The murine CMV (MCMV) immunoevasin m04/gp34 escorts MHC class I (MHC I) molecules to the surface of infected cells where these complexes bind Ly49 inhibitory receptors (IRs) and prevent NK cell attack. Nonetheless, certain self–MHC I–binding Ly49 activating and inhibitory receptors are able to promote robust NK cell expansion and antiviral immunity during MCMV infection. A basis for MHC I-dependent NK cell sensing of MCMV-infected targets and control of MCMV infection however remains unclear. In this study, we discovered that the Ly49R activation receptor is selectively triggered during MCMV infection on antiviral NK cells licensed by the Ly49G2 IR. Ly49R activating receptor recognition of MCMV-infected targets is dependent on MHC I Dk and MCMV gp34 expression. Remarkably, although Ly49R is critical for Ly49G2-dependent antiviral immunity, blockade of the activation receptor in Ly49G2-deficient mice has no impact on virus control, suggesting that paired Ly49G2 MCMV sensing might enable Ly49R+ NK cells to better engage viral targets. Indeed, MCMV gp34 facilitates Ly49G2 binding to infected cells, and the IR is required to counter gp34-mediated immune evasion. A specific requirement for Ly49G2 in antiviral immunity is further explained by its capacity to license cytokine receptor signaling pathways and enhance Ly49R+ NK cell proliferation during infection. These findings advance our understanding of the molecular basis for functionally disparate self-receptor enhancement of antiviral NK cell immunity.
This article is featured in Top Reads, p.
Self recognition receptors provide essential regulation of innate immunity at steady state and in malignancy (1). This is exemplified in the case of germline-encoded self–MHC class I (MHC I)-specific receptors expressed by NK cells. Activating murine Ly49 or human killer Ig-like receptors (KIRs) bind MHC I or MHC I–like molecules and associate with ITAM-containing adaptor molecules. Ligation of NK cell activation receptors can drive NK cell proliferation, cytokine release, and degranulation (2, 3). In contrast, ITIM-bearing Ly49/KIR drive recruitment and activation of phosphatases during recognition of self–MHC I (4). Interactions between these signaling-disparate receptors and their cognate MHC I ligands are essential regulators of NK cell activation and self-tolerance.
Active engagement of self–MHC I inhibitory receptors (IRs) increases NK cell functionality via an educational process referred to as licensing (4). Relative to unlicensed NK cells, licensed NK cells are more responsive to activating receptor signaling pathways and readily reject viral and tumor targets lacking self–MHC I (4, 5). Indeed, to evade licensed NK cell immunity, herpesviruses such as murine CMV (MCMV) employ strategic mechanisms to manipulate MHC I and MHC I–like molecules (6, 7). A notable example is the MCMV immunoevasin m04/gp34, which binds and escorts MHC I to the surface of infected cells, where it can impede both T lymphocyte recognition and missing-self rejection of MCMV targets by licensed NK cells (8–10). Conversely, MCMV infection studies in distinct mouse models have demonstrated that self–MHC I licensing IRs can also provide essential antiviral NK cell immunity. Indeed, we found that licensed NK cells in MHC I Dk-bearing B6 mice with a C57L-derived NKC-Ly49 haplotype (i.e., NKCL-Dk) require the Ly49G2 IR for Dk-dependent licensing, selective expansion, and virus control during MCMV infection (11). Likewise, in MHC I Dd-transgenic B6 mice, interactions between the Ly49A IR and Dd license NK cells and engender essential MCMV protection (12). A mechanistic basis for licensed NK cell–mediated MCMV sensing however remains to be elucidated.
By detecting virus-induced alterations in self–MHC I or MHC I–like molecules during infection, we posited that selected pairs of self-specific Ly49 activation receptors and IRs may together enable more efficient recognition of infected “altered-self” targets by licensed NK cells (13). Indeed, mAb-mediated neutralization of the Ly49R activation receptor prior to MCMV infection was shown to abolish licensed NK cell immunity in NKCL-Dk mice (11), suggesting a role for Ly49R in recognition of MCMV. In contrast, recognition of MCMV gp34/MHC I complexes by Ly49L or Ly49P activation receptors is correlated with MHC I–dependent MCMV control in BALB.K or MA/My mice, respectively (14–17). Because the C57L alleles of Ly49G2 and Ly49R bind Dk (11), and MCMV gp34 is required for efficient surface expression of Dk during infection (10, 17), we predicted that Ly49G2 and Ly49R recognition of MCMV-modified Dk molecules might be an important feature of licensed NK cell sensing of MCMV in vivo.
NK cell receptors, their MHC I ligands, and viral immunoevasins are thought to have coevolved in an intricate arms race whereby host NKR/MHC I gene pairs that enable host survival prevail. In contrast, polymorphic viral immunoevasins are likely selected to thwart host immune defenses. Development of next-generation NK cell replacement therapies will undoubtedly benefit from a deeper understanding of the complex biology of NK cells and the NK cell receptor–ligand interactions at play during viral infection that might affect NK cell reactivity or self-tolerance. In this study, we demonstrate that Ly49R and Ly49G2 bind to altered-self gp34–Dk complexes on MCMV-infected cells. Remarkably, Ly49G2 is required to prevent gp34-mediated evasion of Ly49R+ NK cells during infection. Paired sensing of a shared MHC I ligand by activating and inhibitory Ly49 receptors thus represents an important mechanism of licensed NK cell–mediated viral target detection and host defense.
Materials and Methods
All mouse experiments were performed in accordance with the Animal Welfare Act and approved by the University of Virginia Animal Care and Use Committee.
