Signaling Lymphocytic Activation Molecule Family Receptor Homologs in New World Monkey Cytomegaloviruses

ABSTRACT

Throughout evolution, large DNA viruses have been usurping genes from their hosts to equip themselves with proteins that restrain host immune defenses. Signaling lymphocytic activation molecule (SLAM) family (SLAMF) receptors are involved in the regulation of both innate and adaptive immunity, which occurs upon engagement with their ligands via homotypic or heterotypic interactions. Here we report a total of seven SLAMF genes encoded by the genomes of two cytomegalovirus (CMV) species, squirrel monkey CMV (SMCMV) and owl monkey CMV (OMCMV), that infect New World monkeys. Our results indicate that host genes were captured by retrotranscription at different stages of the CMV-host coevolution. The most recent acquisition led to S1 in SMCMV. S1 is a SLAMF6 homolog with an amino acid sequence identity of 97% to SLAMF6 in its ligand-binding N-terminal Ig domain. We demonstrate that S1 is a cell surface glycoprotein capable of binding to host SLAMF6. Furthermore, the OMCMV genome encodes A33, an LY9 (SLAMF3) homolog, and A43, a CD48 (SLAMF2) homolog, two soluble glycoproteins which recognize their respective cellular counterreceptors and thus are likely to be viral SLAMF decoy receptors. In addition, distinct copies of further divergent CD48 homologs were found to be encoded by both CMV genomes. Remarkably, all these molecules display a number of unique features, including cytoplasmic tails lacking characteristic SLAMF signaling motifs. Taken together, our findings indicate a novel immune evasion mechanism in which incorporation of host SLAMF receptors that retain their ligand-binding properties enables viruses to interfere with SLAMF functions and to supply themselves with convenient structural molds for expanding their immunomodulatory repertoires.

IMPORTANCE The way in which viruses shape their genomes under the continual selective pressure exerted by the host immune system is central for their survival. Here, we report that New World monkey cytomegaloviruses have broadly captured and duplicated immune cell receptors of the signaling lymphocyte activation molecule (SLAM) family during host-virus coevolution. Notably, we demonstrate that several of these viral SLAMs exhibit exceptional preservation of their N-terminal immunoglobulin domains, which results in maintenance of their ligand-binding capacities. At the same time, these molecules present distinctive structural properties which include soluble forms and the absence of typical SLAM signaling motifs in their cytoplasmic domains, likely reflecting the evolutionary adaptation undergone to efficiently interfere with host SLAM family activities. The observation that the genomes of other large DNA viruses might bear SLAM family homologs further underscores the importance of these molecules as a novel class of immune regulators and as convenient scaffolds for viral evolution.

INTRODUCTION

As the immune system has evolved mechanisms to overcome viral infections, viruses have been forced to develop specific tactics to counteract host immune surveillance. Large DNA viruses such as cytomegaloviruses (CMVs), whose genomes range in size from 196 to 242 kbp and have the potential to encode about 200 distinct proteins, can dedicate a substantial part of their genome coding capacity to the production of molecules that blunt antiviral immunity, thereby guaranteeing persistent infections in their hosts (1–3). A part of these molecules exhibits evident sequence similarities to products of host genomes, and thus, they are assumed to have been hijacked by CMVs during coevolution with their hosts (4, 5). While some of these captured genes have maintained or enhanced their original functions, others have diverged to target additional biological processes, particularly immune-related processes. To date, CMV genomes have been shown to encode homologs of major histocompatibility complex class I molecules, the tumor necrosis factor receptor, Fc receptors, cytokines, chemokines, and cytokine and chemokine receptors and employed them to suppress T cell functions, interfere with natural killer (NK) cellular responses, disrupt cytokine/chemokine signaling networks, or evade antibody recognition. Thus, the study of these molecules is turning out to be instrumental in understanding crucial mechanisms of immune regulation and new strategies for their interruption by pathogens.

The signaling lymphocyte activation molecule (SLAM) family (SLAMF) of cell surface receptors, which comprises nine members (SLAMF1 to SLAMF9), is a distinct structural subgroup of the immunoglobulin (Ig) superfamily (6). SLAMF receptors are expressed by a wide range of hematopoietic cells, including T lymphocytes, NK cells, and macrophages, where they regulate several aspects of innate and adaptive immune responses (7, 8). SLAMF molecules are type I transmembrane glycoproteins containing a cytoplasmic tail, with the exception of CD48 (SLAMF2), which is a glycosylphosphatidylinositol (GPI)-anchored protein. The extracellular portion of SLAMF receptors consists of an N-terminal Ig variable domain (V region) lacking the canonical disulfide bond and a C-terminal Ig constant-2-set domain (C region), characterized by conserved cysteines. An exception to this rule is LY9 (SLAMF3), which contains a tandem repeat of two V-region/C-region sets of domains. While most SLAMF molecules are homotypic self-ligand receptors, CD244 (SLAMF4) participates in heterophilic interactions with CD48 (9). The cytoplasmic domain of nearly all SLAMF receptors bears one or more copies of the immunoreceptor tyrosine-based switch motif (ITSM) with the consensus site T-I/V-Y-X-X-V/I, which has a high affinity for the adaptor molecules SAP and EAT-2 and SH2-containing phosphatases (6, 10). Engagement of the N-terminal Ig domains of SLAMF receptors with their cognate ligands results in the recruitment of these intracellular molecules, leading to signaling transduction events that ultimately modulate several immune responses (7, 11). Moreover, coexpression of distinct isoforms of the SLAMF receptors, e.g., alternative cytoplasmic tails of SLAMF1 (12) or functionally distinct isoforms of CD244 and SLAMF6 (8, 13, 14), further refines these responses.

Being pivotal elements of the host immune response, SLAMF receptors might represent appealing targets for viral sabotage. Virus genomes might conceivably encode proteins that minimize or block the expression of SLAMF receptors on the surface of infected target cells, in order to avoid or diminish recognition by their counterreceptors expressed in effector immune cells. Two examples of this class of immune modulators have been reported to date, the Vpu protein of HIV-1, which, via downmodulation of SLAMF6, protects infected CD4⁺ T cells from lysis by NK cells (15), and the m154 protein of murine CMV, which we have recently shown reduces CD48 macrophage surface expression and debilitates antiviral NK-triggered cell responses (16). Additionally, one could envisage that viruses might have captured cellular SLAMF receptors and shaped them to interfere, via various means, with host SLAMF receptor interactions, thus preventing their activities. We have previously described that the human CMV genome encodes UL7, a protein also present in the chimpanzee CMV which contains an Ig-like domain with remarkable structural similarity (31% amino acid sequence identity) to the LY9 N-terminal Ig domain (17). Interestingly, however, we report that, despite mediating adhesion to dendritic cells and attenuation of proinflammatory responses, UL7 is not capable of binding to LY9 or any of the SLAMF receptors, indicating that this viral protein operates through a mechanism unrelated to that of the originally stolen host molecule. Thus, while the presence in viral genomes of SLAM receptor homologs that might function as SLAM decoys is an attractive hypothesis, it still remains to be substantiated.

Here, we report the existence in New World (NW) monkey CMVs of a repertoire of novel versions of SLAMF receptors with the capacity to bind their corresponding cellular counterreceptors, pinpointing an immune subversion strategy not previously described for any other pathogen.

MATERIALS AND METHODS

Identification of CMV homologs.

To search for putative CMV genes encoding proteins homologous to SLAMF members, NCBI's blastn and DELTA-BLAST were used to obtain mRNA and protein sequences, respectively (http://blast.ncbi.nlm.nih.gov).

Sequence analysis.

The sequences of the transcripts of primate and rodent SLAMF6, LY9, and CD48 were downloaded from GenBank or Ensembl (18, 19); transcript identifiers are listed in Table S1 in the supplemental material. The sequences of the open reading frames (ORFs) of squirrel monkey CMV (SMCMV) and owl monkey CMV (OMCMV) were extracted from RefSeq files NC_016448.1 and NC_016447.1, respectively. Multiple-sequence alignments of cDNA sequences were obtained using the T-Coffee program (http://tcoffee.crg.cat) and manually curated. Alignments of the protein sequences were inferred from cDNA alignments. To calculate the percent amino acid sequence identity, pairwise alignments of the CMV and the NW monkey protein sequences were obtained, and positions containing gaps were discarded.

Protein domain and motif prediction.

