Selective Down-Regulation of the NKG2D Ligand H60 by Mouse Cytomegalovirus m155 Glycoprotein

Both human and mouse cytomegaloviruses (CMVs) encode proteinsthat inhibit the activation of NK cells by down-regulating cellularligands for the activating NK cell receptor NKG2D. Up to now,three ligands for the NKG2D receptor, named RAE-1, H60, andMULT-1, have been identified in mice. The resistance of mousestrains to murine CMV (MCMV) infection is determined by theirability to generate an effective NK cell response. The MCMVgene m152, a member of the m145 gene family, down-regulatesthe expression of RAE-1 in order to avoid NK cell control invivo. Here we report that the m155 gene, another member of them145 gene family, encodes a protein that interferes with theexpression of H60 on the surfaces of infected cells. Deletionof the m155 gene leads to an only partial restoration of H60expression on the cell surface, suggesting the involvement ofanother, so far unknown, viral inhibitor. In spite of this,an m155 deletion mutant virus shows NK cell-dependent attenuationin vivo. The acquisition of endo-ß-N-acetylglucosaminidaseH resistance and the preserved half-life of H60 in MCMV-infectedcells indicate that the m155-mediated effect must take placein a compartment after H60 exits from the ERGIC-cis-Golgi compartment.

INTRODUCTION

Top

Introduction

Natural killer (NK) cells are an important defense mechanismagainst pathogens, particularly against viruses belonging tothe herpesvirus family (45, 48). NK cell receptor genes do notundergo somatic recombination and clonal specification (38),and their activation is tightly regulated by a balance of signalingthrough inhibitory receptors specific for major histocompatibilitycomplex (MHC) class I proteins and activating NK cell receptorswith diverse specificities (28). Some activating NK receptorsare specific for viral proteins, such as the m157 protein ofmurine cytomegalovirus (MCMV) and the hemagglutinins of Sendaivirus and influenza virus, which are recognized by Ly49H, NKp44,and NKp46, respectively (4, 5, 31, 46).

NKG2D is a type II C-lectin-like activating NK cell receptorthat was first identified as a member of the NKG2 family (21)and is expressed on all NK cells, as well as on CD8⁺ T cells, ${gamma}$ ${delta}$ T cells, and macrophages (6, 15, 17, 22). NKG2D is a promiscuousreceptor that can recognize a broad spectrum of cell surfaceligands that are distantly related to MHC class I moleculesand are up-regulated on stressed, infected, or transformed cells(11). The known NKG2D ligands on human cells are the MHC classI chain-related molecules (MICA and MICB) (6, 47) and the UL-16binding proteins (ULBP-1, ULBP-2, and ULBP-3) (12), whereasthe mouse NKG2D ligands are H60 (15, 30), retinoic acid earlyinducible gene 1 (RAE-1 ${alpha}$ , -ß, - ${gamma}$ , - ${delta}$ , and - ${varepsilon}$ isoforms)(10), and the recently identified murine UL-16 binding protein-liketranscript 1 (MULT-1) (9). H60 was originally described as aminor histocompatibility antigen recognized by T cells fromC57BL/6 mice in response to BALB.B splenocytes (30), and RAE-1has been shown to play a role during embryonic development (57).

Cytomegaloviruses (CMVs) possess a remarkable range of mechanismsto escape or subvert the immune response (2). Both human CMV(HCMV) and MCMV have developed mechanisms for evading the controlof NK cells by interfering with the expression of NKG2D ligands(36). The HCMV protein encoded by the UL16 gene binds ULBP-1,ULBP-2, and MICB (12), preventing these ligands from being expressedon the surfaces of HCMV-infected cells (16, 55). Based on theirearly susceptibility to MCMV infection, mouse strains can beeither resistant or sensitive to this virus (18, 43). In resistantmouse strains, NK cells become activated via an interactionof Ly49H, an activating NK cell receptor, with the MCMV-encodedm157 protein (3, 46). In contrast, Ly49H-negative mouse strains,including most wild mice, show very low NK activities againstMCMV, rendering them susceptible to this virus (42). The puzzlingfact that Ly49H-negative mice, although being capable of mountingan effective NK cell response against other pathogens (54),are unable to create effective NK cell control of MCMV, hasrecently been explained by the MCMV-driven down-regulation ofcellular ligands for the NKG2D receptor (27). MCMV gp40, a viralglycoprotein encoded by the m152 gene, apart from down-regulatingMHC class I molecules (56), also down-modulates NKG2D ligandsfrom the cell surface (27). The deletion of the m152 gene resultsin the conversion of an NK cell-resistant virus to an NK cell-sensitivevirus strain. A further study by Lodoen et al. (29) revealedthat m152/gp40 down-regulates the expression of RAE-1.

For escape from NK cell control, it is perhaps not sufficientto down-regulate only one of at least three different NKG2Dligands since the remaining ligands might be sufficient to triggerNK cell activation. Therefore, we postulated that in additionto m152/gp40, there are other MCMV proteins that control theexpression of NKG2D ligands other than RAE-1. Here we demonstratethat the m155 MCMV gene product down-modulates the expressionof the H60 protein from the surfaces of infected cells and thatthe deletion of the m155 gene affects virus fitness in vivo.

MATERIALS AND METHODS

Top

Materials and Methods

Cells. NIH 3T3 cells (ATCC CRL1658), CV-1 cells (ATCC CCL70), and thebone marrow stromal cell line M2-10B4 (ATCC CRL1972) were cultivatedin Dulbecco's modified Eagle's medium (DMEM) supplemented with10% fetal calf serum (FCS). Mouse embryonic fibroblasts (MEFs)prepared from BALB/c, BALB/c TAP1^–/–, and BALB/cß2-microglobulin^–/– mice were cultivatedin minimum essential medium (MEM) supplemented with 3% FCS or,alternatively, in DMEM supplemented with 10% FCS.

