P815 (No. TIB-64, haplotype H-2^d) and EL4 (No. TIB-39, haplotype H-2^b) cells were obtained from the American Tissue Culture Collection (ATCC) and cultivated in RPMI supplemented with 5% fetal calf serum (FCS) and antibiotics, or in DMEM with 10% FCS and antibiotics, respectively. Primary murine embryo fibroblasts (MEF) were cultivated in MEM supplemented with 10% FCS and antibiotics. A CD8 T-lymphocyte line (CTLL) specific for the immunodominant viral epitope IE1 (YPHFMPTNL) (65) was generated from spleen-derived memory CD8 T cells of latently infected BALB/c mice by four rounds of restimulation with synthetic peptide (66).

Virus derived from BAC plasmid pSM3fr (67) was used as “wild-type” (WT) virus, mCMV-WT. Recombinant viruses mCMV-Δm152 (40), mCMV-Δm157 (68), and mCMV-m152StopΔm157 (52) have been described previously.

BALB/c, C57BL/6, and C57BL/6-Unc93b1^3d/3d [briefly Unc93b1^3d/3d (69)] mice were bred and housed under specified-pathogen-free (SPF) conditions in the Translational Animal Research Center (TARC) at the University Medical Center of the Johannes Gutenberg-University Mainz, Germany, and at the central animal facility of HZI Braunschweig, Germany.

Recombinant plasmids were constructed according to established procedures, and enzyme reactions were performed as recommended by the manufacturers. Throughout, the fidelity of PCR-based cloning steps was verified by sequencing (GATC, Freiburg, Germany).

Mutagenesis of full-length mCMV BAC plasmid pSM3fr was performed in DH10B by using the two-step replacement method as described (67, 70), resulting in the BAC plasmid pSM3fr_m152.IE+E. For this, the shuttle plasmid pST76K_ie2_m152 was used to integrate the open reading frame (ORF) m152 for ectopic expression under the control of the ie2 promoter. In the first step, the intermediate plasmid pST76K_ie1/3-ie2 was generated by subcloning a PmlI-cleaved 5,557bp fragment of pUCAMB (71), containing nucleotides 181,415 to 186,972 (GenBank accession no. NC_004065) of the mCMV immediate-early (IE) region into the SmaI site of the shuttle plasmid pST76-KSR (70). In a subsequent step, a 1,452bp PCR fragment, encompassing the ie2 promoter and ORF m152, was introduced into the HpaI cleaved vector to generate pST76K_ie2_m152. The fragment was generated by a touchdown PCR with primer pair m152-HpaI-fwd GAAGTTAAC _184,240CATATAAAAGCTGTCCCCCATGCCATTCGA_{184,269-211,468}TCAGACGCGGGCTACTCCCGAAAGAGTAAC_211,439 and m152-HpaI-rev GGAGTTAAC _210,056TGACTAATAAGTTATCTTTATTGTACAAGTGTTGTGTGTTATCCCTGAGCCCATTCCCAG_210,115 (HpaI restriction sites are indicated in bold letters) using ProofStart Taq DNA polymerase (catalog no. 202205; QIAGEN, Hilden, Germany) and cycler conditions as follows: an initial step for 5 min at 95°C was followed by 18 cycles for 30 s at 94°C, 120 s with temperatures decreasing by 1°C per cycle starting from 62°C, and 90 s at 68°C each, followed by 12 cycles for 30 s at 94°C, 120 s at 45°C, and 90 s at 68°C.

Reconstitution and purification of a high-titer virus stock of mCMV-m152.IE+E was performed as described (72).

For in vitro assays, MEF were infected with the indicated viruses at a multiplicity of infection (MOI) of 4 with centrifugal enhancement of infectivity (72–74). Intraplantar infection of 8-to-10-week-old mice was performed in the left hind footpad with 1x10⁵ PFU of the respective virus.

For log-linear in vivo virus growth curves, BALB/c mice were immunocompromised by hematoablative treatment with a single 6.5 Gy dose of total-body γ-irradiation and were infected with the respective virus. Quantification of viral genome load in lungs and spleen was performed on days 2, 4, 6, 8, and 10 by qPCR as described previously (72).

