Introduction
Sjögren's syndrome (SS) is a systemic autoimmune disorder affecting secretory tissue, including the lacrimal and salivary glands (SGs), resulting in keratoconjunctivitis sicca and xerostomia. SS is characterized by mononuclear cell infiltrates in the salivary and lacrimal glands as well as the presence of autoantibodies in serum. Other organ systems may be involved as well and around 5% of the patients develop B cell lymphoma [
1,
2]. There is still an unmet need for an effective treatment of SS.
Anti-tumor necrosis factor (TNF) therapies have been widely and successfully used several chronic autoimmune diseases, such as rheumatoid arthritis (RA) and Crohn's disease. Clinical trials with anti-TNF antibodies and etanercept showed improvement in 60 to 70% of the RA patients [
3,
4]. Patients with SS have been reported to have elevated serum pro-inflammatory cytokine levels compared with normal volunteers [
5,
6] and TNF is also overexpressed in the SGs of SS patients [
7]. However, the use of anti-TNF agents in patients with the autoimmune disease SS has shown conflicting results [
8,
9]. Beneficial results were shown in an open study, while inefficacy of anti-TNF was shown in a randomized, double-blind, placebo-controlled trial.
TNF promotes inflammation by stimulating and inducing other inflammatory cytokines and adhesion molecules and is a key player in the cytokine balance [
4]. In contrast, TNF can also exhibit anti-inflammatory activities, for instance by blocking the development of autoreactive T cells [
10]. Moreover, adoptive transfer of ex vivo TNF treated splenocytes from autoimmune diabetic female non-obese diabetic (NOD) mice into irradiated pre-diabetic male mice prevented the development of hyperglycemia in 80% of the recipients. Recently, a T-cell based mechanism has been proposed to explain the dual effect of anti-TNF therapy in the treatment of autoimmune diseases in which TNF can function as a pro-inflammatory cytokine as well as an anti-inflammatory immunoregulatory molecule by altering the balance of regulatory T cells [
11].
The National Institute of Dental and Craniofacial Research (NIDCR) Sjögren's clinic has previously investigated the efficacy of systemic etanercept treatment in SS patients and could not demonstrate clinical benefit [
12]. Follow-up studies of cytokine levels in these patients before and after treatment revealed no decrease in TNF and other pro-inflammatory cytokines [
13,
14]. The reasons for the failed clinical trials are not well understood, but it is conceivable that the effects would be different if a more localized approach was used. Gene therapy offers the possibility to engineer cells to express therapeutic proteins locally at high levels. Previously, we reported successful gene transfer of interleukin (IL)-10 and vasoactive intestinal peptide (VIP) to mouse SGs [
15,
16]. To investigate the effects of local TNF blockade using gene therapy, we evaluated the effect of a locally expressed TNF inhibitor on the SG function and histopathology in the NOD model of SS.
Materials and methods
Cell lines
Human embryonic kidney 293T cells were grown in DMEM (Invitrogen, Carlsbad, CA, USA). This medium was supplemented with 10% heat-inactivated fetal bovine serum (FBS, Life Technologies, Rockville, MD, USA), 2 mM L-glutamine, penicilline (100 U/ml), and streptomycin (100 μg/ml; Biofluids, Rockville, MD, USA) as previously described [
15]. Human fibrosarcoma (WEHI) cells were grown in RPMI 1640 (Invitrogen). This medium was supplemented with 10% FBS, 2 mM L-glutamine, penicilline (100 U/ml) and streptomycin (100 μg/ml), gentamycin (10 mg/ml; Invitrogen) and 1 M hepes (Invitrogen).
