Myd88 is required for disease development in a primary Sjögren’s syndrome mouse model

Short abstract

Murine model of primary Sjögren’s syndrome establishes Myd88 as a crucial mediator of local and systemic disease manifestations.

Abstract

Sjögren’s syndrome (SS) is an autoimmune disease that often results in diminished exocrine gland function. SS patients also experience systemic disease manifestations, including hypergammaglobulinemia and pulmonary and renal pathoses. MyD88 is a ubiquitously expressed adaptor molecule used by all immune cells that is required for IL‐1 receptor (IL‐1R), IL‐18R, and most TLR signaling. The precise role of MyD88 in SS has not been evaluated, although this adaptor is critical for development of lupus, a related autoimmune disease. This study tested the hypothesis that Myd88‐mediated signaling is required for local and systemic SS manifestations. To this end, we generated NOD.B10Sn‐H2^b/J (NOD.B10) mice that are deficient in Myd88 (NOD.B10^Myd88−/−). We found that NOD.B10 animals that lack Myd88 show reduced exocrine and extraglandular inflammation. Moreover, these animals are protected from loss of salivary flow. Splenocytes from NOD.B10^Myd88−/− mice did not up‐regulate activation markers or secrete IL‐6 in response to a Myd88‐dependent agonist, although BCR signaling remained intact. Finally, IgM, IgG, and anti‐nuclear autoantibodies were reduced in NOD.B10^Myd88−/− mice compared with the parental strain. These data demonstrate that Myd88 is a crucial mediator of local and systemic SS disease manifestations.

Abbreviations

ANA

anti‐nuclear antibody

BL/10

C57BL10/SnJ

DAMP

damage‐associated molecular pattern

DC

dendritic cell

HEp‐2

human epithelial type 2

HSP90α

heat shock protein 90α

IACUC

Institutional Animal Care and Use Committee

Ifnar1

IFN‐α receptor 1

IL‐1R

IL‐1 receptor

M3R

muscarinic 3 receptor

NOD.B10

NOD.B10Sn‐H2^b/J

NOD.B10Myd88+/−

NOD.B10Sn‐H2b/J mice that are heterozygous for Myd88

NOD.B10^Myd88−/−

NOD.B10Sn‐H2^b/J mice that are deficient in Myd88

pSS

primary Sjögren’s syndrome

SMG

submandibular gland

SS

Sjögren’s syndrome

Introduction

SS is an autoimmune disease that often causes loss of saliva and tear production [1]. In addition, patients may experience systemic disease manifestations, including hypergammaglobulinemia, lymphoma, and pulmonary and renal pathoses [1, –, 3]. The mechanisms underpinning SS etiology and the corresponding early and sustained molecular events in SS disease remain poorly understood. MyD88‐mediated signaling pathways contribute to many autoimmune diseases [4]. However, the role of MyD88 signaling in SS pathogenesis remains poorly understood.

Lupus mice that lack Myd88 have attenuated disease [5, 7, 8, –, 10]. This finding is not surprising given the centrality of MyD88 to the immune response and its critical role in inflammation. Although Myd88‐mediated signals are clearly important for lupus pathogenesis, the role of these pathways in SS is poorly understood. Notably, Myd88 is increased in salivary tissue of an SS mouse model [11], and MyD88‐dependent TLRs are up‐regulated and functional in murine and human SS salivary tissue [11, 13, –, 15]. Moreover, IL‐1 and IL‐18, which also rely on MyD88 for signal transduction, are elevated in SS exocrine tissue, saliva, and sera [16, 18, 19, 20, 21, 22, –, 24]. Finally, PBMCs from SS patients display increased responsiveness to MyD88‐dependent TLR ligands [12, 25]. Thus, these data suggest that MyD88‐mediated signaling pathways may be dysregulated in SS.

To determine whether pSS mice deficient in Myd88 were protected from SS disease, we generated NOD.B10 mice that lacked this adaptor. We used the NOD.B10 strain because SS pathogenesis in this model closely recapitulates the human disease, as these animals have a female disease predilection and develop both local and systemic disease spontaneously by 26 wk of age [26, –, 28]. We found that NOD.B10^Myd88−/− females had attenuated disease, as lymphocytic infiltration was reduced in exocrine tissue, as well as in the lung and kidney. In addition, NOD.B10^Myd88−/− mice maintained salivary flow with disease progression and had diminished total and ANA‐specific autoantibody titers. Therefore, Myd88 is a critical mediator of SS pathology, both in exocrine glands and in other peripheral tissues.

