The Mouse Cytomegalovirus G Protein-Coupled Receptor Homolog, M33, Coordinates Key Features of In Vivo Infection via Distinct Components of Its Signaling Repertoire

ABSTRACT

Common to all cytomegalovirus (CMV) genomes analyzed to date is the presence of G protein-coupled receptors (GPCR). Animal models of CMV provide insights into their role in viral fitness. The mouse cytomegalovirus (MCMV) GPCR, M33, facilitates dendritic cell (DC)-dependent viremia, the extravasation of blood-borne infected DCs to the salivary gland, and the frequency of reactivation events from latently infected tissue explants. Constitutive G protein-coupled M33 signaling is required for these phenotypes, although the contribution of distinct biochemical pathways activated by M33 is unknown. M33 engages G_q/11 to constitutively activate phospholipase C β (PLCβ) and downstream cyclic AMP response-element binding protein (CREB) in vitro. Identification of a MCMV M33 mutant (M33_ΔC38) for which CREB signaling was disabled but PLCβ activation was preserved provided the opportunity to investigate their relevance in vivo. Following intranasal infection with MCMV M33_ΔC38, the absence of M33 CREB G_q/11-dependent signaling correlated with reduced mobilization of lytically-infected DCs to the draining lymph node high endothelial venules (HEVs) and reduced viremia compared with wild type MCMV. In contrast, M33_ΔC38-infected DCs within the vascular compartment extravasated to the salivary glands via a pertussis toxin-sensitive, G_i/o-dependent, and CREB-independent mechanism. In the context of MCMV latency, spleen explants from M33_ΔC38-infected mice were markedly attenuated for reactivation. Taken together, these data demonstrate that key features of the MCMV life cycle are coordinated in diverse tissues by distinct pathways of the M33 signaling repertoire.

IMPORTANCE G protein-coupled receptors (GPCRs) act as cell surface molecular “switches” that regulate the cellular response to environmental stimuli. All cytomegalovirus (CMV) genomes analyzed to date possess GPCR homologs with phylogenetic evidence for independent gene capture events, signifying important in vivo roles. The mouse CMV (MCMV) GPCR homolog, designated M33, is important for cell-associated virus spread and the establishment and/or reactivation of latent MCMV infection. The signaling repertoire of M33 is distinct from cellular GPCRs and little is known of the relevance of component signaling pathways for in vivo M33 function. In this report, we showed that temporal and tissue-specific M33 signaling was required to facilitate in vivo infection. Understanding the relevance of the viral GPCR signaling profiles for in vivo function will provide opportunities for future targeted interventions.

INTRODUCTION

A shared feature of beta- and gamma-herpesviruses is the capture of G protein-coupled receptor (GPCR) homologs, signifying central importance for GCPR signaling in the lifecycles of these highly cell-associated viruses (1, 2). Much of our understanding of viral GPCR signaling has come from reporter assays of transiently transfected cells. Several viral GPCR mimics have properties distinct from their cellular counterparts, such as constitutive signaling pathways and differential G protein usage in different cellular contexts (3–6). Functional consequences have included proinflammatory, proliferative, and onco-modulatory signals in vitro that may contribute to various pathologies in vivo (7–10). Viral GPCR signaling is also thought to play a direct role in the virus life cycle (11, 12). While the species-specificity of the beta- and gamma-herpesviruses precludes such analysis of human counterparts, animal models provide the opportunity to gain functional in vivo insights.

The mouse cytomegalovirus (MCMV) GPCR homolog encoded by M33 is conserved among beta-herpesviruses and shares sequence homology with cellular CC chemokine receptors (CKRs) (13). In vitro reporter assays have shown that, unlike most cellular chemokine receptors that engage the pertussis toxin (PTX)-sensitive G_i/o substrate, M33 signals via the G_q/11 pathway activates PLCβ at the membrane as well as distal protein kinase C (PKC)-dependent activation of CREB (14–16). Reports have implicated M33 in promoting pathologies with a chemokine component, including vascular inflammatory disease (5) and chronic transplant rejection (17, 18).

M33 seems to have roles in both the persistence of lytic infection in the salivary gland and reactivation from latent infection (13, 19). Our previous studies demonstrated that MCMV usurps normal dendritic cell (DC) trafficking to facilitate its systemic spread (20). DCs infected via the lung or olfactory mucosa get to the blood by escaping the lymph nodes through high endothelial venules (HEV). Lymph node (LN) escape to the blood by infected DCs was markedly diminished in mice infected with MCMV lacking M33 (ΔM33) or with G protein coupling disabled by an R131Q mutation in the “DRY box motif” of transmembrane region III (M33_NQY) (20). The loss of infected DC spread into the blood exhibited by ΔM33 and M33_NQY viruses was rescued by substituting with the HCMV GPCR homolog US28, which possesses a similar in vitro signaling profile to M33, including engagement of G_q/11 (21, 22). These data suggested that G_q/11-coupled signaling plays an essential role in driving the exit of infected DCs from the LN.