All mice used in this study were bred and maintained at the University of Virginia under specific pathogen-free conditions. B6.Cg-NKCC57L-Dk (NKCL-Dk) and B6.Cg-GO1-Dk (GO1-Dk) mice were previously generated (11, 18). B6.SJL-Ptprca Pepcb/BoyJ mice were purchased (The Jackson Laboratory) and crossed with NKCL-Dk or GO1-Dk mice to generate congenically marked CD45.1.NKCL-Dk and CD45.1.GO1-Dk mice. B6-Prf1tm1Sdz/J mice were purchased (The Jackson Laboratory) and crossed with NKCL-Dk mice to generate perforin-deficient NKCL-Dk mice (Prf1−/−-Dk).
Wild-type (WT) and Dk-deficient M2-10B4 (H-2k/b; American Type Culture Collection) cells were cultured in complete RPMI 1640 media supplemented with 10% FBS and 100 U/ml penicillin/streptomycin. SVEC4-10 (H-2k; American Type Culture Collections) cells were cultured in complete DMEM supplemented with 10% FBS and 100 U/ml penicillin/streptomycin. Ly49-CD3ζ reporter cell lines were generated by lentiviral transduction as described (11). Briefly, C57L-derived CD3ζ-Ly49G2, CD3ζ-Ly49R, Ly49G2WT, or ITIM-mutant Ly49G2Y8F coding sequences were cloned into pCpp2E (provided by Sam Kung) (19) or pCW57-MCS1-P2A-MCS2 (Blast) (a gift of Adam Karpf; Addgene plasmid no. 80921), respectively. Transfers plasmids were then transfected into 293T cells together with psPAX2 and pMD2.G (both provided by Didier Trono (Addgene plasmid nos. 12260 and 12259, respectively) using Lipofectamine 3000. Titered lentiviral supernatants (multiplicity of infection [MOI] of 4) were used to transduce J7 Jurkat T cells (provided by Koho Iizuka) (20). High expressors were sorted or selected with blasticidin (10 μg/ml), respectively, and cultured in RPMI 1640 media supplemented with 10% FBS, 10 mM HEPES, 1 mM sodium pyruvate, 1 μM 2-ME, and 100 U/ml penicillin/streptomycin.
Salivary gland–passaged MCMV (SGV) or tissue culture (TC)–propagated MCMV (Smith strain; American Type Culture Collection) was titered on M2-10B4 monolayers and i.p. injected at indicated doses, as described (21, 22). SGV was used for in vivo virus infections unless indicated otherwise (e.g., for Δm04 MCMV infections, TC-passaged WT and Δm04 MCMV were compared). Ly49R was neutralized using 200 μg of mAb 12A8 (a gift from John Ortaldo, National Cancer Institute, National Institutes of Health, Frederick, MD) given i.p. 72 and 24 h before infection (11). For CD4+ and CD8+ T cell depletion studies, 200 μg of mAbs 2.43 and GK1.5 (provided by Timothy Bullock, University of Virginia) were administered 48 h before and 24 h after infection. Infected spleen genomic DNA was measured for MCMV genomes via quantitative PCR as described (23). For in vitro MCMV infections, M2-10B4 cells were infected with an MOI of 0.5, whereas SVEC4-10 cells were infected with an MOI of 0.1 or 25 as indicated. Adherent monolayers were incubated at 37°C, 5% CO2 in a small volume of diluted virus for 30 min, followed by centrifugation at 800 × g for 30 min at 25°C, as described (17).
Flow cytometry and cell sorting
Splenocytes (single-cell suspension) from uninfected or infected mice were preblocked with Fc receptor blocking mAb 2.4G2 (hybridoma maintained by the University of Virginia Lymphocyte Culture Center). Cell surface staining was performed using fluorophore-conjugated Abs (BD Biosciences, eBioscience, BioLegend, or University of Rijeka Center for Proteomics). Flow cytometry and cell sorting were performed on the Attune NxT (Thermo Fisher Scientific) and Influx (BD Biosciences) cytometers, respectively. Data were analyzed with FlowJo software (version 10.8). The following fluorophore-conjugated or biotin-conjugated Abs were used: CD3ε (145-2C11), CD19 (6D5), CD49b (DX5), NKp46 (29A1.4), Ly49G2 (4D11), Ly49R (12A8), Ly49ROV (4E5), KLRG1 (2F1), CD25 (PC61), H-2Dk (15-5-5), H-2Kk (36-7-5), H-2Kb (AF7-88.5), H-2Db (28-14-8), NKG2D (CX5), p-STAT4 (Y693; clone 38), MCMV m04/gp34 (m04.16), and MCMV m123/IE1 (IE1.01). Live/Dead fixable dyes (Thermo Fisher Scientific) were used to assess cell viability. p-STAT4 was detected in cytokine-stimulated NK cells after cell surface staining. NK cells were fixed with 4% paraformaldehyde for 10 min at 37°C, washed with PBS, permeabilized with ice-cold 100% methanol for 10 min, and washed again with PBS + 2% FBS, as described (24). Cells were stained with the p-STAT4 mAb (Y693; clone 38). M2-10B4 or SVEC4-10 cells were stained for cell surface MCMV gp34 (anti-m04.16 ; mouse IgG2b isotype) and Dk (mAb 15-5-5) followed by staining with an anti-mouse IgG2b-PE secondary Ab and fluorophore-conjugated streptavidin.