Depletion of NK cells or of CD8 T cells was performed 24 h prior to infection by i.v. injection of 25µl rabbit antiserum directed against asialo-GM1 (catalog no. 986-100001; Wako Chemicals, Osaka, Japan) or of 1mg purified antibody directed against CD8 (clone YTS169.4), respectively (75).

To determine viral genome load in lungs and spleen, DNA of infected mice was isolated from the respective tissues with the DNeasy blood & tissue kit (catalog no. 69504; QIAGEN) according to the manufacturer’s instructions. Viral and cellular genomes were quantitated in absolute numbers by M55-specific and pthrp-specific qPCRs normalized to a log₁₀-titration of standard plasmid pDrive_gB_PTHrP_Tdy (72, 76).

Viral transcripts were quantitated from total RNA extracted from infected MEF or from lymph nodes (75), and 500 ng RNA was used as template for RT-qPCR. Absolute quantification of E1 or m152 transcripts using in vitro transcripts as standard has been described previously (77). Spliced E1 transcripts (78, 79) were chosen as a surrogate for viral replication that otherwise would be confounded by inoculum viral DNA. It is important to recall that a PFU of mCMV equals 500 copies of viral genomic DNA (73), so that intraplantar infection with 1x10⁵ PFU corresponds to 5x10⁷ copies, which critically confounds the quantitation of de novo viral DNA replication in the RLN, in particular at early times. E1 (M112-113) expression is essential for viral DNA replication (80) and the quantity of E1 transcripts correlates with the number of infected cells.

Custom peptide synthesis to a purity of > 80% was performed by JPT Peptide Technologies (Berlin, Germany). Synthetic peptides representing antigenic peptides in mouse haplotype H-2^b were M38 (SSPPMFRVP), M45 (HGIRNASFI), M57 (SCLEFWQRV), M122/IE3 (RALEYKNL), m139 (TVYGFCLL), and m141 (VIDAFSRL) (81). Those for mouse haplotype H-2^d were m04 (YPGSLYRRF), m18 (SGPSRGRII), M45 (VGPALGRGL), M83 (YPSKEPFNF), M84 (AYAGLFTPL), M105 (TYWPVVSDI), m123/IE1 (YPHFMPTNL), m145 (CYYASRTKL), and m164 (AGPPRYSRI) (8). The synthetic peptides were used for exogenous loading of target cells in the ELISpot assay.

An interferon gamma (IFNγ) enzyme-linked immunospot (ELISpot) assay was performed for quantification of IFNγ-secreting CD8 T cells after sensitization by peptide-loaded stimulator cells. Frequencies of mCMV-specific CD8 T cells were determined by incubation of graded numbers of immunomagnetically-purified CD8 T cells, derived from the RLN, which is the popliteal lymph node in the case of intraplantar infection, with stimulator cells (P815 or EL4, as it applied) that were exogenously loaded with synthetic peptides at a saturating concentration of 10^-7M (75). IE1 epitope presentation after endogenous antigen processing in infected BALB/c MEF was determined using short-term IE1-CTLL (82) as responder cells. Spots were counted automatically based on standardized criteria using Immunospot S4 Pro Analyzer (CTL, Shaker Heights, OH, USA) and CTL-Immunospot software V5.1.36. Frequencies of IFNγ-secreting cells and the corresponding 95% confidence intervals were calculated by intercept-free linear regression using Mathematica, version 8.0.4.

For a selective arrest of viral gene expression in the IE phase, MEFs were infected in the presence of 50µg/ml cycloheximide (CHX) to block protein synthesis reversibly. At 3 h after infection, the culture medium was replaced by fresh medium containing 5 µg/µl actinomycin D (ActD) as described previously (83).