To obtain cell transfectants, we PCR amplified the hemagglutinin(HA)-tagged H60 open reading frame (ORF) from H60/p7.5k by usingthe forward primer 5'-ACGCGTCGACACCATGGCAAAGGGAGCCACC-3' andthe reverse primer 5'-GTGCGGTCGACGCTCACGCGTAATCTGGAACATCGT-3'and cloned it into the SalI restriction site of pB45-Neo, whichwas kindly provided by E. R. Podack (35). The plasmid was transfectedinto NIH 3T3 fibroblasts by use of the SuperFect transfectionreagent (QIAGEN, Valencia, Calif.) according to the manufacturer'sinstructions. H60-transfected 3T3 cells were selected and culturedin DMEM supplemented with 10% FCS and 500 µg of G418 (Invitrogen,Paisley, Scotland)/ml.

Viruses. A bacterial artificial chromosome (BAC)-derived MCMV, MW97.01,has previously been shown to be biologically equivalent to theMCMV Smith strain (ATCC VR-194 [recently reaccessioned as VR-1399])and is here referred to as wild-type (wt) MCMV (52). For thepreparation of virus stocks, MCMV recombinants were propagatedon MEFs and purified as described previously (7). Titers ofvirus stocks were determined by a standard plaque assay on MEFs(7). Tissue culture-grown virus preparations were used for mouseinoculations.

Site-directed mutagenesis of MCMV BAC. For the generation of ${Delta}$ m155 and ${Delta}$ m155 ${Delta}$ m157 MCMV BACs, a PCR-basedmutagenesis approach was used as described previously (53).The m155 deletions were introduced into either the pSM3fr-16F17(referred to as the wild type) or p ${Delta}$ m157-16F17 (8) MCMV BAC,resulting in p ${Delta}$ m155 and p ${Delta}$ m155-m157, respectively. To delete them155 gene, we introduced a zeocin resistance cassette into theBACs, replacing the MCMV sequence from positions 214443 to 215531(nucleotide positions are numbered according to reference 39).The inserted zeocin resistance cassette was amplified by useof the High Fidelity Expand PCR system (Roche Diagnostics, Mannheim,Germany), with pcDNA4TO (Invitrogen, Paisley, Scotland) as atemplate and with the 5'-m155 (TTTTATTAATCGACGGGAGCGGGGGGACCGGGGTGATCATTTGTATTCGGATCTGATCAGCACGT)and 3'-m155 (TCGTCGAAAATGTCTGTACGAGTATGTGCTCTCCTGCTCTTGATCTAGCACGTGTCAGTCCTGC)primers. The recombinant MCMV BACs were verified by restrictionanalysis and DNA sequencing. MEFs were used for virus reconstitutionfrom recombinant BACs as described previously (52). After reconstitutionof the recombinant viruses, the primary stocks were passagedsix times on M2-10B4 cells to remove the BAC cassette.

The genomes of deletion mutants expressing green fluorescentprotein (GFP), as described in Table 1, were constructed inEscherichia coli strain DH10B by homologous recombination betweenlinear DNA fragments and the MCMV BAC pSM3fr-GFP (32), exploitingthe bacteriophage ${lambda}$ recombination genes red ${alpha}$ , redß,and red ${gamma}$ essentially as described previously (51). Briefly, linearfragments carrying a kanamycin resistance (Kn^r) gene were generatedby PCR, with either pOri6K-F5 (in the case of the BAC MCMV-GFP ${Delta}$ 6)or pGP704-Kan (for all other BACs) as a template. pOri6K-F5and pGP704-Kan contain the Kn^r gene from transposon Tn903 flankedby mutant and wild-type FLP recombinase recognition target sites(FRT), respectively (details of the construction of plasmidspOri6K-F5 and pGP704-Kan will be published elsewhere). The primersused for amplification of the Kn^r gene contained 20 to 22 nucleotides(nt) at their 3' ends that were specific for the Kn^r templateand 50 to 60 nt at their 5' ends that were homologous to thetarget region in the MCMV BAC. The linear fragments were insertedby homologous recombination into the viral target sequence viathe flanking 50- to 60-nt homologies, thereby replacing therespective target gene(s). DNAs from kanamycin-resistant BACclones were tested for correct insertion by restriction analysis.The ranges of the deletions are given in Table 1. Recombinantviruses were reconstituted by transfecting DNAs from the mutatedMCMV BACs into M2-10B4 or MEF cells by electroporation. Briefly,2 to 3 µg of BAC DNA was electroporated into 10⁶ cellsat 250 V and 1,500 µF by use of an Easyject Optima electroporator(Peqlap, Erlangen, Germany).

View this table:
[in this window]
[in a new window]
TABLE 1. MCMV deletion mutants

For reinsertion of the m155 ORF into the ${Delta}$ 6 MCMV genome (lackingthe genes from m144 to m158), a DNA fragment containing theHCMV major immediate early promoter (MIEP) was first amplifiedby PCR with the primers SacI-MIEP.for (5'-GAGGAGCTCCGGGGTCATTAGTTCATAGCCCA-3')and NotI-MIEP.rev (5'-GAGGCGGCCGCCGACCGGTAGCGCTAGCGGATC-3')and with the plasmid pEGFP (Clontech Laboratories, Palo Alto,Calif.) as a template, treated with SacI and NotI, and clonedinto the shuttle plasmid pOri6k-AL, which contains a kanamycinresistance marker, an FRT site, and the bacterial origin ofreplication R6K ${gamma}$ (25). A DNA fragment carrying the m155 ORF wasthen amplified by PCR with the primers m155rei.for (5'-AAGGGATCCGACGCGATACACGTTTGGGATAG-3')and m155rei.rev (5'-AAGGCGGCCGCGTCGAAAATGTCTGTACGAGTATG-3'),treated with BamHI and NotI, and inserted downstream of theMIEP, resulting in the plasmid pOri6Km155MIEP.

The BAC pSM3fr-GFP ${Delta}$ 6.2 carries a deletion comprising nt 203002to 217799 and ORFs m144 to m158. The Kn^r gene in pSM3fr-GFP ${Delta}$ 6.2was removed by FLP-mediated recombination excision, leavinga single FRT site in the resulting BAC, pSM3fr-GFP ${Delta}$ 6.2- ${Delta}$ kan. ThepOri6k-based shuttle plasmid described above was inserted intopSM3fr-GFP ${Delta}$ 6.2- ${Delta}$ kan by FLP-mediated recombination as describedelsewhere (33), leading to the creation of MCMV-GFP ${Delta}$ 6m155Rev.