MATERIALS AND METHODS

Mice

NOD.B10, BL/10, and B6.129P2(SJL)‐Myd88^tm1.1Defr/J mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA). We generated NOD.B10^Myd88−/− animals by crossing NOD.B10 mice with Myd88^−/− mice [B6.129P2(SJL)‐Myd88^tm1.1Defr/J]. We then bred 11 successive generations of heterozygotes back to NOD.B10 mice to generate NOD.B10^Myd88−/− mice. We confirmed that these mice were congenic with the parental NOD.B10 strain using a panel of microsatellite markers (The Jackson Laboratory). Animals were bred and maintained at the University at Buffalo, The State University of New York (Buffalo, NY, USA), and housed in identical conditions. Females were used in all experiments. This study was carried out in accordance with the recommendations of the animal research guidelines, IACUC. The protocol was approved by the University at Buffalo, The State University of New York, IACUC.

Histologic processing and analyses

SMG and lacrimal gland tissues were collected. In addition, lung and kidney were harvested. All tissue was formalin fixed, paraffin embedded, and stained with H&E. Slides were scanned using Aperio software (Leica Biosystems, Buffalo Grove, IL, USA), and ImageJ (NIH, Bethesda, MD, USA) was used to measure the lymphocytic infiltration present in the tissue [29]. This was quantified by the division of the area of infiltration by the total tissue area examined, as described previously [30].

Splenocyte harvest and culture

Spleens were harvested and mechanically disrupted. Cells (5 × 10⁶) were cultured in RPMI 1640 containing 5% heat‐inactivated FBS, 2 mM l‐glutamine, 50 μM 2‐ME, 100 U/ml penicillin, and 100 μg/ml streptomycin in 12‐well plates. Splenocytes were stimulated with LPS (25 μg/ml; Sigma‐Aldrich, St. Louis, MO, USA) or anti‐IgM (10 μg/ml; Jackson ImmunoResearch Laboratories, West Grove, PA, USA) and IL‐4 (20 ng/ml; PeproTech, Rocky Hill, NJ, USA) for 24 h before assay, as described previously [31]. Cells and supernatants were harvested and assayed by flow cytometry and ELISA, respectively.

Flow cytometry

Splenocytes were stained by immunofluorescence for B220 (clone RA3‐6B2; BD Biosciences, Franklin Lakes, NJ, USA), CD23 (clone B3B4; BD Biosciences), CD21/35 (clone 7G6; BD Biosciences), CD4 (clone GK1.5; BD Biosciences), CD86 (clone GL1; BD Biosciences), and CD69 (clone H1.2F3; BioLegend, San Diego, CA, USA). All incubations were performed on ice. In brief, cells were incubated with CD16/32 (BD Biosciences) in staining buffer (2% FBS in PBS) for 30 min, followed by addition of fluorescently labeled antibodies at a 1:1000 dilution for 30 min. Cells were washed once in staining buffer before data acquisition. Data analysis was performed using FlowJo (TreeStar, Ashland, OR, USA).

Sera and saliva collection

Sera and saliva were harvested from female NOD.B10, NOD.B10^Myd88+/−, and NOD.B10^Myd88−/− mice and age‐ and gender‐matched BL/10 controls [32]. Sera were harvested by cardiac puncture immediately following euthanasia and stored at −20°C until use. Saliva was collected for 10 min following intraperitoneal injection with pilocarpine HCl (0.3 mg/100 μl; Sigma‐Aldrich). Saliva was centrifuged briefly and measured by pipette, as described previously [30].

ELISAs

Serum IgM, IgG, IgG1, and IgG2a ELISAs were performed in accordance with the manufacturer’s instructions (Bethyl Laboratories, Montgomery, TX, USA). Murine anti‐ANA IgM and IgG ELISAs (Alpha Diagnostic International, San Antonio, TX, USA) were also carried out in accordance with the manufacturer’s instructions with the following modifications: HRP conjugated to anti‐IgG or ‐IgM, respectively, was used to detect isotype‐specific ANA (SouthernBiotech, Birmingham, AL, USA), as described previously [30].

