|Home | About | Journals | Submit | Contact Us | Français|
The transmission of herpesviruses depends on viral shedding at mucosal surfaces. The salivary gland represents a major site of persistent viral replication for many viruses, including cytomegalovirus. We established a mouse model of salivary gland dysfunction after acute viral infection and investigated the cellular requirements for the loss of secretion. Murine cytomegalovirus (MCMV) infection severely impaired saliva secretion independently of salivary gland virus levels. Lymphocytes or circulating monocytes/macrophages were not required for secretory dysfunction. Dysfunction occurred before glandular inflammation, suggesting that a soluble mediator initiated the disruption of acinar cell function. Despite genetic differences in innate resistance to MCMV, NK cells protected the host against acinar atrophy and the loss of secretions under conditions of an exceedingly low virus inoculum. NK cells also modulated the type of glandular inflammation after infection, as they prevented an influx of Siglec-F+ polymorphonuclear leukocytes (PMNs). Therefore, beyond their recognized role in controlling MCMV replication, NK cells preserve organ integrity and function and regulate the innate inflammatory response within the gland.
Cytomegalovirus (CMV), a betaherpesvirus, disseminates and establishes a persistent infection of the salivary glands (SG) (40). The specialized secretory machinery of glandular epithelial cells is utilized by the virus for highly efficient virion production and excretion in saliva for long periods of time (13, 17, 27). While CMV infection can cause morbidity and mortality in neonates and immunocompromised individuals, infection of immunocompetent individuals is largely asymptomatic (40). Within the SG, the ability of CMV to both evade immune responses and hijack the secretory pathway without pathological consequences is an elegant example of coevolution between virus and host. Indeed, the ability to secrete saliva must be preserved for both virus transmission and the well-being of the host.
Several human viruses have been associated with sicca symptoms. Sicca syndrome may be associated with the herpesviruses CMV and Epstein-Barr virus (EBV) (11, 29, 45, 55). Diffuse infiltrative lymphocytosis syndrome (DILS), a disorder of HIV-positive patients, is characterized in part by sicca symptoms (15). In addition, one study found that submandibular gland (SMG) secretion of early-stage HIV patients is decreased 50 to 60% compared to that of controls (62). Chronic hepatitis C virus (HCV) infection causes sicca syndrome in a proportion of patients and was recently excluded from Sjögren's syndrome (SS) classification criteria, differentiating it as a separate disease entity (39, 48, 56). The mechanism(s) by which viral infections cause secretory dysfunction is unknown.
The loss of saliva secretion, or xerostomia, is a common disorder shared by a diverse group of patients, including those suffering from SS, primary biliary cirrhosis, side effects of drugs or radiation therapy, and viral infection. Of these, SS has been studied in most detail, yet the cause of the dysfunction is not fully understood. Accumulating evidence from humans and mice suggests that dysfunction may not require glandular destruction (8). For example, SS patients often display only a partial destruction of SG tissue, and some respond to pilocarpine, indicating that the remaining tissue is functional. Anti-muscarinic receptor autoantibodies have been documented to inhibit secretory function in mice (42). Other hypotheses proposed to explain SG dysfunction include changes in water channels, neurological abnormalities, and the activation of innate immunity (33). The causes of secretory dysfunction are likely heterogeneous.
Mouse models of virus-induced dysfunction are needed. NZM2328 (here referred to as NZM) mice spontaneously develop systemic lupus erythematosus-like disease features with a female bias, beginning at 5 months of age (60). The infection of NZM mice with murine CMV (MCMV) induces SS-like disease, characterized by severe focal inflammation of the exocrine glands (35). However, because MCMV led to a loss of secretory function in only some of the infected NZM mice, focal inflammation is unlikely a direct cause of organ dysfunction. Here SG secretions were examined at earlier times following MCMV infection in NZM and non-lupus-prone C57BL/6 (B6) mice to establish pertinent models to investigate virus-induced secretory dysfunction and immune responses.