Primary NK cell gene editing
CRISPR RNAs (crRNAs) were selected using the CRISPOR (www.crispor.tefor.net) online platform (25). The first ∼35% of the coding sequence for each gene was prioritized for targeting. Single-guide RNAs (sgRNAs) (Klra18, 5′-GCA GAA CGA GAU GAG GCU CA-3′; Klra7, 5′-GCG UGG UGC UGC AGU UAU CG-3′) (11) were selected for their maximal likelihood of specific targeting with minimal potential to edit off-target genes. Synthetic crRNAs were purchased from Integrated DNA Technologies (www.idtdna.com/CRISPR-Cas9) in Alt-R format. To prepare duplexes, custom Alt-R crRNA and synthetic Alt-R trans-activating crRNA (Integrated DNA Technologies) were reconstituted to 100 μM (100 pmol/μl) with TE (Tris-EDTA) buffer (Life Technologies). Oligonucleotides were mixed at equimolar concentrations in a sterile PCR tube and annealed by heating to 95°C for 5 min in a PCR thermocycler. Annealed duplexes were then removed from the thermocycler and allowed to slowly cool to room temperature. Cas9 ribonucleoprotein (cRNP) complexes were prepared essentially as described (26). Briefly, 1.2 μl (120 pmol) of annealed oligonucleotide duplexes, 0.9 μl of 100 μM Alt-R Cas9 electroporation enhancer (Integrated DNA Technologies), and 3.9 μl of water were added to a sterile strip tube per sample (total volume of 6 μl). Then, 40 pmol of recombinant Alt-R S.p. Cas9 nuclease V3 (Integrated DNA Technologies) was diluted with water to a final volume of 6 μl in a separate sterile strip tube. Six microliters of diluted Cas9 was gently mixed with 6 μl of duplex-enhancer mixture for a total of 12 μl of cRNP complex at a 1:3 molar ratio. The cRNP complex was allowed to incubate for at least 10 min at room temperature. Resting enriched (Miltenyi Biotec mouse NK isolation kit II) spleen NK cells (∼2 × 105 cells/well; 96-well round-bottom plate) were preactivated in complete NK cell media (IMDM containing 10% FBS, 2 mM l-glutamine, 10 mM HEPES, 1× GlutaMAX, 100 U/ml penicillin/streptomycin, and 50 μM 2-ME plus recombinant mouse IL-15 [20 ng/ml, PeproTech]) for 18 h before transfection. IL-15–preactivated NK cells were nucleofected using a method similar to that described for activated primary mouse T cells (27). Two hundred microliters of complete NK cell media was prewarmed per well of a 96-well round-bottom plate. Approximately 5 × 105 NK cells were resuspended in 20 μl of P4 Primary Cell 4D-Nucleofection Solution (Lonza), mixed with 12 μl of cRNP complex and incubated for 2 min at room temperature. The NK cell–cRNP mix was transferred to nucleofection cuvette strips (Lonza) for electroporation using a 4D-Nucleofector X unit (Lonza). Different electroporation pulses were tested (CM137, CM138, DS137, DS138, DS150, DN100, and EH100). Pulse code CM138 was found optimal for gene-editing with minimal loss in activated mouse NK cell viability. After nucleofection, 200 μl of prewarmed NK cell media was added to each cuvette well, and transfected cells were transferred to 96-well round-bottom plates. NK cells were then incubated at 37°C (90 min) before centrifugation and resuspension in complete NK cell media plus IL-15 (20 ng/ml). NK cells were cultured in vitro for 5 d prior to analysis of gene-editing efficiency by flow cytometry, or sorting for use in adoptive transfer experiments.
IL-15–expanded spleen NK cells were flow-sorted into Ly49+ NK subsets and then labeled with respective cell proliferation dyes (CellTrace CFSE or CellTrace Violet [CTV]; Thermo Fisher Scientific). For analysis of cell proliferation, sorted NK cells resuspended in PBS plus FBS (1×) were labeled with an equal volume of freshly prepared proliferation dye (10 μM in PBS), immediately inverted three times, and gently vortexed (10 s). Labeled NK cells were incubated at room temperature for 5 min and then quenched in 10 ml of FBS. For each NK cell subset, 5 × 105 labeled NK cells were i.v. injected per recipient mouse 24 h prior to MCMV infection.
In vitro lymphocyte stimulation and coculture
For analysis of CD25 expression and STAT4 phosphorylation, enriched spleen NK cells were cultured for 24 h in complete NK media with and without 10 ng/ml IL-15 or IL-12 (PeproTech). For analysis of CD25 induction, IFN-β–pretreated (R&D Systems, 500 U/ml, 60 min) splenocytes resuspended in complete NK media plus IL-2 (500 U/ml) were cocultured with M2-10B4 cells pretreated with IFN-β (1000 U/ml) or infected with TC-passaged MCMV (MOI of 1.5) for 24 h prior to flow cytometry analysis. For reporter cell assays, SVEC4-10 cells were pretreated with IFN-β (1000 U/ml) or infected with TC-passaged MCMV (WT or Δm04; MOI of 25). For mAb stimulations, 96-well plates were coated with 10 μg/well anti-Ly49G2 (4D11), anti-Ly49R (12A8), or both mAbs overnight at 4°C. For doxycycline-inducible reporter cell lines, reporters were grown in media with and without 2 μg/ml doxycycline hyclate for 24 h prior to coculture. Ly49-bearing J7 reporter cells (2 × 105) were stimulated for 8 h with target cells (2 × 104) or respective mAbs. LacZ activity was determined using the substrate chlorophenol red–d-galactoside (CPRG), as described (11, 20).
Generation of Dk-deficient M2-10B4 cells
The second exon of the Dk gene was targeted using CRISPR-Cas9 editing. A Dk-specific sgRNA (5′-GCGA GAG AUG AGC CGC GGG UG-3′) was selected using CRISPOR (25) based on published H-2Dk coding sequence (GenBank accession no. M18524.1; https://www.ncbi.nlm.nih.gov/nuccore/M18524.1). The allele-specific sgRNA was selected for maximal likelihood to specifically target Dk with minimal potential to edit related Kk, Kb, and Db genes also expressed by H-2k/b M2-10B4 cells. A 5′ G was appended to the sgRNA to ensure efficient in vitro transcription with the U6 promoter. Dk-specific oligonucleotides (Integrated DNA Technologies) were cloned into pX330-U6-Chimeric_BB-CBh-hSpCas9 (provided by Feng Zhang; Addgene plasmid no. 42230) and then transfected into M2-10B4 cells using Lipofectamine 3000. Single-cell clones (n = 100) obtained by limiting dilution were stimulated with IFN-β (200 U/ml, 18 h) and screened for cell surface Dk, Kk, Kb, and Db expression using flow cytometry. Five Dk-deficient single-cell clones were identified. A single Dk-deficient clonal cell line was analyzed in in vitro assays.