Hep‐2 staining

Sera were diluted 1:10 and incubated with Hep‐2 slides, according to the manufacturer’s instructions (MBL Bion, Des Plaines, IL, USA). ANA were detected with Alexa Fluor 488 anti‐mouse IgG (H+L) or IgM (μ chain; Thermo Fisher Scientific, Waltham, MA, USA), as described previously [30].

Statistics

Trend tests were conducted using Jonckheere’s test with the corresponding P values estimated by permutation (B = 100,000 permutations). One‐sided tests were performed, as decreasing trends (e.g., with respect to a strain ordering of: NOD.B10, NOD.B10^Myd88+/−, and NOD.B10^Myd88−/−) typically constitute the expected departure from the null hypothesis of no trend. A modified Jonckheere’s test was performed for data with replicate measures. Namely, the Jonckheere’s test statistic was evaluated against a permutation distribution in which the group (e.g., strain) labels were shuffled in a way that ensured that replicates maintained the same label. Horizontal lines with asterisks indicating the relative significance of pairwise group comparisons (via more conservative, 2‐sided Mann‐Whitney tests) are also provided. The Jonckheere’s tests constitute the primary statistical hypothesis test, whereas the pairwise P values (which are unadjusted for test multiplicity) are provided for exploratory purposes. The Jonckheere’s tests are referred to hereafter as “trend” tests. Statistical analyses were performed using GraphPad Prism software (GraphPad Software, San Diego, CA, USA), as well as with The R Project for Statistical Computing environment (https://www.r‐project.org/).

RESULTS

NOD.B10^Myd88−/− mice demonstrate reduced prevalence but not severity of salivary and lacrimal gland inflammation

Exocrine gland inflammation is a hallmark of SS patients [1, 33] and is seen in NOD.B10 females with clinical disease, which occurs spontaneously by 26 wk of age [27, 28]. We sought to determine whether NOD.B10^Myd88−/− mice with clinical disease (n = 6) had reduced sialadenitis compared with age‐ and gender‐matched NOD.B10 and NOD.B10^Myd88+/− animals (n = 9 and 11, respectively). We found that inflammation was reduced in SMG tissue from NOD.B10^Myd88−/− mice compared with NOD.B10 and NOD.B10^Myd88−/− animals, although the difference did not reach statistical significance (trend test P value = 0.05259; Fig. 1A and B ). We then assessed dacryoadenitis. Similar to our findings in SMG tissue, NOD.B10 and NOD.B10^Myd88+/− mice exhibited lymphocytic infiltration in lacrimal tissue (n = 12 and 11, respectively). NOD.B10^Myd88−/− mice had diminished inflammation (n = 9), although the differences were not statically significant (trend test P value = 0.08705; Fig. 1C and D). Of note, although almost 80% (7/9) of NOD.B10 and 70% (7/10) of NOD.B10^Myd88+/− mice exhibited some degree of sialadenitis, only 33% (2/6) of NOD.B10^Myd88−/− mice had lymphocytic infiltration of SMG tissue. Similar results were observed in the lacrimal tissue, as dacryoadenitis was observed in almost 90% (8/9) of NOD.B10 and 80% (8/10) of NOD.B10^Myd88+/− mice, whereas 56% (5/9) of NOD.B10^Myd88−/− mice showed lacrimal inflammation. Thus, Myd88‐mediated signals exacerbate exocrine inflammation in salivary and lacrimal tissues in NOD.B10 mice.

Figure 1.

Sialadenitis and dacryoadenitis are reduced in NOD.B10^Myd88−/− mice.

(A and B) SMG and (C and D) lacrimal tissues were harvested from NOD.B10, NOD.B10^Myd88+/− [heterozygous (Het)], and NOD.B10^Myd88−/− [knockout (KO)] female mice at the clinical disease time point. Representative photomicrographs are shown. Original magnification, ×100. Black arrowheads represent lymphocytic infiltration, and white arrowheads indicate ducts. Inflammation was quantified using ImageJ. Horizontal lines represent the mean and sem.