More direct delivery of MCMV to the blood via intraperitoneal (IP) virus inoculation or adoptive transfer of ex vivo infected PEC revealed a second function of M33 – facilitating infected DC extravasation into the salivary glands (23). M33 G protein-dependent signaling again proved essential, but, here, US28 could not compensate and PTX was inhibitory, suggesting that a different aspect of GPCR signaling was needed, specifically G_i/o engagement.

In addition to the lytic spread, M33 was implicated in MCMV reactivation from latency as evaluated by virus recovery from tissue explants of long-term infected mice (24). In this context, HCMV US28 is also substituted for the M33 phenotype (25). Taken together, the data suggested that M33 contributes to host colonization by at least two different GPCR signaling pathways. To define the roles of these pathways more clearly, we analyzed a 38 amino acid carboxy-tail (C-tail) truncation of M33 (designated M33_ΔC38) for which CREB-dependent activation is abolished but phospholipase C β (PLCβ) signaling is retained. The bifurcation of CREB and PLC-β activities exhibited by M33_ΔC38 provided the opportunity to compare the importance of these pathways for M33-dependent phenotypes during acute infection and reactivation from latency.

RESULTS

Separation of membrane-proximal and distal M33 signaling via C-tail truncation.

GPCR signaling is regulated by C-tail modifications, including phosphorylation on serine and threonine residues. M33 induces an increase in phosphatidylinositol turnover (as a measurement of PLCβ activation) and activation of the CREB transcription factor in a gene dose-dependent manner. To understand better how M33 signaling correlates with in vivo viral phenotypes, we compared the ability of C-tail truncated M33 mutants to signal via PLCβ and CREB. The wild type (WT) M33 C-tail and the truncated derivates tested are depicted in Fig. 1A Our previous studies showed that M33_ΔC38 preserves cell surface expression and PLCβ activation but is disabled for CREB signaling, whereas M33_ΔC24 retains it (16) Here, we also analyzed an intermediate 29 amino acid truncation (M33_ΔC29) that preserves the grouped serine and threonine residues lost by M33_ΔC38. The effect of M33 C-tail truncation on CREB-mediated transcription in transfected HEK293 cells, relative to the signaling null M33_NQY mutant, is shown in Fig. 1B. These results confirmed that WT M33 constitutively activated CREB and that this was preserved for M33_ΔC29 but not for M33_ΔC38 (16). Thus, the M33 C-tail region 30 to 38 residues from the C terminus was needed for CREB activation. All the truncation mutants showed detectable PLCβ activation (Fig. 1C). The inositol turnover triggered by M33_ΔC38 was reduced by about 20% compared with the WT and the other C-tail truncations but was still readily detected. Thus, M33_ΔC38 could be used to explore the different roles of M33 CREB and PLCβ activation in MCMV infection.

FIG 1.

Effect of C-tail truncation on M33 CREB activation and inositol turnover. (A) The amino acid sequence of the M33 C-tail is shown with an indication of 24, 29, and 38 amino acid truncations. The location of putative phosphorylation sites is underlined. (B) Constitutive activation of CREB mediated transcription in HEK-293 cells transiently transfected with either the WT M33 or C-tail truncated derivatives at the indicated amounts. Transfection with the M33 signaling null M33_NQY was included as a control and was used as the baseline against which signaling of other constructs was compared. Results are expressed as relative light units (RLU). (C) Constitutive activation of PLCβ in HEK-293 cells transiently transfected with the WT M33 or C-tail truncation mutants at the indicated amounts. Measurement of (myo-3H) inositol phosphorylation is expressed as relative cpm (Rcpm) compared with M33_NQY. For each graph, the mean and standard error are shown (n = 8 to 12).

M33 was not required for infection of myeloid cells.

Previous studies showed that constitutive M33 signaling is not required to infect either mouse fibroblasts or primary bone marrow-derived macrophages (BMM) (13, 26). Because DCs are key to systemic spread in vivo (20), we tested infection of the Mutu DC line, which preserves some functional phenotypes of conventional DCs (27). We compared the growth of EGFP-tagged WT, M33_NQY, and M33_ΔC38 MCMVs using BMM infection as a comparison. Virus growth was quantified over 4 to 5 days pi with a viral input titrated to achieve 20% EGFP⁺ cells at 24 h pi. Mutu DCs proved to be highly permissive for MCMV infection, with output titers exceeding 1000× input (Fig. 2A). In contrast, primary BMM were relatively refractile to infection, needing a 20-fold higher input to achieve 20% infection and producing less new virus (Fig. 2B). Infectivity and productivity for both cell types were equivalent between the WT M33 and the signaling-compromised M33 mutants. Thus, M33 signaling had no detectable effect on MCMV replication in vitro.

FIG 2.