Generation of Δm04 MCMV
The MCMV m04 open reading frame was targeted using CRISPR-Cas9 editing. An m04-specific sgRNA (5′-GAG CAC UGA UAA CGG CAA CGG-3′) was selected using CRISPOR (25) based on the published Smith strain m04 coding sequence (GenBank accession no. GU305914.1; https://www.ncbi.nlm.nih.gov/nuccore/GU305914). The sgRNA was designed to introduce a frameshift mutation upstream of m04’s transmembrane domain, which is required for MHC I association (28). Notably, a 5′ G was appended to the sgRNA to ensure efficient transcription with the U6 promoter. m04-specific oligonucleotides (Integrated DNA Technologies) were cloned into pLentiCRISPR v2 (provided by Feng Zhang; Addgene plasmid no. 52961) and then transfected into 293T cells together with psPAX2 and pMD2.G (both provided by Didier Trono; Addgene plasmid nos. 12260 and 12259, respectively) using Lipofectamine 3000. Titered lentiviral supernatants were used to transduce M2-10B4 cells (MOI of 1) followed by selection in media supplemented with 2 μg/ml puromycin. SGV was propagated in sgRNA/Cas9-coexpressing M2-10B4 cells (MOI of 0.01) as described (29). Plaque-purified Δm04 MCMV was isolated from heterogeneous CRISPR-modified viral supernatants by two rounds of limiting dilution on WT M2-10B4 cells. m04-specific PCR amplicons spanning the anticipated CRISPR/Cas9 cleavage site were generated from clonal virions using m04-specific primers 5′-TCA CTC CCA TGC ACG GAT TA-3′ and 5′-CCT CAT CCG GAG CTG TCA TT-3′. Sequence variants were confirmed and clonal virions were further propagated in WT M2-10B4 cells for use in in vivo studies. TC-passaged MCMV lacking m04 was compared with WT MCMV in a multistep growth curve on M2-10B4 cells (MOI of 0.1). Viral genome copies were quantified from cell lysates and culture supernatants as described (22).
Statistical analyses were performed using GraphPad Prism (version 9.2.0). Significance was assessed using one- or two-way ANOVA in conjunction with Tukey or Holms-Sidak post hoc tests. A Student t test was used to assess the significance for two independent measurements.
Ly49G2+ antiviral NK cells downregulate Ly49R in Dk-bearing mice
Prior work in NKCL-Dk mice established a key role for the Ly49G2 IR in Dk-dependent antiviral immunity mediated by licensed NK cells (11). To further elucidate requirements for licensed NK cell recognition of MCMV, we first compared spleen MCMV titers in immunodeficient NKCL-Dk mice depleted of CD4+ and CD8+ T cells, or lacking perforin (Prf1−/−-Dk), Ly49G2 (GO1-Dk), or Dk itself (i.e., NKCL mice) (Fig. 1A). We observed that NK cells in this mouse model require perforin, host-Dk expression, and Ly49G2 to mediate MCMV control at 90 h postinfection (hpi). Moreover, T cell immunity is not essential early during acute infection.
Ly49R neutralization via blocking Abs was previously shown to abrogate licensed Ly49G2+ NK cell antiviral immunity, suggesting it has a key role in MCMV recognition. The Ly49H activation receptor is likewise required for detection of MCMV m157 in B6 mice, which drives clonal expansion of Ly49H+ NK cells, especially in Prf1-deficient mice (30–32). We employed a similar strategy to assess the licensed NK cell response to MCMV. We found that expansion of antiviral Ly49R+ Ly49G2+ (R+ G2+) NK cells directly corresponded to decreasing Ly49R cell surface expression in infected Prf1−/−-Dk mice (Fig. 1B, 1C, Supplemental Fig. 1A, 1B). Ly49R+ Ly49G2– (R+ G2–) NK cells in contrast declined in frequency and number and did not alter their expression of Ly49R during infection. This selective decrease of Ly49R in licensed NK cells is analogous to m157-driven Ly49H downregulation in antiviral NK cells during acute MCMV infection in B6 mice (31, 33, 34). These results thus suggest that Ly49R engagement of an MCMV-associated ligand gives rise to Ly49Rlo Ly49G2+ (Rlo G2+) NK cells during infection.
We next tested whether MCMV-induced Ly49R downregulation in Prf1−/− mice requires Dk expression in host cells. As shown in (Fig. 1D–F, the decrease in cell surface Ly49R expression in antiviral NK cells occurred exclusively in Dk-bearing mice. We further investigated whether NK cell Ly49R downregulation corresponded with markers of enhanced NK cell functionality during infection. Antiviral NK cells expanding during MCMV infection have been shown to increase expression of the high-affinity IL-2 receptor CD25 and the maturation marker KLRG1 (11, 35, 36). Intriguingly, we found that Rlo G2+ NK cells from Dk mice significantly upregulated KLRG1hi and CD25 expression (Fig. 1F–J). Antiviral Ly49G2+ NK cells thus specifically respond to MCMV-induced changes in host cell Dk expression with Ly49R downregulation, increased activation, and maturation.