NOD.B10^Myd88−/− mice are protected from loss of salivary flow

To determine whether salivary gland function was altered in NOD.B10^Myd88−/− mice, we administered pilocarpine to NOD.B10, NOD.B10^Myd88+/−, and NOD.B10^Myd88−/− females at 8, 12, 16, 20, and 26 wk of age and collected stimulated saliva. We found that both NOD.B10^Myd88+/− and NOD.B10 mice exhibited diminished saliva production with disease progression (trend test P value = 0.01298 and 0.0014, respectively). Strikingly, NOD.B10^Myd88−/− mice maintained salivary flow over time (trend test P value = 0.749; Fig. 2 ). Thus, our data show that Myd88 expression is required for loss of saliva in NOD.B10 and NOD.B10^Myd88+/− mice.

Figure 2.

NOD.B10^Myd88−/− mice maintain salivary flow with disease progression.

Saliva production was quantified from NOD.B10, NOD.B10^Myd88+/−, and NOD.B10^Myd88−/− females at 8, 12, 16, 20, and 26 wk of age. All statistical differences are compared with animals at 8 wk of age. Horizontal lines represent the median and sem (N.S., Not significant; *P < 0.05; **P < 0.01).

NOD.B10^Myd88−/− mice exhibit diminished pulmonary and renal inflammation

NOD.B10 mice display extraglandular disease manifestations, including lymphocytic infiltration of pulmonary and renal tissue that is also observed in a subset of pSS patients [2, 28]. To determine whether Myd88 activation contributes to systemic inflammation in pSS, we harvested lung and kidney tissue from NOD.B10 (n = 12), NOD.B10^Myd88+/− (n = 10), and NOD.B10^Myd88−/− females with clinical disease (n = 8). We first assessed lymphocytic infiltration in pulmonary tissue and observed a highly significant trend (decreasing) with respect to strain ordered by the number of functional Myd88 alleles: NOD.B10, NOD.B10^Myd88+/−, and NOD.B10^Myd88−/− (trend test P value = 0.00008; Fig. 3A and B ). Analyses of renal tissue revealed lymphocytic infiltration was diminished in NOD.B10^Myd88−/− females compared with NOD.B10 and NOD.B10^Myd88+/− mice as well (trend test P value = 0.00092; Fig. 3C and D).

Figure 3.

NOD.B10^Myd88−/− mice show diminished lymphocytic infiltration of pulmonary and renal tissues.

(A and B) Pulmonary and (C and D) renal tissues were harvested from NOD.B10, NOD.B10^Myd88+/− (Het), and NOD.B10^Myd88−/− (KO) female mice at the clinical disease time point. Representative photomicrographs are shown. Original magnification, ×100. Black arrowheads represent lymphocytic infiltration. Inflammation was quantified using ImageJ. Horizontal lines represent the mean and sem (N.S., Not significant; *P < 0.05; **P < 0.01; ***P < 0.001).

Splenic B and T cell populations are similar among NOD.B10, NOD.B10^Myd88+/−, and NOD.B10^Myd88−/− mice

A previous report in a Myd88‐deficient lupus model showed decreased splenic T cell numbers and a slight increase in B cells [5]. To assess splenic B and T populations in our models, we harvested spleens from NOD.B10 (n = 8), NOD.B10^Myd88+/− (n = 10), and NOD.B10^Myd88−/− females with clinical disease (n = 6). We found that there were no differences in either the percentage of total B and T cells or the number of cells among the 3 strains ( Fig. 4A and D , and data not shown). We then examined follicular and marginal zone B cells and found that NOD.B10, NOD.B10^Myd88+/−, and NOD.B10^Myd88−/− females had similar percentages of both subsets (Fig. 4B and C). Of this set of experiments, only the investigation of marginal zone B cells exhibited a statistically suggestive result (trend test P value = 0.07154). Therefore, total B and T cell populations are similar across the 3 strains examined, and the differences in tissue inflammation observed are not due to global reductions in the number of B and/or T cells in NOD.B10^Myd88−/− mice.

Figure 4.

Splenic B and T cell populations are similar in NOD.B10, NOD.B10^Myd88+/−, and NOD.B10^Myd88−/− mice.

Splenocytes were isolated from NOD.B10, NOD.B10^Myd88+/− (Het), and NOD.B10^Myd88−/− (KO) females at the clinical disease time point. Cells were fluorescently labeled with (A) B220. (B) The percentages of follicular (FO; B220⁺, CD23⁺, CD21/35⁻) and (C) marginal zone (MZ; B220⁺, CD23^lo/−, CD21/35⁺) B cells and (D) CD4⁺ T cells are shown. Horizontal lines represent the mean and sem.