Construction, expression and biological activity of plasmid
We previously reported the construction of recombinant Adeno Associated Virus (rAAV)-β galactosidase (rAAV2-LacZ) encoding β-galactosidase [
17]. In this study we used the extra-cellular domain of human 55 kDa Tumor Necrosis Factor Receptor type 1 (hTNFR1; p55) coupled to the Fc-part of mouse Immunoglobulin G1 (IgG1), kindly provided by Dr J. Kolls [
18]. This gene was cloned into the rAAV plasmid containing a Cytomegalovirus (CMV) promoter and the Inverted Terminal Repeat (ITRs) sequences for AAV serotype 2 (AAV2). The resulting plasmid (pAAV2-CMV-hTNFR1-mIgG1) was transfected into 293T cells and secretion of the protein in the supernatant was measured by western blotting and an ELISA-kit for hTNFR1 (R&D systems, Minneapolis, MN, USA). The biological activity was measured in an assay using WEHI fibrosarcoma cells. This assay is based on the cytotoxic effect of TNF-α on WEHI and the neutralizing effect of soluble hTNFR1 on bioactive TNF [
19].
Vector production
To generate rAAV serotype 2 vectors (rAAV2), we used the adenoviral helper packaging plasmid pDG. Plates (15 cm) of approximately 40% confluent 293 T cells were cotransfected with either pAAV-LacZ, pAAV-TNFR1 or pAAV-Luc and pDG according to standardized methods [
20]. Clarified cell lysates were adjusted to a refractive index of 1.372 by addition of Cesium Chloride (CsCl) and centrifuged at 38,000 rpm for 65 hr at 20°C [
21]. Equilibrium density gradients were fractionated and fractions with a refractive index of 1.369 to 1.375 were collected. The titer of DNA physical particles in rAAV stocks was determined by Quantitative-Polymerase Chain Reaction (Q-PCRand the vectors were stored at -80°C. On the day of vector administration to NOD mice, the vector was dialyzed for 3 hr against saline.
Animals
Female NOD mice used in this study were obtained from the University of Florida College of Medicine, Department of Pathology, Immunology and Laboratory Medicine. All the mice were bred and maintained under specific pathogen free conditions in the animal facilities of the National Institute of Molecular Physiology and Therapeutics Branch (MPTB). All the animals were kept under specific conditions as described in Table
1. Animal protocols were approved by MPTB Animal Care and Use Committee and the National Institutes of Health (NIH) Biosafety Committee. All mice received water and food ad libitum. Blood glucose levels were measured by tail cut once a week starting at 12 weeks of age, using a OneTouch monitor (LifeScan, Milpitas, CA, USA). Mice with blood glucose levels >400 mg/dl were treated by subcutaneous injection (1 U/24 h) with long acting Humalin N (Eli Lilly, Indianapolis, IN, USA). The blood sugar level was monitored previously in other studies with NOD mice and does not appear to affect their behavior, feeding activity, or SG activity compared with non-diabetic NOD mice [
15].
Table 1
Specific conditions for NOD mice
Food
| Teklad Global 18% Protein Rodent Diet (2018S), Harlan |
Water
| Autoclaved |
Bedding
| Care fresh (non-bleached) |
Day/Night light cycle
| 14/10 hours |
Racks
| Vented heap-filtered microisolator |
Cage changing
| 1/week |
Cleaning agents
| Clidox 1:18:1 |
Cage set up
| Autoclaved |
Helicobacter
| Tested negative |
MPV
| Tested negative |
MVM
| Tested negative |
MNV
| Tested negative |
rAAV2 vector administration and plasma/saliva collection
Vectors were delivered into the submandibular glands by retrograde instillation as previously described [
15]. Briefly, mild anesthesia was induced by ketamine (100 mg/mL, 1 mL/kg body weight (BW); Fort Dodge Animal Health, Fort Dodge, IA, USA) and xylazine (20 mg/mL, 0.7 mL/kg body weight; Phoenix Scientific, St. Joseph, MO, USA) solution given intramuscularly (im). Ten minutes after im injection of atropine (0.5 mg/kg BW; Sigma, St. Louis, MO, USA), female NOD mice at the age of eight weeks were administered 50 μl vector into both submandibular glands by retrograde ductal instillation (1 × 10
10 particles/gland) using a thin cannula (Intermedic PE10, Clay Adams, Parsippany, NJ, USA). The vector dose was chosen based on previously published results, which showed detectable transgene activity above 10
9 particles/gland [
15]. Saliva collection was done at several time points: baseline (six weeks of age), 16, 20 and 24 weeks of age. Mice were anesthetized as described above and saliva secretion was induced by subcutaneous (sc) injection of pilocarpine (0.5 mg/kg BW; Sigma-Aldrich, St. Louis, MO, USA). Stimulated whole saliva was gravimetrically collected for 20 minutes from the oral cavity with a hematocrit tube (Drummond Scientific Company, Broomall, PA, USA) placed into a preweighed 0.5 ml microcentrifuge tube, and the volume was determined by weight as previously described [
15]. The presented saliva data are the result of two independent experiments (N = 20 for LacZ and N = 19 for TNFR1:IgG). Blood was collected at the saliva collection time points by retro-orbital plexus bleeding, from which plasma was separated by centrifugation for five minutes in an eppendorf tube centrifuge. Plasma was stored at -80°C until further analysis.