Natural killer (NK) cells are essential in tumor and viral immunity (57, 63). Upon stimulation, NK cells produce major cytokines and chemokines and display enhanced cytotoxicity. Moreover, NK cells provide critical viral control after MCMV infection (3, 4, 7, 22). There is also evidence that NK cells serve an immunoregulatory role. First, NK cell-derived interleukin-10 (IL-10) was shown previously to temper pathogenic CD8+ T cell responses after MCMV infection of perforin 1-deficient mice, although IL-10 was not detectable in NK cells of wild-type (WT) mice (23). In addition, NK cell-derived IL-10 can dampen IL-12 release by dendritic cells after systemic infection with disseminating pathogens like Toxoplasma gondii (37). Given their vital response to MCMV, we hypothesized that NK cells might protect the host from destructive inflammatory responses targeting the salivary gland. To address this, we examined the effect of MCMV on SG during the first week of infection, including sialadenitis and secretory dysfunction, in lupus- and non-lupus-prone mouse strains. The consequences of viral infection in SG were further investigated with immune cell-deficient mice, which included NK cells, T and B lymphocytes, and other inflammatory leukocytes under study, in a model of virus-induced SG dysfunction. Data from these experiments revealed a novel regulatory role of NK cells, which are needed to limit SG inflammation and tissue damage following viral infection.
M. McDuffie and S. M. Fu generated NZM2328.129S7-Rag1tm1Mom/J (NZM.Rag1) mice in the systemic lupus erythematosus (SLE) Specialized Center of Research (SCOR) at the University of Virginia (UVA) by using a speed congenic approach (59). Rag1 genotyping was performed according to instructions of the Jackson Laboratory. A deficiency in splenic T and B cells was confirmed by staining for CD3 and CD19 followed by flow cytometric analysis. C57BL/6J (B6) and B6.CCR2−/− (B6.129S4-Ccr2tm1Ifc/J) mice were purchased from The Jackson Laboratory. All mice were bred and maintained under specific-pathogen-free conditions at UVA, which is AAALAC accredited, and managed with the Jackson Laboratory Colony Management System (JCMS, version 4.1.2). Mice were infected at 8 to 18 weeks, prior to spontaneous kidney disease in NZM2328 mice. Unless noted otherwise, female mice were used due to sex-related differences in SG secretory function. Animal studies were approved and conducted in accordance with Animal Care and Use Committee oversight.
Salivary gland MCMV (SGV) (Smith strain) was serially passaged in weanling BALB/c mice; viral titers were determined on NIH 3T3 cell monolayers. Tissue culture (TC)-derived MCMV (TC-MCMV) was collected from infected 3T3 monolayer supernatants, and titers were determined on 3T3 and 3T12 cell monolayers. For UV inactivation, MCMV was exposed to short-wave UV light at a distance of 8 cm for 30 min. UV-inactivated MCMV (UV-MCMV) was verified on 3T3 monolayers. Mice were infected intraperitoneally (i.p.) with 105 PFU SGV, 5 × 104 PFU TC-derived MCMV, or low-dose MCMV (100 PFU in NZM mice or 4,000 PFU in B6 mice) in 0.2 ml sterile phosphate-buffered saline (PBS), unless indicated otherwise. The effects of individual SG virus stocks differed slightly, as shown in the figures. All cumulative results consisted of pooled data using a given virus stock. The MCMV burden in tissues was determined by quantitative real-time PCR (qPCR) as described previously (61). All qPCR sample measurements were performed in triplicate. Results are reported as a ratio relating the number of MCMV ie1 copies detected to the number of endogenous cellular β-actin copies.
Salivary secretion was measured after induction by the i.p. injection of pilocarpine (MP Biomedicals) into unanesthetized NZM (0.5 mg/kg of body weight) and B6 (1.0 mg/kg) mice as described previously (24, 35). Saliva volumes are expressed as the total volume produced in 25 min unless indicated otherwise.
The procedure for the isolation of salivary gland leukocytes was modified from a procedure described previously(30). Submandibular gland (SMG) tissues were carefully excised, removed of all periglandular lymph nodes, and placed into 3% newborn calf serum (NCS) in Dulbecco's modified Eagle's medium (DMEM) on ice. Pooled glands (from 2 to 4 mice) were minced and digested with type IV collagenase (0.9 mg/ml; Sigma) at 37° for 15 min with periodic agitation. EDTA was then added to the mixture (0.05 M final concentration) and incubated for an additional 5 min. Digested tissue was pelleted (400 × g for 5 min) and resuspended in cold 3% NCS in DMEM plus 0.005 M EDTA. A single-cell suspension of leukocytes was obtained after straining (70-μm nylon cell strainer) SMG tissues and washing with 3% NCS in PBS prior to plating for flow cytometry.
