C57L-derived MHC I D^k-disparate congenic mouse strains R7 and R2 (referred to as D^k and non-D^k, respectively) were previously generated and described (39). C57Bl/6 (B6)-derived IFNAR-KO mice (B6.129S2-Ifnar1^tm1Agt/Mmjax) were obtained from the Mutant Mouse Resource & Research Centers. CD45.1 (45.1) mice (B6.SJL-Ptprc^aPepc^b/BoyJ) were purchased from The Jackson Laboratory. Congenic B6 mice harboring the natural killer gene complex (NKC) haplotype from C57L mice (B6.NKC^c57l; referred to here as NKC^l) were generated and described previously (13). NKC^l mice lack the Ly49h gene and, consequently, Ly49H⁺ NK cells (13). NKC^l and IFNAR-KO mice were crossed to generate B6.Cg-Nkc^c57l-129S2-Ifnar1^tm1Agt (aka B6.NKC^l IFNAR-KO). IFNAR wild-type (IFNAR-WT) and IFNAR-heterozygous (IFNAR-het) littermates from the crosses were included as comparators in experiments. This study was carried out in accordance with the recommendations of the Animal Welfare Act and the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the University of Virginia Animal Care and Use Committee.

Mice were i.p. injected with salivary gland passaged MCMV (Smith strain ATCC VR1399; 5 × 10⁴ PFU/mouse, or as indicated). For depletion of pDC, mice were i.p. injected twice with 250 μg monoclonal antibody (mAb) 927 (kindly provided by Marco Colonna) (40) 48 and 24 h prior to infection. For depletion of G2⁺ NK cells, mice were i.p. injected with 200 μg mAb 4D11 (hybridoma kindly provided by Wayne Yokoyama) 48 h prior to infection. For blocking IFNAR, mice were i.p. injected with 1 mg mAb MAR1-5A3 (Leinco Technologies, Inc.) before infection and 500 μg every 24 h thereafter. For blocking Ly49H receptors, mice were i.p. injected with 200 μg mAb 3D10 (hybridoma also provided by Wayne Yokoyama) 24 h prior to infection.

Disrupted spleen tissues were digested with collagenase D (Roche) and processed with Falcon cell strainers essentially as described (41). Splenocytes were Fc blocked with mAb 2.4G2 and stained with mAb purchased from eBiosciences, BioLegend, BD Pharmingen, and Miltenyi including 2G9 (MHC II), 53-6.7 (CD8), MAR1-5A3 (IFNAR1), 145-2C11 (CD3), 6D5 (CD19), P84 (SIRPα), N418 (CD11c), HL3 (CD11c), 129c1 (mPDCA), M1/70 (CD11b), GK1.5 (CD4), RA3-6B2 (B220), 53-2.1 (Thy1.2), A20 (CD45.1), 104 (CD45.2), and 30-F11 (pan-CD45). LIVE/DEAD^® Viability Dye (Invitrogen) was used for dead cell exclusion. Samples were run on a BD Canto II with Diva acquisition software and analyzed using FlowJo v10. Detailed gating strategies are illustrated in supplemental material (Figures S1 and S2 in Supplementary Material). Cell numbers were quantified via a series of calculations. Whole spleens and spleen fractions processed for flow cytometry were weighed to determine their mass. The number of splenocytes per flow sample was determined using a hemocytometer, and this was converted to splenocytes/g tissue. FlowJo data were then used to determine the number of DC/total splenocytes. This allowed us to calculate the number of DC/g tissue and then adjust for the total mass of the whole spleen.

A population of CD11c⁺ MHC II⁺ cells from infected mice was observed that displayed significantly increased CD11b expression and higher side scatter values. Thus, these cells were regarded as inflammatory monocytes/inflammatory DC and were excluded from analyses, as indicated in Figures S1 and S2 in Supplementary Material.

Blood was collected via tail bleed or postmortem cardiac puncture. Serum was isolated by allowing blood to clot followed by centrifugation. IFNα quantitation was performed with VeriKine Mouse IFN Alpha ELISA kits (PBL) according to the manufacturer’s instructions.