Ly49R downregulation in response to MCMV-modified Dk marks proliferating antiviral effectors
We next sought to elucidate whether Ly49R downregulation is associated with the activation or expansion of virus-responsive Ly49G2+ NK cells. To test this, we transferred Dk-licensed R+ G2+ NK cells into congenically marked Ly49G2-deficient GO1-Dk host mice (Fig. 2A). Although donor NK cells homed to the spleens of PBS-injected and MCMV-infected mice, we specifically observed proliferation and expansion of donor NK cells in response to MCMV infection (Fig. 2B, 2D). Intriguingly, whereas R+ G2+ NK cells transferred into PBS-injected recipients maintained their Ly49R expression, we observed progressive Ly49R downregulation on Ly49G2+ NK cells proliferating during MCMV infection (Fig. 2E, 2F). Notably, Ly49R downregulation coincided with upregulation of CD25 and KLRG1 on proliferating NK cells (Fig. 2G, 2H). These results showing Ly49R downregulation on proliferating antiviral Ly49G2+ NK cells correspond with those demonstrating selective NK cell activation and maturation during MCMV infection.
We reasoned that Ly49R downregulation might result from NK cell sensing of MCMV-induced alterations in Dk molecules on infected targets. Alternatively, MCMV-induced inflammatory cytokines might facilitate increased host cell Dk expression and subsequent Ly49R downregulation. To address this, we compared the effect of poly(I:C)- versus MCMV-induced increases in host cell Dk expression on R+ G2+ NK cells. We injected NKCL-Dk mice with poly(I:C) or MCMV and measured accumulation of Rlo G2+ NK cells. Whereas both treatments similarly increased cell surface Dk expression on splenic host cells (Supplemental Fig. 1C), the accumulation of Rlo G2+ NK cells was specific to MCMV infection (Supplemental Fig. 1D, 1E). Because Ly49G2 also binds Dk, we reasoned that the IR could have prevented Ly49R engagement of poly(I:C)-induced Dk in NKCL-Dk mice. However, Ly49G2-deficient GO1-Dk Rlo NK cells likewise accumulated during MCMV infection, but not following poly(I:C) treatment (Supplemental Fig. 1F, 1G). Considering that Ly49R downregulation specifically occurred in response to Dk-bearing targets (Fig. 1D–F), these findings suggest that loss of Ly49R expression during infection results from its engagement of an MCMV-modified Dk ligand, thereby giving rise to experienced Rlo G2+ antiviral NK effectors.
MCMV gp34 is required for infected cell recognition by Ly49R
The MCMV immunoevasin gp34 has been shown to bind distinct polymorphic MHC I molecules (28) and can enhance cell surface Dk expression in MCMV-infected cells (10, 17). To test whether gp34–Dk complexes might represent an MCMV-modified Dk ligand of Ly49R, we generated a CRISPR mutant MCMV strain deficient for m04/gp34 (Supplemental Fig. 2A, 2B). Notably, WT MCMV and Δm04 MCMV strains exhibited similar rates of in vitro replication and virion release by M2-10B4 cells (Supplemental Fig. 2C). We further evaluated cell surface Dk and gp34 expression in MCMV-infected (MOI of 0.5) WT or Dk-deficient M2-10B4 cells (Supplemental Fig. 2D). As expected, Δm04 MCMV-infected M2-10B4 lacked cell surface gp34 expression. Likewise, we did not observe expression of gp34–Dk complexes in H-2k SVEC4-10 cells infected with Δm04 MCMV (MOI of 0.1) (Supplemental Fig. 2E). Whereas we found a direct correspondence between MCMV IE1 and gp34-Dk expression in SVEC4-10 infected with WT virus, IE1 expression in Δm04-infected cells coincided with loss of surface Dk expression (Supplemental Fig. 2E).
To determine whether gp34 facilitated Ly49R stimulation, we engineered a Jurkat reporter cell line (J7) to express chimeric CD3ζ-R receptors (Fig. 3A). We then analyzed the ability of SVEC4-10 cells pretreated with IFN-β or infected with WT or Δm04 MCMV to stimulate CD3ζ-R reporters. WT MCMV-infected targets consistently triggered CD3ζ-R reporters, whereas Δm04 MCMV-infected targets did not (Fig. 3B). A key difference was that cell surface gp34–Dk complexes were prevalent on the WT MCMV-infected targets (Fig. 3C). These data demonstrate that Ly49R recognition of MCMV-infected targets is facilitated by gp34. Taken together, our findings suggest that MCMV gp34 enables Ly49R to recognize virus-modified Dk molecules on MCMV-infected cells, which can lead to selective expansion of antiviral Rlo G2+ NK cells during MCMV infection.
Ly49G2+ antiviral NK cells require both Ly49R and Ly49G2 for proliferation and expansion during MCMV infection
We hypothesized that Ly49R is required for mature licensed NK cell virus sensing in vivo. To test this, we ablated Ly49R’s expression in IL-15–expanded primary NK cells using Klra18/Ly49r-specific CRISPR/cRNP complexes and then transferred a 1:1 mix of sorted WT R+ G2+ (CFSE-labeled) and mutant Rnull G2+ (CTV-labeled) NK cells into Ly49G2-deficient GO1-Dk mice prior to MCMV infection (Supplemental Fig. 3A, 3B). Unexpectedly, we found that donor Rnull G2+ NK cell proliferation and CD25 upregulation were comparable to those of donor WT R+ G2+ NK cells during MCMV infection (Supplemental Fig. 3C–E). However, R+ G2+ NK cells exhibited greater representation by percentage in the spleen following infection (Supplemental Fig. 3C, 3F). Thus, although Ly49R is not explicitly required to expand IL-15–primed antiviral Ly49G2+ NK cells during MCMV infection, Ly49R signaling is nonetheless essential for optimal accumulation of Ly49G2+ antiviral effectors.