Monoclonal antibodies (MAbs) 2.4G2 and PK136 were purified from spent-cell-free hybridoma supernatants (Lymphocyte Culture Center, UVA). The following antibodies were used to stain cells prior to flow cytometry: CD45-allophycocyanin (APC)-Alexa Fluor 750, CD11b-Alexa Fluor 647, DX5-fluorescein isothiocyanate (FITC) (eBioscience), Ly6C-FITC, NK1.1-phycoerythrin (PE), CD3e-APC, CD19-APC-Cy7 (BD Biosciences), and anti-mouse NKp46-PE (R&D Systems). Cell surface Fc receptors (FcRs) were blocked with 2.4G2 prior to staining with the indicated MAbs on ice. Dead cells were excluded from analysis by staining with either 7-aminoactinomycin D (7-AAD) (Sigma) or Live/Dead fixable violet dye (Invitrogen). Fixed cells were run on a FACScan or FACSCanto II instrument (BD Biosciences) and analyzed with FlowJo software (version 8.0; Tree Star).
For NK cell depletion experiments, NZM2328 (350 μg) and B6 (200 μg) mice were injected i.p. with MAb PK136 or control PBS 2 days prior to infection. Since splenic NK cell levels are diminished at 48 h after MCMV infection in resistant and susceptible mice (7, 52), and because MAb PK136 treatment further delays splenic NK cell recovery after MCMV infection (10), we monitored NK cell recovery kinetics following MAb PK136 treatment. At day 7 after viral infection, we observed that <5% of splenic DX5+ CD3− cells had recovered in PK136-treated NZM females. However, ~33% of splenic NKp46+ CD3− cells had recovered in PK136-treated NZM males after low-dose MCMV infection. In PK136-treated B6 females, ~13% of NKp46+ CD3− cells had recovered after low-dose MCMV infection. Thus, differences in NK depletion/recovery rates were likely due to sex- and strain-specific effects in addition to the infectious MCMV dose. To deplete monocytes/macrophages, liposomes containing dichloromethylenediphosphonic acid (Cl2MDP) (Sigma) were prepared as described previously (54). NZM mice were injected intravenously (i.v.) with 0.2 ml of the liposome suspension 1 h prior to MCMV infection. At 24 h post-Cl2MDP injection, <10% of CD11b+ Ly6Chi monocytes were observed in peripheral blood mononuclear cells (PBMC) compared to control (PBS-injected) mice.
SMG were fixed with 10% phosphate-buffered formaldehyde. The UVA Research Histology Core stained paraffin-embedded tissue sections (5 μm) with hematoxylin and eosin (H&E). An expert pathologist blindly evaluated inflammation and pathology.
Tissue sections were fixed in acetone-ethanol (1:1 solution) and blocked with avidin and biotin in 3% bovine serum albumin (BSA) (Vector Laboratories). Sections were incubated with anti-Ly6G (0.5 μg/ml) or anti-Siglec-F (0.1 μg/ml) primary antibody followed by biotinylated goat anti-rat IgG (0.5 μg/ml) (BD Biosciences). Excess background staining was blocked by using 3% hydrogen peroxide in 10 mM sodium azide. The reaction was amplified by using a TSA biotin system kit (Perkin-Elmer) and visualized by using streptavidin-Alexa Fluor 594 (Invitrogen). Sections were stained with secondary antibody alone as a negative control (not shown).
Results were analyzed with GraphPad Prism software. The nonparametric Kruskal-Wallis test with Dunn's multiple-comparison test was used unless indicated otherwise. A one-way analysis of variance (ANOVA) with Tukey's multiple-comparison test was used on transformed values when assumptions could be fulfilled. A two-tailed Student t test was used for comparisons of two groups. A P value of <0.05 was considered significant.
Previously, we reported that MCMV causes severe SS-like lesions to develop in the SG of NZM females by day 56 after infection when viral latency is established (35). Our results suggested that virus-induced inflammatory lesions might be associated with SG dysfunction. However, although MCMV infection plainly triggered chronic inflammation, SG dysfunction was not strictly correlated (35; our unpublished data). Thus, other virus-induced factors are responsible for sicca symptoms.
Inasmuch as sicca is a major clinical feature of SS and of SS secondary to viral infection, we examined the kinetics of MCMV-induced SG dysfunction in lupus- and non-lupus-prone animals following viral infection. Although the average amounts of saliva induced after MCMV differ, diminished secretory output was already evident by day 7 after MCMV infection in NZM and non-lupus-prone B6 mice (Fig. 1A). Other non-lupus-prone strains of mice, including NZW mice, displayed similarly impaired functions after MCMV infection (data not shown). However, considerably less infectious virus was required to significantly impair SG function in NZM mice (Fig. 1A). The data imply that SG dysfunction and viral burden generally corresponded in both strain types within days after MCMV infection.