B cells from NOD.B10^Myd88−/− mice are activated by BCR cross‐linking but are not responsive to LPS

We then sought to assess B cell function in NOD.B10^Myd88−/− mice. We harvested spleens from NOD.B10 (n = 6), NOD.B10^Myd88+/− (n = 5), and NOD.B10^Myd88−/− females with clinical disease (n = 5). We cultured cells in media alone, with the Myd88‐dependent agonist LPS and Myd88‐independent stimuli anti‐IgM and IL‐4. We measured expression of the B cell activation markers CD69 and CD86 using flow cytometry. We found that B cells derived from NOD.B10 and NOD.B10^Myd88+/− mice showed significant up‐regulation of both activation markers following stimulation with LPS or anti‐IgM and IL‐4, as expected. In contrast, B cells from NOD.B10^Myd88−/− mice up‐regulated CD69 and CD86 in response to BCR cross‐linking but failed to do so in the presence of LPS ( Fig. 5A and B ).

Figure 5.

B cells from NOD.B10^Myd88−/− mice maintain responsiveness to BCR cross‐linking but are not activated by LPS.

Splenocytes were isolated from NOD.B10 (n = 6), NOD.B10^Myd88+/− (Het; n = 5), and NOD.B10^Myd88−/− (KO) females at the clinical disease time point (n = 5). (A) Cells were cultured for 24 h, and expression of CD69 (A) and (B) CD86 was assessed by flow cytometry (gray shading, isotype control; blue line, media alone; red line, LPS; yellow line, anti‐IgM + IL‐4). Representative results are shown. (C) Cells were cultured for 24 h, as indicated, and supernatants harvested. IL‐6 was quantified by ELISA. All samples were analyzed in duplicate. Horizontal lines represent the mean and sem (**P < 0.01).

We then quantified IL‐6 production by stimulated splenocytes isolated from the 3 strains. We found that supernatants from NOD.B10 and NOD.B10^Myd88+/− splenocytes produced high levels of IL‐6 when stimulated with either LPS or anti‐IgM and IL‐4. We did not detect any IL‐6 in the supernatants of NOD.B10^Myd88−/− splenocytes following LPS treatment. Therefore, production of IL‐6 by NOD.B10 and NOD.B10^Myd88+/− splenocytes was elevated compared with that from NOD.B10^Myd88−/− mice (trend test P value = 0.00003). However, we detected similar IL‐6 levels following stimulation with anti‐IgM and IL‐4 in supernatants derived from NOD.B10 and NOD.B10^Myd88−/− mice (Mann‐Whitney P value = 0.1) and between supernatants isolated from NOD.B10^Myd88+/− and NOD.B10^Myd88−/− splenocytes (Mann‐Whitney P value >0.9999; Fig. 5C). Therefore, NOD.B10^Myd88−/− B cells respond normally to BCR cross‐linking but fail to respond to Myd88‐dependent TLR agonism.

Total and autoreactive antibodies are reduced in NOD.B10^Myd88−/− mice with clinical disease

Finally, we sought to determine whether total and autoreactive IgM and IgG titers were altered in NOD.B10 mice that lack Myd88. We harvested sera from NOD.B10 (n = 10), NOD.B10^Myd88+/− (n = 9), and NOD.B10^Myd88−/− females with clinical disease (n = 7) and age‐ and gender‐matched BL/10 controls (n = 5). We found that total IgM levels were reduced in NOD.B10^Myd88−/− females compared with NOD.B10, NOD.B10^Myd88+/−, or BL/10 animals (trend test P value = 0.00007, for the ordering: NOD.B10, NOD.B10^Myd88+/−, NOD.B10^Myd88−/−; Fig. 6A ). IgG titers of NOD.B10^Myd88−/− mice were also diminished compared with NOD.B10 and NOD.B10^Myd88+/− animals (trend test P value = 0.00002, for the ordering: NOD.B10, NOD.B10^Myd88+/−, NOD.B10^Myd88−/−), although there was no difference between IgG levels of NOD.B10^Myd88−/− mice and BL/10 controls (Mann‐Whitney P value = 0.3). NOD.B10 and NOD.B10^Myd88+/− mice also displayed higher IgG levels than BL/10 controls (Mann‐Whitney P values = 0.003 and 0.002, respectively; Fig. 6B). Finally, we assessed IgG1 and IgG2a levels in NOD.B10 and NOD.B10^Myd88−/− females with clinical disease. Both IgG1 and IgG2a were significantly elevated in NOD.B10 mice compared with those lacking Myd88 (trend test P values = 0.00218 and 0.00052, respectively; Fig. 6C and D). However, there was no appreciable difference in the IgG1:IgG2a ratio between the 2 strains (data not shown, Mann‐Whitney P value = 0.4).

