Processing and Presentation of Murine Cytomegalovirus pORFm164-Derived Peptide in Fibroblasts in the Face of All Viral Immunosubversive Early Gene Functions

Abstract

CD8 T cells are the principal effector cells in the resolution of acute murine cytomegalovirus (mCMV) infection in host organs. This undoubted antiviral and protective in vivo function of CD8 T cells appeared to be inconsistent with immunosubversive strategies of the virus effected by early (E)-phase genes m04, m06, and m152. The so-called immune evasion proteins gp34, gp48, and gp37/40, respectively, were found to interfere with peptide presentation at different steps in the major histocompatibility complex (MHC) class I pathway of antigen processing and presentation in fibroblasts. Accordingly, they were proposed to prevent recognition and lysis of infected fibroblasts by cytolytic T lymphocytes (CTL) during the E phase of viral gene expression. We document here that the previously identified MHC class I D^d-restricted antigenic peptide ²⁵⁷AGPPRYSRI²⁶⁵ encoded by gene m164 is processed as well as presented for recognition by m164-specific CTL during the E and late phases of viral replication in the very same cells in which the immunosubversive viral proteins are effectual in preventing the presentation of processed immediate-early 1 (m123-exon 4) peptide ¹⁶⁸YPHFMPTNL¹⁷⁶. Thus, while immunosubversion is a reality, these mechanisms are apparently not as efficient as the term immune evasion implies. The pORFm164-derived peptide is the first noted peptide that constitutively escapes the immunosubversive viral functions. The most important consequence is that even the concerted action of all immunosubversive E-phase proteins eventually fails to prevent immune recognition in the E phase. The bottom-line message is that there exists no immune evasion of mCMV in fibroblasts.

Many viruses have evolved mechanisms to avoid recognition by the various effectors of the innate and adaptive immune system (for a review, see reference 1). Cytomegaloviruses (CMVs) have proved to be particularly inventive in that respect. Even though human cytomegalovirus (hCMV) and murine cytomegalovirus (mCMV) have arrived at somewhat different molecular solutions to subvert immune control, interference at various steps with antigen processing and presentation in the major histocompatibility complex (MHC) class I pathway is a common trait in evading control by CD8 T cells (for reviews, see references 18 and 66). The two viruses share homologous genes but differ in gene clusters unique to either virus. Specific adaptation to the host immune system is suggested by the finding that genes involved in the manipulation of murine MHC class I glycoprotein trafficking are unique to mCMV (indicated by lowercase m, according to the nomenclature introduced by Rawlinson et al. [46]). These are the early (E) genes m04 and m06, members of the m02 gene family, and m152, a member of the m145 gene family. Likewise, immune evasion genes of hCMV interfering with human MHC (HLA) class I molecule trafficking belong to the US2 and US6 gene families, which have no sequence homologs in mCMV (8, 18, 46).

Gene m152 of mCMV was the first immune evasion gene identified for CMVs (64, 70). Its gene product, gp37/40, mediates retention of peptide-loaded MHC class I β2-microglobulin complexes in the endoplasmic reticulum (ER) and the ER-Golgi intermediate compartment (9, 69, 70). As a consequence, processed antigenic peptides can be isolated from infected cells, but they are not presented at the cell surface (9). Deletion of gene m152 in a recombinant mCMV was found to enhance the susceptibility to immune control in vivo (37). The m04 gene product gp34 binds to MHC class I molecules. The complex can be immunoprecipitated from the cell surface (35) and, in a distinct form, from a pre-Golgi compartment (31). It was originally speculated that m04-gp34 may restore the display of surface class I molecules for silencing natural killer cells in order to also cope with innate immunity (35). In accordance with this hypothesis, Oliveira et al. recently mapped the protection against natural killer cell-mediated immune surveillance of mCMV to the m02 gene family (43). In addition, m04 is negatively implicated in CD8 T-cell-mediated immune surveillance. As exemplified recently by Kavanagh et al., m152-gp37/40 is sufficient for the retention of certain MHC class I molecules, D^b in their specific example, whereas assistance by m04-gp34 was found to be required for evading K^b-restricted recognition (30). Finally, gp48 encoded by gene m06 appears to be particularly efficient in downregulating cell surface MHC class I expression by targeting MHC class I molecules in the ER and rerouting them for lysosomal degradation (57). However, the proposed role of m06-gp48 in subverting the presentation of antigenic peptides is not yet fully explored. For the L^d-restricted IE1 (gene m123-exon 4) peptide ¹⁶⁸YPHFMPTNL¹⁷⁶ (55) that is presented in the immediate-early (IE) phase but not in the E phase (10, 50), the E-phase evasion phenotype is reverted by deletion of the m152 gene in recombinant mCMV ΔMC95.21 (37). This implies that genes m04 and m06 are not significantly involved in the prevention of IE1 peptide presentation during the E phase. Altogether, one should expect that a concerted action of all three immune evasion genes protects against recognition by CD8 T cells in the E phase of viral gene expression.