To explore this, we measured MCMV in SG and spleen to determine if viral burden was correlated to secretory dysfunction. As expected, MCMV in SG was readily detected in NZM and B6 mice, corresponding to the infectious dose (Fig. 1B), although MCMV reached slightly higher levels in NZM mice (Fig. 1B). These data demonstrate highly efficient MCMV dissemination to SG in both strains. In contrast, a disproportionate increase in the viral load was observed for spleens of NZM mice (Fig. 1C), consistent with the lack of highly efficient NK-mediated MCMV resistance in NZM2328 mice (43). Altogether, these data demonstrate that SG function is very sensitive to viral infection in lupus-prone and non-lupus-prone strains within days after MCMV infection.
MCMV virus stocks are routinely prepared from SG homogenates, which might contain inflammatory cytokines and SG antigens. Thus, SG dysfunction following MCMV infection might have resulted from the introduction of soluble factors. To address this, B6 mice were treated with UV-inactivated SGV. As shown in Fig. 2A, SG function remained intact in UV-MCMV-infected mice, but a significant drop in amounts of saliva occurred with productive viral infection. Moreover, TC-derived MCMV, which was entirely free of SG factors, significantly impaired SG function when NK cells were depleted (Fig. 2B). Although TC-MCMV actually enhanced day 7 saliva in one experimental cohort, this finding was inconsistent. We conclude that viral infection and replication, not an unknown soluble SGV homogenate factor, were necessarily linked to SG dysfunction.
Because MCMV efficiently seeds and initiates high-level viral replication within SG acinar epithelial cells during the first week of infection (5), we further examined the kinetics of secretory dysfunction to determine whether it parallels an increase in levels of MCMV in SG. Remarkably, NZM mouse SG function was already impaired by day 2 after high-dose infection, despite low or undetectable levels of MCMV in SG tissue (Fig. 3A and B). An equivalent impairment was observed for infected B6 mice, before MCMV was detectable in SG (see Fig. 6C and D). These data indicated that SG dysfunction was independent of the MCMV burden within SG. Nonetheless, systemic virus was readily detectable at day 2 in spleens of MCMV-infected NZM and B6 mice (Fig. 3B and see Fig. 6D), which suggested that impaired SG function corresponded best with early viremia.
To pursue this, we used low-dose viral infection as another strategy to assess the effect of MCMV on gland function. Interestingly, SG secretions were unaffected for 14 days following infection with 100 PFU (Fig. 4A), yet high levels of MCMV were readily detected in SG tissue after low-dose infection (Fig. 1B and and4B).4B). Moreover, although levels of MCMV in SG were comparable after low-dose (day 14) and higher-dose (day 7) infections at the indicated times, tissue function was significantly impaired only by the higher-dose infection when MCMV was detectable in spleen (Fig. 1B and and4B).4B). Altogether, these data demonstrated that SG function correlated with systemic viremia (spleen MCMV load) rather than viral burden in SG tissue.
Activated lymphocytes with a prominent role in early MCMV defense infiltrate into SG within days following viral infection (5; our unpublished data). Because SG dysfunction was most evident following higher-dose MCMV infection, when both innate and adaptive immune systems are activated, tissue-infiltrating lymphocytes might have contributed deleterious effects. To address this question, we measured SG secretions after high-dose infection in mice lacking lymphocytes. As shown in Fig. 5A and B, MCMV-induced SG dysfunction was just as severe in NK cell-depleted, T and B cell-deficient, and control mice, indicating that these cell types were not responsible for the functional impairment. Comparable saliva outputs in control NZM and Rag1 mice (Fig. 5B) further suggested that T and B cells have no impact on the homeostatic capacity of SG to secrete saliva. Moreover, saliva outputs of control NZM and Rag1 mice were similar when all animals under study were broadly compared across different experiments (Fig. 1 and and33 to to6),6), but a controlled study to address T and B cell effects on normal secretion was not carried out. Although the level of viral burden was not elevated in spleens of NK cell-depleted NZM mice (compare Fig. 5C and and1C),1C), significantly higher levels of MCMV in SG (mean, 0.854 versus 0.217; P < 0.01) indicated that NK cells are needed for the control of the virus in SG. The amounts of MCMV were comparable in spleens and SG of NZM and Rag1 mice (compare Fig. 5C with 1B and C). Although T cells have been shown to mediate MCMV resistance in other strains (28), the effect was not seen in “susceptible” NZM mice, perhaps due to the high viral dose given. Taken together, the data demonstrate that lymphocytes were not essential for MCMV-induced SG dysfunction.