Figure 6.

Total and ANA‐specific antibodies are diminished in NOD.B10^Myd88−/− mice.

Sera were harvested from NOD.B10, NOD.B10^Myd88+/− (Het), and NOD.B10^Myd88−/− (KO) females at the clinical disease time point and age‐ and gender‐matched BL/10 controls. (A) Total IgM and (B) IgG titers, (C) IgG1, (D) IgG2a, and ANA‐specific (E) IgM and (F) IgG were quantified by ELISA. Hep‐2 staining for (G) IgM and IgG was performed. Original magnification, ×200. One representative animal of 4 examined from each strain is shown. Horizontal lines represent the mean and sem (*P < 0.05; **P < 0.01; ***P < 0.001).

We then assessed ANA levels by both ELISA and Hep‐2 staining. ANA levels increase with clinical disease in NOD.B10 females [26, 28]. ANA‐specific IgM ELISA results revealed diminished titers in NOD.B10^Myd88−/− mice compared with NOD.B10 and NOD.B10^Myd88+/− animals (trend test P value = 0.00008; Fig. 6E). We then examined ANA‐specific IgG levels. We found that these were considerably more variable than those observed for IgM, and as a result, we were unable to detect significant differences among the 3 groups. However, ANA‐specific IgG tended to be higher in both NOD.B10 and NOD.B10^Myd88+/− mice compared with Myd88‐deficient animals (trend test P value = 0.05471; Fig. 6F). These findings were corroborated by Hep‐2 staining, as sera from NOD.B10 and NOD.B10^Myd88+/− showed significant autoreactivity that was diminished in sera from NOD.B10^Myd88−/− females (Fig. 6G). Thus, pSS mice that lack expression of Myd88 have diminished total and autoreactive antibody production.

DISCUSSION

Results from this study show that NOD.B10 females that lack Myd88 are protected from SS exocrine‐related and extraglandular disease manifestations. Specifically, NOD.B10^Myd88−/− mice exhibit diminished lymphocytic infiltration of salivary, lacrimal, pulmonary, and renal tissues. Moreover, Myd88‐deficient pSS mice are protected from loss of salivary flow. Although animals display normal splenic B and T cell populations, B cells derived from NOD.B10^Myd88−/− mice fail to respond to LPS, although activation marker expression and cytokine secretion induced by BCR cross‐linking are similar in the NOD.B10, NOD.B10^Myd88+/−, and NOD.B10^Myd88−/− strains. Finally, levels of total and ANA‐specific antibodies are decreased in NOD.B10^Myd88−/− mice with clinical disease compared with NOD.B10 and NOD.B10^Myd88+/− animals. Altogether, these data demonstrate that Myd88 is essential for development of pSS disease.

MyD88 is used by both innate and adaptive immune cells. Historically, adaptive immune dysfunction was thought to be the prime driver of SS, although more recent studies show that the innate immune system is equally crucial in mediating disease [34]. Accordingly, work in many different SS mouse models demonstrates an important role for innate immune activation in disease pathogenesis [35, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, –, 48]. For example, the IL‐14α‐transgenic pSS model has increased lymphotoxin α in the salivary tissue, even before the arrival of lymphocytes in the gland [43]. This suggests that activation of the innate response precedes the adaptive in salivary tissue. In addition, treatment of SS‐prone mice with a potent activator of innate immunity—Freund’s incomplete adjuvant—induced salivary hypofunction and exacerbated autoantibody production [38]. Therefore, data from numerous SS models provide strong evidence for dysregulated innate immune activation in the development and progression of SS.