IE1-CTLL (5x10⁵ cells) were co-cultivated with 1x10⁵ mCMV-infected, IE phase-arrested MEFs (BALB/c, haplotype H-2^d) for 5h at 37°C in the presence of brefeldin A (BD GolgiPlug, final concentration 1:1,000; catalog no. 555029; BD Biosciences). Thereafter, the CTLL cells were fixed, permeabilized with BD Cytofix/Cytoperm (catalog no. 554722, BD Biosciences) and stained with FITC-conjugated anti-mouse IFNy antibody (clone XMG1.2, catalog no. 554411, BD Biosciences) for cytofluorometric (CFM) analysis performed with flow cytometer Cytomics FC500 and CXP analysis software (Beckman Coulter).

An mCMV ORF library of expression plasmids spanning the entire mCMV genome (81) was used for ORF-specific stimulation of ex vivo isolated CD8 T cells with transfected SV-40 fibroblasts, followed by CFM detection of intracellular IFNγ.

The expression of mCMV proteins was detected by Western blot analysis as described (40). In brief, lysates of infected MEF were prepared and 30µg of total protein amount was subjected to separation by 12.5% SDS-PAGE, followed by blotting onto polyvinylidene difluoride membrane and protein labeling with the respective antibodies. The following antibodies were used: m152 (M3D10, monoclonal antibody, kindly provided by E. Kremmer, Helmholtz Zentrum München, Munich, Germany), IE1 (Croma 101, monoclonal antibody, kindly provided by S. Jonjic, University of Rijeka, Rijeka, Croatia), and IE2 (αIE2-N, polyclonal antibody, rabbit, Peptide Specialty Laboratories, Heidelberg, Germany). Detection of antibody binding was visualized by chemiluminescence using the ECLplus Western blotting detection system (catalog no. RPN2132; Amersham Bioscience, Little Chalfont, United Kingdom) and Lumi-Film (catalog no. 11666657001, Roche Applied Science, Mannheim, Germany).

To evaluate statistical significance of differences between two independent sets of data, the unpaired t-test with Welch’s correction of unequal variances was used. Differences are considered statistically significant for P values <0.05 and highly significant for P values <0.001. In cohort analyses of viral epitope-specific CD8 T cells by the ELISpot assay, differences are considered statistically significant if 95% confidence intervals do not overlap.

Virus doubling times (vDT) and their 95% confidence intervals (CI) were calculated from the slopes of log-linear regression lines determined by linear regression analysis (84–86). Calculations were performed with Graph Pad Prism 10 (Graph Pad Software, San Diego, CA). It should be noted that vDT values vary between different organs but are a constant for each organ independent of the viral replication parameter tested, that is, identical for viral genomic DNA copy numbers measured by qPCR, infectious virus expressed as PFU, or numbers of infected tissue cells determined by immunohistological detection of viral proteins (85).

Conclusions on the mechanism of CD8 T-cell priming are usually based on measuring the magnitude of the primary immune response. This includes the tacit assumption that antigen presentation requirements are identical for sensitization of naïve CD8 T cells by antigen presentation in the immunological synapse (87, 88), which is the initial priming event that requires pAPCs, and for subsequent clonal expansion. Accordingly, the terms of “priming” and “primary response” are usually used as synonyms, not least because the initial priming event is difficult to access. Thus, while a response indicates a successful initial priming, one must keep in mind that one actually looks at the combined result of priming and subsequent clonal expansion.

To provide evidence for or against either of the two pathways of antigen presentation, we studied here the mCMV-specific CD8 T-cell response in an RLN draining a site of local infection, specifically in the popliteal lymph node after intraplantar infection. The RLN is the anatomical structure where direct priming is said to occur in the peripheral interfollicular region (89–91). Our work builds on a previous study (75) in which we have shown that mCMV rapidly reaches the RLN, infects cells in the peri-subcapsular sinus region, and generates virus-specific effector CD8 T cells as early as by day 3 after virus exposure. Notably, this finding is in perfect accordance with the more recent study by Reynoso and colleagues (92) showing that lymph node conduits rapidly transport virions to infect pAPCs in the RLN paracortex followed by rapid direct T-cell priming within the T-cell zone.