Ig domains were determined from annotations in the Conserved Domain Database (20). Saimiri boliviensis SLAMF6 (sbSLAMF6) residues involved in homophilic interactions were extrapolated from the human SLAMF6 sequence (21). Signal peptides and transmembrane regions were predicted by using the SignalP (v4.1) and TMHMM (v2.0) programs, respectively. The LY9 signal peptide of Callithrix jacchus was deduced from the one for human LY9 that is already known (22). Protein N-glycosylation and O-glycosylation sites were identified by using the NetNGlyc (v1.0) and NetOGlyc (v4.0) programs, respectively. These bioinformatics prediction tools are available at http://www.cbs.dtu.dk/services/SignalP/ (SignalP, v4.1), http://www.cbs.dtu.dk/services/TMHMM-2.0 (TMHMM, v2.0), http://www.cbs.dtu.dk/services/NetNGlyc (NetNGlyc, v1.0), and http://www.cbs.dtu.dk/services/NetOGlyc (NetOGlyc, v4.0).

Maximum likelihood phylogenetic analysis.

The evolutionary history was inferred by using the maximum likelihood method based on the Hasegawa-Kishino-Yano 1985 (HKY85) model (23) and applying a bootstrap test of 1,000 replicates (24). For SLAMF6 and the CD48 homologs, we considered a DNA alignment of both Ig domains in the phylogenetic analysis. For LY9 homologs, only their first N-terminal Ig domain was used. SLAM family genes of the mouse and rat were used as an outgroup. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1,000 replicates) is shown next to the branches of the consensus tree (see Fig. 2). An initial tree(s) for the heuristic search was obtained automatically as follows. When the number of common sites was <100 or less than one-fourth of the total number of sites, the maximum parsimony method was employed; otherwise, the BIONJ method with the maximum composite likelihood (MCL) distance matrix was used. A discrete gamma distribution was used to model evolutionary rate differences among sites (5 categories). The trees were drawn to the same scale, and branch lengths were measured as the number of substitutions per site. All codon positions were included. All positions with less than 80% site coverage were eliminated. That is, gaps with alignments of less than 20%, missing data, and ambiguous bases were allowed at any position. Evolutionary analyses were conducted in the MEGA5 program (25).

FIG 2.

Phylogenetic analysis using the maximum likelihood method based on the HKY85 model. The resulting bootstrap consensus trees inferred from 1,000 replicates are shown. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test is shown next to the branches. The trees are drawn to the same scale, with branch lengths being measured as the number of substitutions per site. (A) Phylogenetic tree of the Ig1 and Ig2 regions of the annotated primate SLAMF6 genes and the viral homolog S1; (B) phylogenetic tree of the Ig1 region of the annotated primate LY9 genes and the viral homolog A33; (C) phylogenetic tree of the Ig1 and Ig2 regions of the annotated primate CD48 genes and the viral homologs S30, S31, A43, A44, and A45. The corresponding SLAMF genes from the mouse and rat were used as an outgroup.

Bayesian phylogenetic analysis.

The SLAMF6 capture moment was estimated by applying BEAST (v1.7.5) with the Bayesian random local clocks method (26), using the cDNA alignments previously obtained. For the alignment of the two Ig domains (Ig1 and Ig2), BEAST was run 16 times. For the alignment of the complete coding sequence, BEAST was run 8 times. Two consensus trees, the tree obtained with Ig1 and Ig2 and the tree obtained with the complete coding sequence, were generated by combining the results of the 16 and 8 runs using the LogCombiner (v1.7.5) program of the BEAST software package. The HKY85 model (23) was used for nucleotide substitution with a discrete gamma distribution (5 categories) to model evolutionary rate differences among sites. Priors for the time to the most recent common ancestor of Primata (71.5 million years ago [mya], standard deviation [SD] = 5.0 mya), Catarrhini (30.0 mya, SD = 0.5 mya), and Callithrix-Saimiri (21 mya, SD = 1.0 mya) were set according to the bibliography (27, 28).

Estimation of evolutionary rates.

The number of nonsynonymous substitutions per nonsynonymous site (dN), the number of synonymous substitutions per synonymous site (dS), and the dN/dS ratio were estimated using the maximum likelihood method implemented in the CodeML program of PAML (v4.7) (29) by applying the free ratio model (one ratio for each branch) and the empirical estimate of the equilibrium codon frequencies (number of codon frequencies = 3).

Cell culture and viral infections.

The cell lines 300.19 (mouse pre-B cell), NS-1 (mouse myeloma), COS-7 (green monkey fibroblast), and HEL299 (human embryonic lung fibroblast) were obtained from the American Type Culture Collection (ATCC). The owl monkey kidney cell line OMK 637-69 was from Sigma-Aldrich. COS-7 and HEL299 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 2 mM glutamine, 1 mM sodium pyruvate, 50 U of penicillin per ml, 50 g of streptomycin per ml, and 10% fetal bovine serum. 300.19 and NS-1 cells were cultured in RPMI 1640 medium supplemented as indicated above plus the addition of 0.05 mM 2-mercaptoethanol to the 300.19 cell culture medium. OMK 637-69 cells were cultured in minimum essential medium supplemented with 2 mM glutamine, 50 U of penicillin per ml, 50 g of streptomycin per ml, and 10% fetal bovine serum. SMCMV was obtained from ATCC (VR-1398), and OMCMV was kindly provided by A. Davison (Medical Research Council, University of Glasgow Center for Virus Research, Glasgow, United Kingdom) (30). Viral stocks were prepared by infecting HEL299 cells with SMCMV or OMK 637-69 cells with OMCMV at a low multiplicity of infection (MOI). Cell supernatants were recovered when the maximum cytopathic effect was reached and cleared of cellular debris by centrifugation at 1,700 × g for 10 min. SMCMV and OMCMV titers were determined by standard plaque assays on HEL299 cells and OMK 637-69 cells, respectively. All the infections included a centrifugal enhancement-of-infectivity step (31).

Antibodies.

The anti-human LY9 monoclonal antibody (MAb; clone 1.84) recognizing the second Ig domain of human LY9 and the anti-hemagglutinin (anti-HA) MAb used for flow cytometry have been described previously (22, 32). The anti-human IgG (antibody 29.5, Fc specific) was previously generated in our lab (University of Barcelona) and, when indicated, was biotinylated with biotin amidocaproate N-hydroxysuccinimide ester (Sigma-Aldrich). The anti-human IgG (Fc specific) labeled with peroxidase (POD) was from Sigma-Aldrich, the anti-mouse IgG labeled with phycoerythrin (PE) was from Jackson ImmunoResearch, and the streptavidin-PE conjugate was from BD Biosciences. The biotinylated anti-HA and the rabbit anti-HA used for the Western blotting assay were from Sigma-Aldrich and Cell Signaling, respectively. The anti-rabbit IgG labeled with POD was from Promega, and the anti-mouse IgG labeled with POD was from Sigma-Aldrich. The anti-actin MAb was purchased from MP Biomedicals. The anti-S1 MAb [clone S1.1.188, IgG1(κ)] was generated by fusing an NS-1 myeloma cell line with spleen cells from a BALB/c mouse immunized three times with 300.19 cells stably transfected with HAS1 (300.19-HAS1 cells). The hybridoma was subcloned at least three times. The S1.1.188 MAb was purified using an Affi-Gel protein A MAPS II kit (Bio-Rad) from concentrated supernatants obtained from the culture of hybridomas in Integra CL 1000 flasks (Integra Biosciences AG). A characterization of the S1.1.188 MAb is shown in Fig. S2 in the supplemental material. In 300.19-HAS1 cells, this MAb recognizes the S1 protein by immunoprecipitation (see Fig. S2A in the supplemental material), Western blotting (detecting the glycosylated but not the deglycosylated protein; see Fig. S2B in the supplemental material), immunocytochemistry (see Fig. S2C in the supplemental material), and flow cytometry (see Fig. S2D in the supplemental material). In addition, the S1.1.188 MAb is able to block the binding of sbSLAMF6-Fc to S1 (see Fig. S2E in the supplemental material). All procedures involving animals and their care were approved (protocol number CEEA 308/12) by the Ethics Committee of the University of Barcelona (Barcelona, Spain) and were conducted in compliance with institutional guidelines as well as with national laws and policies (Generalitat de Catalunya decree 214/1997, DOGC 2450) and international laws and policies (33).

Plasmid constructions.