Several studies demonstrate a key role for Myd88 in lupus pathogenesis. Accordingly, mice with hyperactive Myd88 signaling in the DC compartment develop a lupus‐like disease characterized by spontaneous B and T cell activation and production of autoantibodies. Strikingly, deletion of Myd88 in DCs completely reverses this inflammatory phenotype [49]. Corroborating work demonstrates a causative role for Myd88 in DC and B cell‐mediated tissue damage in lupus [6], and Myd88 is crucial for autoantibody generation in this disease [5, –, 7, 50]. MyD88 expressed by the T cells also contributes to autoimmunity. MyD88 is a key regulator of Th17 proliferation, and Myd88‐deficient T cells show defective Th17 and Th1 responses [51, 52]. Disease is also attenuated in animals with Myd88‐deficient T cells in a mouse model of multiple sclerosis [52]. Thus, dysregulated MyD88 signaling in hematopoietic cells plays a causative role in numerous autoimmune diseases.

These studies have important implications for SS pathogenesis, as it remains unknown as to whether Myd88 signaling in the hematopoietic or stromal compartment mediates disease. Of note, these possibilities are not mutually exclusive, as both may be required for robust disease development. Because salivary gland epithelial cells may be activated in a MyD88‐dependent manner [13], activation of MyD88‐dependent signaling cascades in salivary tissue may be the prime driver of both local and systemic disease. Accordingly, diminished salivary gland inflammation in NOD.B10^Myd88−/− mice may result in the relatively low total and ANA‐specific autoantibody levels observed (Fig. 6E–G), as salivary tissue may be an important niche for B cell activation in this disease [53]. Indeed, corroborating data from a recent study suggest that salivary inflammation may drive systemic disease in pSS [54].

In contrast, immune cells may become activated in the periphery and enter the salivary tissue stochastically in SS. Support for this scenario is provided by recent work in a related NOD model (NOD.H‐2h4), demonstrating that blockade of CD40 ligand before disease onset abrogates splenic germinal center formation and prevents salivary gland inflammation and autoantibody production later in life [55]. Data from human studies provide support for the importance of early peripheral immune dysregulation in disease, as antibody production is reported to occur years before onset of salivary dysfunction in pSS patients [56, 57].

Importantly, some autoantibodies have pathogenic function in SS [58]. Similar to humans, NOD.B10 animals produce anti‐M3R‐specific IgG1 autoantibodies that inhibit salivation [59]. As our NOD.B10 animals were protected from loss of salivary flow, and IgG1 autoantibodies were decreased significantly in NOD.B10^Myd88−/− mice (Figs. 2 and 6C), it is possible that this protection is mediated by diminished autoantibody production rather than by local effects of Myd88 deficiency within the salivary tissue itself. Further studies are needed to establish the relative contributions of Myd88 signaling in exocrine tissue and in the hematopoietic compartment to SS initiation and progression.

In addition to anti‐M3R, many different autoantibodies are seen in SS patients [58]. We measured ANA‐specific autoantibodies and found that ANA‐specific IgM was decreased in all NOD.B10^Myd88−/− mice examined. ANA‐specific IgG was also reduced, although the difference was not significant as a result of the variation observed among the groups. This is in contrast to results in lupus models, where Myd88 is essential for ANA production (vide infra). Although we do not know the reason for the disparate findings in our pSS model, it is possible that these autoantibodies arise through distinct disease‐specific mechanisms that have yet to be unraveled. Therefore, whereas Myd88 contributes to autoantibody generation in both diseases, the way in which Myd88‐dependent pathways drive pathogenic B cell activation in lupus and SS may be highly disease specific.

Several insightful studies in lupus models demonstrate that Myd88 is required for production of ANA, anti‐Smith, and anti‐RNA autoantibodies [5, 6, 8, 10]. However, other autoantibodies are reduced in these models (rheumatoid factor and anti‐dsDNA) but are not eliminated entirely, suggesting that certain autoantibodies are generated by both Myd88‐dependent and ‐independent pathways [5, 6, 10]. Importantly, these data show that the defects in autoantibody production are indeed a result of MyD88‐mediated signaling and are not caused by nonspecific defects in B cell activation and class switching [5]. Therefore, studies in lupus models demonstrate that MyD88 is essential for specific disease manifestations, although Myd88‐independent events are also required for robust disease development.