Histologic assessment of submandibular glands
Submandibular glands were removed for histologic analysis from NOD mice at the time of sacrifice, 24 weeks of age, and placed overnight in 10% formalin. After fixation, the tissues were dehydrated in ethanol series and embedded in paraffin according to standard techniques. Sections were cut at 5 μm with 50 μm between consecutive sections and subsequently stained with hematoxylin and eosin (H&E). Histopathologic scoring was performed using the focus score system in which a focus is defined as an aggregate of 50 or more lymphocytes. A total of three sections per SG were scored and the results were calculated and expressed as foci per 4 mm2. The focus scores were assessed blindly by three different examiners (JV, HY, NR) and the mean scores were determined.
Immunohistochemistry
Infiltrating lymphocytic cells were phenotyped on frozen SG sections by immunohistochemistry with antibodies directed towards CD4 (clone L3T4, eBioscience, San Diego, CA, USA), CD8 (clone 53-6.7, eBioscience), CD19 (clone ID3, BD, Breda, The Netherlands), and CD138 (clone 281-2, BD, Breda).
Sections were blocked with endogenous peroxidase with 1% H
2O
2. Aspecific binding was reduced by blocking with PBS (phosphate buffered saline)/1% BSA (Bovine Serum Albumin) + 10% goat serum for 30 minutes at room temperature (RT). After blocking, the sections were incubated with primary antibody at 4°C overnight (O/N). Stainings were visualised with a horseradish peroxidise (HRP)-labelled goat-anti-rat secondary antibody (Southern Biotechnology, Birmingham, AL, USA) and developed by AEC substrate (Dako, Glostrup, Denmark). Concentration- and isotype-matched control antibodies were included as negative control. All sections were coded and randomly analyzed by computer-assisted image analysis. For all markers, 18 high-power fields were analyzed. The images of the high-power fields were analyzed using the Qwin analysis system (Leica, Cambridge, UK), as described previously [
22]. Positive staining of the cellular markers was expressed as the number of positive cells/mm
2.
Quantification of cytokine profiles
The levels of several cytokines were determined after extraction of soluble protein from the SGs. After measuring the wet weight, SGs were homogenized in protease buffer PBS and complete protease inhibitor cocktail (Roche Applied Science, Indianapolis, IN, USA). The excessive connective tissue and large aggregate debris were removed by 15 minutes centrifugation at 1500 × g. Total protein in the supernatant was determined with BCA™ protein assay kit (Pierce, Rockford, IL, USA) according to the manufacturer's instructions. Human sTNFR1 was measured by a commercial available ELISA kit (R&D Duokit, Minneapolis, MN, USA). mIL-4, mIL-5, mIL-6, mIL-10, mIL-12p70, mIL-17, MCP-1, mTGF-β1, mIFN-γ and mTNF-α were measured commercially using SearchLight proteome assay (Pierce Biotechnology, Woburn, MA, USA). This assay is a multiplexed sandwich ELISA procedure for detecting multiple cytokines in the same minimal sample. The same analytes were determined in plasma samples. Lower detection limits for this assay are: mIL-4: 1.2 pg/ml, mIL-5: 2.3 pg/ml, mIL-6: 5.5 pg/ml, mIL-10: 1.6 pg/ml, mIL-12p70: 0.78 pg/ml, mIL-17: 1.6 pg/ml, MCP-1: 0.78 pg/ml, mTGF-β1: 6.8 pg/ml, mIFN-γ: 7.8 pg/ml, mTNF-α: 3.1 pg/ml.