M33 is not required for infection of myeloid cells. Mutu DC (A) or bone marrow-derived macrophages (BMM; B) were infected with either EGFP-tagged WT MCMV, M33_NQY, or M33_ΔC38. Mutu DCs were infected at an MOI of 0.5 and BMMs were infected at an MOI of 5, resulting in 20% infection of both cell types at 24 h pi as assessed by FACS. The unadsorbed virus was removed by washing 4× in a culture medium. Cultures were stored at −80°C at designated intervals up to 96 to 120 h pi and virus titers were subsequently determined by plaque assay on MEF. Data are expressed as the mean ± standard error of the mean (SEM) of 3 independent samples. The dashed line represents the limit of detection of the plaque assay.

M33_ΔC38 was unable to mobilize MCMV-infected DCs from the LN into the blood.

Next, we tested the capacity of M33_ΔC38 MCMV (PLCβ⁺ CREB⁻) to mobilize DCs in vivo compared to the WT or the signaling of the null M33_NQY mutant. Following intranasal virus inoculation (10⁶ plaque-forming units [PFU]), we quantified lung and salivary gland infectivity at day 5 pi, which corresponded to the peak WT MCMV amplification in the lung, detectable viremia, and the initial spread of DCs to the salivary glands. We measured spread to the blood by qPCR. Infectivity titers in the lungs were equivalent between all the viruses with M33_ΔC38 being somewhat higher, but it was not significant (Fig. 3A). However, M33_NQY and M33_ΔC38 infections showed no detectable viremia (Fig. 3B) or spread to the salivary glands (Fig. 3C).

FIG 3.

M33_ΔC38 was unable to mobilize MCMV infected DCs from the LN to the blood. BALB/c mice were infected intranasally with either WT MCMV, M33_NQY, or M33_ΔC38 (10⁶ PFU; n = 4/group). Five days later, virus loads were quantified in the lungs (A), blood (B), and salivary glands (SG; C). Systemic spread of MCMV M33_NQY and M33_ΔC38 was not detected. Mediastinal lymph nodes (MLN) from BALB/c mice infected 3 days previously with EGFP-tagged, WT, M33_NQY, and M33_ΔC38 MCMVs (10⁶ PFU; n = 4/group) were stained for markers of HEV (CD31 and peripheral node addressin [PNAd]) and scanned for the relative position of EGFP⁺ cells (D). Open arrows show EGFP⁺ cells distal from HEVs, and filled arrows show EGFP⁺ within or abutting HEV. The white bar denotes 50 μM. The relative proportion of EGFP⁺ cells associated or not with HEV (HEV⁺ or HEV⁻, respectively) was quantified (E). WT MCMV showed significantly more EGFP⁺ cells associated with HEV. BALB/c mice were administered intranasally with a high virus dose of either WT MCMV M33_NQY or M33_ΔC38 (6 × 10⁶ PFU; n = 3/group) and lungs (F), MLN (G), blood (H), and SG were harvested on days 5, 8 and 12 pi. M33_NQY and M33_ΔC38 MCMVs viremias were significantly reduced compared with the WT MCMV, but only M33_NQY MCMV failed to colonize the SG. Symbols show individual samples, and bars denote means. The dashed line indicates the detection limit of the plaque assay. NS, not significant; **, P < 0.01; ***, P < 0.001.

To identify where the spread of the infection was blocked, we examined lung-draining mediastinal LN (MLN) sections of mice infected with EGFP-tagged WT MCMV, M33_NQY, or M33_ΔC38. After lung infection, essentially all infected cells in the MLN were DCs (20). With WT MCMV, we observed infected (EGFP⁺) cells mostly adjacent to or within CD31⁺ peripheral node addressin (PNAd)⁺ HEV (Fig. 3D; filled arrows). In contrast, few M33⁺_NQY or M33⁺_ΔC38 cells were associated with HEV (Fig. 3D; open arrows). Quantifying this, scanning 4 sections from 4 independent MLN for each virus group, showed a significantly higher ratio of HEV-associated with infected cells with the WT compared to either mutant (black or gray bars in Fig. 3E, respectively; P < 0.001). Thus, the primary infection defect of M33_ΔC38 MCMV was in the exit of DCs from the LN via HEV.

To determine whether the block in LN escape by M33⁺_NQY or M33⁺_ΔC38 DCs could be overcome by higher viral loads, we intranasally infected animals with an inoculum of 6 × 10⁶ PFU and measured virus titers over a longer time course. We then harvested the lungs, mediastinal lymph nodes (MLNs), blood, and salivary glands at days 5, 8, and 12 after infection. With this higher inoculation dose, day 5 lung titers were increased by 3 to 4 orders of magnitude (comparing Fig. 3F with Fig. 3A). The lung titers then diminished 10-fold every 3 to 4 days. All viruses exhibited equivalent titers at each time point (Fig. 3F). Tracing infection from the lung, viral titers in the MLN were low and equivalent (Fig. 3G). We detected MCMV DNA in the blood at day 5 for all viruses. However, M33_NQY levels were markedly reduced compared with the WT, and, while M33_ΔC38 levels were higher than M33_NQY, they were still significantly lower than the WT MCMV (Fig. 3H). Spread to the salivary glands was detected for the WT MCMV and M33_ΔC38 at day 8, and infection was further amplified for both by day 12 pi. However, M33_ΔC38 levels remained significantly lower (Fig. 3I). We did not detect salivary gland colonization by M33_NQY (16, 26).