Our work is consistent with findings in lupus models, in that Myd88‐independent pathways are also important in SS disease. Agonism of TLR3, which signals independently of Myd88, mediates SS‐like disease, and TLR3 activation is also demonstrated in human salivary tissue derived from SS patients [13, 36, 37, 60, –, 63]. The importance of Myd88‐independent pathways in SS disease is illustrated further by studies in Ifnar1‐deficient animals. To generate these animals, the well‐established pSS model, B6.Aec1Aec2, was bred to Ifnar1^−/− mice [40, 64]. The resultant strain, termed B6.Aec1Aec2Ifnar1^−/−, was protected from sialadenitis and dacryoadenitis [40]. Moreover, the animals did not exhibit reduced salivary flow with disease progression [40]. Of note, autoantibody titers were similar between the parental strain and B6.Aec1Aec2Ifnar1^−/− animals [40], suggesting a dichotomy in the requirement for Ifnar1 in SS disease manifestations in this model.

Significantly, our data corroborate these findings, as lacrimal and salivary inflammation and loss of salivary flow occurred in the absence of Myd88 in a minority of animals examined (Figs. 1 and 2). However, NOD.B10^Myd88−/− mice were nearly completely protected from pulmonary and renal inflammation and demonstrated significantly diminished titers of total and autoreactive antibodies (Figs. 3 and 6). Taken together, these findings suggest that specific SS disease manifestations may arise through distinct MyD88‐dependent and ‐independent signaling networks.

The IL‐1R, IL‐18R, and numerous TLRs all use MyD88 to transduce inflammatory signals [65, 66]. Although MyD88 is clearly essential for robust SS pathogenesis, it is not clear, at present, whether 1 of these pathways is predominant or whether all 3 coalesce to promote disease. Receptors for IL‐1 family members are present on numerous cell types implicated in SS, including macrophages, DCs, and epithelium [67]. Although our data show that there is no difference in IL‐1β levels in sera derived from NOD.B10 and NOD.B10^Myd88−/− mice with clinical disease (data not shown), it is possible that IL‐1 family members may be important mediators of tissue inflammation and that further work is required to clarify the role of these cytokines in disease.

TLR activation is another potential mechanism whereby MyD88‐dependent pathways may contribute to SS development and progression. TLRs are ubiquitously expressed receptors, and almost all TLRs signal via MyD88 [68]. Evidence suggests that heightened TLR signaling may be an important disease mechanism in SS patients [12, 13, 25, 61], although further studies are needed to establish the ligands that give rise to TLR activation in SS. Endogenous ligands, termed DAMPs, can activate both innate and adaptive immune cells via TLRs [69]. DAMPs include many diverse types of molecules, including HSPs, extracellular matrix proteins, and saturated fatty acids [70]. Numerous studies show that activation of TLRs by DAMPs has pathologic consequence in autoimmunity [69, 71, 72, –, 74], although this has not been evaluated in depth in SS.

It is intriguing to speculate about DAMPS as a source of MyD88 activation in SS. Although the role of DAMP‐induced inflammation is not established in SS, studies show that calprotectin (S100A8/A9) is elevated in SS patients compared with healthy controls [75, 76]. Both S100A8 and S100A9 are DAMPs that activate TLR4 [77, 78]. Moreover, HSP90α is elevated in SS patients with high fatigue [79]. HSP90 may facilitate TLR9 activation by binding CpG oligonucleotides [80]. Finally, proteolytic degradation of biglycan is observed in NOD.B10 salivary tissue [81], and fibronectin is dysregulated in SS mice and patients [82, 83]. Both biglycan and fibronectin induce inflammation via TLR2 and TLR4 [69, 71, 84]. Of note, studies in a lupus model show that biglycan activates both Myd88‐dependent and ‐independent signaling pathways [85]. Thus, DAMP‐mediated activation of MyD88‐dependent signaling pathways may be an important mechanism that contributes to the chronic inflammatory landscape that is characteristic of SS [86].

The etiology of SS is complex, and many factors likely contribute to disease development. We demonstrate for the first time that Myd88 expression is crucial for many SS‐related disease manifestations. Thus, our data suggest that therapeutic targeting of MyD88‐dependent signaling pathways may have therapeutic utility for SS patients.

AUTHORSHIP

J.K., L.G.M., and J.M.K. provided substantial contributions to the conception or design of the work. D.P.G. made substantial contributions to the analysis of data for the work. J.M.K. and D.P.G. drafted the work, and J.K. and L.G.M. revised it critically for important intellectual content. J.K., L.G.M., D.P.G., and J.M.K. gave final approval of the version to be published and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

DISCLOSURES

The authors declare no conflicts of interest.