proteins behaved similarly to full-length m138. To further eliminate a role for the cytoplasmic tail in this response, we examined the mislocalization of B7-1 in the presence of Δtailm138 by confocal microscopy. The cytoplasmic tail of m138 was not required to mislocalize B7-1 to intracellular punctate compartments (Fig. 8⇓C). Therefore, neither the cytoplasmic tail or the transmembrane domain of m138 was required for modulation of B7-1 expression.
The cytoplasmic tail and transmembrane (TM) domain of m138 are not required for modulation of B7-1 expression. A, Diagram of m138 truncation mutants. B, CHO-B7-1 were cotransfected with full-length m138, Δtailm138, or ΔCD4TMtailm138 and GFP. Forty-eight hours following transfection, the cells were stained with anti-B7-1 or an isotype control Ab, and assessed by flow cytometry. The histograms display GFP+ cotransfected cells and represent three independent experiments. C, Confocal microscopy analysis of CHO expressing B7-1-GFP, in the presence or absence of full-length m138 or Δtailm138.
Discussion
CMV actively suppresses the expression of costimulatory molecules in DC. This is an effective immune evasion strategy, given that infection correlates with an inability of CMV-infected DC to promote T cell activation (36, 37, 38, 39). At the cellular level, CMV promotes the loss of cell surface expression of CD40, CD54, CD83, B7-1, and B7-2 in various DC subsets (25, 36, 37, 38, 39, 40). Consistent with these studies, we also observed the abrogation of both B7-1 and B7-2 expression upon MCMV infection. In all cases of CMV-mediated immune suppression of DC function, the specific viral gene(s) responsible has not been identified. A possible exception is MCMV m147.5, which encodes a 23-kDa protein referred to as mod B7-2. Mod B7-2 is responsible for B7-2 loss from the cell surface of RAW 264.7 macrophages (25). Its role in DC, its mechanism of action, or the fate of B7-2 in its presence is unknown. In this study, we describe the specific interference of B7-1 cell surface expression by the MCMV protein m138 in both cell lines and primary DC. To our knowledge, this is the first demonstration of the impact of a viral immune evasion gene in primary DC.
Although MCMV modulates both B7-1 and B7-2 from the cell surface, m138 and m147.5 specifically interfere only with B7-1 or B7-2, respectively, and do so independently of each other. Given that B7-1 and B7-2 both bind the same T cell coreceptors, CD28 and CTLA4, this specificity is, at first glance, unexpected. Both B7-1 and B7-2 are type I transmembrane glycoproteins that possess a V-type and C-type Ig superfamily domain (41). Despite their functional overlap, however, they are only 25% identical in amino acid sequence, and consequently possess rather divergent structures (42, 43, 44, 45). Although B7-1 forms a homodimer at the cell surface (42, 43), B7-2 does so only upon receptor-induced clustering (45). In addition, B7-2 contains a longer cytoplasmic tail. Thus, the structural features that distinguish B7-1 and B7-2 most likely explain the presence of individual and specific modulators in the MCMV genome.
The MCMV gene responsible for B7-1 down-modulation was identified as m138. m138 was previously reported as the fcr-1 MCMV gene, encoding a putative FcR. It was postulated that FcR-1 (m138) expression at the MCMV-infected cell surface would prevent the recognition of MCMV-infected cells by circulating anti-MCMV Ab (33). This role was questioned, however, when the attenuated growth observed in vivo for Δm138 MCMV was not restored in B cell-deficient mice (46). Hence, the absence of Ab did not alter the course of infection with Δm138 MCMV. This prompted a re-evaluation of a role for m138 as an FcR that has remained elusive until now. Our experiments are also inconsistent with FcR function, given that m138 was not definitively shown to be expressed at the cell surface and its absence did not promote loss of Ig binding at the infected cell surface. Therefore, despite its putative role as an FcR, the clear consequences of m138 expression for the cell surface display of B7-1 suggest an attractive alternative for the major role of m138 in the immune evasion activities in MCMV infection. There is emerging evidence to support the notion that a single viral protein can exert several distinct immune evasion functions (47, 48), potentially in a cell type-specific manner. Of interest, while this manuscript was in preparation, a role for m138 in the down-modulation of NKG2D ligands MULT-1 and H60 was described (49). Consequently, the complex immune evasion strategies used by MCMV may explain the divergent functions of m138.