M33_ΔC38 rescued infected DC extravasation into the salivary glands.

WT MCMV colonizes the salivary glands via CD11c⁺ DCs, and the MCMV CC chemokine, MCK-2, facilitates subsequent infection of aquaporin V⁺ acinar epithelial cells (28). Because our previous studies using a US28-substituted ΔM33 CMV failed to colonize the salivary glands despite normal viremia, our observation that M33_ΔC38 colonized the salivary glands by increasing viremia suggested that the extravasation function of M33_ΔC38 was still intact (20). To test this further, we sought to compare WT and M33_ΔC38 MCMV salivary gland colonization from an equivalent level of viremia. As a first approach, we infected mice IP with EGFP-tagged WT MCMV, M33_ΔC38, or M33_NQY.

We measured infectivity in the spleen, blood, and salivary glands after 5 days to match the peak of acute infection. Spleens and blood showed equivalent infections between the various viruses (Fig. 4A and B). While MCMV M33_NQY failed to colonize the salivary glands, the WT MCMV and M33_ΔC38 showed equivalent titers (Fig. 4C). MCMV also prominently infects brown adipose tissue (BAT) via an M33-independent migration of infected macrophages (22). Immunohistochemistry showed equivalent EGFP⁺ cell numbers in BAT for all three viruses (black histograms; Fig. 4D). In contrast, only WT and M33_ΔC38 MCMV showed DC-dependent colonization of the salivary glands (gray histograms; Fig. 4D). EGFP⁺ cell numbers in the salivary glands of M33_ΔC38-infected mice were approximately 60% of the WT (P < 0.0003), suggesting a reduced efficiency of extravasation.

FIG 4.

M33_ΔC38 rescues extravasation of infected DCs to the salivary glands. BALB/c mice were infected IP with WT MCMV, M33_NQY, or M33_ΔC38 (10⁶ PFU, n = 4/group). Five days later, virus loads were quantified in spleens (A), blood (B), and salivary glands (SG; C). Spleen and blood titers were equivalent. Only WT MCMV and M33_ΔC38 were detected in the salivary glands. Brown adipose tissue (BAT) and SG harvested from BALB/c mice infected IP 5 days previously with EGFP-tagged WT MCMV, M33_NQY, or M33_ΔC38 (10⁶ PFU, n = 4/group) were assessed histologically for the percentage of sections positive for the presence of EGFP⁺ cells in brown adipose tissue (BAT; black histograms) and the salivary glands (SG; gray histograms, n = 4/GP) (D). The number of EGFP⁺ sections per number of sections counted is shown above each histogram. Significant differences between the rate of detection in BAT versus SG for each virus are shown. ****, P < 0.0001; ***, P < 0.001. Immunohistochemistry of salivary glands harvested from mice infected with either EGFP-tagged WT MCMV or M33_ΔC38 at 5 or 12 days pi (E). At 5 days pi, EGFP⁺ cells colocalized with CD11c. By day 12 pi, colocalization with aquaporin V was detected (filled arrows). The white bar denotes 25 μM. Salivary gland titers tracked over 60 days pi in BALB/c mice following IP infection (10⁶ PFU) with either WT MCMV, M33_NQY, or M33_ΔC38 (F). Differences in titers between WT and M33_ΔC38 MCMVs at each time point in the salivary glands are indicated. *, P < 0.05. MCMV M33_NQY was not detected at any time point. The dashed line indicates the limit of detection of the plaque assay. The distribution and cell surface phenotype of adoptively transferred peritoneal cells infected ex vivo with either WT MCMV or M33_ΔC38 (G). The proportion of CD11c⁺ cells detected in the salivary glands for both viruses was significantly higher than BAT, lungs, and liver (P values ranged from 0.0432 to 0.0013).

EGFP⁺ cells in the salivary glands at day 5 exhibited punctate CD11c⁺ staining typical of MCMV infected DCs (Fig. 4E, left panel). By day 12 pi, EGFP⁺ salivary gland cells were aquaporin V⁺ for both WT and M33_ΔC38 MCMV, indicating infection transfer to acinar epithelial cells (Fig. 4E, right panel). Thus, while MCMV M33_ΔC38 showed somewhat reduced DC extravasation into the salivary glands compared to WT, it showed no defect in subsequent infection transfer (Fig. 4E).

To test M33_ΔC38 MCMV persistence, we repeated the IP infection and quantified the salivary gland titers over 2 months. M33_NQY infection was not detected at any time. MCMV M33_ΔC38 failed to reach peak WT titers. However, infectivity persisted for longer (Fig. 4F).