For the generation of the m155Rev revertant virus, the m155coding sequence (nt 214434 to 215567 [numbered according toreference 39]) was first amplified from pSM3fr by use of them155Rfor (5'-GGGGACTAGTGGGTGATCATTTGTAGACG-3') and m155Rrev(5'-GGAAGGTACCTCTTGATCGCTTGTGCCTA-3') primers and then subclonedinto the pGPS1.1 vector (New England Biolabs) by the use ofSpeI and KpnI sites, resulting in pGPS-m155. The m155 ORF, togetherwith the adjacent kanamycin resistance cassette of pGPS-m155,was amplified by use of the 5'-m155REV (5'-GAGTGTCATAATTGTTTTATTAATCGACGGGAGCGGGGGGACCGGGGTGACGTCAAGTCAGCGTAATGCT-3')and 3'-m155REV (5'-GACTTTCGTCGAAAATGTCTGTACGAGTATGTGCTCTCCTGCTCTTGATCGCTTGTGCCTA-3')primers and the High Fidelity Expand PCR system (Roche), andthis linear fragment was inserted into the ${Delta}$ m155 BAC by RecERecT(ET) recombination (53). ET recombination restored the m155ORF and introduced a kanamycin resistance cassette between thestop codon (nt 214434 to 214436) and the predicted polyadenylationsite (nt 214398 to 214404) of the m155 gene (numbered accordingto reference 39). Note the difference between the MCMV-GFP ${Delta}$ 6m155Revand m155Rev viruses: the first one lacks all of the genes fromthe m144-m158 region, except for m155, whereas the latter virushas the m155 gene orthotopically reinserted into the ${Delta}$ m155 mutantvirus.

The growth kinetics in NIH 3T3 cells of all recombinant virusesused for this study were indistinguishable from those of wtMCMV.

Production of recombinant vaccinia viruses. For the generation of a recombinant vaccinia virus (VV), theH60 cDNA sequence missing its intracellular domain (GenBankaccession no. AF084643) was PCR amplified without its intracellulardomain from the pcDNAI H60 plasmid. Using the forward primer5'-CGGGATCCCGAAGACCATGGCAAAGGGAGCCACC-3' and the reverse primer5'-GGGGTACCCCTCACGCGTAATCTGGAACATCGTATGGGTATTTTTTCTTCAGCATACACC-3',we added the HA epitope sequence C-terminally. The PCR productswere cloned into 5' BamHI and 3' KpnI restriction sites of plasmidp7.5K131 (44). For the creation of a recombinant vaccinia virusbearing the m155 ORF, the cDNA was HA tagged by PCR amplificationfrom pOri6Km155MIEP with the forward primer 5'-CGTGGATCCACCATGTCTGTACGAGTATGTGCTC-3'and the reverse primer 5'-TGAGGATCCTCAAGCGTAGTCCGGGACGTCGTACGGGTATTTGTAGACGGGCGG-3'.The m155-HA ORF was cloned into the BamHI restriction site ofp7.5K131. The constructs were used for the generation of recombinantvaccinia viruses expressing H60 or m155 by homologous recombinationwith the vaccinia virus strain Copenhagen. Vaccinia virus recombinantswere selected by infection of thymidine kinase-negative 143cells in the presence of 100 µg of bromodeoxyuridine/mlas described previously (50).

Flow cytometry. NIH 3T3 cells were mock treated or infected with MCMV (2 PFU/cell)and then were trypsinized at 12 h postinfection. CV-1 cellswere mock treated or infected with VV (multiplicity of infectionof 3/cell) and then were harvested 14 h after infection by theuse of 2 mM EDTA. Both cell types were washed in phosphate-bufferedsaline (PBS) supplemented with 1% bovine serum albumin and 0.1%NaN₃ and then stained with either a phycoerythrin (PE)-NKG2Dtetramer (27), the rat anti-H60 monoclonal antibody (MAb) clone205326 (kindly provided by J. P. Houchins, R&D Systems,Minneapolis, Minn.), or the rat anti-RAE-1 ${alpha}$ ß ${gamma}$ MAb CX1,kindly provided by L. L. Lanier (29). After a washing step,bound antibodies were visualized by the addition of PE-labeledgoat anti-rat immunoglobulin G (IgG) (Caltag Laboratories, Burlingame,Calif.) or fluorescein isothiocyanate (FITC)-labeled goat anti-ratIgG (Sigma-Aldrich, Munich, Germany). Cells incubated with PE-streptavidinserved as a negative control for cells stained with the PE-NKG2Dtetramer, and a second antibody served as a negative controlfor cells stained with the anti-H60 and anti-RAE-1 ${alpha}$ ß ${gamma}$ MAbs. After being stained, the cells were analyzed with a BectonDickinson FACScan instrument and gated for propidium iodide-negativecells. The infection of H60-3T3 transfectant cells was controlledby intracellular staining with the MAb CROMA 229, which recognizesthe MCMV gp48 antigen (41), and after washing, bound antibodieswere visualized by the addition of FITC-labeled goat anti-mouseIg (BD Pharmingen, San Diego, Calif.).

Metabolic labeling of cells and immunoprecipitation. Subconfluent layers of cells were labeled with [³⁵S]methionine(Amersham Pharmacia Biotech, Freiburg, Germany) at a concentrationof 500 µCi/ml at 37°C for 30 min and then chased inthe presence of 10 mM unlabeled methionine. After being washedwith ice-cold PBS, the cells were lysed in 1 ml of lysis buffer(140 mM NaCl, 5 mM MgCl₂, 20 mM Tris [pH 7.6], 1 mM phenylmethylsulfonylfluoride, 0.1 M leupeptin, 1 µM pepstatin A) containing1% (wt/vol) digitonin (Calbiochem-Novabiochem, La Jolla, Calif.)or 1% (vol/vol) IGEPAL (Sigma-Aldrich) for 20 min and then centrifugedat 1,300 x g for 30 min.