Bone marrow transplantations were performed essentially as described (25). Briefly, B6 mice were irradiated twice with 5.5 Gy (11 Gy total) over a span of 3 h. Twenty-four hour later, i.v. injection of ~4.5 × 10⁶ total donor BM cells was performed, either IFNAR-KO (45.2), -WT (45.1), or a 1:1 mix. Recipient mice were maintained on sulfate drinking water for 3 weeks following irradiation. Peripheral blood analysis was performed at 4 weeks to assess BM chimerism. After 8 weeks, recipients were 3D10 treated to block the Ly49H receptor and i.p. infected 24 h later. Splenocytes were analyzed 3 days postinfection (d.p.i.). Due to limited flow cytometer channels in the BM chimera experiments, we defined CD8⁺ DC by low SIRPα and CD4 expression (28) to evaluate cDC subsets side by side. However, results were further confirmed by a separate analysis for CD8⁺ DC specifically (Figure S3 in Supplementary Material).

Virus levels were measured by determining the ratio of viral genomes to β-actin in DNA isolated from tissue samples. DNA was isolated using the Gentra Puregene tissue kit according to the manufacturer’s instructions. Isolated DNA was analyzed by quantitative real-time PCR as described previously (42).

Since our samples sizes are not large enough to verify normality of the data, significance was determined with the non-parametric Mann–Whitney (two groups) or Kruskal–Wallis (more than two groups) tests using GraphPad Prism and XLSTAT software. If the Kruskal–Wallis test rejected the null hypothesis, a Conover–Iman post hoc test was run for multiple comparisons using XLSTAT. Significance is represented as follows: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Error bars denote SD. Dots on graphs represent individual animals.

Previous work has shown that Ly49H⁺ NK cells can preserve splenic cDC numbers during MCMV infection (5, 8, 9). However, the role of self-MHC I licensed NK cells in cDC protection has been minimally explored. We therefore examined the effect of D^k-licensed G2⁺ NK cells on cDC populations during MCMV infection. Similar to prior work (12), we observed that total splenic cDC numbers declined by 3 d.p.i. in all mice (Figure 1A). The majority of loss occurred in the CD8⁺ and CD4⁺ DC subsets, whereas DN DC were largely unaffected (Figure 1B). However, overall cDC loss was less prominent in D^k mice with licensed G2⁺ NK cells, suggesting that the virus control mediated by these cells can enhance cDC retention. Both CD8⁺ and CD4⁺ DC numbers were higher in D^k mice compared to non-D^k, but DN DC numbers were equivalent across all mice (Figure 1B). These results suggest that distinct cDC subsets are differentially sensitive to MCMV infection, and, as a result, licensed G2⁺ NK cell control differentially enhances subset numbers.

We further examined the importance of licensed NK cell-mediated control by specifically depleting G2⁺ NK cells prior to MCMV infection. Without Ly49G2⁺ NK cells, cDC loss was exacerbated in D^k mice compared to isotype-treated control mice; this was similar to the results obtained for non-D^k versus D^k mice (Figure 1C compared to Figure 1A). Individual cDC subset loss also mirrored the patterns from D^k and non-D^k mice. Reductions in cDC were observed in all infected mice, with CD8⁺ and CD4⁺ DCs constituting the entirety of cDC loss, while DN DC numbers remained unchanged (Figure 1D compared to Figure 1B). Loss of CD8⁺ and CD4⁺ DC was aggravated by the absence of G2⁺ NK cells, further demonstrating that an effective licensed NK cell response to MCMV can promote cDC retention. We considered the possibility that Ly49G2 depletion itself could have resulted in cDC loss, but no differences in DC numbers were detected following G2 depletion in the absence of infection (Figure S4A in Supplementary Material). We are also confident that protection is a Ly49G2-specific effect in these mice since all non-G2 NK cell subsets were preserved in G2-depleted mice (Figure S4B in Supplementary Material). Thus, CD8⁺ and CD4⁺ DCs are specifically sensitive to MCMV-induced loss, which is amplified in mice without licensed NK cell-mediated virus control.

Since licensed NK cell-mediated virus control did not fully protect cDC populations, we further investigated the basis of cDC loss. It has been suggested that excessive IFN-I production from pDC could lead to cDC loss during infection (5, 10, 11). In this context, our data implied that licensed NK cells might reduce IFN-I levels enough to blunt cDC loss, but not enough to allow complete retention. Indeed, both D^k and non-D^k mice exhibited large increases in serum IFNα during early infection, but D^k mice produced lower amounts overall than non-D^k mice (Figure 2A). These data fit the hypothesis that high IFN-I could account for the residual cDC loss observed in D^k mice.