As a second approach to the M33_ΔC38 assay from the starting point of an equivalent viremia, we infected peritoneal cells. Peritoneal cells are comprised mainly of macrophages but also have smaller populations of DCs. We flushed the peritoneal cavity of uninfected mice, recovered the cells by centrifugation, and infected them with GFP-tagged WT or M33_ΔC38 MCMV (multiplicity of infection (MOI) = 3). After 24h, the infected cells (>95% infected as assessed by EGFP⁺ fluorescence) were transferred intravenously (IV) into naive mice (10⁷ cells each). At 18 h posttransfer, we culled mice and perfused animals with saline to remove cells from the intravascular compartment that would interfere with the detection of extravasated cells. We harvested the livers, lungs, BAT, and salivary glands. Quantification of EGFP⁺ cells on sections of the liver, BAT, and lung and salivary glands showed 58 to 73% of cells were CD11c⁺ (DC) with most of the remainder being CD11b⁺ (macrophages) (Fig. 4G). In contrast, at least 93% of the EGFP⁺ cells on the salivary gland sections were DCs. WT and M33_ΔC38 MCMV both showed this pattern. Thus, this second, more direct approach to monitoring infected cell extravasation supported the idea that M33 with PLCβ, but not CREB signaling, was impaired for the exit of the LN but was able to drive DC extravasation only slightly less efficiently than the WT.

M33_ΔC38-dependent DC extravasation into the salivary glands depended on G_i/o.

Our previous studies showed that M33-dependent DC extravasation into the salivary glands was sensitive to the action of pertussis toxin (PTX), which is a selective G_i/o. inhibitor (23). We investigated the sensitivity of M33_ΔC38-driven DC migration by treating mice with PTX then infecting IP with WT or M33_ΔC38 MCMV. Spleens, lungs, and salivary glands were harvested 4 days later. PTX treatment augmented virus titers in the spleen and lungs for both WT and M33_ΔC38 MCMVs, perhaps due to its immunosuppressive effects. However, it suppressed SG colonization, demonstrating that, for both viruses, G_i/o-coupled signaling was required (Fig. 5).

FIG 5.

M33_ΔC38-dependent extravasation of DCs to the salivary glands is dependent on G_i/o engagement. BALB/c mice were infected IN with either WT MCMV or M33_ΔC38 (10⁶ PFU) and were treated with PTX (n = 4 per group). After 4 days, virus loads from the spleen (A), lungs (B), or salivary glands (SG; C) were quantified by plaque assay. Symbols show individual tissues, and bars show means. The sensitivity limit of the plaque assay was 50 PFU/organ, shown as a dashed line in (C). PTX treatment significantly increased spleen and lung infection and decreased salivary gland infection. *, P < 0.05; **, P < 0.01.

M33_ΔC38 was disabled for reactivation from latency.

Finally, we investigated the ability of M33_ΔC38 to reactivate from latency. We used spleen explants from mice infected long-term with MCMV to quantify reactivation. BALB/c adult mice were infected IP with 10⁶ PFU of either WT MCMV, M33_NQY MCMV, or M33_ΔC38 MCMV (n = 5 per virus). Three months later, spleen explant cultures were established (48 cultures per virus) and tested twice weekly for reactivation by plaque assay of supernatants for another 8 weeks. Salivary glands were also harvested and an absence of detectable lytic infection for both viruses was confirmed by plaque assay. Of the spleen explant cultures, 58% reactivated for WT MCMV, 24% for M33_NQY, and 10% for M33_ΔC38 (Fig. 6A). Thus, M33_ΔC38 was at least as compromised as the signaling null M33_NQY mutant for reactivation from latency.

FIG 6.

M33_ΔC38 is disabled for reactivation from latency. Cumulative reactivation curves of spleen explant cultures taken from BABL/c mice latently infected with either WT MCMV, M33_NQY, or M33_ΔC38 (10⁵ PFU; IP). Spleens (8 to 12 per virus group) were harvested 12 weeks later, were finely dissected, and 1 mm³ pieces were distributed in a 48-well plate. Supernatants were harvested once per week for 8 weeks and assayed for infectious virus on MEFs using centrifugal enhancement. Comparisons of the proportion of reactivated to nonreactivated cultures of the M33 mutants with the WT MCMV were made by Fisher’s Exact Test.

DISCUSSION

MCMV spreads systemically via infected myeloid cells (20, 29). The salivary glands are distinct from other tissues in their infection being seeded almost exclusively by DCs. We demonstrated previously that MCMV delivered to either the lungs or olfactory mucosa mobilizes infected DCs back to the blood via lymph node HEV then to the salivary glands (20, 23). Such DC trafficking is remarkable in that infected DCs ignored the inflammatory retention signals within the LN and lacked signals in salivary glands.

Given that DC migration is regulated by chemokines (30), it is not surprising that a viral GPCR is essential to disrupt it. We and others have shown that MCMV salivary gland colonization requires constitutive G protein-dependent signaling by M33 (16, 19). Here, the observed separation of PLCβ/CREB activating pathways in HEK cells in vitro guided us to investigate how these components of GPCR signaling might be used by MCMV in vivo.