Determination of autoantibodies
Plasma samples of 24 wk old NOD mice were analyzed for autoantibodies against nuclear antigens (ANA), Sjögren's syndrome (SS) A/Ro and SSB/La. ANA (total Ig), the autoantibody against SSA/Ro (total Ig) and SSB/La (total Ig) were measured by a commercial available ELISA kit (Alpha Diagnostic International, San Antonio, TX, USA) according the manufacturer's protocol.
Statistical analysis
Differences in salivary flow, cytokines and lymphocytic cell type characterization among experimental groups were assessed using the non-parametric Wilcoxon's ranksum test. The focus scores and autoantibodies were assessed using unpaired Student's t-test to compare differences between groups. Non-parametric correlations were assessed using Spearman's Rho test and was performed with SPSS for Windows (SPSS version 16.0.02, Chicago, IL, USA) and all the other analyses were performed with GraphPad Prism statistical software (GraphPad Software Inc. version 4.02, La Jolla, CA, USA) using a P value less than 0.05 as statistically significant.
Discussion
In order to better understand the local role of TNF and the effect of TNF soluble receptor based therapies on SG activity, we have stably expressed soluble TNFR1:IgG fusion protein locally in the SGs of NOD mice and followed stimulated saliva flow, sialadenitis, cytokine production, autoantibody levels, and insulin-dependent diabetes mellitus (IDDM) over 18 weeks. This local therapy resulted in decreased stimulated saliva flow and reduced Th1, Th2 and Th17 cytokine levels in the SG accompanied by increased levels of Th1 and Th2 cytokines in plasma.
The reduced SG activity after local TNF blockade suggests a protective role for TNF in SG function. TNF has been reported to exhibit anti-inflammatory activities, for instance by blocking the development of autoreactive T cells or by altering the balance of regulatory T cells [
10,
11]. Recently, the anti-inflammatory effects of TNF were reported to be mediated through TNFR2 in tolerogenic TGF-β treated antigen presenting cells (APC) and blocking of TNFR1 signaling enhanced the ability of APCs to secrete TGF-β in response to TGF-β exposure [
29]. Consistent with this notion, TNFR1:IgG treatment resulted in increased TGF-β1 tissue levels. The unaltered IFN-γ and MCP-1 levels suggest an IFN-γ independent pathway. In plasma we found an opposite pattern after TNFR1:IgG treatment, with increased levels of IL-4, IL-12p70 and, IFN-γ and decreased expression of TGF-β1. Of interest, IL-12p70 (together with IFN-γ) has been reported to be elevated in serum of SS patients [
30]. Moreover, recently we have reported salivary dysfunction in IL-12 transgenic mice, providing a new model of SS [
31]. The systemic decrease of TGF-β1 levels found in the present study is in agreement with the SS-like lymphoproliferation seen in TGF-β KO mice [
32]. Our data indicate that systemic upregulation of IL-12p70 and downregulation of TGF-β1 may play an important role in SS and these cytokines might promote SG dysfunction after anti-TNF treatment. The change in SG function after TNFR1:IgG treatment is in line with previous observations showing that under certain circumstances patients on anti-TNF therapy may develop additional autoimmune complications (for example, demyelinating events) [
33‐
36]. TNF blockers also have been shown to induce antinuclear antibodies (ANA) and antibodies directed against double stranded (ds) DNA in autoimmune diseases, such as RA, spondyloarthritis (SpA) and systemic lupus erythematosus (SLE) [
27,
28].