In the specific case of CMV infections, however, viral interference with the MHC-I pathway of direct antigen presentation was expected to prevent or at least severely inhibit CD8 T-cell priming. To test this prediction, we analyzed the mCMV-specific CD8 T-cell response in Unc93b1^3d/3d mice. This strain on C57BL/6 genetic background is known to lack endosomal TLR3, 7, and 9 signaling and is impaired in exogenous antigen processing, resulting in a blockade of cross-presentation (69, 93–96). Accordingly, CD8 T-cell priming in these mice ought to largely depend on direct presentation. Previous reports on mCMV infection in this mutant mouse strain revealed an impaired cytokine production but comparable frequencies of M45-specific hepatic CD8 T cells (69, 97). Herein, we compared the CD8 T-cell response in C57BL/6 WT and Unc93b1^3d/3d mice not just for the M45 epitope but for a panel of known mCMV peptides presented in the H2^b haplotype ( Figure 1 ).

Overall, no qualitative differences in the immunodominance patterns were found. To our surprise, the magnitude of the CD8 T-cell response to some of the peptides tested, most markedly for M45, was even slightly higher in Unc93b1^3d/3d mice, but certainly not lower. This finding is consistent with data reported for C57BL/6 mice in which cross-presentation was suppressed by CpG pretreatment (61). In summary, the magnitude of the primary CD8 T-cell response is surprisingly not at all inhibited by viral interference with direct antigen presentation, which is known to be effective at the cellular level following infection with mCMV-WT in several cell types tested, including pAPCs, and in mice of haplotypes H-2^b and H-2^d (35, 98).

Given the remarkable finding that a CD8 T-cell response occurs despite genetic prevention of cross-presentation and despite viral interference with direct antigen presentation, we wondered whether the magnitude of the response would at least benefit from improved direct antigen presentation. For this, we compared infection with mCMV-WT and the immune evasion gene deletion mutant mCMV-Δm152 both under conditions with an open ( Figure 1A ) or closed ( Figure 1B ) antigen cross-presentation pathway in C57BL/6 and Unc93b1^3d/3d mice, respectively. The result was most striking: for the entire panel of viral epitopes tested, improved direct antigen presentation failed to improve the response, regardless of whether the antigen cross-presentation pathway was accessible or not. For C57BL/6 mice, earlier work by the group of A.B. Hill (98, 99) has already shown that an enhancement of direct antigen presentation by deletion of the viral key immune evasion gene m152 has little impact on the magnitude of the CD8 T-cell response, having suggested that priming may rather depend on antigen cross-presentation. Based on the data in Unc93b1^3d/3d mice, however, this explanation has now become obsolete, and we are faced with the riddle that the CD8 T-cell response in mice with the genetic background of C57BL/6 is resilient and buffered in that it neither depends on antigen cross-presentation nor does it appear to be notably influenced by enhanced direct antigen presentation.

Selectively in mice with the genetic background of C57BL/6, the viral protein m157 restricts viral replication by serving as an activatory ligand of Ly49H⁺ NK cells (100, 101). It has been shown that the activation of Ly49H⁺ NK cells by m157 suppresses the mCMV-specific CD8 T-cell response, most likely by reducing the number of infected pAPCs available for direct antigen presentation (102). Thus, Ly49H⁺ NK cells are a relevant player that certainly contributes to the magnitude of the CD8 T-cell response in C57BL/6 mice.

For clarification, we investigated the impact of NK cells on the CD8 T-cell response in the draining RLN measured on day 7 after intraplantar infection of C57BL/6 mice, and correlated the response magnitude with viral replication at this site on day 3 ( Figure 2 ). In the presence of NK cells and CD8 T cells, mCMV-WT and mCMV-Δm152 induced similar CD8 T-cell responses ( Figure 2A1 ), consistent with the preceding, independent experiment ( Figure 1A ). Notably, transcription of viral gene E1, which serves as a surrogate for viral replication, was identical for both viruses and on a low level ( Figure 2A2 ). In contrast, pan-NK cell depletion prior to infection led to an increase in the response magnitude for the immunodominant viral peptides M45, M57, and m139, but only when direct antigen presentation was enhanced by deletion of m152 ( Figure 2B1 ). As it was predicted, depletion of NK cells led to a greatly enhanced viral replication in the RLN ( Figure 2B2 ) compared to the control group with no NK-cell depletion ( Figure 2A2 ). Deletion of m152 resulted in a slight, but statistically significant, reduction in viral replication compared to WT virus infection ( Figure 2B2 ). This reduction was caused by recently primed CD8 T cells, as the difference was abolished by the depletion of CD8 T cells ( Figure 2C ).