The 38 amino acid C-tail truncation of M33 did not reduce BMM or Mutu DC infection in vitro or infected DC mobilization to LN in vivo, which is consistent with these functions being M33-independent. It did compromise infected DC mobilization within and their exit from the LN. Even with large viral inoculums, the viremic load of mice infected with the M33_ΔC38 mutant was significantly lower than the WT. High dose IN infection of M33_ΔC38 revealed that extravasation of DCs to the salivary glands was preserved. This result was opposite of what we have observed previously when MCMV M33 was replaced with US28 and, thus, was made dependent on US28 for DC escape from the LN. Despite high levels of viremia, US28-infected DCs could not spread to the salivary glands. The extravasation mechanism may be species-specific. Although M33_ΔC38 MCMV never reached the peak salivary titers of the WT, detectable infectivity lasted for longer. Because MCMV clearance from the salivary gland is CD4 T cell- and interferon-γ dependent (31, 32), a simple explanation may be that lower peak titers recruit less dramatic antiviral responses.

Viral CKRs have demonstrated variations to the classical picture of G α protein and downstream pathway engagement (3). M33 is homologous to CC CKRs. It remains an orphan receptor and unlike most cellular CKRs that signal via G_i/o α subunits and protein kinase A (PKA), M33 signals constitutively via G α_q/11, activating PLCβ and PKC-dependent CREB (14). Typically, PLCβ triggers calcium mobilization and activates PKC. However, PLCβ activation and inositol trisphosphate (IP₃) accumulation by M33 fails to elicit an intracellular calcium flux. Instead, it seems to activate PKC via diacylglycerol (15). G protein-independent mechanisms are also possible, but because they remain functional in the absence of M33 G_q/11 signaling (the M33_NQY mutant) (15), they likely did not contribute to the in vivo DC phenotypes investigated here.

CREB activation may also proceed via GRKs that target intracellular serine and threonine residues of GPCR C-tails and intracellular loops. Normally this disrupts the heterotrimeric G protein complex, and GRK-dependent desensitization is important for leukocyte sensing of chemokine gradients (33) M33 can be phosphorylated by GRK2 in HEK293 cells (14). CREB activation was preserved after deleting 7 of the 11 M33 serine/threonine residues (M33_ΔC29) but not after deleting a further 3 (M33_ΔC38). Thus, those 3 residues may be functionally important GRK targets for DC exit from the LN.

The ability of M33_ΔC38-infected DCs to colonize the salivary glands demonstrates CREB signaling is not required for this role. Moreover, DC colonization was sensitive to PTX, demonstrating G_i/o engagement by M33. While M33-G_i/o interactions have not been detected in transformed cell lines in vitro, in vivo signaling in DCs could be different. It would not be surprising for M33 signaling pathways to depend strongly on cell type and tissue context because this has been observed for other GPCRs (34). An unknown M33 ligand could promote G_i/o engagement, which is similar to the differential G protein usage in HCMV US28 (35). Alternatively, the possibility exists that M33 constitutive G_q/11 signaling may activate G_i/o signaling of other GPCRs or heterodimerize with different host GPCRs to drive signals through different pathways depending on tissue context (36–38).

Taken together, M33-dependent MCMV spread operates in a coordinated and tissue-specific manner that is specific for infected DCs and facilitates efficient routing of MCMV exit to the salivary gland. Pathways from entry to exit for mucosal and IP infection are summarized in Fig. 7. Mucosal entry resulted in the infection and mobilization of DCs to the draining LN and was independent of M33. Within the LN, DC “escape” from the LN HEV (Fig. 7, red arrow) is facilitated by M33 engagement with G_q/11 and is correlated activation of CREB rather than membrane-proximal PLC-β. DCs that enter the blood through IP infection directly or via mucosal infection extravasate to the salivary glands via engagement of G_i/o and PLC-β activation (Fig. 7, blue arrow). Dissemination of DCs other tissues (e.g., BAT, liver, and lungs) is independent of M33 (Fig. 7, dashed arrow). Infected macrophages that arise from IP infection also traffic to distal sites in a M33-independent manner.

FIG 7.

Summary of M33-dependent and M33-independent pathways of MCMV dissemination. Routes of MCMV systemic dissemination following either mucosal or IP infection are depicted. Dashed arrows show dissemination pathways independent of M33. (1) Following MCMV infection at the mucosa, DCs take MCMV to the draining lymph nodes. (2) Within the LN, M33 G_q engagement facilitates DC mobilization to the blood via high endothelial venules (HEV) for which M33 PLC-β activation is redundant (red arrow). (3) IP MCMV infection delivers infected DCs and macrophages (MΦ) to the blood. (4) Blood-borne infected DCs or MΦ colonize tissues independently of M33, including brown adipose tissue (BAT), liver, and lungs. (5) The salivary glands are almost exclusively colonized by infected DCs, for which M33 G_i/o engagement is required (blue arrow).