HAS1, HAA33, and HAA43 were obtained by chemically synthesizing (GenScript) DNA sequences corresponding to full-length viral proteins without their corresponding signal peptides flanked by BglII and PstI restriction sites at the 5′ and 3′ ends, respectively, and inserting them in frame with the HA at the N-terminal ends of the proteins into the mammalian expression vector pDisplay (Invitrogen). In HAA33, a silent point mutation was included in the chemically synthesized A33 DNA sequence to eliminate the internal PstI restriction site. HAA33-LY9, expressing human LY9 in which its N-terminal Ig domain was swapped for the Ig N-terminal domain of A33, was constructed as follows: a PCR product was generated using HAA33 as a template and primers XhopDisplay (5′-CTCGAGATGGAGACAGACACACTCCTG-3′) and A33IgVRev (5′-GAATTCCTCCTTGGTGGTGACTTGGGA-3′). The first primer incorporates an XhoI restriction site, and the second one incorporates an EcoRI restriction site. The resulting PCR product was inserted into the pGEM-T vector (Promega), and the XhoI-EcoRI fragment was excised and subsequently cloned into XhoI-EcoRI-digested pCI-neoHLY9. HAA43-Tm, expressing the ectodomain of A43 fused to the platelet-derived growth factor receptor transmembrane (Tm) domain, was generated by employing a PCR product obtained using HAA43 as a template and primers A43TmpDFor (5′-AGATCTGTTAATATAGTAAATGCCCTT-3′) and A43TmpDRev (5′-GTCGACTCCTGTCTGACCACATCCCGA-3′) containing BglII and SalI restriction sites, respectively. The PCR product was inserted into the pGEM-T vector, and a BglII-SalI fragment was then excised and inserted into the mammalian expression vector pDisplay (Invitrogen) that had been digested with BglII and SalI. An sbSLAMF6-Fc fusion protein containing the CD33 leader peptide and the Fc region of human IgG1 was obtained by chemically synthesizing the DNA sequences of the two Ig domains of S. boliviensis SLAMF6 with a splice donor site sequence flanked by BamHI restriction sites and cloning them into the pCI-neo Fc vector. pCI-neo Fc is derived from the pCI-neo vector (Promega) after introduction of the CD33 leader peptide and the Fc region of human IgG1 from the mammalian expression vector signal pIg-Tail (R&D Systems) within the XbaI and NotI restriction sites and disruption of its original BamHI site. In sbSLAMF6-Fc, a silent point mutation was included in the chemically synthesized sbSLAMF6 DNA sequence to eliminate the internal BamHI restriction site. S1-Fc and A33-Fc fusion proteins were constructed by employing a PCR product amplified either from HAS1 using primer set S1FcFor (5′-AGATCTCGTCTCCCAAACCAGCTCCACC-3′) and S1FcRev (5′-AGATCTACTTACCTGTCTCTGTAGACATATAGGCCAC-3′) for S1-Fc or from HAA33 using primer set A33FcFor (5′-AGATCTTAAGAATTGCTTAGCTCACACA-3′) and A33FcRev (5′-AGATCTACTTACCTGTAGCATAGCTACAAAACCGCGC-3′) for A33-Fc. The four primers contain a BglII restriction site. In addition, S1FcRev and A33FcRev include the splice donor site sequence necessary for the correct expression of these fusion proteins. The resulting PCR products were inserted into the pGEM-T vector (Promega). Subsequently, the BglII fragments were excised and cloned in frame into a BamHI-digested pCI-neo Fc vector. The Aotus trivirgatus CD244 (aoCD244)-Fc and aoLY9-Fc fusion proteins, containing the two N-terminal Ig domains of owl monkey CD244 and LY9, respectively, were constructed as follows: first, PCR products were generated using as the template DNA extracted from OMK 637-69 cells and primer sets based on regions flanking the second and third exons of CD244 or LY9 of C. jacchus or S. boliviensis. The transcript annotations of CD244 considered were those of C. jacchus (GenBank accession number XM_008984755.1) and S. boliviensis (GenBank accession number XM_010348557.1). The transcript annotation of LY9 considered was that of C. jacchus (GenBank accession number XM_002760233.1). The primers used were Intron12B4Sa/ClxFor (5′-CCTCCCACTACCTTCAGCTTG-3′) and Intron32B4Sa/ClxRev (5′-TTCAGCCCCTGTTATAGCCCA-3′) for amplification of sequences expanding from the second exon to the third exon of CD244, AotLY9intron1For (5′-TCTAGCTGCTCCTCCAAGTCC-3′) and AotLY9intron2Rev (5′-ATCAAAACACAGAAGCTCTCT-3′) for sequences corresponding to the second exon of LY9, or AotLY9intron2For (5′-GGATCTGAGCATCTTCCCATG-3′) and AotLY9intron3Rev (5′-TGGAATTCCCTCAAGGGCCAA-3′) for sequences corresponding to the third exon of LY9. The resulting PCR products were inserted into the pGEM-T vector and sequenced (see Fig. S1 in the supplemental material). Splicing by overhang extension (SOE) PCRs were then performed to join the sequences coding for the first and second Ig domains of each molecule using these generated plasmids as the templates for the first PCRs and employing primers based on their sequencing: primers BglIIAot2B4FcFor (5′-AGATCTCCAAGAATGCCAAGTTCCAGCTGACCAC-3′) and BglIIAot2B4exon3Rev (5′-AGATCTACTTACCTGTAAATCTGAATTCCTGACGGGCATTCTGACAA-3′) as external primers for CD244, primers SOEAot2B4exon2/3iso1For (5′-CAGGTTTCTGTATTTGATACAGTTGAGAAG-3′) and SOEAot2B4exon3/2iso1Rev (5′-CTTCTCAACTGTATCAAATACAGAAACCTG-3′) as internal primers for CD244, primers BglIIAotLY9FcFor (5′-AGATCTAGAGGACTCAGCCCCAACAGAGGTG-3′) and BglIIAotLY9FcRev (5′-AGATCTACTTACCTGTGGCTCCTGGATCTGTACAGAACTGCCCAGCACG-3′) as external primers for LY9, and primers SOEAotLY9exon3/2Rev (5′-CTTGCAGCTGCTCATAGACGGACAGGG-3′) and SOEAotLY9exon2/3For (5′-CCCTGTCCGTCTATGAGCAGCTGCAAGAGC-3′) as internal primers for LY9. The four external primers contain a BglII restriction site, and in addition, the reverse external primers include the splice donor site sequence necessary for the correct expression of these fusion proteins. Products from the second PCRs were inserted into the pGEM-T vector, and then the BglII fragments were excised and cloned in frame into a BamHI-digested pCI-neo Fc vector. The identities of all recombinant plasmids were confirmed by DNA sequencing.

Transfections and generation of Ig fusion proteins.

The stable cell line 300.19-HAS1 was obtained by transfecting 300.19 cells with 5 μg of plasmid expressing HAS1 with an Amaxa cell line Nucleofector kit V, followed by G418 selection and subcloning. COS-7 cells were transiently transfected with 5 μg of plasmids expressing HAA33, HAA43, HAA33-LY9, or HAA43-Tm using an Amaxa cell line Nucleofector kit R according to the manufacturer's protocol. To obtain S1-Fc, aoLY9-Fc, aoCD244-Fc, A33-Fc, and sbSLAMF6-Fc, COS-7 cells were transiently transfected as indicated above with the corresponding expression vectors. The supernatants containing the Ig fusion proteins were concentrated using an Amicon Ultra-15 centrifugal filter unit (Millipore). Quantification of Fc fusion proteins was performed by sandwich enzyme-linked immunosorbent assay using anti-human Fc IgG MAb and anti-human IgG (Fc specific) POD.

Flow cytometry analysis.

Flow cytometry was performed using standard procedures. Cells were stained with the corresponding MAbs, followed by anti-mouse IgG labeled with PE. Fc fusion protein stainings were performed using 2 to 8 μg of Ig fusion protein, followed by incubation with anti-human IgG (Fc specific) and then anti-mouse IgG labeled with PE. An irrelevant Fc fusion protein was used as a negative control. When stated, cells were preincubated with 2 μg/ml of the MAb indicated below for 30 min at 37°C before addition of the Fc fusion protein. Samples were analyzed using a FACSCalibur flow cytometer (BD Biosciences) and FlowJo software (TreeStar Inc.).