The lysates were incubated for 1 h at 4°C with 0.5 µgof anti-HA (Sigma-Aldrich) or anti-L^q (28-14-8s) or 1 µgof the anti-H60 MAb. Immunoprecipitation was performed as describedpreviously (14, 20). In brief, immune complexes were retrievedwith protein A- or protein B-CL-4B Sepharose (Amersham PharmaciaBiotech) (60 µl of buffer-Sepharose slurry [1:1] for 1h at 4°C). The Sepharose beads were washed three times witha buffer containing 0.2% (vol/vol) IGEPAL, 10 mM Tris-HCl (pH7.6), 140 mM NaCl, and 2 mM EDTA, twice with a buffer containing0.2% (vol/vol) IGEPAL, 10 mM Tris-HCl (pH 7.6), 500 mM NaCl,and 2 mM EDTA, and once with 10 mM Tris-HCl (pH 7.6). The immunecomplexes bound to Sepharose beads were resuspended in 50 mMphosphate buffer (pH 5.5) containing 0.1% (wt/vol) sodium dodecylsulfate (SDS), 0.1% (vol/vol) IGEPAL, 0.2 mM phenylmethylsulfonylfluoride, and 0.1 M 2-mercaptoethanol. For selective immunoprecipitationof cell surface proteins, cells were metabolically labeled for120 min before being transferred to 4°C, and antibodieswere added to the cell layer for 30 min. Unbound antibodieswere removed by two rounds of washing with PBS. After cell lysis,the precipitation of immune complexes was performed as describedabove. Sepharose-bound immune complexes were mock treated orincubated with 2 mU of endoglycosidase H (endo H; Roche Diagnostics,Mannheim, Germany) at 37°C overnight. Digestion was stoppedby the addition of sample buffer, and the immune complexes wereeluted from Sepharose by heating at 94°C for 5 min. Theprecipitates were analyzed by SDS-11.5% polyacrylamide gel electrophoresis(SDS-11.5% PAGE). Dried gels were exposed to Kodak BioMax MRfilms for 1 to 3 days.

Animals, infection conditions, detection of infectious MCMV in tissues, and statistical evaluation. The BALB/c (H-2^d) and congenic BALB.B6-Cmv1^r (H-2^d) mice usedfor this study were housed and bred under specific-pathogen-freeconditions at the Central Animal Facility of the Medical Faculty,University of Rijeka, in accordance with the guidelines containedin the International Guiding Principles for Biomedical ResearchInvolving Animals. The ethical committee at the University ofRijeka approved all animal experiments described here. Six-to 8-week-old female mice were used for experiments.

Mice were injected intravenously with 4 x 10⁵ PFU (BALB/c mice)or 5 x 10⁵ PFU (BALB.B6-Cmv1^r mice) of wt MCMV or a recombinantvirus in 500 µl of diluent. Organs were collected 4 daysafter infection, and viral titers were determined by a standardviral plaque-forming assay performed on MEFs (40). The statisticalsignificance of differences between experimental groups wasdetermined by the Mann-Whitney exact rank test. Viral titers(from groups x and y) were considered significantly differentfor P values (x versus y) of <0.05 (one-sided).

Depletion of NK cell subsets in vivo. The depletion of NK1.1^– cells (from BALB/c mice) was performedby intraperitoneal injections of a rabbit antiserum to asialo-GM1(Wako Chemicals, Osaka, Japan) at a dose of 25 µl 24 to2 h before infection. The depletion of NK1.1⁺ cells (BALB.B6-Cmv1^r)was done with the MAb PK136 (26) at a concentration of 1 mg/mouseinoculated intraperitoneally 24 to 2 h before infection. Theefficacy of depletion was assessed by cytofluorometric analysesof spleen cells by use of a PE-conjugated MAb directed againstmouse NK1.1 molecules (BD Pharmingen) and the biotin-labeledanti-mouse pan-NK cell MAb DX5 (BD Pharmingen).

RESULTS

Top

Results

MCMV genes in addition to m152 regulate the expression of NKG2D ligands. MCMV gp40, a viral glycoprotein encoded by the m152 gene, down-modulatesthe expression of RAE-1 on MCMV-infected cells and thus modulatesrecognition and virus control by NK cells (27, 29). Consideringthe facts that MCMV-infected cells show a complete absence ofNKG2D ligands on the plasma membrane, as shown by staining withan NKG2D tetramer, and that gp40 down-regulates RAE-1 but notH60 (29), we surmised that there must exist an MCMV gene(s)which down-regulates H60. To study the effect of MCMV on H60,we chose a cell line that constitutively expresses H60 on thecell surface and which is permissive for MCMV. Using H60-specificmonoclonal antibodies, we screened several cell lines and selectedNIH 3T3 cells for further studies (data not shown). To discriminateinfected from uninfected cells, we took advantage of a recombinantMCMV expressing GFP (32) by gating GFP-positive cells in flowcytometry analyses. The infection of NIH 3T3 cells with theMCMV-GFP virus resulted in a strong down-modulation of NKG2Dligands from the surfaces of infected cells compared to thoseon uninfected NIH 3T3 cells (Fig. 1A). Identical results werealso observed with wt MCMV, which does not express GFP (datanot shown). Cells infected with MCMV-GFP ${Delta}$ 6, a mutant that lacksthe genes m144 to m158, however, remained positive for NKG2Dstaining. Because MCMV-GFP ${Delta}$ 6 also lacks the m152 gene, the lackof down-modulation of NKG2D ligands could be assigned to theexpression of RAE-1 molecules on the surfaces of infected cells.To separate the role of m152 and to narrow down the genomicregion encoding new NKG2D ligand regulators, we tested the mutantsMCMV-GFP ${Delta}$ 6S2 and MCMV-GFP ${Delta}$ 6S3, characterized by deletions of m149to m153 and of m154 to m157, respectively (Table 1). The resultsshowed that neither of these two mutants was able to down-regulateNKG2D ligands to the level of MCMV-GFP, although the intensityof NKG2D staining of cells infected with these viruses was notthe same as that observed for cells infected with the MCMV-GFP ${Delta}$ 6virus (Fig. 1A). We concluded that m152 cannot be the only MCMVgene in the region defined by the mutant MCMV-GFP ${Delta}$ 6 which isinvolved in the down-regulation of NKG2D ligands and that atleast one additional gene involved in down-modulation of theseligands should be located within the m154-m157 region.

View larger version (21K):
[in this window]
[in a new window]
FIG. 1. The m155 protein is involved in down-regulation of NKG2D ligands. NIH 3T3 cells were infected with 2 PFU of the indicated GFP-positive viruses per cell or were left uninfected. Twelve hours after infection, the cells were collected and analyzed for the expression of surface NKG2D ligands by staining with the PE-NKG2D tetramer. Cells incubated with streptavidin-PE were used as a negative control (thin line). Each histogram represents 10,000 gated propidium iodide-negative, GFP-negative (uninfected), or GFP-positive (infected) cells. (A) Regional mutants. (B) Single mutants.