However, the immune evasion theory has never adequately addressed the original finding that polyclonal cytolytic T lymphocytes (CTL) derived from draining lymph nodes of BALB/c (H-2^d) and C57BL/6 (H-2^b) mice during acute infection lysed infected fibroblasts in the E and late (L) phases (52, 53). In addition, presentation of antigenic peptides during the E phase on BALB/c fetal fibroblasts was indicated by the finding that in vivo protective and ex vivo cytolytic CTL isolated from pulmonary infiltrates during mCMV pneumonia after syngeneic experimental bone marrow transplantation (BMT) recognized E-phase target cells with preference and in an MHC-restricted manner (22).

In accordance with these earlier findings, H-2^d-restricted antigenic peptides were recently identified in the E proteins M84-p65 (26), M83-pp105 (23), and pORFm18-derived protein (21), and, notably, also in the immune evasion protein m04-gp34 (24). Thus, surprisingly, gp34 is positively and negatively implicated in immune control. However, these peptides gave no explanation for the recognition of E-phase target cells by CTL, as none of those was consistently found to be presented in a detectable quantity during the E phase on infected fibroblasts when tested with the corresponding peptide-specific CTL lines (CTLL) (R. Holtappels, unpublished data).

In the present report, this long-standing enigma is resolved. We have recently identified an antigenic peptide in pORFm164, namely, peptide ²⁵⁷AGPPRYSRI²⁶⁵, that is presented by D^d (25). Notably, this peptide turned out to be as immunodominant as the IE1 peptide (25), the one previously regarded by everyone in this field as the only immunodominant mCMV peptide in the H-2^d haplotype. We show here that the pORFm164-derived peptide is not only processed but is also presented during the E and L phases of viral gene expression in the very same cells in which presentation of the correctly processed IE1 peptide is prevented by the immunosubversive functions. Thus, even a concerted action of all immune evasion proteins of mCMV cannot prevent the recognition of infected cells by CTL.

MATERIALS AND METHODS

Generation of peptide-specific CTLL.

Female BALB/c mice (H-2^d haplotype) were used as donors of memory T cells for the generation of CTLL. The mice were primed at the age of 8 weeks by intraplantar infection with 10⁵ PFU of sucrose gradient-purified, cell culture-propagated mCMV strain Smith ATCC VR-194 (38), which was recently reaccessioned as strain Smith ATCC VR-1399. At ca. 3 months after inoculation, a time at which the acute infection is cleared and viral latency is established in immunocompetent donors (48), spleen cells were isolated and used as responder cells for repetitive stimulations with synthetic peptides. The precise conditions of in vitro restimulation with peptides IE1 ¹⁶⁸YPHFMPTNL¹⁷⁶ and m164 ²⁵⁷AGPPRYSRI²⁶⁵ as well as the in vitro and in vivo functional properties of the resulting IE1-CTLL and m164-CTLL were described in detail recently (25).

Target cells and cytolytic assay.

Target cells were either infected BALB/c fetal fibroblasts, referred to as mouse embryo fibroblasts (MEF), or P815 (H-2^d) mastocytoma cells pulsed for 30 min at 20°C with the indicated molar concentrations of synthetic peptides (JERINI Bio Tools GmbH, Berlin, Germany) or with naturally processed peptides present in eluate fractions of a high-pressure liquid chromatography (HPLC) separation of acidic cell extracts (see below). MEF were infected in the second tissue culture passage under conditions of centrifugal enhancement of infectivity (29, 38) with 0.2 PFU of mCMV per cell, which equals a multiplicity of infection of 4 PFU* per cell (with the asterisk indicating the 20-fold enhanced infectivity), so that almost every cell is infected. Infected cells were used in the three kinetic phases of viral gene expression, referred to as the IE, E, and L phases (27, 32, 53).

(i) IE targets.

The culture medium was replaced by fresh medium containing 50 μg of cycloheximide (CH) per ml (Sigma) to block protein synthesis reversibly, and the infection was performed 15 min later. At 3 h after the infection, the CH medium was discarded. The cells were washed thoroughly with medium containing 5 μg of actinomycin D (ActD) per ml (Sigma) and incubated for 2 h in this ActD medium to block E-phase transcription.

(ii) E targets.

At 15 min prior to infection, the culture medium was replaced by medium containing 250 μg of phosphonoacetic acid per ml (Sigma) to block DNA synthesis. The cells were harvested at 16 h after the infection.

(iii) L targets.