Immunoprecipitations, N-glycosidase treatments, and Western blot analyses.

Immunoprecipitations and Western blot analyses were performed as previously described (16). 300.19 and 300.19-HAS1 cells were surface labeled with biotin (Sigma-Aldrich) and lysed. Cell lysates were precleared 3 times for 30 min each time using protein G-Sepharose (GE Healthcare) and immunoprecipitated by incubation with an anti-HA–agarose conjugate (Sigma-Aldrich) or protein G-Sepharose with anti-S1 MAb. Concentrated supernatants of COS-7 cells nontransfected or transfected with either HAA33 or HAA43 were immunoprecipitated by incubation with anti-HA–agarose conjugate. Immunoprecipitates were washed, eluted, and treated when indicated with an N-glycosidase F deglycosylation kit (Roche Diagnostics GmbH) following the manufacturer's instructions. Samples from immunoprecipitates or cell extracts (from 300.19 or 300.19-HAS1 cells) that had previously been lysed and quantified by use of a bicinchoninic acid protein assay kit (Thermo Scientific) were subjected to SDS-PAGE in 10% acrylamide gels and subsequently transferred to nitrocellulose membranes (Protran). Membranes were incubated with streptavidin-POD conjugate (Roche Diagnostics GmbH) when analyzing HAS1 immunoprecipitates, biotinylated anti-HA followed by streptavidin-POD when analyzing HAA33, or rabbit anti-HA followed by anti-rabbit IgG labeled with POD when analyzing HAA43. When analyzing 300.19 and 300.19-HAS1 cell lysates, the membranes were incubated with either rabbit anti-HA followed by anti-rabbit IgG labeled with POD or anti-S1 MAb followed by anti-mouse IgG labeled with POD. Anti-actin MAb was used as a loading control. Blots were developed using a SuperSignal West Pico chemiluminescent substrate (Pierce) according to the manufacturer's protocol.

Immunofluorescence microscopy.

For immunofluorescence microscopy assays, cells (300.19, 300.19-HAS1, and HEL299 cells) were cultured on glass coverslips in 24-well tissue culture plates. In the case of 300.19 cells, coverslips coated with poly-l-lysine (Sigma-Aldrich) were used. HEL299 cells were either mock infected or SMCMV infected at an MOI of 0.2 for 3 days. Cells were washed in phosphate-buffered saline (PBS), fixed and permeabilized in ice-cold acetone, or fixed with 4% formaldehyde and subsequently blocked with 20% rabbit serum (Linus) and 6% fetal bovine serum in PBS. Samples were incubated with the primary anti-S1 MAb followed by a secondary antibody, goat anti-mouse IgG (H+L)–Alexa Fluor 555 (Life Technologies). Nuclei were counterstained with a Hoechst reagent (Invitrogen). The samples were mounted in ProLong gold antifade reagent (Invitrogen). Fluorescence images were taken with a Leica TCS SL confocal spectral microscope (Leica Microsystems GmbH, Heidelberg, Germany).

RT-PCR.

HEL299 and OMK 637-69 cells were infected with SMCMV and OMCMV, respectively, at an MOI of 0.5 for the times indicated below. Total RNA was isolated from either uninfected or infected cells by the TRIzol reagent method (Invitrogen). Reverse transcription-PCR (RT-PCR) was then carried out using a SuperScript III first-strand synthesis system for RT-PCR (Invitrogen) according to the manufacturer's protocol. Briefly, RNA samples were treated with RNase-free DNase I (Promega) for 30 min at 37°C, and the DNase was inactivated at 65°C for 10 min. The RNA was reverse transcribed using oligo(dT) primers at 50°C for 50 min, and reactions were terminated by heating at 85°C for 5 min. The reverse-transcribed products were treated with RNase H for 20 min at 37°C and amplified with specific primers. The primers used for the different genes and the sizes of the amplified fragments were as follows: primers S1FcFor and S1FcRev or primers S1FcFor and S1RevNotI (5′-GCGGCCGCTCAGAAAATCAAGGTTTCTGC-3′) for the S1 gene (632-bp or 768-bp fragment, respectively), primers S30FcBamHIFor (5′-GGATCCAGACGAATCTATTTATATCGTA-3′) and S30FcBamHIRev (5′-GGATCCACTTACCTGTGTATCGCTGAGAAAGTGGGAC-3′) for the S30 gene (614-bp fragment), primers S31FcBamHIFor (5′-GGATCCAATTTTCACATCTCAAGATCCAGATGATTACAGA-3′) and S31FcBamHIRev (5′-GGATCCACTTACCTGTTAACGATAATAAGAGGGACGAAGA-3′) for the S31 gene (638-bp fragment), primers IE1SMCMVFor (5′-TGCTACATTACAAGTTTGGGC-3′) and IE1SMCMVRev (5′-GTCAAAACACAACTGTCTCTT-3′) for the SMCMV IE1 gene (360-bp fragment), primers A33FcFor (5′-AGATCTTAAGAATTGCTTAGCTCACACA-3′) and A33FcRev (5′-AGATCTACTTACCTGTAGCATAGCTACAAAACCGCGC-3′) for the A33 gene (593-bp fragment), primers A43FcFor (5′-AGATCTCGTTAATATAGTAAATGCCCTT-3′) and A43FcRev (5′-AGATCTACTTACCTGTCAGTGCTTTAATTCCTGTCTG-3′) for the A43 gene (599-bp fragment), primers BglAotineA44For (5′-AGATCTAATCTTCCTGTTAACGCCATT-3′) and A44hLy9EcoRIRev (5′-GAATTCGTGATCTTGTTGGCTATTGTC-3′) for the A44 gene (300-bp fragment), primers A45FcBglFor (5′-AGATCTAGAACCTACACCGAGTCTCTTG-3′) and A45FcBglRev (5′-AGATCTACTTACCTGTATGGTATGTACGTATTGATTC-3′) for the A45 gene (637-bp fragment), primers IE1OMCMVFor (5′-ACAGCATCTAGGTTGCCAAGA-3′) and IE1OMCMVRev (5′-GTCAGTGGAGCCCCAAGTCAA-3′) for the OMCMV IE1 gene (590-bp fragment), and primers GAPDHFor (5′-CACCAACTGCTTAGCACCCC-3′) and GAPDHRev (5′-CCATCACTGCCACCCAGAAGA-3′) for the GAPDH (glyceraldehyde-3-phosphate dehydrogenase) gene (101-bp fragment). PCRs were performed under the following conditions: 1 cycle at 94°C for 5 min; 30 cycles of 1 min at 94°C, 1 min at 51°C, and 1 min at 72°C; and 1 cycle at 72°C for 10 min. Control reactions carried out in the absence of reverse transcriptase were used to assess the specific detection of RNA. Amplified products were separated on a 1% agarose gel and visualized by SYBR green staining (Life Technologies) or by use of RedSafe nucleic acid staining solution (iNtRON Biotechnology Inc.). RT-PCR-amplified fragments were cloned into the pGEM-T vector, and their identities were confirmed by DNA sequencing.

Nucleotide sequence accession numbers.

The newly identified nucleotide sequences reported here (see Fig. S1 in the supplemental material) were deposited in GenBank under the following accession numbers: KT340627 (exon 2 and exon 3 of Aotus trivirgatus LY9 with partial flanking introns) and KT340628 (exon 2, intron 2, and exon 3 of Aotus trivirgatus CD244 with partial flanking introns).

RESULTS

Identification of SLAMF receptor homologs in NW monkey CMVs.

For an initial scanning of putative CMV homologs of SLAMF genes, we used the transcript and protein sequences of the human SLAMF members in NCBI BLAST searches over the CMV sequence data set available in GenBank (18). In addition to UL7 sequences common to human and chimpanzee CMVs (17), we found SLAMF homologs in two CMVs infecting NW monkeys: SMCMV, which infects Saimiri sciureus (the Guyanese squirrel monkey), and OMCMV, which infects Aotus trivirgatus. We identified seven putative SLAMF homologs, which are located in the terminal regions of the viral genomes (Fig. 1A), similar to other immunomodulatory CMV genes (34).

FIG 1.