Ly49R+ NK cells lacking endogenous Ly49G2, or those from Ly49G2-deficient GO1-Dk mice, exhibit severe defects in NK cell proliferation during MCMV infection (11). However, it is unknown whether Ly49G2’s essential role is limited to NK cell development, or whether it dynamically affects Ly49R+ NK cell effector functions. Thus, we ablated Ly49G2 from primary Ly49R+ NK cells using Klra7/Ly49g2-specific cRNP complexes to assess its role in antiviral NK cells during MCMV infection (Fig. 4A, 4B). Whereas both sorted subsets of Ly49R+ NK cells exhibited progressive downregulation of Ly49R coinciding with NK cell division during infection, mutant G2null NK cells displayed significantly less accumulation in the spleen and impaired proliferation in comparison with WT G2+ NK cells (Fig. 4C–G), possibly stemming from less effective CD25 upregulation during cell division (Fig. 4H). Antiviral NK cells thus require sustained Ly49G2 expression to efficiently accumulate during MCMV infection. We infer that direct recognition of MCMV-modified Dk ligands by Ly49R facilitates the expansion of Rlo G2+ effectors with enhanced antiviral activity.
Ly49G2 is required to counter MCMV gp34-mediated immune evasion
Previous studies have shown that MHC I–gp34 complexes can engage Ly49 IRs to facilitate MCMV evasion of licensed NK cell antiviral immunity (10, 17). Nonetheless, Ly49G2 is critical for cytotoxic control of MCMV in NKCL-Dk mice (Fig. 1A). We thus questioned whether the Ly49G2L receptor also binds gp34–Dk complexes. To test this, we generated J7 reporter cells expressing chimeric CD3ζ-G2 receptors (Fig. 5A) and evaluated whether they can be stimulated by SVEC4-10 cells pretreated with IFN-β or infected with WT or Δm04 MCMV. Whereas IFN-β–treated or WT MCMV-infected targets readily stimulated CD3ζ-G2 reporters, Δm04 MCMV-infected targets were much less effective (Fig. 5B). Remarkably, although WT MCMV-infected cells exhibited significantly lower cell surface Dk expression relative to IFN-β–treated cells (Fig. 5C), CD3ζ-G2 reporters were equivalently stimulated by both targets. MCMV gp34–Dk complexes thus represent a bona fide ligand of the Ly49G2L IR.
Ligand binding can promote IR clustering and ITIM phosphorylation, driving recruitment and activation of SH2 domain–containing phosphatases (4). We thus tested whether Ly49G2 inhibits Ly49R signaling in CD3ζ-R reporter cells engineered to express doxycycline-inducible G2WT or ITIM-mutant G2Y8F receptors (Supplemental Fig. 4A). Whereas Ly49G2 crosslinking impeded Ly49R stimulation in dox-induced CD3ζ-R.G2WT reporter cells (Supplemental Fig. 4B), it had no effect on CD3ζ-R.G2Y8F reporters (Supplemental Fig. 4C). Likewise, induction of G2WT receptors inhibited Ly49R engagement of gp34–Dk complexes on MCMV-infected SVEC4-10 cells, in contrast to G2Y8F receptors (Supplemental Fig. 4D–F). These results demonstrate that Ly49G2 is functionally competent to inhibit Ly49R signaling.
We then evaluated gp34’s role in Dk- and Ly49G2-dependent NK cell antiviral immunity. Whereas MCMV gp34 did not significantly alter viral burden in NKCL-Dk, Prf−/−-Dk, or NKCL mice (Fig. 5D), GO1-Dk mice were selectively susceptible to gp34-expressing MCMV (Fig. 5D). These data thus demonstrate that MCMV gp34 facilitates immune evasion in the absence of Ly49G2-mediated Dk sensing. Nonetheless, higher Δm04 virus levels in GO1-Dk than NKCL-Dk spleen tissues further establish a gp34-indpendent role of Ly49G2 in MCMV control (Fig. 5D). Indeed, we observed striking Ly49G2- and Dk-dependent accumulation of NK cells in response to WT or gp34-deficient MCMV strains (Fig. 5E, 5F). Ly49G2-mediated self–MHC I sensing is thus a focal point of antiviral NK cell immunity in this model system.
We reasoned that Ly49G2 inhibitory signals might protect Ly49R+ NK cells from disarming in response to gp34-Dk molecules displayed by MCMV-infected targets. Indeed, mature NK cells can decrease their responsiveness to activation receptor stimulation within days of transfer into MHC I–deficient hosts (37), presumably due to lack of engagement of self-specific IRs. We thus sought to elucidate whether MCMV gp34 disarms Ly49R+ NK cells in the absence of Ly49G2. To test this, we blocked Ly49R in NKCL-Dk or Ly49G2-deficient GO1-Dk mice by administering the Ly49R-specific 12A8 mAb. Whereas this treatment fully abrogated MCMV resistance in NKCL-Dk mice, it did not result in higher GO1-Dk spleen virus titers (Fig. 5G). Thus, despite its essential role in Dk-dependent antiviral immunity (11), Ly49R signaling is ineffectual in mice lacking the Ly49G2 IR.