MEF were infected for ca. 24 h in the absence of metabolic inhibitors. At 15 min before cell harvest, ActD (5 μg per ml) was added to arrest viral gene expression.

After the incubations, the cells were harvested and radioactively labeled with Na₂⁵¹CrO₄ for the 4-h chromium release assay performed in 96-well microcultures according to standard procedures. Effector cells in the assay were cell lines IE1-CTLL and m164-CTLL. Data represent the mean values of triplicate assay cultures for each of the indicated effector/target (E/T) cell ratios.

Analysis of naturally processed antigenic peptides.

The IE-, E-, and L-phase-infected MEF described above were also used for the acid extraction and HPLC separation of naturally processed peptides performed as originally described by Falk et al. (13), with some modification (24). In brief, 1 × 10⁸ to 3 × 10⁸ cells were acidified to a pH of 2 with trifluoroacetic acid and were then homogenized. The extract was cleared by ultracentrifugation, and low-molecular-weight compounds in the supernatant were concentrated by size exclusion chromatography (Sephadex G-25 column; Pharmacia) followed by solid-phase extraction (C₁₈ reversed-phase unit; SepPak, Waters, Germany) and vacuum centrifugation. Separation of peptides was performed by HPLC using a SuperPac Sephasil C₁₈ 5-μm reversed-phase column (Pharmacia). Peptides were eluted at a flow rate of 0.8 ml per min on a linear acetonitrile gradient. As shown previously, the m164 peptide elutes with a peak in fraction 22, and the IE1 peptide elutes with a peak in fraction 28 (25). Extraction efficacies for the two peptides were determined by measuring the percent peptide recovery after HPLC separation of an extract from uninfected MEF that was supplemented with synthetic peptide.

Aliquots of the HPLC fractions and the indicated dilutions thereof were used for pulsing P815 target cells for the cytolytic assay. Relative quantities of naturally processed IE1 and m164 peptides were estimated from the dilutions of the respective HPLC peak fractions and the molar concentrations of the corresponding synthetic peptides required for half-maximal lysis of P815 target cells by IE1-CTLL and m164-CTLL, respectively.

Analysis of gene expression by RT-PCR.

The kinetics of viral gene expression in infected MEF were assessed by reverse transcriptase (RT) PCRs. MEF were infected at a multiplicity of infection of 4 PFU* per cell (see above), a dose that corresponds to 100 viral genomes per cell (38). Under such a condition, >90% of the cells are infected, as verified by immunofluorescence specific for the intranuclear IE1 protein pp89 of mCMV (34). For each time point in the kinetics, three replicate six-well plate MEF monolayer cultures were harvested, and the cells were pooled for analysis. Metabolic inhibitors to arrest viral gene expression in the IE and E phases were used essentially as described above for the preparation of target cells. Specific conditions are given in the legends to the figures for the particular experiments. Highly purified polyadenylated RNA was isolated from the cell lysate by using oligo(dT)-coated superparamagnetic 50-nm-diameter microbeads (μMACS mRNA isolation kit and column type-μ; Miltenyi Biotec Systems, Bergisch-Gladbach, Germany) followed by digestion of contaminating DNA as described in greater detail previously (23). Throughout, oligo(dT) priming was used for reverse transcription. Reactions were carried out by using an automated thermal cycler (GeneAmp PCR System 9700; Perkin-Elmer Applied Biosystems, Norwalk, Conn.). RT-PCRs specific for transcripts of viral genes ie1 (m123-exon 4), m04, and m152, as well as for the cellular gene hprt (hypoxanthine phosphoribosyltransferase), were described in previous reports (23, 24). Amplification products were visualized by standard procedures of 2% (wt/vol) agarose gel electrophoresis with ethidium bromide and UV-light staining, Southern blotting, hybridization with the respective γ-³²P-end-labeled oligonucleotide probe, and autoradiography.

(i) PCR specific for m164 transcripts.

Primers and probe for amplification and detection of m164 cDNA are shown in the map in Fig. 1A. For cycles 2 through 29, the time-temperature profile was as follows: denaturation for 30 s at 96°C, annealing for 1 min at 58°C, and elongation for 1 min at 72°C. In the first cycle, denaturation was performed for 3 min at 95°C. In the last cycle (cycle 30), the elongation time was extended to 5 min.

FIG. 1.