Genome localization and protein characterization of New World monkey CMV homologs of SLAMF genes. (A) Localization and coding direction of each CMV homologous ORF along its corresponding genome (SMCMV or OMCMV). Yellow, the SLAMF6 homolog (S1); orange, the LY9 homolog (A33, biexonic); blue, the CD48 homologs (S30, S31, A43, A44, and A45); black and gray lines, unique long and unique short regions, respectively. Terminal and internal repeats (terminal repeat of the unique long region [TRL], terminal repeat of the unique short region [TRS], internal repeat of the unique long region [IRL], internal repeat of the unique short region [IRS]) are also indicated. (B) Protein domains of SLAMF6, LY9, CD48, and their SMCMV and OMCMV homologs and the percent amino acid sequence identity between the viral homologs and their cellular counterparts for each domain. Yellow, the signal peptide (SP); blue or green, the Ig domains (Ig1, Ig2, Ig3, Ig4); gray, the S31 and A45 stalks; orange, the transmembrane region (TM); white, the cytoplasmic tail (CT); fuchsia, the endoplasmic reticulum membrane domain of CD48 that is cleaved off and replaced by the GPI anchor (GPI); light blue, a region of A33 that may be part of the signal peptide or that may be part of the Ig1 domain, as the limit between SP and Ig1 is unclear in this molecule. The sequence of the last part of S1 (transmembrane region and cytoplasmic tail) was compared to the C. jacchus SLAMF6 sequence, because the corresponding region of sbSLAMF6 (marked by dotted lines) is still incomplete. Green marks under Ig2, Ig3, and Ig4 of LY9, fragments that are aligned with Ig2 of A33; black triangles inside cytoplasmic tails, ITSM motifs; blue and red lines below each protein, predicted N- and O-glycosylation sites, respectively. Proteins and domains are drawn to scale. Protein lengths are indicated at the right for all proteins except sbSLAMF6, because the length of the incomplete fragment is unknown.

Characterization of the proteins encoded by the SMCMV and OMCMV homologs of SLAMF genes.

Proteins predicted to be encoded by all of these homologous genes showed the typical Ig-like membrane receptor structure: a type Ia membrane protein with a signal peptide cleaved out in the endoplasmic reticulum, two Ig domains (Ig1 and Ig2), a transmembrane region, and a cytoplasmic tail.

We aligned the transcript and protein sequences of the viral genes and the corresponding primate genes (see Fig. S3 in the supplemental material). Figure 1B plots the different domains of the seven CMV proteins and the percent amino acid sequence identity for each domain compared to the available protein sequence of the primate evolutionarily closest to the CMV host. It should be noted that we currently have SLAMF protein sequences for only two NW monkeys, the genomes of which have been sequenced: Saimiri boliviensis (Bolivian squirrel monkey) and Callithrix jacchus (marmoset) (35).

S1, the product of the first ORF of SMCMV (Fig. 1A), is a protein of 285 amino acids (aa) that exhibits an overall identity of 77% with sbSLAMF6, including nearly complete identity on its two Ig domains (97% on Ig1, 93% on Ig2; Fig. 1B). Moreover, 5 out of the 6 residues that are conserved between human SLAMF6 and sbSLAMF6 Ig1 and that are involved in the homophilic interaction are present in S1 (see Fig. S4 in the supplemental material) (21).

A33, a two-exon ORF of OMCMV encoding a protein of 296 aa, is a homolog of LY9 (Fig. 1). A remarkable feature of this viral protein is the presence of two Ig domains instead of the 4 Ig domains (Ig1 to Ig4) characteristic of LY9 molecules (Fig. 1B). Ig1 of A33 showed a high percent amino acid sequence identity (71%) with Ig1 of C. jacchus LY9 (cjLY9). Interestingly, Ig2 of A33 contains fragments (at the start) similar to Ig2 of cjLY9 and others (at the end) similar to the Ig4 domain of cjLY9. Notably, the cytoplasmic tail of A33 is very short (15 aa), whereas the tail of LY9 is long (182 aa).

In addition, both CMVs contain genes homologous to CD48: S30 and S31 in SMCMV and A43, A44, and A45 in OMCMV (Fig. 1). Interestingly, the proteins encoded by these viral genes have a transmembrane region, whereas CD48 is a GPI protein (Fig. 1B) (36). In general, the percent homology of the Ig domains is considerably lower than that of the Ig domains of S1 and A33. However, Ig1 of A43 has a high percent identity to Ig1 of sbCD48 (80%). The cytoplasmic tails of S30 and A44 are nearly absent, with only 4 aa and 2 aa, respectively. On the other hand, S31, A43, and A45 have intracellular regions of 22 aa, 31 aa, and 20 aa, respectively.

We predicted the N- and O-glycosylation sites along the extracellular amino acid sequences of sbSLAMF6, cjLY9, and sbCD48 and their homologous CMV proteins (Fig. 1B; see also Fig. S4 in the supplemental material). In general, we observed a tendency for a higher number of N-glycosylation sites in the CMV homologs. Even S1, the most conserved homolog, showed two new N-glycosylation sites. Moreover, S31 and A45 presented long stalks containing abundant threonine and serine residues and were predicted to have high numbers of O-glycosylation sites. Finally, in their intracellular regions, none of the viral homologs had the ITSM motifs characteristic of the cytoplasmic tail of most SLAMF receptors (Fig. 1B; see also Fig. S4 in the supplemental material) (7).

Phylogenetic analysis of the CMV homologs of SLAMF genes.

To unravel the genesis of these CMV genes and support their homologous nature, we calculated, using the maximum likelihood method, phylogenetic trees from the DNA sequences of the Ig domains of these genes and primate SLAMF6, LY9, and CD48 sequences (Fig. 2). The seven CMV sequences were clearly clustered with the SLAMF members of the NW monkeys. Given that all except A33 are monoexonic and considering the calculated trees, we hypothesize that the origin of these CMV genes was gene capture from the host by retrotranscription at different stages of virus-host coevolution. The first exon of A33 and its splicing were likely added after the capture of LY9 from the host. The lengths of the tree branches suggest that the CD48 homologs are the product of the oldest capture, whereas the most recent capture generated S1.

In order to estimate the moment of SLAMF6 capture that led to S1, we used a Bayesian method with random local clocks implemented in BEAST (26) that tries to solve the problem of having different nucleotide substitution rates among the various lineages, particularly when virus sequences are included. The result suggests a really recent capture, 0.57 or 1.3 mya, depending on whether we consider only the two Ig domains or the entire coding sequence, respectively (see Fig. S5 in the supplemental material). This is long after the appearance of the first Saimiri species, 16 mya or before (28). The SLAMF6 mRNA was almost completely copied since, apart from some homology still present in the signal peptide and cytoplasmic tail of S1, the 90 nucleotides downstream of the S1 stop codon bear 94% identity with the nucleotide sequences of three fragments from the 3′ untranslated region of SLAMF6 (see Fig. S3 in the supplemental material).

Our phylogenetic trees hint at a unique gene-capture event that would explain the emergence of the five CD48 homologs. This event took place before the split of the two CMVs and probably before the split between the two species of NW monkeys. The captured gene suffered a further gene duplication episode in the CMV ancestor of SMCMV and OMCMV, giving rise to the precursors of S30/A43-A44 and S31/A45. After CMV speciation, A43 and A44 appeared via an additional gene duplication only in OMCMV. This explains the resemblance between S31 and A45 (similar sequences, long highly O-glycosylated stalks, matching N-glycosylation sites) and their parallel position in the genome (Fig. 1).

The substantial evolution of the CMV SLAMF homolog genes after being captured contrasts with the high conservation of their Ig domains.

After capture, the new genes suffered substantial changes in some parts of the coding sequence. Even the most recently captured gene, S1, showed deletions in its signal peptide and cytoplasmic regions with respect to the sequence of SLAMF6 (see Fig. S3 in the supplemental material). Moreover, long deletions took place in the Ig2, Ig3, and Ig4 regions of the LY9 retrocopy that created A33, thus resulting in a unique Ig2 domain (Fig. 1; see also Fig. S3 in the supplemental material). In addition, modifications in the CD48 homologs generated new transmembrane and cytoplasmic domains and stalk regions in S31 and A45. These substantial alterations contrast with the high conservation of the Ig domains (especially Ig1), indicated by the percent amino acid sequence identity (Fig. 1) and strongly supported by our subsequent evolutionary rate analysis. We calculated the number of nonsynonymous substitutions per nonsynonymous site (dN), the number of synonymous substitutions per synonymous site (dS), and the dN/dS ratio between the human SLAM gene and the NW monkey gene(s) or the CMV gene(s) by considering the full coding sequence or only the Ig domains. As shown in Table 1, whereas the dS values of the CMV homologs were at least 14-fold higher than those of the NW monkey homologs, dN values showed an increase of 5-fold or less. This was especially notable for Ig domains; for instance, dN was practically the same for S1 and NW monkey SLAMF6, while the dS of S1 was 19-fold higher. These values indicate that stronger functional constraints are operating on the Ig domains, particularly Ig1, and suggest that the binding to the ligand may be partially or totally preserved.