Ly49G2 licensing of cytokine responsiveness and Ly49R together enhance CD25 upregulation in antiviral NK cells during MCMV infection
Primary NK cells cultured in IL-12 and antiviral NK cells responding to MCMV infection require IL-12R/STAT4 signaling to efficiently upregulate CD25 (35). We reasoned therefore that a selective increase in CD25 expression in R+ G2+ antiviral NK cells during MCMV infection might be related to enhanced sensitivity to IL-12R signaling. To test this, we enriched splenic NK cells from Dk- or Ly49G2-disparate mice and stimulated them with different cytokine conditions. IL-12 stimulation resulted in selective increases in CD25 expression and p-STAT4 in R+ G2+ antiviral NK cells from mice with Dk and Ly49G2 (Fig. 6A, 6B). Conversely, CD25 expression and p-STAT4 were substantially lower in NK cells from mice lacking either Dk or Ly49G2 (Fig. 6A, 6B). These data suggest Dk licensing of Ly49G2+ NK cells can enhance activation downstream of the IL-12R, thereby promoting CD25 upregulation of R+ G2+ antiviral NK cells during infection. To this end, we cocultured NKCL-Dk splenocytes with control IFN-β–treated or MCMV-infected M2-10B4 cells and evaluated CD25 expression on R+ G2+ versus R+ G2– NK cells. Although both subsets of NK cells specifically increased CD25 expression in response to MCMV-infected targets, R+ G2+ NK cells had significantly higher CD25 expression in comparison with their R+ G2– counterparts (Fig. 6C). Thus Ly49G2’s licensing effect on IL-12R signaling and subsequent CD25 upregulation may confer a proliferative advantage to R+ G2+ NK cells responding to MCMV infection.
Our findings in Prf1−/−-Dk mice raised the possibility that Ly49R engagement of MCMV target ligands may promote CD25 upregulation in Rlo G2+ NK cells. To pursue this, we blocked Ly49R in NKCL-Dk mice using the Ly49R-specific 12A8 mAb. Ly49R-blocked G2+ NK cells from MCMV-infected mice were detected using the anti-Ly49ROV mAb (4E5) as described (Fig. 6D) (11, 38). ROVlo G2+ NK cells from Ly49R-blocked mice failed to differentially acquire increased CD25 or KLRG1hi expression in comparison with cIgG-treated mice (Fig. 6E, 6F). Furthermore, when gating on CD25+ subsets, Ly49R-blocked ROV+ G2+ NK cells exhibited significantly lower CD25 mean fluorescence intensity (Fig. 6G, 6H). Thus, endogenous antiviral R+ G2+ NK cells rely on Ly49R activation receptor signaling to increase CD25 upregulation, maturation, and survival during MCMV infection.
In this study, we provide mechanistic insight into how recognition of self–MHC I by functionally disparate NK cell receptors regulates the responsiveness of antiviral NK cells responding to MCMV infection. Prior work has shown that selected Ly49 activating and IRs bind MCMV gp34-associated MHC I complexes in vitro (10, 14, 16, 17), suggesting that altered-self ligands might contribute to NK cell detection of MCMV in vivo. Indeed, we found that the Ly49R activation receptor selectively binds to altered-self gp34–Dk complexes at the surface of MCMV-infected target cells, but not IFN-β–induced Dk molecules. Ly49R thus may be sensitive to conformational changes in Dk molecules induced by gp34, or another MCMV immunoevasin such as MATp1 (17), or possibly an MCMV peptide. In contrast, we found that the C57L allele of Ly49G2 binds both self and altered-self Dk molecules. Inasmuch as neither Ly49RL nor Ly49G2L possesses putative amino acid residues (R/M223, L234, and N244) thought required for binding gp34–MHC I complexes (14), these receptors may contact gp34 differently than other Ly49 receptors (e.g., Ly49P, Ly49D2, or Ly49L). These findings further suggest that coincident recognition of altered-self by coexpressed, “paired” activating and IRs synergistically enhances NK cell–mediated virus sensing.
MCMV gp34–MHC I engagement of self–MHC I licensing IRs is posited to interfere with NK cell immunity to MCMV by facilitating evasion of the NK cell missing-self response (10, 17). Remarkably, we show that gp34 does not impede R+ G2+ antiviral NK cells from controlling MCMV infection. Rather, our findings suggest that altered-self gp34-Dk sensing by Ly49G2 is necessary for Ly49R+ NK cell immunity to MCMV. Ly49G2 is required to prevent gp34-mediated immune evasion, as evidenced by selective gp34-dependent MCMV spread in the Ly49G2-deficient GO1-Dk mice. Preferential control of Δm04 MCMV in GO1-Dk animals hints that another missing-self IR may have a role in Dk-dependent antiviral immunity. Alternatively, gp34 may be able to disarm NK cells by tolerizing signaling through gp34-Dk–specific activation receptors such as Ly49R or Ly49P. Nonetheless, our data demonstrate that altered-self complexes facilitate evasion of these recognition systems. Paired altered-self detection via signaling-disparate MHC I recognition receptors thus may underpin MCMV sensing via R+ G2+ antiviral NK cells and increased effector functionality, including proliferation, expansion, and activation during infection.
Inasmuch as prior studies indicate that Ly49 IRs facilitate missing-self recognition of MCMV (10, 12, 17), our findings suggest that R+ G2+ NK cell MCMV sensing is mechanistically distinct. Neutralization of Ly49R via blocking Abs revealed its role in driving Ly49G2+ NK cell antiviral immunity (11), yet Ly49R is ineffectual in this context without coexpression of the Ly49G2 IR. In contrast, transgenic expression of Ly49H is sufficient for MHC I–independent MCMV resistance (39). These distinct requirements for antiviral immunity might suggest a difference in the magnitude or quality of activating signals propagated by the DAP12 signaling adaptor. Whereas Ly49H has a relatively high affinity for m157 (40), Ly49R reporter cells are modestly stimulated by gp34–Dk complexes on MCMV-infected cells. Because lower, albeit not significantly so, spleen viral titers were observed in Ly49R-blocked GO1-Dk mice in comparison with control IgG-treated controls, our data might suggest that ongoing interaction with altered-self ligands resulting in sustained Ly49R–DAP12 signaling in the absence of Ly49G2 may serve to disarm NK cells during MCMV infection. This interpretation is in line with previous studies showing that active engagement of MHC I molecules by self-specific IRs is essential to maintain NK cell reactivity toward target cells with altered MHC I expression levels (37, 41). Moreover, sustained signaling through selected DAP10/12-dependent receptors has been shown to render NK cells broadly hyporesponsive to activation receptor stimulation (42–44). Thus, Ly49G2 inhibitory signals may engender NK cell reactivity during MCMV infection by preventing tolerance induction via virus-specific activation receptors. Alternatively, active cis-engaged Ly49G2 receptors in licensed NK cells may sequester SH2-domain–containing phosphatases away from Ly49R activation receptor signaling substrates (L. Schmied, T.T. Luu, J.N. Søndergaard, S. Meinke, D.K. Mohammad, S.B. Singh, C. Mayer, G. P. Casoni, M. Chrobok, H. Schlums, et al., manuscript posted on medRxiv, DOI: 10.1101/2022.03.08.483415), thus further amplifying Ly49R signaling output. Regardless of the mechanism, we infer that Ly49G2 augments Ly49R+ NK cell recognition of viral infection and antiviral immunity via altered-self Dk sensing.