Temporal expression of gene m164. (A) Location map of gene m164 and of primers and probe for the PCR amplification and Southern blot detection of a 431-bp cDNA fragment. The HindIII restriction map and the map positions refer to the sequence of mCMV strain Smith ATCC VR-194 (12, 46) (GenBank accession no. MCU68299). (B) Kinetics and drug sensitivity of m164 transcription. RT-PCR was performed with purified poly(A)⁺ RNA isolated from MEF at the indicated times after infection in the presence or absence of metabolic inhibitors. ActD, 5 μg/ml; CH, 100 μg/ml; PAA, phosphonoacetic acid (250 μg/ml). An ethidium bromide-stained gel is shown on the left; M, 100-bp size markers. The corresponding Southern blot autoradiograph obtained after hybridization with the γ-³²P-end-labeled oligonucleotide probe is shown on the right. Asterisk, poly(A)⁺ RNA retested for Fig. 2B; arrow, faint signal visible at 1.5 h postinfection.

(ii) PCR specific for m06 transcripts.

Nucleotide positions of primers and probe refer to the mCMV Smith strain genomic sequence according to Rawlinson et al. (46; GenBank accession no. MCU68299 [complete genome]). The sequences of the primers and probe are as follows: forward primer, 5′-(n5401)-GATGACTTTCCAGATGGGAGAATC-(n5424)-3′; reverse primer, 5′-(n5580)-CGGTTTATCTCGGTACGCGAAACG-(n5557)-3′; probe, 5′-(5489)-ATGCCTCCGTCGACTAACGAAACG-(n5512)-3′. The time-temperature profile for the amplification of m06 cDNA was as described above for m164 cDNA, except that the annealing temperature was 64°C.

RESULTS

Kinetics of mCMV m164 gene expression in infected fetal fibroblasts.

The antigenic pORFm164-derived peptide ²⁵⁷AGPPRYSRI²⁶⁵ (25) has been identified by the reverse immunology approach, that is, by functional screening of synthetic nonapeptides containing an MHC class I (D^d in this specific case) binding motif (for an overview, see reference 45). Gene m164 encodes a putative glycoprotein of 427 amino acids with a sequence-deduced molecular mass of 46.6 kDa (46) and two potential membrane-spanning α-helices (25). Except for its role in antiviral immunity, there exist no experimental data on the glycosylation, subcellular localization, or functional role in the viral life cycle.

Immunodominance of the IE1 peptide ¹⁶⁸YPHFMPTNL¹⁷⁶ (22, 54, 55) was previously explained by expression of the IE1 protein pp89 in the IE phase, that is, prior to the expression of E-phase immune evasion proteins. As documented in our recent work, the m164 peptide compares to IE1 with respect to in vivo immunogenicity (25). The expression kinetics of gene m164 were therefore important and relevant to its immunological properties. In the original work by Keil et al., abundant expression under IE conditions, defining the major IE region, was mapped to the HindIII K and L junction (32), now known to comprise the enhancer-flanking genes ie1/3 and ie2 (33, 40, 41). IE transcripts also hybridized to fragment HindIII E (32), which was the first evidence for IE genes being present in the gross region where m164 is located (for a map, see Fig. 1A). Very recently, the group of T. E. Shenk has revisited the gene expression kinetics of mCMV by using an open reading frame DNA microarray. The analysis defined a cluster of IE genes in HindIII E, namely, genes m166/7, m168, and m169, whereas IE cDNA did not hybridize to m164 (B. A. Wing, E. P. Browne, and T. E. Shenk, Abstr. 26th Int. Herpesvirus Workshop, Regensburg, Germany, abstr. 1.08, 2001; T. E. Shenk, personal communication). Based on that information, we expected to find m164 transcripts with E- or L-phase kinetics.

The analysis was performed with RT-PCR (Fig. 1). Notably, m164 transcripts were detected as a faint signal at 1.5 h postinfection (Fig. 1B), which is considered an IE time (32). The possibility that this signal results from amplification of RNA packaged into virions, as shown by Bresnahan and Shenk for several RNA species of hCMV (6), was ruled out by the finding that the 1.5-h transcript was not generated in cells infected in the presence of inhibitor ActD. By definition, the IE phase of herpesvirus replication is characterized by transcription in the absence of de novo protein synthesis (27, 28). To our surprise, m164 transcripts were clearly detectable at 3 h after infection in the presence of CH, even though the comparison to the 3-h signal obtained in the absence of CH indicated a partial sensitivity of the transcription to the block of de novo protein synthesis.

Basal level of m164 transcription under IE conditions.