To further examine the role of IFN-I, we specifically depleted pDC before infection. We examined susceptible (non-D^k) mice since the mechanism of cDC loss should be similar in D^k and non-D^k settings if IFN-I toxicity is responsible. Furthermore, cDC protection should be readily detectable in a susceptible setting since loss is profound in these mice. As expected, the α-pDC mAb treatment effectively reduced the pDC population and serum IFNα levels (Figures 2B,C). The pDC depletion also resulted in increased numbers of cDC, but the majority of the effect was due to an expanded DN DC compartment (Figures 2D,E). Neither CD8⁺ nor CD4⁺ DC numbers were fully restored by the removal of pDC. These results suggest that pDC and pDC-derived IFN-I are unlikely to be responsible for the loss of CD8⁺ and CD4⁺ DC, although they may restrict DN DC during infection.

Although pDC are responsible for the bulk of IFN-I released prior to 48 h p.i., additional cell types also produce IFN-I during the course of MCMV infection (e.g., cDC, stromal cells, macrophages) (43). Hence, these could contribute to cDC loss in the absence of pDC if cDC loss is truly dependent on IFN-I signaling. We therefore broadly impaired IFN-I responses by blocking IFNAR signaling in non-D^k mice. Receptor occupancy was assessed by staining DC with fluorophore-conjugated α-IFNAR. As expected, IFNAR-blocked DC were much less effectively stained than DC from isotype-treated mice (Figure 3A). Inhibition of IFNAR signaling was verified by measuring mPDCA upregulation on DC and B cells, since surface mPDCA expression is highly responsive to IFN-I stimulation (44). Cells from IFNAR-blocked mice expressed much lower levels of mPDCA in comparison to isotype-treated controls (Figure 3B). Expression of the activation markers CD69 and CD86 was also much lower on T cells, B cells, and/or DC from IFNAR-blocked mice (Figure S5 in Supplementary Material).

Similar to pDC depletion, we observed a greater number of cDC in IFNAR-blocked mice (Figure 3C). However, this again stemmed largely from an increase in DN DC (Figure 3D). Although higher CD8⁺ DC numbers were also detected in IFNAR-blocked spleens, their overall retention was negligible when compared to numbers observed in mock-infected mice (Figure 3D compared to PBS in Figures 1 and 2). CD4⁺ DC numbers were unaffected by IFNAR blockade and declined equally in all mice during MCMV infection (Figure 3D compared to PBS in Figures 1 and 2). In aggregate, these results indicated that the loss of CD8⁺ and CD4⁺ DCs during MCMV infection is largely IFN-I independent, but IFN-I signaling specifically limits DN DC numbers.

To exclude the possibility of low-level IFN-I signaling despite IFNAR blockade, we repeated the experiments in IFNAR-deficient (IFNAR-KO) C57BL/6 (B6) mice. Although we had thus far examined cDC in C57L-derived mice, C57L and B6 mice are genealogically related (45), but encode distinct Ly49 receptors in their respective NKC haplotypes (46–49). Notably, the B6 NKC encodes Ly49H, which mediates MCMV resistance (19–23), whereas C57L mice lack Ly49H expression. We therefore mitigated this NKC difference in our experiments by Ly49H blockade or congenic replacement of the B6 NKC with the C57L NKC (NKC^l mice; see Materials and Methods). When cDC loss was assessed in the B6 background, isotype-treated, Ly49H-sufficient mice retained total cDC, CD8⁺ DC, and DN DC numbers (Figure 4A). Unexpectedly, although, they exhibited a strong trend toward loss of CD4⁺ DC, emphasizing the sensitivity of this population to infection-induced loss (Figure 4A). The reduced number of CD4⁺ DC was apparently compensated for by a slight increase in DN DC—thereby maintaining total cDC numbers. B6 mice lacking Ly49H-mediated resistance (α-Ly49H and NKC^l) displayed pronounced overall loss of cDC numbers (Figure 4A). This included significant decreases in CD8⁺ and CD4⁺ DC but, unlike C57L-derived mice, also included significant loss of DN DC numbers (Figure 4A). These data showed that, when B6 mice lack Ly49H-mediated resistance, MCMV-induced cDC loss is at least similar to that observed on the C57L background and, potentially, more profound given the DN DC data.