Our previous studies suggested a role for M33 in either the establishment or the reactivation from latency based on the frequency of reactivation in tissue explants (24, 26). While M33_ΔC38 MCMV being as disabled as M33_NQY for reactivation does not prove a need for CREB, they are consistent with reports of CREB transcription factors binding to the reactivation-associated MIE control region of HCMV (39) and with a report of CREB activation by UL33 promoting HCMV reactivation (40). However, given that reactivation rates correlate with M33 mutations, which preserve systemic spread (24, 26), an equally plausible conclusion is that tissue colonization during acute infection of the naive host is directly proportional to the latent load and hence the reactivation potential of explanted tissues.

MCMV provides a useful model to uncover the importance of chemokines and chemokine receptors in the CMV life cycle because we know that natural infection disseminates via DCs and that transmission to naive hosts depends on DCs taking infection to the salivary glands (41). Data from the studies presented here showed how these key features of the viral life cycle were coordinated within one protein. Given the proven “druggability” of GPCRs coupled with the technology to preferentially modulate specific signaling pathways through biased agonism, the viral GPCRs present viable targets to ablate undesirable consequences of their signaling (42, 43). Understanding the role of intracellular pathways critical for the physiological functions of virus-infected cells is thus important for the design of such targeted interventions.

MATERIALS AND METHODS

Mice.

BALB/c female mice were derived from the Animal Resource Centre (Perth, Australia). They were infected with MCMV at 6 to 8 weeks old either intranasally (IN) or intraperitoneally (IP). Peritoneal cells were harvested from uninfected mice 72 h after IP inoculation using 3% wt/vol thioglycolate solution. All experiments were approved by the University of Queensland Animal Ethics Committee in accordance with the Australian National Health and Medical Research Council guidelines (projects 388/12 and 207/18).

Plasmids.

WT M33 and modified derivatives were generated by PCR-mutagenesis and cloning into the EcoRI site of pcDNA3. C-terminal M33 truncations were generated by amplification of M33 incorporating a stop codon. All constructs were verified by sequencing. The ΔC24, and ΔC38 truncations have been described previously (16). Here, we included a ΔC29 C-tail mutant, which was generated by amplification of M33 incorporating a stop codon (underlined in the forward and reverse oligonucleotide) before final 29 C-terminal amino acids: forward primer: (5′-TTCGAATTCGGGTCGAGATGGACGTC-3′; reverse primer: 5′-GGAATTCGCGGCCGCTCAGGCGCGGTGTGGCGTCGAG-3′). The EcoRI site introduced to facilitate cloning into pcDNA3 is shown in italics, the introduced stop codon is underlined and the start codon of M33 is doubly underlined.

Viruses and cells.

The K181 (Perth) MCMV strain was for these studies and is referred to as WT MCMV. We have described previously a LACZ disruption of the M33 ORF (designated ΔM33) (13). The M33 single point mutant (M33_NQY) that disrupts G protein coupling and the M33 38 amino acid C-tail truncation (M33_ΔC38) were derived from ΔM33LACZ by selecting and purifying “white” plaques from β-galactosidase⁺ “blue” plaques. To facilitate histological detection, an EGFP cassette inserted in the m131/129 intron was used to generate EGFP⁺ derivatives (20). All viruses were propagated on NIH 3T3 fibroblasts (ATCC CRL-1658) and quantified by plaque assay on primary mouse embryonic fibroblasts (MEF) prepared from CD1 mouse embryos at days 13 to 17. HEK 293 cells (ATCC CRL1573) were used for luciferase reporter and inositol turnover assays. The above primary and continuous cells were cultured in Dulbecco’s Modified Eagle’s medium (DMEM), which was supplemented with 2 mM glutamine, 100 IU/mL penicillin, 100 μg/mL streptomycin, 10% heat-inactivated fetal calf serum, and 50 μM 2-mercaptoethanol (designated “supplements”). Bone marrow macrophages (BMM) were recovered from the femurs and tibias of BALB/c adult female mice. They were cultured for 7 days in Rosewell Park Memorial Institute (RPMI) medium with supplements and 20 ng/mL macrophage colony-stimulating factor (m-CSF; Peprotech). The Mutu DC line ((27); a kind gift from Michael Chopin, Walter and Eliza Hall Institute, Melbourne) was cultured in Iscove’s Modified Dulbecco’s Medium (IMDM) with supplements. All cell culture medium was purchased from In Vitro Technologies.

Immunostaining.