We might have detected a basal level of m164 gene expression not requiring a transactivation by newly synthesized IE proteins. This could result from a basal activity of the promoter or from activation of the promoter by a virion protein delivered to the cells through the entry process. Figure 2 shows a number of relevant controls. The first objection to be addressed was a possible leakiness of the translational block. The effects of increasing CH concentrations as well as of prolonged periods of preincubation of the cells in CH medium prior to infection were tested using a freshly prepared CH solution (Fig. 2A). None of these modifications of the protocol prevented the generation of m164 transcripts. At the highest dose of CH tested, namely, at 1 mg/ml, the signal in the gel was slightly diminished, but this very high dose also reduced transcription from the cellular gene hprt, indicating the onset of a toxic effect. Again, unlike the transcription from the IE gene ie1 (m123-exon 4), m164 transcription was found to be partially sensitive to the inhibition. This finding indicates that m164 gene expression differs from the expression defined for a classical IE gene. A second objection to be addressed was the detection of contaminating viral DNA. As shown in Fig. 2B, the signal clearly depended on reverse transcription. In conclusion, while m164 does not precisely behave like a classical IE gene, one cannot neglect the fact that there exists a basal transcription under IE conditions.

FIG. 2.

Basal transcription of gene m164 in the IE phase. (A) Different protocols of metabolic inhibition. Throughout, ActD (5 μg/ml) was added for 1 h at 3 h postinfection. RT-PCRs specific for genes m164, ie1 (m123-exon 4), and hprt were performed with purified poly(A)⁺ RNA isolated from MEF at 4 h after infection. Ethidium bromide-stained gels are shown; M, 100-bp size markers. Lanes 1, negative control (uninfected MEF); lanes 2, 3-h infection in the absence of CH; lanes 3, 3-h infection in the presence of CH (100 μg/ml); lanes 4, presence of CH (100 μg/ml) for 40 min prior to infection and for 3 h postinfection; lanes 5, presence of CH (100 μg/ml) for 2 h prior to infection and for 3 h postinfection; lanes 6, presence of CH (200 μg/ml) for 2 h prior to infection and for 3 h postinfection; lanes 7, infection in the presence of CH (1 mg/ml) for 2 h prior to infection and for 3 h postinfection. (B) Signal dependence on reverse transcription. Poly(A)⁺ RNA from the experiment shown in Fig. 1 (lane marked by an asterisk) was tested by m164- and hprt-specific PCRs after reverse transcription (RT) or without preceding reverse transcription (Ø). An ethidium bromide-stained gel is shown; M, 100-bp size markers.

Expression of immune evasion genes in the E phase.

The immune evasion genes m04 (35), m06 (57), and m152 (70) were characterized as E genes. Since the previous studies were not performed with RT-PCR, we revisited the temporal expression of these three genes to get an authentic comparison to the kinetics of m164 expression (Fig. 3). Gene m04 is clearly an E gene, as first transcripts were detected at 4 h after infection and were absent after infection in the presence of CH. Gene m152 is expressed earlier than m04, with a signal already apparent at 2 h. Since this transcription was prevented by CH, m152 is an E gene, too. It is worth emphasizing that the metabolic inhibitor protocol that successfully prevented m152 transcription was precisely the same as the one that had failed in blocking m164 transcription. Notably, m06 behaved like m164 in this respect. Expression was detected at 2 h but was only partially sensitive to inhibition of protein synthesis. It may be informative to note that mCMV open reading frame DNA microarray analysis did not identify m06 as an IE gene (T. E. Shenk, personal communication).

FIG. 3.

Temporal transcription patterns of immune evasion genes. RT-PCRs specific for genes m04, m06, and m152 resulting in cDNA fragments of 324, 180, and 506 bp, respectively, were performed with purified poly(A)⁺ RNA isolated from MEF at the indicated times after infection in the absence or presence of metabolic inhibitors. ActD, 5 μg/ml; CH, 100 μg/ml. Southern blot autoradiographs obtained after hybridization with the corresponding γ-³²P-end-labeled oligonucleotide probes are shown.

Processing and presentation of the pORFm164-derived peptide.

The previously described CTLL specific for the IE1 (¹⁶⁸YPHFMPTNL¹⁷⁶) and the m164 (²⁵⁷AGPPRYSRI²⁶⁵) peptides, referred to as lines IE1-CTLL and m164-CTLL, respectively (25), were used to measure antigen processing and presentation in MEF during the IE, E, and L phases of viral gene expression (Fig. 4). Detection of an antigenic peptide in HPLC-fractionated acidic extracts from infected cells (13) shows that it has been generated at the proteasome (for reviews, see references 36 and 68), while lysis of infected cells proves that the MHC-peptide complex is presented at the cell membrane. In accordance with previous work (9, 10, 49), the IE1 peptide was here found to be processed and presented in the IE phase. In contrast, the basal transcription of m164 observed during the IE phase (see above) did not lead to an efficient generation of the m164 peptide. Accordingly, m164-CTLL did not detect the peptide during the IE phase. During the E and L phases, the IE1 peptide was efficiently generated by processing of the IE1 protein, but it was insufficiently presented. This finding reflects the E-phase immune evasion originally described by the group of U. H. Koszinowski (10, 50). Subsequent studies had identified the m152 gene product gp37/40 as the molecule responsible for retention of IE1 peptide-loaded MHC class I complexes in the ER and ER-intermediate Golgi compartment (9, 64, 69, 70), which explains the lack of presentation of correctly processed IE1 peptide. Lack of IE1 peptide presentation in our specific experiment thus proved that gp37/40 was functional. Importantly, peptide m164 was found to be efficiently processed as well as presented in the very same cells in which the presentation of IE1 was prevented. The message is twofold. First, immunosubversive mechanisms that effectively prevent presentation of the L^d-restricted IE1 peptide simultaneously fail to prevent presentation of the D^d-restricted m164 peptide. Second, m164 is now identified as a phenotypic E gene for which bulk expression occurs in the E phase, even though some transcripts can already be detected in the IE phase. The situation is reminiscent of the γ-1 genes, which are transcribed in the E phase, whereas the bulk of protein synthesis occurs in the L phase.