To address whether other infiltrating cell types might be the cause of MCMV-induced SG dysfunction, we further examined cellular infiltrates in SMG. After low-dose MCMV infection, a relatively mild mononuclear infiltrate was observed (Fig. 4C), whereas more severe mononuclear and polymorphonuclear leukocyte (PMN) cell infiltrates were associated with higher-dose infection (Fig. 3C). In accordance with these data, SMG acini were largely normal with intact basement membranes at day 7 after low-dose MCMV infection, despite occasional patches of infiltrating leukocytes. Mononuclear infiltrates became even more distinct at longer times (day 14) after low-dose infection. Nonetheless, zones of healthy acini were still prevalent, consistent with a normal functional output. Thus, histological analysis revealed that saliva secretion was not impaired by the presence of a formidable mononuclear inflammatory response in SMG.
To explore the question of whether secretory dysfunction occurs before inflammatory cell infiltration into SMG, further histological analysis was performed on SMG after high-dose infection. At day 2, when SG function was already compromised (~20% of normal levels) (Fig. 3A), SMG acini were healthy, without signs of atrophy or necrosis, and inflammatory cell infiltrates were not observed (Fig. 3C). However, the acini were somewhat enlarged and had a “foamy” appearance, i.e., lighter in color with prominent vacuoles (Fig. 3C). These changes were observed across entire regions of tissue, affecting most of the gland. At day 4, small areas of infiltrating leukocytes were observed near blood vessels surrounding the ducts, and the acini showed signs of atrophy and shrinkage, but the granular convoluted tubules (GCTs) were unchanged (Fig. 3C). At day 7, cellular infiltration was more severe and widespread; it included both mononuclear and PMN cells. Ducts and GCTs were intact, but many acini were atrophic with dense nuclei or necrotic. Thus, the data demonstrate that the onset of functional impairment did not correspond to the appearance of infiltrating cells in SG following MCMV infection.
We next performed flow cytometric analysis of SG CD45+ cells to further characterize immune cells that might contribute to secretory dysfunction at day 2 after high-dose infection. CD11b+ macrophages bearing F4/80 and major histocompatibility complex (MHC) class II represented a major resident leukocyte population of mouse SG (Fig. 6A and B and data not shown). Three major cell types were promptly recruited into SG tissue after infection: neutrophils (CD11bhi Ly6Cint SSChi), lymphocytes (CD11b− Ly6Clo CD45hi), and inflammatory monocytes (CD11b+ Ly6Chi SSClo). Whereas neutrophil infiltration varied in different experiments, and few neutrophils accumulated in SG of B6 mice (Fig. 6A), and because lymphocytes had been previously excluded (Fig. 5), we further examined inflammatory monocytes (CD11b+ Ly6Chi SSClo), which were consistently found in infected SG of NZM and B6 mice on day 2 (Fig. 6A and B). Monocytes emigrating from blood have been implicated in the delivery of MCMV to SG (34), and inflammatory monocytes can cause tissue damage following other viral infections, including influenza virus infection (14). Intriguingly, inflammatory monocytes were present in SG when function was impaired (Fig. 6C and E), suggesting that these cells might cause deleterious effects associated with SG dysfunction.
To address this question, we examined SG-infiltrating cells in mice deficient in the chemokine receptor CCR2, which is required by monocytes/macrophages for homing to sites of inflammation (2, 20, 21). However, the role of CCR2 in monocyte/macrophage recruitment into SG is still unknown. Surprisingly, we observed comparable frequencies of CD11b+ Ly6Chi SSClo cells in SG tissues of both strains (Fig. 6A). Thus, the recruitment of inflammatory monocytes into SG following MCMV infection was unaffected by the CCR2 deficiency.
As an alternate strategy, we investigated the role of CD11b+ Ly6Chi SSClo cells in SG dysfunction using clodronate liposomes to deplete phagocytic cells, including intravascular monocytes and macrophages (50, 54), prior to viral infection. CD11b+ Ly6Chi cell depletion in SG was confirmed by flow cytometry (Fig. 6B). However, this maneuver was not adequate to deplete tissue-resident CD11b+ Ly6Clo F4/80+ macrophages (Fig. 6B) (our unpublished data).
Clodronate treatment significantly increased virus levels in spleen and SG (Fig. 6F), but it had no effect on the ability of viral infection to cause secretory dysfunction (Fig. 6E). We conclude that the recruitment of clodronate-sensitive inflammatory monocytes to the gland was not essential to impair SG function.