Given its importance in driving Dk-dependent antiviral immunity, it was surprising that Ly49R ablation had little, if any, effect on MCMV-induced proliferation of IL-15–activated primary Ly49G2+ NK cells. Notably, IL-15 primes the PI3K/mTOR pathway in NK cells, resulting in enhanced signaling downstream of cytokine receptors and enhanced cytokine production following activation receptor stimulation (45). IL-15 treatment might render Ly49G2+ NK cells more sensitive to stimulatory signals via other ITAM-linked activation receptors or cytokine receptors such that Ly49R is not explicitly required to induce expansion of cytokine-primed Ly49G2+ NK cells. Indeed, high-dose IL-15 priming can rewire NK cell metabolism such that NK cell activation is no longer dependent on glucose-driven oxidative phosphorylation (46). Nonetheless, we consistently observed that Rnull G2+ NK cells were underrepresented in comparison with their WT counterparts after several days of infection, which suggests Ly49R signaling may be needed to support NK cell survival during MCMV infection. Conversely, ablation of Ly49G2 in IL-15–activated Ly49R+ NK cells revealed its intrinsic role in enhancing proliferation. These data further demonstrate that cytokine priming is insufficient to compensate for Ly49G2 deficiency. We conclude that IR-mediated licensing of NK cell functionality is essential in MHC I–dependent antiviral immunity mediated by NK cells.
We provide evidence that signaling by Ly49G2 and Ly49R contributes to selective upregulation of CD25 on licensed NK cells responding to MCMV infection. Our data further suggest that Ly49G2 drives CD25 upregulation in response to IL-12 by promoting STAT4 phosphorylation. Although self-specific IR ITIM signals have previously been shown to enhance NK cell stimulation downstream of activation receptor ligation (47), our study implies an additional role for licensing receptors in amplification of cytokine receptor signaling. The PI3K-AKT-mTOR pathway is critical for priming of NK cells with cytokine (45), and SHP-1 enhances the basal mTOR reactivity of licensed NK cells (48). Ly49G2 may amplify licensed NK cell responses to cytokine receptor signaling by modulating the abundance, localization, or phosphorylation status of downstream phosphatases (13, 49, and L. Schmied et al., manuscript posted on medRxiv, DOI: 10.1101/2022.03.08.483415). The resulting differential sensitivity to IL-2, IL-12, or IL-15 cytokines may enable rapid accumulation of R+ G2+ antiviral effectors during infection, which is supported by recent studies showing that IL-2 and IL-15 nonredundantly drive proliferation or survival of MCMV-specific Ly49H+ cells (50).
A hallmark of this MCMV detection mechanism is scrutinization of a shared self MHC I molecule by functionally disparate paired MHC I recognition receptors. We predict that such paired recognition systems may be a conserved feature of NK cell immunity to viral infection. Indeed, CMV employs several MHC I mimics or MHC I–binding immunoevasins to evade cytolysis by NK cells, including MCMV m157, m04, m12, and HCMV UL40 (7). Notably, each of these proteins has been shown to exhibit high mutability (51–56), and both activating and ITIM-bearing NK cell receptors have been implicated in direct recognition of these molecules displayed by virally infected cells (10, 14, 55, 57–60). It follows that a combined effect of polymorphism in the viral genome, in the altered-self ligands, and in the signaling disparate self-receptors themselves likely dictates the extent to which a given receptor pair is beneficial or suppressive to the antiviral NK cell response (61). We propose that efficient altered-self recognition via activation receptors and licensing IRs is critical for paired-self receptor–mediated NK cell sensing of viral infection. Further study thus is warranted to understand the contexts in which paired-self receptor recognition can be harnessed to improve disease outcomes.
We thank Awndre Gamache, Jing Huang, Flávio da Silva Mesquita, Ariel Hay, and Abhinav Arneja for technical support. We thank Timothy Bullock, John Ortaldo, C. John Luckey, Koho Iizuka, and Sam Kung for providing key reagents. We thank Awndre Gamache, William Nash, and Jessica Annis for helpful discussion and comments on the manuscript. We are grateful for services provided by the University of Virginia Flow Cytometry and Genetically Engineered Murine Model Cores.
This work was supported by the National Institute of Allergy and Infectious Diseases Grants R01 AI050072 and R21 AI141943-01A1 (to M.G.B.), the Division of Nephrology, Department of Medicine and the Beirne Carter Center for Immunology Research, University of Virginia. J.M.C. received support from National Institute of Allergy and Immunology Training Grant T32 AI007496.
The online version of this article contains supplemental material.
J.M.C. and M.G.B. designed the study; J.M.C., K.H.D., and R.B.C. performed experiments and collected data; J.M.C. and M.G.B. analyzed the data; J.M.C., P.P., and M.-L.H. contributed new reagents/analytic tools; and J.M.C. and M.G.B. wrote the paper.
The authors have no financial conflicts of interest.