FIG. 4.

Temporal pattern of antigen processing and presentation. For the processing graphs, naturally processed peptides were acid extracted from infected MEF in the IE, E, and L phases and were separated by reversed-phase HPLC. The indicated aliquots of the HPLC fractions were used to pulse P815 target cells. The cytolysis assay was performed with IE1-CTLL and m164-CTLL as effector cells at an E/T ratio of 15:1. For the presentation graphs, recognition of infected MEF in the IE, E, and L phases was tested in the cytolysis assay with IE1-CTLL and m164-CTLL at the graded E/T ratios indicated.

Comparative quantitation of processed IE1 and m164 peptides.

How does the m164 peptide manage to become presented? We considered it unlikely that a particular biochemical property of the m164 peptide-D^d complex could confer resistance to all three immune evasion mechanisms specified by genes m04, m06, and m152. Rather, an abundance of peptide-loaded D^d molecules could saturate the inhibitory capacity of the immunosubversive proteins so that a number of molecules sufficient for recognition by CTL can reach the cell surface. Since only peptides that are bound to MHC molecules are exempt from complete proteolytic degradation (9, 14), the amount of peptide in the extracts reflects the number of peptide-loaded MHC molecules. The hypothesis of “inhibitor saturation” raised above predicted a molar excess of m164 peptide over IE1 peptide in the E extract of the experiment shown in Fig. 4. From the dilutions of the HPLC peak fractions 28 and 22 containing naturally processed IE1 and m164 peptides, respectively, a difference was not obvious (Fig. 4). However, one has to take into account the sensitivities of the two CTLL used in this particular assay. Figure 5 shows a comparative quantitative analysis for the two peptides. Line IE1-CTLL exerted half-maximal lysis of target cells loaded with synthetic IE1 peptide at a concentration of ca. 8 × 10⁻¹² M, and half-maximal lysis of target cells loaded with naturally processed IE1 peptide at a dilution of peak fraction 28 of ca. eightfold (Fig. 5A). Thus, the concentration of IE1 peptide in the undiluted fraction 28 was ca. 6.4 × 10⁻¹¹ M. A calculation performed likewise for the m164 peptide gave a peptide concentration of ca. 1.4 × 10⁻⁹ M in the undiluted fraction 22 (Fig. 5B). A precise calculation should include the amount of peptide contained in the neighboring positive fractions as well as the extraction efficacies, which are close to 100% for the m164 peptide and ca. 40% for the IE1 peptide (not shown). If we think in orders of magnitude, we can conclude that there was a ca. 10-fold excess of m164 peptide over IE1 peptide in the E extract used for the experiment shown in Fig. 4.

FIG. 5.

Relative quantitation of IE1 and m164 processing in the E phase. (A) Quantitation of the IE1 peptide. (B) Quantitation of the m164 peptide. The sensitivities of lines IE1-CTLL and m164-CTLL were determined by a cytolysis assay performed with P815 target cells pulsed with the indicated molar concentrations of the cognate synthetic peptides (left panels). The E/T ratio was 15:1. The very same CTL were used as effector cells in a cytolysis assay performed with P815 target cells that were pulsed with the indicated dilutions of HPLC peak fractions 28 and 22 (data are from Fig. 4) containing naturally processed peptides IE1 and m164, respectively (right panels). The E/T ratio was 15:1. Dotted lines, peptide concentration or dilution of the HPLC peak fraction required for half-maximal lysis (background subtracted) of the target cells; dashed lines, regression lines calculated by the least-squares method; dots, data included in the calculation of the regression lines; circles, data for upper and lower plateau lysis not included in the calculation of the regression lines. The molar concentration of the respective peptide in the undiluted HPLC fraction is calculated from the abscissa values of the intersection points between half-maximal lysis and the corresponding regression lines by multiplying the molar concentration of synthetic peptide with the dilution of the HPLC fraction.