NK cells have a critical role in MCMV resistance (57). Since secretory dysfunction best correlated with systemic viremia after MCMV infection (Fig. 1), we considered that NK cells might protect against SG disease. To test this, we measured saliva secretions in mice given only low-dose viral infection so that SG function would be maintained. NK cells were depleted with MAb PK136; this treatment had no effect on SG function (the mean saliva volume of B6 mice was 66 ± 20 μl) (Fig. 2B). As expected, virus resistance in spleen and SG required NK cells (Fig. 7A and B). In addition, SG function was not significantly impacted after low-dose MCMV infection in B6 or NZM mice. However, saliva secretions were significantly diminished in NK cell-depleted B6 and NZM mice in comparison to uninfected and NK cell-replete infected controls (Fig. 2B and and8A).8A). These data suggested that NK cell-mediated MCMV control is required to retain SG secretory function after viral infection. We speculated that if a similar NK cell protective effect were operative in B6 mice, then NK cell-depleted B6 mice might also exhibit significant glandular sensitivity after viral infection, comparable to that of NZM mice. We tested this with NK cell-depleted B6 mice given a low dose of MCMV commensurate with low-dose MCMV infection in NZM mice. As shown in Fig. 8B and C, SG function was impaired when MCMV became readily detectable in spleen. We conclude that NK cell-mediated MCMV control is needed to maintain normal SG function after MCMV infection.
We then asked whether adaptive lymphocytes also protect the animal from secretory dysfunction after infection. To pursue this question, we further examined SG function after low-dose MCMV infection in T cell- and B cell-deficient Rag1 mice. As expected, the splenic MCMV load was significantly higher in Rag1 than in WT mice of either the B6 or NZM background at day 7, but MCMV levels in the SG were comparable (Fig. 7A and B). Higher average saliva outputs in uninfected and infected Rag1 mice implicated possible T and B cell involvement in normal SG functions (Fig. 8A); however, this was not a significant finding, consistent with data shown in Fig. 5. More importantly, saliva secretion was largely maintained after infection of Rag1 mice on either background (Fig. 8A). These data clearly demonstrate that T, NKT, and B cells are not involved in protecting SG function after low-dose MCMV infection. Rather, saliva secretion levels were significantly decreased in NK cell-depleted control and infected Rag1 mice (Fig. 8A). Thus, we conclude that NK cells protect against SG secretory dysfunction, independent of adaptive immunity, and that T and B cells do not mediate SG dysfunction following low-dose MCMV infection.
We considered that NK cells might preserve SG function by regulating glandular inflammation or preventing glandular damage. To address this question, we examined SMG tissue sections for inflammation and tissue pathology. After low-dose infection, SG of B6 mice displayed moderate inflammation (Fig. 7C, middle left). Mostly lymphocytic cells were clustered in perivascular and periductal regions, including some apoptotic lymphocytes. The tissue architecture was largely intact, although some acini were atrophic or smaller than those of controls. The extent of inflammation was similar in NK cell-depleted mice (Table 1), but inflammatory cells were distributed throughout the interstitium and were in contact with acinar cells, which is in contrast to the clusters of lymphocytes observed for NK cell-replete B6 mice (Fig. 7C). Notably, NK cell depletion prior to infection led to drastic glandular changes. The majority of acini were atrophic or necrotic, and the extent of atrophy was more severe (Fig. 7C, bottom left, and Table 1). These data demonstrate that NK cells limited glandular atrophy after infection, which correlated with secretory dysfunction at day 7 (Fig. 8A).
To determine whether T and B cells also contributed to the observed changes, we examined SG from B6.Rag1 mice. After low-dose infection, SG from B6.Rag1 mice displayed mild acinar atrophy without inflammation (Fig. 7C, middle right), whereas severe PMN inflammation in large cell clusters, including some cells that were apoptotic, was in contact with acini in SG of NK cell-depleted B6.Rag1 mice (Fig. 7C, bottom right). We further examined SG tissue sections by immunofluorescence to ascertain the identity of the PMNs within SG from NK cell-depleted mice. Interestingly, SG-infiltrating cells observed in NK cell-depleted Rag1 mice stained positive for Siglec-F, a marker of eosinophils (49, 65), but not for Ly6G, a marker of neutrophils (V. A. Carroll, S. Sainz, and M. G. Brown, unpublished data). Altogether, the data demonstrate that NK cells regulated the type and severity of SG inflammation and atrophy after MCMV infection in immunocompetent strains as well as in strains without adaptive lymphocytes.