TABLE 1.

Evolutionary rates

Pair of genes	dN	dS	dN/dS
Human SLAMF6 (Ig1 + Ig2) and:
SMCMV S1 (Ig1 + Ig2)	0.1267	1.9549	0.0648
Squirrel monkey SLAMF6 (Ig1 + Ig2)	0.1252	0.1019	1.2286
Marmoset SLAMF6 (Ig1 + Ig2)	0.1188	0.1024	1.1604
Human SLAMF6 (all) and:
SMCMV S1 (all)	0.1759	1.3947	0.1261
Squirrel monkey SLAMF6 (all)	0.1241	0.0947	1.3104
Marmoset SLAMF6 (all)	0.1215	0.0928	1.3092
Human LY9 (Ig1) and:
OMCMV A33 (Ig1)	0.2117	2.5949	0.0816
Marmoset LY9 (Ig1)	0.1265	0.1105	1.1451
Human LY9 (all) and:
OMCMV A33 (all)	0.4104	1.4099	0.2911
Marmoset LY9 (all)	0.0883	0.0983	0.8979
Human CD48 (Ig1 + Ig2) and
OMCMV A43 (Ig1 + Ig2)	0.3856	64.2131	0.0060
OMCMV A44 (Ig1 + Ig2)	0.7289	4.2712	0.1706
OMCMV A45 (Ig1 + Ig2)	1.0314	55.3423	0.0186
SMCMV S30 (Ig1 + Ig2)	0.8245	55.1124	0.0150
SMCMV S31 (Ig1 + Ig2)	0.9440	24.5025	0.0385
Squirrel monkey CD48 (Ig1 + Ig2)	0.1888	0.1623	1.1631
Marmoset CD48 (Ig1 + Ig2)	0.1908	0.1320	1.4457

S1 interacts with SLAMF6 and functions as a homophilic receptor.

We first sought to explore the properties of S1 in more detail. The viral gene was tagged with the HA epitope in its N terminus to produce HAS1 and used to transfect 300.19 cells and generate a stable cell line (300.19-HAS1). Flow cytometry analysis using an anti-HA antibody showed that S1 is expressed at high levels on the cell surface (Fig. 3A). Since SLAMF6 is its own ligand, we investigated whether sbSLAMF6, expressed as an Ig fusion protein comprising its extracellular domain and the Fc region of human IgG1 (sbSLAMF6-Fc), was able to interact with the stable 300.19-HAS1 transfectant. As illustrated in Fig. 3B, the sbSLAMF6-Fc fusion protein efficiently recognized 300.19-HAS1 cells, while no binding to 300.19 cells could be detected. The specificity of the interaction was further supported by the finding that an irrelevant control fusion protein failed to recognize 300.19-HAS1 cells (Fig. 3B). In addition, we explored, through the construction of an S1-Fc fusion protein, whether S1-S1 homophilic interactions could take place. Notably, Fig. 3B shows that S1 has retained the capacity to associate with itself.

FIG 3.

Expression, cellular localization, and ligand interaction capacities of S1. (A) Flow cytometry analysis of 300.19 cells nontransfected (left) or stably transfected with HAS1 (300.19 HAS1; right) and stained with anti-HA MAb (α-HA MAb; shaded histogram) or an isotype control (open histogram) followed by anti-mouse IgG-PE. (B) 300.19 cells nontransfected (left) or stably transfected with HAS1 (300.19 HAS1; right) were incubated with sbSLAMF6-Fc, S1-Fc, or an unrelated negative control-Fc fusion protein (CTL-Fc) followed by anti-human Fc IgG and anti-mouse IgG-PE and then analyzed by flow cytometry (shaded histograms). Open histograms represent isotype controls. In panels A and B, the percentage of positive cells is indicated in each histogram. (C) 300.19 cells nontransfected (300.19) or stably transfected with HAS1 (300.19 HAS1) were surface labeled with biotin, lysed, and then immunoprecipitated with an anti-HA MAb. When indicated, immunoprecipitates were treated with N-glycosidase F (N-Gly F). Samples were subjected to SDS-PAGE and subsequent Western blot analysis using streptavidin-POD. Molecular masses (in kilodaltons) are marked.

To gain further insights into the S1 protein, we immunoprecipitated the viral product from biotinylated 300.19-HAS1 cells with an anti-HA MAb and analyzed it by Western blotting. A single broad band of 61 to 74 kDa was detected (Fig. 3C). Digestion with N-glycosidase F enhanced the migration of the specific band to 34 kDa (Fig. 3C), thus confirming that S1 undergoes an extensive posttranslational N-glycosylation process.

A33 and A43 are capable of binding to LY9 and CD244, respectively.

We then extended the analysis to A33 and A43, the two OMCMV ORFs exhibiting the highest homology to SLAMF members. The proteins encoded by both genes were tagged with the HA epitope in their N-terminal end to generate the HAA33 and HAA43 constructs and used to transiently transfect COS-7 cells. As depicted in Fig. 4A, both viral proteins were detected, although in modest amounts, at the cell surface. To assess whether A33 and A43 could mediate interactions with their corresponding cellular counterparts, LY9 and CD244, respectively, we first ensured a high level of cell surface expression of both viral molecules. To this end, we transfected COS-7 cells with HAA33-LY9, a chimeric molecule in which the Ig1 domain of human LY9 was exchanged for the Ig1 domain of A33, or with HAA43-Tm, a chimeric molecule expressing the A43 ectodomain fused to the platelet-derived growth factor receptor transmembrane region (Fig. 4B). As the Aotus trivirgatus genome has not been sequenced, for these assays we PCR amplified DNA sequences corresponding to the two N-terminal Ig domains of LY9 and CD244 from the owl monkey kidney cell line OMK 637-69 and generated with them constructs that allowed the expression of the Fc fusion proteins aoLY9-Fc and aoCD244-Fc, respectively. Figure 4C shows that aoLY9-Fc fusion protein was capable of selectively binding, although at low levels, to COS-7 cells expressing HAA33-LY9 but not to COS-7 cells. By means of constructing an A33-Fc fusion protein, we explored as well whether A33 could self-associate. Unlike S1, which is able to mediate homophilic interactions, we could not detect a significant association of A33 to itself (data not shown). In the case of A43, the aoCD244-Fc fusion protein largely and specifically recognized the HAA43-Tm expressed in COS-7 cells (Fig. 4C). Moreover, this interaction could also be detected even when low concentrations of A43 were present at the cell surface in COS-7 cells transfected with HAA43 (data not shown). Altogether, these results indicate that both A33 and A43 are capable of interacting with their cognate cellular counterparts.

FIG 4.

Expression and ligand-binding capacities of A33 and A43. (A) Flow cytometry analysis of COS-7 cells nontransfected (COS-7) or transfected with HAA33 (COS-7 HAA33) or HAA43 (COS-7 HAA43) and stained with anti-HA MAb (α-HA MAb; shaded histograms) or an isotype control (open histograms) followed by anti-mouse IgG-PE. (B) COS-7 cells nontransfected (COS-7) or transfected with HAA33-LY9 (COS-7 HAA33-LY9) or HAA43-Tm (COS-7 HAA43-Tm) were stained as indicated with a MAb recognizing either the second Ig domain of human LY9 or the HA epitope (α-LY9 MAb and α-HA MAb, respectively; shaded histograms) or an isotype control (open histograms) followed by anti-mouse IgG-PE. (C) Flow cytometry of COS-7 cells nontransfected or transfected as stated in the legend to panel B and incubated with either aoLY9-Fc, aoCD244-Fc, or an unrelated negative control-Fc fusion protein (CTL-Fc), as indicated, followed by anti-human Fc IgG and anti-mouse IgG-PE (shaded histograms). Open histograms represent isotype controls. The percentage of positive cells is indicated in each histogram.