The m155 gene product down-regulates H60. Mutants lacking single ORFs in the m154-m157 region were generated(Table 1) and tested for the capacity to down-modulate NKG2Dligands. As shown in Fig. 1B, all of the tested mutants stilldown-modulated the expression of NKG2D ligands, with the exceptionof MCMV-GFP ${Delta}$ m155, which showed an NKG2D staining pattern similarto that of MCMV-GFP ${Delta}$ 6S3. To characterize the NKG2D ligand(s)regulated by the m155 gene product, we directly compared thecell surface expression levels of RAE-1 ${alpha}$ , -ß, and - ${gamma}$ and of H60 in cells infected with wild-type and mutant virusstrains (Fig. 2). In accordance with previously published results(29), MCMV-GFP ${Delta}$ 6 and MCMV-GFP ${Delta}$ m152 did not affect the surfaceexpression of RAE-1 in these cells (Fig. 2). In contrast, MCMV-GFP ${Delta}$ 6S3and MCMV-GFP ${Delta}$ m155 still down-modulated RAE-1 expression, indicatingthat the m155 product does not affect this ligand. Stainingwith an anti-H60 MAb revealed that the down-modulation of theexpression of H60 by MCMV (Fig. 2 and 3) is strong, althoughnot complete, as one would have predicted based on the stainingwith the NKG2D tetramer (Fig. 1). Infection with either MCMV-GFP ${Delta}$ 6or MCMV-GFP ${Delta}$ 6S3 partially reconstituted the surface expressionof H60. Importantly, a single deletion mutant demonstrated thatm155 gene expression is required for the down-modulation ofH60 from the cell surface (Fig. 3). The reintroduction of them155 gene into the genome of a mutant with a larger deletion(MCMV-GFP ${Delta}$ 6) yielded MCMV-GFP ${Delta}$ 6m155Rev and served to confirm thatthe m155 protein interacts with H60 cell expression. MCMV-GFP ${Delta}$ 6m155Revlacks all of the genes from the m144-m158 region, with the exceptionof m155. In cells infected with this mutant, H60 was again down-modulatedto the level observed for MCMV-GFP-infected cells. However,the results repeatedly showed that the expression level of H60on cells infected with either the MCMV-GFP ${Delta}$ 6 or MCMV-GFP ${Delta}$ m155virus was significantly lower than that in uninfected cells.This indicates that another viral function is also involvedin the down-regulation of H60.

View larger version (21K):
[in this window]
[in a new window]
FIG. 2. The m155 protein is responsible for down-regulation of H60. NIH 3T3 cells were infected with 2 PFU of GFP-positive viruses per cell or were left uninfected. Twelve hours after infection, the cells were collected and stained with an anti-RAE-1 ${alpha}$ ß ${gamma}$ or anti-H60 MAb, followed by PE-conjugated goat anti-rat IgG. Cells incubated with the secondary antibody in the absence of the primary antibody were used as a negative control (thin line). Each histogram represents 10,000 gated propidium iodide-negative, GFP-negative (uninfected), or GFP-positive (infected) cells.

View larger version (55K):
[in this window]
[in a new window]
FIG. 3. MCMV does not affect maturation of the H60 protein. (A) NIH 3T3 cells were infected with H60-VV and metabolically labeled at 8 h p.i. with [³⁵S]methionine. The samples were chased in the presence of unlabeled methionine as indicated, lysed in 1% IGEPAL buffer, and immunoprecipitated with anti-HA antibodies coupled to protein A-Sepharose. Prior to separation by SDS-11.5% PAGE, the lysates were mock treated (–) or digested with endo H (+). wt VV-infected cells were used as a negative control. (B and C) H60-3T3 cells were mock treated or infected with 2 PFU per cell of either wt or ${Delta}$ m155 MCMV. At 14 h p.i., the samples were subjected to pulse-chase labeling with [³⁵S]methionine as indicated. Immunoprecipitation was performed with anti-HA (B) or anti-L^q (C) antibodies bound to protein A-Sepharose. Prior to SDS-PAGE, the lysates were mock treated (–) or digested with endo H (+). R, resistant to endo H; S, sensitive to endo H; D, digested with endo H; *, Gp48_S and gp48_D MCMV m06/gp48 protein coprecipitating with L^q (41). (D) H60-3T3 cells were infected with wt MCMV or not infected and then analyzed at 18 h p.i. for the expression of surface and intracellular H60 by the use of anti-H60 and anti-HA MAbs, respectively. Each histogram represents 10,000 GFP-negative (uninfected cells) or GFP-positive (infected cells) cells. Cells incubated with the secondary antibody in the absence of the primary antibody were used as a negative control (thin line).

H60 is an MHC class I-like protein which matures independent of TAP and ß2m. The H60 sequence predicts a type I transmembrane glycoproteinof 335 amino acids comprising a signal sequence, a luminal domain,a transmembrane domain, and a cytosolic domain (30). The ectodomainincludes seven potential N-linked glycosylation sites. To date,studies on the maturation and posttranslational modificationsof H60 are still lacking (30). In order to monitor the maturationof the protein, we constructed a recombinant vaccinia virusexpressing HA epitope-tagged H60 (H60-VV). After 8 h of infectionwith H60-VV, NIH 3T3 cells were metabolically labeled for 30min with [³⁵S]methionine, and the label was then chased for30 min, 1 h, and 8 h. H60 was immunoprecipitated from lysatesby the use of anti-HA antibodies coupled to protein A-Sepharose.The immunoprecipitated molecules were mock treated or incubatedwith endo H and then separated by SDS-PAGE. Wild-type vacciniavirus (wt VV)-infected cells were used as a negative control.As shown in Fig. 3A, 46- and 70-kDa bands were detected in pulse-labeledsamples. The treatment of immune complexes with endo H, whichcleaves high-mannose N-linked glycans that have not been processedinto complex glycans by enzymes resident in the medial Golgi,resulted in a shift of the 46-kDa protein to a band of approximately27 kDa. The size of the deglycosylated H60 protein was in accordancewith its published sequence (30). Most of the newly synthesizedH60 molecules were readily processed into the endo H-resistantform of about 70 kDa within 30 min of the chase. A substantialloss of the 70-kDa band occurred within 8 h of the chase. Ahalf-life of approximately 4 to 8 h was calculated by densitometryof the bands obtained in three independent experiments, includingthe one shown in Fig. 3A. N-linked glycosylation of all sevenpredicted sites in the H60 glycoprotein was observed by partialdigestion of the immunoprecipitated H60 with endo H (data notshown).