Organs were fixed in 1% formaldehyde, 10 mM sodium periodate, 75 mM l-lysine (18 h, 4°C), equilibrated in 30% sucrose (18 h, 4°C), and then frozen in optimal cutting temperature compound (OCT). Sections (6 μm) were blocked with PBS containing 0.1% Triton X-100/5% normal goat serum and then incubated (18 h, 4°C) with primary antibodies to either CD68 (rat MAb, FA-11), CD11b (rat MAb, M1/70; Abcam) peripheral node addressin (PNAd; rat MAb MECA-79; Abcam), CD31 (rat MAb MEC 7.46; Abcam), CD11c (hamster MAb HL-3; BD Pharmingen), or aquaporin V (AQPV; rabbit pAb; Alamone Labs). After incubation with primary antibodies, the sections were washed three times in phosphate-buffered saline (PBS), incubated (1 h, 23°C) with combinations of Alexa Fluor 488-, Alexa Fluor 568-, or Alexa Fluor 647-conjugated goat pAb (Abcam) then washed three times in PBS, stained with 4′,6′-diamidino-2-phenylindole (DAPI), and mounted in ProLong Gold (Life Technologies). Green fluorescent protein (GFP) expression from MCMV-GFP-infected cells was visualized directly. Images were acquired with a Zeiss LSM510 microscope and analyzed with Zen software.

Luciferase reporter assays.

HEK-293 were seeded (5 × 10⁴ cells/well) in 96-well black and white isoplates (PerkinElmer) using DMEM lacking antibiotics but supplemented with 2 mM glutamine and 0.5% heat-inactivated fetal calf serum. One day after seeding, cells were cotransfected with pcDNA3-based M33 receptor constructs (0.625 to 10 ng DNA) and 6 ng of pFA2-CREB (Pathdetect transreporting system) diluted in Opti-MEM containing Lipofectamine 2000 reagent (Invitrogen). The Opti-MEM medium was aspirated 6 h following transfection and replaced with DMEM lacking antibiotics but supplemented with 2 mM glutamine and 0.5% heat-inactivated fetal calf serum. Following another 24 h of incubation, the cells were assayed for luminescence (PerkinElmer Steadylite^TM plus) using an Envision Multimode Plate Reader. The signaling null M33_NQY mutant served as baseline controls.

Phosphatidylinositol turnover assay.

HEK-293 cells were seeded (3.5 × 10⁴ cells/well) in poly-l-lysine-coated 96-well plates and cultured overnight in a growth medium (2 mM glutamine, 100 IU/mL penicillin, 100 μg/mL streptomycin, and 10% heat-inactivated fetal calf serum). The cells were cotransfected the following day with receptor constructs at 2.5, 5, or 10 ng/well, and the promiscuous Gα subunit GαΔ6qi4myr (a gift from Mette Rosenkilde, University of Copenhagen) at 10 ng/well using the serum-free medium Opti-MEM and Lipofectamine 2000 at 10 μl/mL. At 6 h posttransfection the cells were transferred to a normal growth medium. Twenty-four hours later, the cells were labeled with 5 μCi of [myo-³H] inositol per well and incubated overnight at 37°C. The following day, the cells were washed once in assay buffer (20 mM HEPES, 140 mM NaCl, 5 mM KCl, 1 mM MgSO₄, and CaCl₂) and incubated for 90 min at 37°C in assay buffer containing 10 mM LiCl. Subsequently, the cells were incubated on ice in cold formic acid (10 mM) for 30 min. A total of 20 μL of the lysis solution was subsequently transferred to a new white 96-well plate and mixed with 80 μL of freshly diluted scintillation proximity assay YSi beads at 12.5 mg/mL. The plates were vigorously shaken at room temperature for 30 min, centrifuged at 1,500 rpm for 5 min, left overnight at room temperature, and counted the following day using a TopCounter (Packard) after a final round of centrifugation.

Quantitative PCR (qPCR).

MCMV genomic coordinates 111,123-111,347 (per the published Smith Strain (44)) were amplified by PCR (LightCycler 480 SYBR green; Roche Diagnostics) from extracted DNA (Wizard genomic DNA purification kit; Promega) and converted to genome copies by comparison with plasmid standards amplified in parallel. Oligonucleotides specific for MCMV included: (forward [5’CCTAGGGAGCCTCATCCTCT3’] and reverse [5’ATCGAGCGTGAGGTACAGGT3’]). The mouse titin gene (forward [5′AAAACGAGCAGTGACGTGAGC3′] and reverse [5′TTCAGTCATGCTGCTAGCGC3′]) was used to normalize for genomic DNA.

Ex vivo reactivation.

Adult female BALB/c mice were infected IP with 10⁵ PFU of either WT MCMV, M33_NQY, or M33_ΔC38. After 3 months, spleens were harvested, dissected into 1 mm³ pieces, and single pieces were placed in the wells of a 48-well plate containing DMEM with supplements. The medium was changed twice weekly. After 1 week, the medium was replaced with DMEM containing 2% fetal calf serum (FCS). Reactivation of lytic virus from explant cultures was examined by plating supernatant on MEF and scoring for evidence of a cytopathic effect. The time postexplant for evidence of reactivation was scored for each well.

Statistical analysis.

Multiple experimental groups were compared by analysis of variance (ANOVA) with Tukey’s multiple-comparison test. Single comparisons were performed using two-tailed unpaired Student's t test or Fischer’s Exact Test. For comparisons between data for which all values were “zero” for one group, a Mann-Whitney test was used. Differences were considered significant at >95% confidence.