DISCUSSION

Immune evasion mechanisms of CMVs have attracted great interest in recent years (for reviews, see references 18 and 66). Even though it was not always intended by the authors of the work on immune evasion genes, many readers took the message that these mechanisms would explain how CMVs finally avoid recognition by the immune system. This interpretation of the data is clearly in conflict with the undisputed medical experience that CMV disease is a prototypic example of a disease that affects primarily the immunocompromised host (reviewed in reference 7). Early findings in the murine model have identified antiviral CD8 T cells as the principal antiviral effector cells in the control of in vivo CMV replication (51, 56, 62). In experimental models of mCMV infection after syngeneic or MHC class I disparate BMT, endogenously reconstituted CD8 T cells were found to be recruited to infected lungs (2, 22), where they confined the infection to focal inflammatory infiltrates (44). The protective function was proven by two complementary approaches as follows: (i) prevention of CD8-T-cell infiltration by selective in vivo depletion of the CD8 subset resulted in uncontrolled mCMV replication and disseminated interstitial pneumonia (44), and (ii) CD8 T cells isolated from the focal pulmonary infiltrates prevented mCMV disease in infected indicator recipients upon adoptive cell transfer (2, 44). In accordance with the murine model, control of hCMV disease after clinical BMT correlated with the reconstitution of CD8 T cells (58), and preemptive cytoimmunotherapy with CD8 T-cell lines reduced the incidence of hCMV disease in BMT patients (60, 65). Apparently, there is a conflict between the protective in vivo function of CD8 T cells and the immune evasion mechanisms that were supposed to prevent antigen presentation in the MHC class I pathway (for a critical discussion, see reference 47).

A first idea for solving this conflict was based on the finding that the then-known immunodominant antigenic peptides of CMVs were derived from proteins that are likely to be unaffected by immune evasion proteins. Specifically, for mCMV, the immunodominant IE1 protein pp89 (11, 22, 34, 54) is expressed prior to the E-phase immune evasion proteins, and for hCMV, the immunodominant UL83 protein pp65 (15, 39, 67) is a virion tegument protein that is processed and presented after virus entry before and even in the absence of viral gene expression (59). However, this explanation cannot apply to recently identified antigenic peptides of mCMV that are specified by E genes m04 (24), m18 (21), M45 (presented by D^b) (M. C. Gold, M. Munks, M. Wagner, U. H. Koszinowski, A. B. Hill, and S. P. Fling, submitted for publication; A. B. Hill, personal communication), M83 (23), and M84 (26). The corresponding proteins must have successfully primed an immune response during infection of immunocompetent mice, since it would otherwise not have been possible to select CTLL from memory T cells derived from the spleen.

How can these E-phase peptides prime an immune response? One answer was suggested by Hengel et al., who showed that the IE1 peptide, which is presented in fibroblasts during the IE phase but not in the E phase, is also presented in macrophages during the E phase (20). Thus, immune evasion mechanisms appear to operate with different efficacies in different cell types. Antigen-presenting cells, such as infected macrophages (16, 17, 63) and infected dendritic cells (3), could present peptides during the E phase of viral gene expression to prime the immune response. In fact, the E-phase peptides listed above were not presented by infected fibroblasts during the E phase (R. Holtappels, unpublished data; A. B. Hill, personal communication). For most of these E peptides it remains to be tested whether they are presented, like the IE1 peptide, during the E phase in infected macrophages. So far, this experiment has been done only for the D^b-restricted M45 peptide, with the result that it is not the case (Gold et al., submitted; A. B. Hill, personal communication). A currently popular idea is the initiation of an antiviral immune response by so-called cross-presentation or cross-priming (5, 61), which means that uninfected antigen-presenting cells take up, process, and present viral antigens derived from infected cells (4; Gold et al., submitted; A. B. Hill, personal communication). This mechanism may “fool” the immune system by priming of a useless response, since the respective CD8 T cells will not recognize the antigens during the E phase on infected fibrocytes and other target cells of the in vivo infection, such as endothelial cells and various types of epithelial cells, including hepatocytes and pneumocytes.

However, this hypothesis is inconsistent with the in vivo antiviral protection elicited by E-phase antigens. CTLL specific for peptides m04, M83, and M84 (m18-CTLL and M45-CTLL have not been tested) were found to control mCMV replication in various host tissues upon preemptive cytoimmunotherapy of CMV disease in immunodeficient recipients (23, 24). In accordance with this finding, genetic immunization with gene M84 and gene m04 expression plasmids protected recipients from mCMV (42; D. H. Spector, personal communication). Taken together, this is firm evidence for the in vivo presentation of E peptides by infected stromal and parenchymal tissue cells. There is also a possible explanation for this. Hengel et al. have shown that pretreatment of fibroblasts with gamma interferon (IFN-γ) can overcome the inhibition of IE1 peptide presentation in the E phase (19). Whether this also applies to E protein-derived peptides needs to be tested. Since CMV infection in vivo is associated with a strong inflammatory response (22, 44), the pro-inflammatory cytokine IFN-γ could indeed precondition the tissue cells for the presentation of antigenic peptides in the E phase.