Considering that H60 is a nonpolymorphic MHC class I-like glycoprotein,we studied H60 biogenesis, processing, and transport in transporterassociated with antigen processing 1 (TAP-1)- and ß2-microglobulin(ß2m)-deficient MEFs after its expression by H60-VV.The data showed a maturation pattern identical to the one obtainedfor NIH 3T3 cells, indicating that H60 matures in a TAP- andß2m-independent manner (data not shown).

Maturation of H60 in MCMV-infected cells. To study the maturation of H60 upon MCMV infection, we generateda stable NIH 3T3 H60 transfectant. H60-3T3 cells were mock treatedor infected with either wt MCMV or the ${Delta}$ m155 deletion mutantand analyzed by immunoprecipitation 14 h after infection (Fig.3B). Immunoprecipitation of pulse-labeled L^q molecules was performedas a control (Fig. 3C). The maturation pattern of H60 in transfectants(data not shown) was comparable to that in H60-VV-infected cells(Fig. 3A). A comparison between wt MCMV-infected, ${Delta}$ m155 MCMV-infected,and uninfected H60-3T3 cells revealed no difference regardingthe stability of the H60 protein or its ability to reach theendo H-resistant form. In contrast, the virus caused a retentionof MHC class I L^q molecules in an endo H-sensitive form uponboth wt MCMV and ${Delta}$ m155 MCMV infection (Fig. 3C). Thus, the m155product alters neither the maturation of H60 prior to its transitthrough the ER-Golgi intermediate compartment (ERGIC)-cis-Golgicompartment nor the half-life of the endo H-resistant form ofH60.

To exclude the possibility that m155 has no effect on H60 expressedby H60-3T3 cells, we performed a fluorescence-activated cellsorting analysis of the MCMV-infected H60-3T3 cells. In agreementwith the results presented in Fig. 2, the surface expressionof H60 was significantly reduced in MCMV-infected cells (Fig.3D). Intracellular staining with anti-HA antibodies, however,revealed that HA-tagged H60 was detectable in MCMV-infectedH60-3T3 cells, supporting the data obtained by immunoprecipitation.

Although we cannot completely rule out the possibility thatH60 molecules derived from a few residual uninfected cells contributeto the background for immunoprecipitation, we assume that ifm155 caused a degradation of the molecule, the quantity of theremained band would be significantly lower than that in uninfectedcells. It is worth noting that the radioactivity was quantifiedand equalized in all of the samples prior to their being loadedinto gels. Altogether, the intracellular staining of HA-taggedH60 and its immunoprecipitation strongly suggest against a significantinfluence of m155 on the stability of the H60 protein.

In an additional attempt to monitor the fate of the H60 proteinin infected cells, we performed a selective immunoprecipitationof surface-resident H60 molecules from H60-3T3 cells that weremetabolically labeled for 120 min starting at 12 h postinfection(p.i.). The complete absence of membrane-exposed H60 moleculesin wt MCMV-infected H60-3T3 cells, but not in uninfected and ${Delta}$ m155 MCMV-infected cells, suggests a rapid effect of m155 onnewly synthesized surface-destined H60 molecules that do notreach the plasma membrane (data not shown). Immunoprecipitationof H60 from the supernatants of MCMV-infected cells did notdetect any H60 protein (data not shown). Taken together, theseresults reveal that the loss of H60 from the surfaces of infectedcells is not due to shedding of the H60 glycoprotein from thecell surfaces of MCMV-infected cells but to an altered intracellulardistribution of H60.

To assess the effect of the isolated m155 gene on H60 expression,we constructed a recombinant vaccinia virus expressing HA epitope-taggedm155 (m155-VV) and coinfected CV-1 cells either with H60-VVand wt VV or with H60-VV and m155-VV. Infection with H60-VVresulted in a high level of H60 expression on the cell surface(Fig. 4A). The surface expression was not altered upon coinfectionwith wt VV but was significantly lower upon coinfection withm155-VV. Neither uninfected nor wt VV-infected CV-1 cells showedany membrane H60 staining. These data clearly show that m155gene expression is sufficient for the down-regulation of H60from the cell surface. It is worth noting that the down-modulationof H60 was not complete, suggesting that there is a thresholdlimit for the cellular mechanism involved.

View larger version (58K):
[in this window]
[in a new window]
FIG. 4. Down-regulation of H60 by coinfecting CV-1 cells with m155-VV and H60-VV. CV-1 cells were infected with H60-VV or wt VV or coinfected with H60-VV and wt VV or with H60-VV and m155-VV and then were analyzed at 12 h p.i. (A) Cells were collected and stained with an anti-H60 MAb followed by FITC-conjugated anti-rat IgG. Cells incubated with the secondary antibody in the absence of the primary antibody were used as a negative control (thin line). Each histogram represents 10,000 gated propidium iodide-negative, GFP-positive cells. (B) Cells were metabolically labeled for 30 min and chased for the indicated times, and the lysates were immunoprecipitated with anti-HA antibodies coupled to protein A-Sepharose. Prior to separation by SDS-11.5% PAGE, the samples were mock treated (–) or digested with endo H (+). R, resistant to endo H; S, sensitive to endo H; D, digested with endo H. Arrows indicate H60 shifts in the presence of m155.