B7-1 is a critical costimulatory signal required for T cell immunity. B7-1-deficient APC display a significant reduction in their ability to promote T cell activation (34). Indeed, the loss of B7-1 at the cell surface, upon m138 expression, inhibited the ability of DC2.4 to promote optimal CD8+ T cell responses in vitro. We would therefore expect m138, in concert with other MCMV-encoded modulators of costimulatory and MHC class I molecule expression, to act together to effectively suppress antiviral CD8+ T cell immunity in vivo. Unexpectedly, attenuation of Δm138MCMV growth in vivo was not restored upon T cell (or NK cell) depletion (46). A direct examination of m138 function in vivo is complicated by its potential role in cell to cell spreading of the virus similar to that of its functional FcR homologues encoded by HSV (50, 51) and pseudorabies virus (52). m138 may also participate in other B7-1-mediated functions. In addition to a role in T cell responses, B7-1 can evoke, independently of CD28 engagement, signaling events in the APC itself (53, 54). For example, triggering of B7-1 blocks B cell proliferation and Ab production (55). Other examples include the induction of B7-1 in kidney podocytes, resulting in an increase in glomerular filtration and transient proteinuria (56) and the induction of B7-1 expression on keratinocytes by inflammatory stimuli (57). Therefore, the function of costimulatory molecules extends beyond their traditional role in T cell signaling and may explain the broader expression pattern of these molecules on cell types other than APC (reviewed in Ref. 19). m138 may consequently interfere with B7-1 function both in the context of an immune response and in potentially undiscovered roles that the virus must manipulate for successful propagation in vivo.
Host proteins suffer different fates in the presence of viral immune evasion proteins. Expression can be suppressed at the level of transcription, or the protein itself can be targeted via retention, inhibition of function, or enhanced degradation. In the case of B7-1, MCMV m138 does not shut off its transcription, but hijacks newly synthesized protein early during biosynthesis. This may occur in the ER given that m138 is localized both in the ER and in a lysosomal compartment. Unlike KSHV K5, which targets B7-2 already present at the cell surface (20), m138 recruits B7-1 early in the secretory pathway. The mechanism most likely requires an association between m138 and B7-1, and presumably involves the recruitment of other proteins. The involvement of accessory proteins such as the adaptor protein complexes or mannose 6-phosphate receptor remains to be investigated. m138 redirects B7-1 to a LAMP-1+ compartment. Of interest is that neither the m138 or B7-1 proteins acquire complex glycan modifications, despite their lysosomal localization. This is also the case for TLR9, which is recruited from the ER to an endosomal compartment without concomitant conversion of TLR9-associated glycans to Endo H resistance (58).
Mislocalization of host proteins to lysosomes is also promoted by other viral proteins, including MCMV m6/gp48 (14), HIV Nef (59), and HHV-7 U21 (60). For both m6/gp48 (14) and Nef (61, 62), lysosomal mislocalization is mediated by consensus-sorting signals in their cytoplasmic domains. These motifs recruit adaptor protein complexes that form part of the endocytic sorting machinery. In contrast, the mechanism of m138 function did not require the presence of its predicted transmembrane or cytoplasmic domains. Therefore, m138-mediated B7-1 mislocalization is reminiscent of the action of HHV-7 U21, in which its lumenal domain is sufficient for targeting of MHC class I to a lysosomal compartment (63).
In summary, we have identified the first viral protein that targets and disables the function of B7-1, a critical costimulatory molecule. MCMV m138, although possessing Ig-binding capacity, is the protein responsible for modulation of the cell surface expression of B7-1. Therefore, through the manipulation of B7-1, rather than acting as an FcR, m138 participates in the complex mechanisms of immune evasion activity that is ultimately responsible for the establishment of persistent herpesvirus infection. An understanding of the interference of B7-1 by a viral protein also provides insight into targeting and turning off B7-1 expression in contexts other than viral immunity.
Acknowledgments
We thank Drs. Howard Hang and Rene Maehr for critical reading of the manuscript, and Dr. Brendan Lilley for excellent scientific advice.
Disclosures
The authors have no financial conflict of interest.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 J.D.M. was supported by a CJ Martin Fellowship. M.W. was supported by a Longterm Fellowship from the Human Frontiers Science Organization. Y.M.K. was supported by a Leukemia & Lymphoma Society Fellowship. M.E.P. was funded by Fonds de la Recherche en Sante du Quebec.
↵2 Address correspondence and reprint requests to Dr. Hidde L. Ploegh, Whitehead Institute for Biomedical Research, 9 Cambridge Center, Cambridge, MA 02142-1479. E-mail address: ploegh{at}wi.mit.edu
↵3 Abbreviations used in this paper: MCMV, murine CMV; DC, dendritic cell; BMDC, bone marrow-derived DC; CHO, Chinese hamster ovary; DC2, D2SC/1 DC; EEA-1, early endosomal Ag-1; Endo H, endoglycosidase H; ER, endoplasmic reticulum; HA, hemagglutinin; ICOSL, inducible costimulatory molecule ligand; KSHV, Kaposi’s sarcoma-associated herpesvirus; LAMP-1, lysosomal-associated membrane glycoprotein-1; MOI, multiplicity of infection; PD-L, programmed death ligand; PDI, protein disulfide isomerase; Tfn, transferrin.
- Received April 28, 2006.
- Accepted September 18, 2006.
- Copyright © 2006 by The American Association of Immunologists