We crossed B6.IFNAR-KO and NKC^l mice to generate B6.NKC^l IFNAR-KO animals, which lack both Ly49H and IFNAR expressions (IFNAR-KO). This also yielded IFNAR-WT and IFNAR-Het littermates (Figure S6 in Supplementary Material), which were used as comparators. Upon challenge with MCMV, none of the mice displayed enhanced cDC retention, regardless of IFNAR expression status (Figure 4B). This contrasts somewhat with results from C57L mice, which maintained cDC numbers following IFNAR blockade due to an increase in DN DC. The lack of cDC retention was accompanied by a corresponding failure to increase DN DC numbers (Figure 4B). It is unclear why DN DC behaved differently in these IFNAR-KO mice, but such results could indicate that these cells are programmed differently when IFNAR signaling is absent throughout development rather than transiently blocked on mature cells. Alternatively, additional genetic modifiers could have altered DN DC behavior in this setting. Nonetheless, CD8⁺ and CD4⁺ DC loss was also comparable across all IFNAR genotypes (Figure 4B). The results demonstrated that cDC loss progressed similarly in the presence or absence of IFNAR signaling.

Global blockade and genetic ablation of IFN-I signaling can broadly alter cytokine profiles, immune cell responses, and overall host resistance to viral infection—all of which could influence DC numbers (3, 43, 50, 51). Thus, we considered that such reduced virus control could induce additional mechanisms of DC loss and interfere with our ability to detect IFN-I-dependent toxicity. To verify the degree of IFN-I-intrinsic toxicity in cDC populations, we generated a panel of BM chimeras with varying combinations of IFNAR expression. B6 mice were used as BM transplant recipients to preserve WT IFNAR signaling in stromal cells. These cells are primary targets of MCMV and IFN-I signaling in this compartment is important for limiting initial virus spread (4, 52, 53). B6 hosts were reconstituted with IFNAR-WT BM (CD45.1⁺), IFNAR-KO BM (CD45.2⁺), or a 1:1 mix to generate hematopoietic cell compartments with full, null, or half IFNAR-signaling capability (Figure 5A). As expected, IFNAR-KO B cells from IFNAR-KO and mixed chimeras lacked IFNAR expression (Figure 5B) and failed to upregulate mPDCA in response to MCMV (Figure 5C). Thus, IFNAR-KO cells remained IFN-I insensitive in all chimeric settings.

As expected, manipulation of IFN-I signaling resulted in increased virus levels (Figures 6A,B). Therefore, we also assessed MCMV genome levels in the spleens of the BM chimeras. At 3 d.p.i., the spleens of all chimeric mice exhibited similar viral burdens (Figure 6C). Hence, a difference in virus levels was not a complicating factor in the analysis of chimeric mice. We first analyzed total splenic cDC without distinguishing between CD45.1 and CD45.2 cells (bulk cDC). The cDC were substantially reduced in all infected chimeras when compared to uninfected controls (Figure 7A). However, despite the fact that cDC decreased in all infected mice, KO chimeras exhibited a significant retention over both mixed and WT settings. This, again, was solely due to retention of the DN DC subset since there was no appreciable increase in the number of CD8⁺ or CD4⁺ DC in KO chimeras (Figure 7A). Thus, hematopoietic IFNAR deficiency does not protect CD8⁺ and CD4⁺ DC, though it is sufficient to maintain DN DC numbers at levels equal to those seen in uninfected mice.

Direct comparison of CD45.1 and CD45.2 DC in mixed BM chimeras revealed that none of the subsets preferentially retained IFNAR-KO cells during infection (Figures 7B–D). However, IFNAR-KO DN DC trended toward higher numbers and did not differ significantly from either WT or KO populations in uninfected mice, which could indicate an intrinsic retention benefit. Overall, the data show that, during MCMV infection, DN DC numbers are the most responsive to IFN-I regulation, while profound CD8⁺ and CD4⁺ DC loss can occur independently of IFN-I.