In order to biochemically monitor the maturation pathways ofboth the m155 and H60 proteins, we coinfected CV-1 cells withH60-VV and wt VV or with H60-VV and m155-VV. Both proteins wereimmunoprecipitated from cell lysates by the use of anti-HA antibodiescoupled to protein A-Sepharose 14 h after infection (Fig. 4B).Infection with wt VV was used as a negative control. The datashowed the simultaneous conversion of endo H-sensitive to -resistantforms of H60 (70 and 46 kDa, respectively) and the m155 glycoprotein(85 to 100 and 70 kDa, respectively). In accordance with theresults obtained upon infection of H60-3T3 cells with wt MCMV(Fig. 3B), m155-VV neither caused any change in the half-lifeof the H60 glycoprotein nor affected its ability to reach theendo H-resistant form. Remarkably, there was a slight shiftin the migration properties of the glycosylated H60 proteinin the presence of m155 (Fig. 4B, arrows), consistent with analtered posttranslational modification of H60.

m155 contributes to MCMV virulence by compromising NK cell-mediated control. We next studied whether the absence of m155 affects virus controlby NK cells in vivo. We expected the deletion of the m155 geneto result in viral sensitivity to NK cells in vivo. In orderto avoid an effect of GFP expression on the immune responseto MCMV, we constructed a single m155 deletion mutant and adouble m155 m157 deletion mutant without GFP. BALB/c mice weredepleted of NK cells or left untreated, and virus titers weredetermined by a plaque assay 4 days after infection. The ${Delta}$ m155virus was indeed attenuated compared to wt MCMV, and this attenuationwas reversed by the depletion of NK cells (Fig. 5A). A confirmationof the specificity of the effect of m155 came from an experimentin which the m155 gene was reintroduced into the ${Delta}$ m155 genome(m155Rev). Reinsertion of the m155 gene resulted in m155Revvirus resistance to NK cell control in vivo (Fig. 5B). Next,we tested the ${Delta}$ m155 virus in BALB.B6-Cmv1^r mice, which expressthe natural killer gene complex of C57BL/6 mice in the BALB/cgenetic background (Fig. 5C). Since these mice are positivefor Ly49H, the role of m155 could only be tested in a viruslacking m157 to avoid natural killer gene complex activationvia Ly49H. In accordance with our recent results (8), the deletionof the m157 gene resulted in a gain of virulence and in resistanceto NK cell control. Furthermore, the results revealed that NKcell activation via the NKG2D receptor is blocked as well. Asexpected, the deletion of the m155 gene led to sensitivity ofthe ${Delta}$ m155 ${Delta}$ m157 mutant to NK cell-mediated control. Like the casefor BALB/c mice, this attenuation could be reversed by the depletionof NK cells. Interestingly, 2 weeks after infection, the virustiters in the salivary glands of BALB/c mice infected with the ${Delta}$ m155 virus were significantly lower than the titers after infectionwith the control virus (Fig. 6). Given that virus control inthe salivary glands is CD8⁺ T cell independent (23), one possibilityfor these results is that the attenuation of virus growth insalivary glands is related to the enhanced virus sensitivityto NK cells.

View larger version (21K):
[in this window]
[in a new window]
FIG. 5. Deletion of the m155 gene sensitizes MCMV to NK cell-mediated control. NK cell-depleted or undepleted BALB/c (A and B) and BALB.B6-Cmv1^r (C) mice were injected intravenously with 4 x 10⁵ or 5 x 10⁵ PFU, respectively, of the indicated viruses. The mice were sacrificed 4 days after infection, and the viral titers in the organs were determined. Titers in organs of individual mice are shown; horizontal bars indicate the median values. The differences in viral titers between NK-depleted and undepleted groups of mice are indicated by shaded areas. Open circles, no depletion; gray circles, NK cell depletion; DL, detection limit.

View larger version (15K):
[in this window]
[in a new window]
FIG. 6. The ${Delta}$ m155 virus is attenuated in salivary glands. BALB/c mice were injected intravenously with 4 x 10⁵ PFU of the indicated viruses. The mice were sacrificed 12 days after infection, and the viral titers in salivary glands and lungs were determined. Titers in organs of individual mice are shown; horizontal bars indicate the median values.

DISCUSSION

Top

Discussion

In this article, we have reported data for m155, a member ofthe MCMV m145 gene family which encodes a protein that modulatesthe expression of H60, one of the three cellular ligands formurine NKG2D that have been described so far. NKG2D is an activatingreceptor expressed on all NK cells, CD8⁺ T lymphocytes, and ${gamma}$ ${delta}$ T cells (6, 15, 17, 22). Disruption of the m155 gene leadsto an enhanced antiviral NK cell response and an attenuationof virus growth in vivo. Thus, similar to m152/gp40, which regulatesMHC class I alleles and RAE-1 family members, the role of them155 protein seems to be to prevent NK cell activation duringinfection. Since the expression of neither RAE-1 nor MULT-1(S. Jonjic, unpublished data) is affected by m155, it appearsthat MCMV devotes specialized proteins to the control of eachof the NKG2D ligands. By using recombinant vaccinia virusesthat express H60 or m155, we were able to demonstrate that m155does not require the presence of other MCMV proteins to down-modulatesurface H60. However, our results showed that m155 is not theonly gene responsible for the down-regulation of surface H60:the fact that the down-modulation of this protein was observedin cells infected with the ${Delta}$ m155 mutant points to the existenceof another viral function that regulates H60. Nevertheless,the functional significance of this second viral function invivo is questionable since it cannot override the successfulNK cell control of the ${Delta}$ m155 virus. The in vivo results are consistentwith the recent finding that an MCMV possessing a transposoninsertion at the m155 ORF showed a growth deficit and attenuationin SCID mice (1).

Our data show that neither the ability of H60 to reach the endoH-resistant form nor its stability is altered upon MCMV infection.Therefore, the failure of H60 to reach the cell surface is dueto a mechanism that operates beyond the ERGIC-cis-Golgi compartmentrather than to an alteration in the kinetics of internalizationafter it reaches the cell surface. Although it is unlikely,it even remains possible that H60 accumulates at the cell surfacein an altered conformation or in association with another proteinthat masks both its function and its detection by specific MAbsand NKG2D tetramer. This is not without precedent and reflectsthe example we described for m04/gp34 binding to MHC class Imolecules (24). Clearly, the remarkable reduction of membraneexpression in spite of the apparent maturation of the glycoproteinrequires further experimentation aiming at the identificationof the compartment to which H60 is targeted in the presenceof m155.