Yet even this explanation did not provide the complete answer. When CD8 T cells were isolated from pulmonary infiltrates during focal mCMV pneumonia, they exerted ex vivo CTL activity (not requiring in vitro cultivation) against infected fibroblasts, which (it is important to note) had not been pretreated with IFN-γ. Interestingly, infected fibroblasts were most susceptible to lysis by polyclonal MHC-restricted lung infiltrate CTL during the E phase (22). This was a clear indication of the presentation of an antigenic peptide(s) during the E phase. Actually, the finding that fibroblasts are recognized by CTL during the E and L phases of mCMV gene expression is not at all new. Early findings, long fallen into oblivion, had documented the recognition of fibroblast targets of H-2^d and H-2^b haplotypes during the E and L phases of mCMV replication by polyclonal mCMV-specific CTL (52, 53). Somewhat later, Del Val et al. reported the existence of two CTL clones that recognized H-2^d fibroblasts in the E phase (10). For one of these clones, an antigen presented by L^d was genetically mapped to the EcoRI fragment F located within HindIII fragment E. This is noteworthy, as gene m164 is located in the same region. Unfortunately, despite extensive screening of combinatorial peptide libraries for EcoRI fragment F by the group of U. H. Koszinowski in 1989 and 1990 (unpublished data), the corresponding peptide remained unidentified.

The present report shows that the recently identified, D^d-restricted m164 peptide (25) behaves like the unidentified E peptide. We have thus found the first antigenic peptide of mCMV that is processed and presented in fibroblasts during the E and L phases, with no need for preconditioning of the cells with IFN-γ. Relevantly, this presentation occurred in the very same cells in which presentation of the IE1 peptide was prevented. This proves that immunosubversive mechanisms were operative in these cells, and it proves that the m164 peptide escaped the concerted action of all immune evasion proteins. This finally explains why polyclonal CTL can recognize target cells in the E and L phases.

Kavanagh et al. (30) have recently provided evidence for MHC allele-specific function of immune evasion proteins. Specifically, expression of m152-gp37/40 sufficed for preventing D^b-restricted recognition, whereas assistance by m04-gp34 was needed for evading K^b-restricted recognition. Since the IE1 peptide, which is a target of immune evasion proteins in the E phase, is presented by L^d, whereas the evasion-resistant m164 peptide is presented by D^d, one might come up with the idea that our findings reflect an MHC allele-specific difference such that the D^d molecule is not targeted by immune evasion proteins. This possibility can easily be refuted. First, biochemical data have clearly shown an m152-gp37/40-mediated retention of D^d molecules in an endo-β-N-acetylglucosaminidase H-sensitive pre-Golgi compartment. This retention was found to be as efficient as for L^d molecules (9). Second, D^d-restricted peptide m04 is not presented in the E and L phases (24). Third, even though we do not know Del Val's peptide (see above) in its amino acid sequence, it is presented in the E and L phases by L^d (10).

A possible explanation for the presentation of the m164 peptide is provided by the relative quantitation of peptides during the E phase (Fig. 5). The 10-fold excess of m164 peptide over the IE1 peptide in the E phase opens the possibility that the m164 peptide-D^d complexes saturate the inhibitory capacity of the immunosubversive proteins, while the less abundant IE1 peptide-L^d complexes do not. Many parameters determine the amount of antigenic peptide that can be extracted from infected cells. These include the amount and kinetic stability of the antigenic protein, the efficacy of protein cleavage and peptide liberation at the proteasome, the efficacy of peptide transport into the ER and of N-terminal trimming, as well as the MHC binding affinity. Which one applies to the m164 peptide awaits further analysis. Most probably, however, a combination of reasons will explain the abundance of m164 peptide-D^d complexes. While the data are in accordance with the hypothesis of inhibitor saturation, it should be emphasized that the abundance of the m164 peptide is only one explanation that appears to make sense. There may well exist other reasons for the privileged presentation of the m164 peptide during the E phase.

Conclusion.

This work has solved a long-standing conflict in mCMV immunology. Presentation of the m164 peptide during the E and L phases of the viral replication cycle explains why fibroblast target cells are recognized and lysed by CTL despite the expression of immunosubversive viral proteins.