The low cell surface levels of both A33 and A43 (Fig. 4A) could be indicative of a proteolytic cleavage of these molecules and the shedding of their soluble extracellular domains. Accordingly, stimulation with the phorbol 12-myristate 13-acetate led to a substantial decrease in the amounts of the two viral proteins at the cell surface (data not shown). We further analyzed this possibility by examining the medium of COS-7 cells transiently transfected with either HAA33 or HAA43 through immunoprecipitation with an anti-HA antibody and subsequent Western blot analysis. As shown in Fig. 5A, a released major A33 protein with a molecular mass of 54 to 60 kDa could be readily detected. A minor band possibly corresponding to a different glycosylated protein form of 65 kDa was also observed. In a similar way, A43 was also found to be shed into the medium as a protein with a molecular mass of 47 to 50 kDa (Fig. 5B). The sensitivity of soluble A33 and A43 to N-glycosidase F (Fig. 5A and B), leading to single proteins with molecular masses of 23 kDa and 29 kDa, respectively, corroborated the highly glycosylated nature of the two viral proteins. These results were in agreement with the 12 potential N-linked glycosylation sites that A33 bears and the 7 ones present in A43 (Fig. 1B).

FIG 5.

A33 and A43 are shed from the cell surface. The supernatants of COS-7 cells nontransfected (negative control; COS-7) or transfected with HAA33 (COS-7 HAA33) (A) or with HAA43 (COS-7 HAA43) (B) were immunoprecipitated with an anti-HA MAb. When indicated, immunoprecipitates were treated with N-glycosidase F (N-Gly F). Samples were subjected to SDS-PAGE and subsequent Western blot analysis using biotinylated anti-HA MAb and streptavidin-POD for HAA33 or rabbit anti-HA followed by anti-rabbit IgG-POD for HAA43. Molecular masses (in kilodaltons) are marked.

Expression of SMCMV and OMCMV homologs of SLAMF genes during infection.

We next asked whether these viral genes are expressed in the context of the infection. We started to explore this issue for the SMCMV SLAM homolog genes by performing RT-PCR analyses using the human embryonic lung fibroblast cell line HEL299, shown to be permissive for SMCMV (37). Total RNA was extracted from either mock-infected or SMCMV-infected HEL299 cells for 3 days and reverse transcribed before being amplified with primers specific for S1, S30, S31, and the SMCMV immediate early gene IE1 as a control. As illustrated in Fig. 6A, while transcripts for the three SMCMV SLAM homologs (S1, S30, and S31) were undetectable in mock-infected samples, they could be clearly identified in SMCMV-infected cells. As expected, the presence of the RT-PCR product corresponding to the spliced IE1 transcript was also observed. Subsequently, we sought to detect the expression of OMCMV genes A33, A43, A44, and A45 during infection of OMK 637-69 cells, carrying out RT-PCR assays similar to those described above using primers specific for these genes and the OMCMV immediate early gene IE1 as a control. Likewise, specific transcription of the four SLAM homologs and the IE1 gene of OMCMV was observed in infected cells (Fig. 6B). In all cases, the absence of PCR products in control reaction mixtures lacking reverse transcriptase verified the identities of the amplified fragments as cDNAs (data not shown). In addition, the RT-PCR products were confirmed by DNA sequencing. Therefore, the data indicate that the SMCMV and the OMCMV SLAM homologs analyzed get expressed during the viral infection cycle.

FIG 6.

Detection of viral SLAMs in SMCMV- or OMCMV-infected cells. (A) HEL299 cells were mock infected (mock) or infected with SMCMV at an MOI of 0.5. Total RNA was harvested at 3 days postinfection, treated with DNase, and reverse transcribed by using oligo(dT). PCRs were performed utilizing primer sets specific for S1, S30, S31, SMCMV IE1, and GAPDH as described in Materials and Methods. Amplified products were separated on 1% agarose gels and visualized by SYBR green staining or the use of RedSafe nucleic acid staining solution. The amplified fragments obtained in the different reactions are shown. Sizes were as expected for each primer set (see Materials and Methods for details). Specific PCR-amplified products were not detected in control reactions in which no reverse transcriptase was added during the RNA reverse transcription reaction (data not shown). (B) Same as described in the legend to panel A, except that OMK 637-69 cells were mock infected (mock) or infected with OMCMV, and RT-PCRs were performed using primer sets specific for A33, A43, A44, A45, OMCMV IE1, and GAPDH. (C) Same as described in the legend to panel A, except that HEL299 cells were mock infected (mock) or infected with SMCMV for the indicated times and RT-PCRs were performed using primer sets specific for S1, SMCMV IE1, and GAPDH. (D) HEL299 cells were mock infected or infected with SMCMV at an MOI of 0.2 for 3 days. Cells were fixed with 4% formaldehyde (a to f) or fixed and permeabilized with acetone (g to l) and stained with anti-S1 MAb followed by an anti-mouse IgG-Alexa Fluor 555. The cells were examined under a microscope at 405 nm (Hoechst) and at 510 to 560 nm (Alexa Fluor 555). Shown are representative cells from cultures stained for S1 (b, e, h, and k), nuclei stained with the Hoechst reagent (a, d, g, and j), and overlaid images (c, f, i, and l). Magnifications, ×40; bars, 10 μm. (E) Flow cytometry analysis of HEL299 cells mock infected (mock) or infected with SMCMV at an MOI of 5 for 3 days. Cells were stained with anti-S1 MAb S1.1.188 (shaded histograms) or an isotype control (open histograms) followed by anti-mouse IgG-PE. The percentage of positive cells is indicated in each histogram.

Next, we chose to examine more in depth the expression of S1 over the course of the infection. To this end, we first performed RT-PCR assays on HEL299 cells infected with SMCMV for different times (6 h, 24 h, 48 h, 72 h, 96 h, and 120 h). As shown in Fig. 6C, S1 transcription could be first detected at 24 h after infection, and it slightly increased by 48 h and continued during the infectious cycle (Fig. 6C). As expected, under these conditions, IE1 transcription was readily observed from 6 h after infection. We then generated an MAb (S1.1.188) by immunizing mice with 300.19 cells expressing S1. The anti-S1 MAb specifically recognizes an epitope of the ectodomain of this viral protein (see Fig. S2 in the supplemental material for a characterization of MAb S1.1.188). We used this MAb (S1.1.188) to assess the expression and localization of S1 within the infected cells by confocal microscopy of HEL299 cells infected with SMCMV for 3 days. Consistent with the results obtained with the viral protein when it was independently expressed (Fig. 3A and C), under nonpermeabilizing conditions the S1 protein could be also definitely identified at the infected cell surface (Fig. 6D, panels e and f), while no specific staining was detected in uninfected cells (Fig. 6D, panels b and c). Imaging of permeabilized SMCMV-infected HEL299 cells revealed specific S1 staining throughout the cytoplasm with a punctate distribution, and a substantial proportion of the protein was located at the plasma membrane (Fig. 6D, panels k and l). In addition, the surface expression of S1 in the infected cells was corroborated by flow cytometry (Fig. 6E). Taken together, we concluded from these experiments that during infection S1 is expressed in significant amounts, being able to reach the cell surface.

DISCUSSION

Horizontal gene transfer from host cells is thought to be a key process that contributes to viral adaptation and evolution. Comparative genome sequence analysis allows the reconstruction of the history of the captured viral genes and their evolutionary relationships with their cellular homologs (38). In this study, we report the presence of a number of homologs of SLAMF genes in NW monkey CMV genomes. Specifically, a SLAMF6 homolog and two CD48 homologs were identified in SMCMV and an LY9 homolog and three CD48 homologs were identified in OMCMV. Our phylogenetic analyses revealed that SLAMF6, LY9, and CD48 were acquired by retrotranscription at three different moments of virus-host coevolution by the NW monkey CMVs. The S1 cell surface molecule, whose sequence displays homology of 93 to 97% with sbSLAMF6 in its two Ig domains, is the product of the most recent capture event, which was 1.3 mya or less, and represents the first evidence for the capture of a host gene by CMV involving an entire gene, including part of the 3′ untranslated region. Only the viral homologs of interleukin-10 and dihydrofolate reductase present in the genomes of some gammaherpesviruses exhibit higher protein sequence homology with their ancestral host genes (39–41), though in those cases the proteins are considerably shorter than SLAMF6 (178 aa, 187 aa, and 332 aa for human interleukin-10, dihydrofolate reductase, and SLAMF6, respectively). Our analysis also indicates that the five viral CD48 homologs are the product of a unique gene-capture event that presumably occurred before the split between the Saimiri and Aotus species (more than 19 mya) (28), which was followed by two duplication episodes. Alternatively, the capture could have taken place after Saimiri and Aotus speciation and before CMV speciation (the split between SMCMV and OMCMV). It would be necessary to evaluate more sequences of Saimiri and Aotus species to determine more precisely when the capture event occurred.