Upon first impression, it may be irritating that not all viral epitopes show the same response pattern. Epitope hierarchy, however, is a general observation and is explained by differences in many consecutive steps in the MHC-I pathway of antigen processing and presentation. Key variables, which are even used in epitope prediction algorithms, include the efficacy of peptide-generating proteasomal cleavage and peptide binding affinity to the presenting MHC-I molecule. Cell surface density of pMHC-I complexes determines the cooperative TCR binding avidity and thus the intensity and duration of TCR signaling. This in turn defines the proliferation rate of CD8 T cells for clonal expansion and thus, finally, the response magnitude. As a consequence, differences that do not reach statistical significance for subdominant viral epitopes can reach statistical significance for dominant viral epitopes. For this reason, we always tested a panel of epitopes, and conclusions must be drawn from the overall picture rather than from single epitopes. Our data did not reveal a critical influence of the type of the presenting MHC-I molecule, which are K^b and D^b in C57BL/6 mice.

These findings were essentially reproduced in an independent experiment using the alternative approach of testing the influence of m152 on the response magnitude in infected C57BL/6 mice in the absence specifically of Ly49H⁺ NK cell activation via Ly49H-m157 ligation, instead of pan-NK cell depletion ( Figure 3 ). For this, we compared response magnitude and viral replication in the RLN after infection with mCMV-Δm157, expressing m152, and mCMV-m152StopΔm157, lacking m152 expression, both in the absence of Ly49H⁺ NK cell activation. Notably, deletion of m152 increased the overall CD8 T-cell response ( Figure 3A1 ), more or less paralleling the findings after pan-NK cell depletion ( Figure 2B ). Under these conditions, too, early recognition of infected cells is indicated by a reduction in viral replication in the RLN after infection with mCMV-m152StopΔm157 ( Figure 3A2 ). Again, this antiviral control in the absence of Ly49H⁺ NK cell activation is mediated by recently primed CD8 T cells, as depletion of CD8 T cells led to an increase in viral replication in the RLN ( Figure 3B compared to Figure 3A2 ).

The results of the two approaches are consistent in having demonstrated that early viral replication in the RLN of C57BL/6 mice is mainly controlled by Ly49H⁺ NK cells. While an enhanced direct antigen presentation did not improve the overall CD8 T-cell response magnitude in the presence of the NK-cell response ( Figure 2A1 ), depletion or missing activation of NK cells disclosed an improvement ( Figures 2B1 , 3A1 ). Importantly, a high CD8 T-cell response corresponded to low intranodal viral replication, and a low response corresponded to high viral replication ( Figures 2B , 3A ). This clearly argues against antigen cross-presentation for which the opposite should apply: high virus replication and thus high amounts of antigenic proteins available for uptake by uninfected pAPCs ought to correspond to a high CD8 T-cell response and, accordingly, low virus replication ought to correspond to a low CD8 T-cell response. In summary, although masked by the strong activity of Ly49H⁺ NK cells, CD8 T cells in C57BL/6 mice are primed primarily by direct antigen presentation.

Since the mCMV-specific CD8 T-cell response in C57BL/6 mice is masked by the strong response of Ly49H⁺ NK cells, we decided to switch to the analysis of the mCMV-specific CD8 T-cell response in BALB/c mice, which do not express Ly49H and accordingly lack this functionally dominant subset of NK cells. Based on the evidence that the CD8 T-cell response in C57BL/6 mice is driven by direct antigen presentation (see above), and assuming that this is also the case in BALB/c mice, we reasoned that the best evidence for direct antigen presentation would be to show that enhanced and inhibited direct antigen presentation correspond to a high and low CD8 T-cell response, respectively. Increased direct antigen presentation compared to infection with the WT virus is achieved by deletion of the immune evasion gene m152 in virus mCMV-Δm152. For a more strongly inhibited direct antigen presentation compared to infection with the WT virus, we constructed a “super-evasion” virus mCMV-m152.IE+E as a new study tool.