An understanding the mechanism(s) of secretory dysfunction is fundamental to the development of effective treatments for xerostomia. This report examined key factors which might impair SG function after acute viral infection in a mouse model of SG disease. MCMV infection in two immunocompetent mouse strains indicated distinct early (day 2) and later (day 7) phases of SG tissue impairment. We found that inflammatory monocytes and lymphocytes were not required for the observed loss of function. On the contrary, NK cells, but not T or B cells, provided significant protection from SG dysfunction after viral infection. Our results support a novel regulatory role of NK cells in limiting tissue inflammation and disease.
In the earliest period, secretion is likely inhibited by an unknown serum factor as a result of an inflammatory response, since (i) the level of splenic MCMV, an indicator of systemic viremia, was high; (ii) the MCMV level in SG was low or undetectable at this point, rendering direct infection of acini an unlikely explanation; and (iii) histologically, the acinar structure remained normal, and the gland was largely free of inflammatory cells, yet global changes in the size and appearance of acini were observed. Recently, another group studying acute MCMV infection of BALB/c mice reported liver damage to be a possible cause of SG dysfunction (18). Other explanations for dysfunction after acute MCMV infection were tested previously by Krane et al. but with negative results; i.e., there was no improvement in secretion after splenectomy or hydration and no correlation with glandular levels of aquaporin 5, a water channel important for secretion (19). We expand on these results to show that monocytic cell depletion via clodronate liposomes did not rescue secretion, indicating the these cells are not the source of a hypothetical serum factor. Instead, these cells added protection in the spleen, in agreement with previously reported findings for other mouse strains (12, 26). The possibility remains that SG-resident macrophages not depleted by liposome treatment or the glandular tissue itself is a source of inflammatory mediators responsible for hypofunction. For example, the few infected macrophages or epithelia may produce a cytokine(s) which acts on neighboring cells. Indeed, the fact that inflammatory monocytes are recruited to the gland at day 2 predicts a change in the local SG environment. The early loss of secretory function at day 2 prior to atrophy is in line with a nonapoptotic model of the hypofunction of SS. In this model, the immune-mediated inhibition of secretion initiates atrophy, rather than the immune-mediated destruction of acini leading to a loss of function (8, 33).
The identity of a serum factor(s) responsible for secretory dysfunction remains to be found. The innate response to viral infection mediated by systemic inflammatory cytokines is an attractive candidate for causing dysfunction. Indeed, a decrease in saliva production was demonstrated after the activation of the innate immune response by the repeated systemic administration of a Toll-like receptor 3 (TLR3) agonist (9, 32). Secretion was restored once treatment was discontinued, suggesting that systemic inflammatory signals were required. The kinetics of induction of innate cytokines such as type I interferon (IFN), IL-6, and tumor necrosis factor (TNF) in serum after MCMV infection have been characterized (41, 44, 46) and immediately precede the secretory defect observed on day 2. We hypothesize that a serum factor disrupts the normal secretion signaling pathway within acini, a carefully orchestrated process initiated by the activation of muscarinic receptors, the generation of inositol triphosphate (IP3), and the release of calcium from intracellular stores and modulated by key small molecules such as cyclic ADP-ribose and NO, among others (25, 38).
The loss of secretory function was most severe at day 7 after infection. In this later phase, the SG was heavily infiltrated with inflammatory cells, and many acini were atrophic or necrotic (Fig. 3C, bottom right). Interestingly, MCMV-induced SG dysfunction on day 7 was lymphocyte independent. In addition, dysfunction was independent of virus levels, since normal secretion was observed for SG with high levels of virus (Fig. 4). These data suggest that the early control of systemic virus replication prevented dysfunction. Once innate defenses were overwhelmed after MCMV infection or in a genetically susceptible host, a damaging systemic inflammatory response ensued. Systemic cytokines at levels above a certain threshold may initiate widespread acinar hypertrophy, as noted on day 2 after infection. Over the next few days, the response of SG cytokines/chemokines to infection may lead to the recruitment of inflammatory cells. Atrophy was observed with the entry of inflammatory cells on days 4 to 7. This was most evident in the comparison of SG histologies at day 7 after infection. Severe mixed inflammation and widespread acinar atrophy were characteristic of infection, while lower-dose MCMV infection had a relatively mild but focal mononuclear infiltrate and preserved tissue architecture. We cannot determine whether acinar atrophy at day 7 is a downstream consequence of early systemic factors or due to the actions of inflammatory cells such as macrophages or PMNs. In summary, the early control of virus replication “sets the stage” to limit glandular inflammation, damage, and loss of function.