The interaction of NKG2D with its cognate ligands on infectedcells elicits a potent NK cell response (37). Despite possessinga functional NKG2D receptor, Ly49H-negative mice, includingmost wild mice, elicit only a weak NK cell response toward MCMV(13, 42, 49) and are therefore quite susceptible to this viruscompared to resistant mouse strains possessing Ly49H, an activatingNK cell receptor that recognizes the MCMV-encoded protein m157(4, 46). We have recently shown that Ly49H-negative mice areconstitutively able to mount an effective NK cell response toMCMV but that this activity is prevented by MCMV interferencewith the expression of NKG2D ligands (27). The gp40 proteinencoded by the m152 MCMV gene down-modulates the expressionof RAE-1 (29). Consequently, the deletion of the m152 gene resultsin the expression of RAE-1 and in viral sensitivity to NK cellsin vivo. Both the m152 and m155 genes belong to the MCMV m145gene family, which has 11 members, some of which have immunomodulatoryfunctions (e.g., m152 and m157) (39, 46). Notably, we also describedthe m152 product as a protein that prevents the presentationof viral peptides to CD8⁺ T cells by retaining MHC class I moleculesin an unknown compartment with ERGIC-cis-Golgi properties (56).

NKG2D ligand expression is restricted in normal cells and tissuesto avoid becoming a target for NK cells. Transformation, infection,and stress lead to the induction of these proteins on the cellsurface and thereby induce NK cell activation (11). It is notsurprising that viruses have evolved mechanisms to evade NKcells by down-modulating ligands for the NKG2D receptor. AlthoughMCMV induces the transcription of RAE-1 genes, the translocationof the protein to the cell surface is prevented by gp40 (29).One would expect MCMV to have evolved strategies to down-modulatethose immune receptor ligands whose genes are induced by infection.MCMV does not induce H60 gene expression (29), but nevertheless,the protein is down-modulated from the surfaces of infectedcells. Perhaps MCMV infection does induce H60 expression invivo in other cell types that have not been tested so far. Notethat H60 mRNA is expressed in resting splenic cells and dendriticcells as well as in activated splenocytes (30), which are potentialtargets for infection by MCMV. Another possibility is that thebasal level of H60 expression is sufficient and perhaps necessaryfor the fine-tuning of NK cell activation. MCMV down-modulatesMHC class I molecules on infected cells to escape cytotoxicCD8-T-cell lysis. This may already suffice as an NK activatingsignal. Therefore, if the down-modulation of MHC class I moleculesskews the balance of NK-interacting proteins displayed on thesurface of an infected cell towards the activation of NK cells,an obvious viral countermeasure would be to compromise the activationof NK cell receptors by down-modulating their ligands. Thisscenario is in fact supported by the findings presented here.Because the NKG2D receptor is also expressed on T cells andmacrophages, it is also possible that interference with thefunction of NKG2D ligands represents an efficient and perhapsessential requirement for virus-mediated attenuation of bothinnate and specific immune responses relevant to the establishmentof life-long infections and/or successful virus transmission.

What is the reason for the existence of more than one ligandfor the NKG2D receptor? One possibility to account for thisobservation is the differential activation of these ligandsbetween cells and tissues (11). Hamerman et al. have recentlyshown that signaling through Toll-like receptors in macrophagesinduces the transcription of RAE-1 but not of H60 and MULT-1genes (19). Furthermore, different ligands may be differentiallyexpressed during tissue and cell development and may have distinguishablebiological functions. Although the effects of MCMV immune evasionmechanisms on NK cell activity have been well documented invitro and in vivo, the importance of these virus-encoded functionsin tissue-specific pathology is still unknown. The virus-encodedimmune evasion functions that promote persistence may representa viral virulence factor, particularly in an organ with a limitedcapacity for self-renewal. One observation that is relevantto this point is the recently described in vivo phenotype ofthe Ly49H NK receptor and its ligand MCMV m157, which manifestin the spleen and the lungs but not in the liver (8). Furthermore,the role of viral evasion genes in modulating the expressionof NK cell ligands such as RAE-1, H60, and MULT-1 in developingtissues like the central nervous system (CNS) is unknown. Ligandsfor NK cells such as RAE-1 can be detected in the embryonicCNS but appear to be down-regulated in cells in the adult CNSand other adult organs (10, 15, 34). It remains to be determinedwhether cytomegaloviruses use immunoevasion genes to explicitlylimit immunopathology in vital tissues in order to preservethe vital functions of their specific hosts.

ADDENDUM IN PROOF

Top

Addendum in proof

Another paper presenting the evidence for m155-mediated regulationof H60 was recently published (M B. Lodoen, G. Abenes, S. Umamoto,J. P. Houchins, F. Liu, and L. L. Lanier, J. Exp. Med. 200:1075-1081).Our work showing the effect of the m145 gene on the expressionof the third known mouse NKG2D ligand, MULT-1, was recentlypublished (A. Krmpotic, M. Hasan, A. Loewendorf, T. Saulig,A Halenius, T. Lenac, B. Polic, I. Bubic, A. Kriegeskorte, E.Pernjak-Pugel, M. Messerle, H. Hengel, D. H. Busch, U. H. Koszinowski,and S. Jonjic, J. Exp. Med. 201:211-220).

ACKNOWLEDGMENTS

We thank J. P. Houchins and L. L. Lanier for generously providingthe anti-H60 and anti-RAE1 ${alpha}$ ß ${gamma}$ MAbs and D. H. Busch forproviding the NKG2D tetramer. We are indebted to A. Zimmermannfor technical advice and to J. Trgovcich for critically readingthe manuscript. We also thank D. Rumora for technical assistanceand M. Fritz-Boukhatem for organizing the experimental mousefacility.

This work was supported by Croatian Ministry of Science grants0062004 (S. Jonjic) and 0062007 (A. Krmpotic) and by DeutscheForschungsgemeinschaft grants SFB 455 (U. H. Koszinowski), EUQLRT-2001-01112 and SFB421 A8 (H. Hengel), and ME1102/2-1 (M.Messerle).

FOOTNOTES

* Corresponding author. Mailing address: Department of Histology and Embryology, Faculty of Medicine, University of Rijeka, B. Branchetta 20, 51000 Rijeka, Croatia. Phone: 385 51 651 206. Fax: 385 51 651 176. E-mail: jstipan{at}medri.hr.

Present address: Institute for Virology, Heinrich-Heine-UniversityDüsseldorf, Düsseldorf, Germany.

REFERENCES

Top