There was no statistically significant change in focal infiltration of the gland or autoantibody levels in our study. Similarly, we previously found that SG activity can be affected by the local expression of immunomodulatory proteins independent of the focus score [
16]. Additionally, phenotyping of the infiltrating lymphocytes showed a trend to increased positive cell counts for plasma cells, CD4+ and CD8+ T lymphocytes in TNFR1:IgG treated SGs. Our data suggest that i) expression of TNFR1:IgG does not impair the ability of lymphocytes to migrate to areas of inflammation and ii) there is no clear cut correlation between decreased saliva flow and the severity of inflammation within the glands. A large body of literature has determined that in rodents, AAV2 based vectors only trigger a minimal and largely transient immune response. Although, we have not specifically tested for the generation of anti-AAV2 antibodies or cytotoxic T lymphocytes (CTLs) in this study, the SG activity and infiltrate scores were very similar for both untreated mice and those cannulated with AAV2 LacZ vector, suggesting the AAV2 vector itself has minimal response. In other studies we have also used a vector that does not encode a transgene such as LacZ and found similar results (data not shown).
There is at present no generally accepted animal model of SS, which represents all the features of this condition. However, in addition to its utility in studying IDDM, the NOD mouse can also develop other autoimmune conditions such as gender- and age-specific mononuclear gland infiltrates and exocrine dysfunction of salivary and lacrimal glands that resemble Sjögren's syndrome [
37]. We have previously used this model successfully to demonstrate the stimulatory effect of IL-10 [
15] and vasoactive intestinal peptide gene transfer [
16] on SG activity. The development of autoimmune disease in these mice can, however, be affected by environmental and experimental conditions and is also dependent on the specific substrain of the NOD mice [
38‐
40]. Over the last two years we have maintained a colony of NOD mice under defined conditions that develop sialadenitis. Consistent with our recent observations [
38], the mice developed SG infiltrates and autoantibodies without developing a spontaneous loss of SG activity. Also, other NOD-derived strains have been reported as SS animal models, such as NOD.B10-H2b [
41] and C57BL/6.NOD-Aec1 Aec2 [
42]. Each of these strains has advantages compared with NOD/LTJ mice and as they become more widely studied, may be useful models for examining the long term effects of immunomodulatory proteins which is difficult in NOD/LTJ mice. The results presented here also show that local TNFR1:IgG treatment administered to the SG does not affect the incidence of diabetes in this model. Several studies have indicated that systemically increasing TNF levels can prevent the onset of IDDM in NOD mice [
11,
43]. Potential mechanisms include the effect of TNF on signal transduction through the T-Cell Receptor (TCR), including NF-κB expression, and/or a direct effect of TNF on T-cell apoptosis in the periphery [
11,
44]. On the other hand, systemic TNFR1:IgG treatment prevented diabetes in NOD mice [
45]. Taken together, the role of TNF in the pathogenesis of diabetes in NOD mice is still controversial [
43]. Conceivably, the systemic levels of TNFR1:IgG in our study were too low to influence the disease proves in the pancreas. In contrast to TNFR2, the type 1 TNF receptor can bind both TNF-α and lymphotoxin (LT)-α. The results shown in our experiments are likely the result of binding both cytokines. At this point, we are not able to distinguish between both effects and therefore cannot conclude that the effect on gland activity is caused by blocking TNF-α alone. Further mechanistic experiments are needed to explain the exact role of anti-TNF in SS.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
JLV and JAC designed the study. JLV, HY and NR acquired the data. JLV, HY, PPT and JAC analyzed, interpreted and evaluated the data. JLV, MRK, PPT and JAC prepared the manuscript.