Accordingly, we show that NK cells are critical for protection against the MCMV-induced loss of secretory function after low-dose infection. NK cells preserved SG secretion in both lupus- and non-lupus-prone mouse strains after acute infection. We envision that NK cells mediate their effect by restraining MCMV during acute infection and thus limiting a damaging systemic inflammatory response, as documented previously for sera of infected mice (51). It is likely that NK cells act in organs other than SG to limit damage, as a recent report showed that circulating NK cells are not recruited to the gland after acute MCMV infection, and SG-resident NK cells are hyporesponsive in vivo and ex vivo (53). The timing of NK cell virus control may be critical for protection from organ damage. NK cells are innate immune responders, while MCMV-specific T cells require 4 to 5 days to develop (31, 41, 47). The antiviral effect of adaptive immunity in the spleen was not sufficient to protect the gland, since Rag1 mice did not succumb to secretory dysfunction after low-dose infection. Only when NK cells were depleted did Rag1-deficient animals lose secretory function. This finding informs us that tissue damage was independent of T and B cells, since atrophy of acini and associated dysfunction were observed for NK cell-depleted Rag1 mice.
It is interesting that although NK cell-mediated MCMV resistance is generally less effective in NZM than in B6 mice, NK cells in both strains were needed to support secretory function following low-dose MCMV infection. These results suggest that NK cells may regulate potential damage inflicted during an inflammatory response by cytokine cross talk with other cells. In this regard, it was suggested that NK cell cytokine responses can be protective in other disease models. NK cell IFN-γ in the lung was shown previously to limit interstitial fibrosis after bleomycin treatment (16), and IL-22 produced by NK cells prevented weight loss in dextran sulphate sodium (DSS)-induced colitis (64) and increased survival after Citrobacter rodentium infection (6). A recent study further demonstrated that NK cells produce IL-10 after infection with various systemic pathogens, which in turn diminished IL-12 production by dendritic cells (37). This immunoregulatory circuit might explain how NK cell-mediated protection restricts inflammation and further limits SG dysfunction (58).
This report highlights the importance of the innate immune system in controlling virus-induced SG dysfunction and later organ damage. Classic immune players such as monocytes and lymphocytes were dispensable for virus-induced dysfunction, directing us toward alternative mechanisms. In immunocompetent animals, it is clear that a high systemic viral load is associated with dysfunction in our model. Therefore, antiviral drugs may be an effective treatment for virus-induced xerostomia. In point of fact, the prevalence of DILS has significantly decreased among cohorts of HIV patients since the introduction of highly active antiretroviral treatment (HAART) (1, 36).
A common characteristic of the viruses associated with sicca syndromes is their ability to infect for life, either through persistent infection (HIV and HCV) or in a latent/reactivating state (EBV). Direct inflammatory signals in response to viral products may be a causative agent for secretory dysfunction, including times of viral persistence and reactivation. Thus, keeping virus replication at bay is critical. In the case of MCMV, our previous study of NZM mice showed that a subset of animals displayed secretory dysfunction 6 months after infection despite the fact that all mice exhibited severe focal sialadenitis (35). Although virus was undetectable in most of these animals by PCR, formal reactivation studies have not been carried out. The ability of the host to both restrict virus replication and limit innate inflammatory responses to reactivating virus will likely influence susceptibility to virus-induced dysfunction.
In conclusion, MCMV infection initiated two stages of early dysfunction: the first stage occurred with intact glandular architecture and was independent of infiltrating inflammatory monocytes, and the second stage was characterized by glandular atrophy and was independent of lymphocytes. These observations support the view that SG dysfunction does not necessarily require immune cell-mediated destruction and point to other systemic inflammatory mediators that affect the homeostasis of secretion. Importantly, we show a novel regulatory role for NK cells in preventing tissue damage and secretory dysfunction mediated by innate immunity after MCMV infection. It remains to be seen whether NK cell regulation occurs via a simple dampening of the viral load or through another immunoregulatory pathway. MCMV infection serves as a useful model of virus-induced secretory dysfunction and will be helpful in dissecting the role of NK cells in regulating a pathogenic innate immune response.
This work was supported by Public Health Service grant AI050072 (M.G.B.) from the NIAID, a Sjögren's Syndrome Foundation award (V.A.C.), and the Division of Rheumatology and Immunology, Department of Medicine, University of Virginia.
We thank Shu Man Fu and Marcia McDuffie for kindly providing NZM2328 and NZM.Rag1 mice. We also thank Shu Man Fu for insightful discussion, Claudia Rival for help with immunofluorescence, and the UVA Research Histology Core for embedding and staining tissue sections.
Published ahead of print 7 December 2011