In mCMV-WT infection, m152 is expressed quite early in the Early (E) phase of the viral replication cycle (24, 39). Antigens expressed even earlier, that is, in the Immediate-Early (IE) phase, may profit from a head start advantage of presentation before immune evasion can operate. After infection with mCMV-m152.IE+E, m152 is expressed from the IE phase onward.

Ectopic expression in the IE phase was achieved by insertion mutagenesis placing ORF m152 under the control of the ie2 enhancer-promoter, thereby disrupting the ie2 gene. Expression as an E phase protein occurred from its authentic position in the mCMV genome ( Figure 4A ). For verifying immune evasion operating already in the IE phase, viral gene expression in BALB/c-derived fibroblasts was metabolically arrested in the IE phase ( Figure 4B1 ). This led to an enhanced synthesis of IE proteins IE1 and IE2, and absence of the m152 protein, after infection with either mCMV-WT or the deletion mutant mCMV-Δm152, whereas all three glycosylation isoforms of m152 (40) were strongly expressed and IE2 was absent after infection with mCMV-m152.IE+E ( Figure 4B2 , ). Presentation of the antigenic peptide IE1 (YPHFMPTNL, presented by L^d) was tested functionally with an IE1 epitope-specific CTLL (IE1-CTLL) used as responder cells in an IFNγ-based ELISpot assay with IE phase-arrested infected cells as stimulator cells ( Figure 4B3 ) as well as by intracellular cytofluorometric staining of IFNγ in IE1-CTLL cells sensitized by IE phase-arrested infected cells ( Figure 4B4 ). In both assays, IE1 peptide was presented without significant difference after infection with either mCMV-WT or the mCMV-Δm152 deletion mutant, since m152 is not expressed in the IE phase anyway. In contrast, presentation was blocked in IE phase-arrested cells after infection with mCMV-m152.IE+E ( Figures 4B3, B4 ).

Since infected cells in vivo are not normally arrested in the IE phase, we studied the kinetics of m152 transcription in untreated cells infected with either mCMV-WT, expressing m152 only in the E phase and onward, or mCMV-m152.IE+E, expressing m152 additionally already in the IE phase ( Figure 5A ). The transcription data were then correlated with presentation of the antigenic IE1 peptide detected by sensitization of IE1-CTLL cells ( Figure 5B ). In accordance with the concept of constructing mCMV-m152.IE+E, m152 was expressed faster and inhibited antigen presentation earlier compared to infection with mCMV-WT. Specifically, at both times chosen, the IE1 peptide was not detectably presented by cells infected with mCMV-m152.IE+E, whereas it was always presented after infection with mCMV-Δm152. Under conditions of “physiological” immune evasion gene expression, as it applies to infection with mCMV-WT, the IE1 peptide was presented early in the time course but almost absent later ( Figure 5B ). We thus conclude that direct antigen presentation is largely inhibited at early and later times in cells infected with mCMV-m152.IE+E, whereas it is not inhibited at any time after infection with mCMV-Δm152, and only at later times after infection with mCMV-WT.

BALB/c mice are competent in both antigen presentation pathways and therefore theoretically have the choice between direct antigen presentation and antigen cross-presentation. This raised the question of which pathway is used as the canonical pathway for priming of naïve CD8 T cells and for subsequent clonal expansion or whether the pathways can replace each other in case of need. Assuming that direct antigen presentation is the preferred mode of priming, one expects a response magnitude in a rank order defined by the quantity of pMHC-I complexes on the surface of infected APCs and thus reciprocal to the strength of immune evasion. Specifically, the response should be strongest after infection with mCMV-Δm152, intermediate after infection with mCMV-WT, and lowest after infection with mCMV-m152.IE+E.