Recent studies have shown that after eukaryotic gene duplication there is an accelerated sequence divergence in the novel copy created (42). Moreover, a CMV gene resulting from capture of a host gene is expected to experience an even stronger divergence, in view of the higher nucleotide substitution rate of double-stranded DNA viruses (43). However, as our results indicate, this higher divergence might be hampered by functional constraints in those protein domains that are preserved for viral advantage. In that respect, we observed exceptional preservation of viral Ig regions, particularly of the N-terminal Ig domain, of several of these molecules. These preserved sequences included for example, critical residues, such as amino acids Ser-39, Phe-42, His-54, Thr-56, and Ser-90, which are involved in SLAMF6 ligand binding (see Fig. S4 in the supplemental material) (21). Accordingly, we show that the three viral proteins analyzed herein in more detail, S1, A33, and A43, maintain the ligand-binding capacities of their cellular counterparts, being able to interact with SLAMF6, LY9, and CD244, respectively. It is of note that these molecules constitute the first description of homologs of SLAMF receptors conserving the binding abilities encoded by the genome of a pathogen.

Thus, it is expected that these viral SLAMF homologs retaining their natural ligand-binding capacities may act by interfering with the functions of their corresponding cellular counterparts, thereby influencing the course of infection. This interpretation is consistent with the finding of functional isoforms of several SLAMF genes with antagonistic activities (8, 12, 13, 44). SLAMF6, LY9, and CD244 are differentially expressed on the surface of a wide variety of hematopoietic cells, including CMV-permissive macrophages and dendritic cells, and effector NK and cytotoxic CD8⁺ T cells known to play a central role in the control of CMV infection. Among other activities, these SLAM receptors participate in the regulation of cytokine and cytotoxic cellular responses. As a consequence, CMV might have tailored the stolen host SLAMs on target cells to sabotage SLAM recognition by their counterreceptors in cytotoxic cells. Accordingly, we report that the acquired cellular LY9 and one of the CD48 genes have been specifically modified and engineered by OMCMV to yield two soluble viral glycoproteins, A33 and A43, clearly detected in the supernatants of transfected cells. The existence of these soluble forms of CMV SLAMF homologs may well represent a mechanism evolved by the virus to locally deliver decoy molecules that directly block the corresponding SLAMF receptors on effector cells. Thus, through the production of such SLAM decoy receptors, NW monkey CMVs could be preventing NK and cytotoxic T cell recognition of infected cells, further promoting viral persistence in their hosts. In this line, a soluble form of CD48 has been detected in plasma and its levels have been documented to be elevated in patients with lymphoid leukemias and arthritis (45). Moreover, Elias et al. (46) have reported that CD48 expression is downregulated in patients with acute myeloid leukemia due to oncogenic fusion proteins PML-RARA and AML1-ETO, thereby failing to sustain binding of the human activating NK receptor CD244 and ultimately resulting in reduced killing by NK cells. Similarly, the Vpu protein of HIV (15) and the m154 protein of murine CMV (16) have been shown to diminish the expression of SLAMF6 (Vpu) and CD48 (m154) on the surface of infected cells, leading to impaired NK-triggered cell responses. Thus, altogether these findings underscore the importance of both oncogenic and viral proteins targeting SLAMF members to escape immune detection.

We show, on the other hand, that S1 gets expressed on the surface of the infected cell. The surface density of some SLAM receptors on the target cell, such as CD48, has been proposed to be one of the factors that dynamically regulates the predominance of activating or inhibitory signals on effector cells (47, 48). Thus, increased CD48 density on the target cell can make expression of the adaptor protein SAP limiting, providing negative signals downstream of CD244 on NK or CD8⁺ T cell effector cells. In this context, it is also possible that the expression of high levels of the membrane-bound viral SLAMF receptor homolog S1 on the surface of infected cells by SMCMV results in the inhibition of effector cell functions after counterreceptor triggering, as postulated for the CD48-CD244 interaction.

Additional clues as to the modes of action used by some of the identified viral homologs might arise not only from their binding properties but also from their unique structural features compared to those of their host counterparts. It must be taken into account that despite the striking sequence similarities with their cellular homologs, viral SLAMF receptors have significantly altered specific parts of their structure, plausibly to hinder cell signaling. The reported CMV SLAMF receptor homologs lack recognizable signaling motifs in their cytoplasmic tails, including the ITSMs and other regions that interact with adaptor proteins, known to be essential for SLAM functionality (6). The GPI anchor in CD48 is critical for its association with the lipid rafts and its recruitment into the immunological synapse (49). Importantly, one of the hallmarks of the viral CD48 homologs is the replacement of their GPI anchors by a transmembrane region and a cytoplasmic tail, modifications that can result in altered localization and stability at the cell surface. Moreover, it is worth noting the size changes that some of these SLAMF homologs have undergone. They are either substantially trimmed down in size, as is the case for A33, which contains only two of the four Ig domains present in LY9, or significantly expanded, as shown by two of the CD48 homologs, S31 and A45, which display long and highly O-glycosylated stalks. The latter alterations may well represent a steric impediment at the contact interface between effector and target cells. In this connection, elongated forms of CD48 have been shown to be profoundly inhibitory by disrupting the symmetry of the immunological synapse (50). Finally, carbohydrate modifications in cell surface receptors markedly change the properties and modulate the functions of these glycoproteins, as illustrated for CD244, where the sugar content has been reported to have an important impact on CD48 binding and NK cell responses (51). The fact that the viral SLAMF homologs are predicted to be more extensively glycosylated than the corresponding cellular SLAMF receptors may be of relevance not only to preserve their stability or prevent unwanted interactions but also for functional recognition.

However, it is also plausible that some of the CMV SLAMs have evolved to additionally target other immune receptors or ligands. In particular, this could apply to the more divergent viral CD48 homologs or even to A33. The capacity of A33 to bind to its cellular counterpart LY9 is low, and in contrast to S1, A33 does not interact homophilically. In this regard, while UL7, a more distant LY9 homolog of human CMV, retains the ability to mediate adhesion to dendritic cells, it does not associate with itself or any SLAMF receptor (17). This process is reminiscent of the evolution of Ig superfamily members, where proteins have been proposed to evolve from an ancestral homotypic adhesion molecule and only to later, via gene duplication and divergence, change into more complex heterotypic interaction molecules (52). Similarly, the presence of the more divergent CD48 homolog copies, S30, S31, A44, and A45, in the NW monkey CMV genomes presumably evidences the emergence of new functional properties, revealing that the SLAM structure is a convenient scaffold for viral evolution.

Moreover, CMV pathogenesis is dependent on the efficient hematogenous spread of the virus to various host cells (53). Thus, it might also be possible that CMV would be making use of the adhesion properties of some of these cell surface-exposed molecules captured from the host to facilitate its dissemination via cell-to-cell spread, getting access to uninfected SLAM-containing permissive myeloid cells. The activities and mechanisms that each CMV SLAM homolog uses to impart its effects are most likely different. Understanding how each of these viral molecules operates should provide novel and valuable insights into viral pathogenesis and into the way viruses subvert host immune responses.

Importantly, the potential exploitation of virus genome-encoded SLAMF homologs does not seem to be unique to NW monkey CMVs. Indeed, our extensive sequence similarity searches indicate that a number of complex and evolutionarily diverse DNA viruses infecting a variety of hosts, including humans, contain recognizable SLAMF homologs incorporated within their genomes. For instance, putative CD48 homologs can be found in poxviruses (e.g., F5 in squirrel pox virus), adenoviruses (e.g., ORF11 in turkey adenovirus 5), and additional herpesviruses (e.g., E52 in elephant herpesvirus 1A), further supporting the notion that harboring SLAMF homologs confers an evolutionarily selective advantage to viruses.

Altogether, the results presented here reinforce the relevance of SLAMF receptors in antiviral immune responses and introduce viral SLAMF homologs as an important new class of immune modulators. Moreover, these viral proteins might provide key tools for better comprehending the functions and mechanisms of action of the SLAMF receptors in immunity.