The results of the experiment ( Figure 6A ) did not match this prediction. For all antigenic peptides tested, the magnitude of the CD8 T-cell response was essentially the same for the extremes of lowest and highest direct antigen presentation after infection with mCMV-m152.IE+E and mCMV-Δm152, respectively ( Figure 6A ). It seemed to be an obvious conclusion that direct antigen presentation is not the mode of priming. However, the results for mCMV-WT do not fit this interpretation, as the CD8 T-cell response to this virus was higher compared to the two immune evasion virus mutants for the epitope panel tested ( Figure 6A , Supplementary Figure 1 ) and was broader in terms of the epitope specificity repertoire determined by using a viral genome-wide ORF expression library ( Supplementary Figure 1 ). So, if direct presentation plays no role at all, as the two antipodal mutants suggest, why is the response after infection with mCMV-WT the best, with immune evasion and direct antigen presentation being in between?

Based on our experience with the C57BL/6 model ( Figures 2 , 3 ) and our published previous work in the BALB/c model (75), we quantitated replication of the three viruses in the draining RLN, the site where priming and clonal expansion take place, on day 3 after infection ( Figure 6B ). As an important control, replication differences caused by genetically-determined viral replicative fitness were excluded by showing identical replication of the three viruses in immune-deficient mice ( Supplementary Figure 2 ). Therefore, replication differences in the RLN must reflect differences in immune control. Notably, unlike the sizes of the CD8 T-cell responses, the intranodal viral replication fulfilled the logic, that is, highest replication corresponds to strongest immune evasion of mCMV-m152.IE+E, lowest replication corresponds to weakest immune evasion of mCMV-Δm152, and intermediate replication corresponds to intermediate immune evasion of mCMV-WT ( Figure 6B ).

A possible contribution of recently primed virus-specific CD8 T cells to the control of intranodal virus replication was tested by CD8 T-cell depletion prior to infection ( Figure 6C ). It may be instructive that our previous work has already shown the presence of antiviral effector CD8 T cells in the RLN after 3 days of infection (75). In accordance with almost missing antigen presentation on cells infected with mCMV-m152.IE+E, intranodal viral replication was not detectably controlled by CD8 T cells, whereas, in accordance with optimal antigen presentation on cells infected with mCMV-Δm152, intranodal viral replication was almost prevented when CD8 T cells were present. Again, results for mCMV-WT were in between.

In summary, up to this point, the diametrically different levels of intranodal viral replication of mCMV-m152.IE+E and mCMV-Δm152 in BALB/c mice reflect missing and strong antiviral control, respectively, by the just generated effector CD8 T cells. Surprisingly, the result is almost the same low level of CD8 T-cell response, while the intermediate level of immune evasion by mCMV-WT results in the best response, generating high numbers of antiviral CD8 T cells exported from the RLN for controlling viral replication at distant sites of viral pathogenesis.

Before arriving at a conclusion, it was important to consider the possibility that direct antigen presentation and antigen cross-presentation may not be mutually exclusive. An alternative explanation for the comparable magnitude of response following high and low direct antigen presentation could be that the response to mCMV-Δm152 is actually driven by high direct antigen presentation, whereas the response to the super-evasion virus mCMV-m152.IE+E may be due to antigen cross-presentation being used as an alternative pathway, aided by large amounts of antigenic material derived from many dying cells. This tempting idea, though, was refuted by comparing the response to the two viruses in the cross-presentation deficient Unc93b1^3d/3d mice, revealing no notable difference for the panel of viral epitopes tested ( Figure 7A ).

As m152 also impacts on the NK-cell response by downmodulating ligands of the activatory NK-cell receptor NKG2D (47–51), high and low NK-cell responses to mCMV-Δm152 and mCMV-m152.IE+E, respectively, could indirectly affect the CD8 T-cell responses to these two viruses differentially. However, except for a minor difference in the case of viral epitope M45, the magnitude of the CD8 T-cell responses to the two extremes of immune evasion remained comparable after pan-NK cell depletion under conditions of absent antigen-cross presentation in Unc93b1^3d/3d mice ( Figure 7B ). As a consequence, all our results in both C57BL/6 and BALB/c mice must be explained on the basis of direct antigen presentation.