SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Objective

To investigate a potential immunomodulatory effect of the 60-kd heat-shock protein (Hsp60) on experimental spontaneous Sjögren's syndrome (SS).

Methods

Seven-week-old nonobese diabetic (NOD) mice were immunized with eukaryotic Hsp60 or an Hsp60-derived peptide (amino acid residue [aa] 437–460). At 21 weeks of age, nondiabetic mice were investigated for salivary gland inflammation, exocrine function, and extraglandular disease manifestations. In addition, biomarker profiles comprising 87 analytes in serum and 75 in saliva were analyzed.

Results

In mice immunized with Hsp60 and aa 437–460, SS-related histopathologic features were significantly reduced compared with NOD controls. In addition, 50% of Hsp60-injected mice and 33% of aa 437–460–injected mice retained normal exocrine function. Both treatments induced similar changes in biomarker profiles. Notably, levels of circulating interferon-γ–inducible 10-kd protein (IP-10) and eotaxin were decreased significantly after treatment. Anti–type 3 muscarinic acetylcholine receptor (anti-M3R) IgG1, interleukin-10, and leptin discriminated best between the different treatment groups. Successful prevention of hyposalivation was accompanied by quantitative alterations in 36 biomarkers, of which 19 mediators of inflammation declined to levels comparable with those found in BALB/c mice. Low secreted vascular endothelial growth factor A was the most accurate predictor of successful prevention of hyposalivation. Low salivary granulocyte chemotactic protein 2 was identified as the best predictor of normal secretory function across the strains.

Conclusion

Immunization with Hsp60 and its peptide aa 437–460 led to inhibition of SS in NOD mice. Comprehensive analyses revealed specific biomarker signatures capable of predicting treatment group and treatment outcome. Molecules involved in inflammatory chemotaxis, neovascularization, and regulatory pathways caused the differences displayed by the biomarker profiles.

Certain autoantigens have been shown to exhibit remarkable capabilities for modulating experimental autoimmunity (1). Heat-shock proteins (HSPs), which possess such properties, are ubiquitously expressed chaperones, and also have been proven to be highly immunogenic (2). Exceptionally conserved throughout evolution, molecular mimicry between HSPs of microbial and mammalian origin has been suggested as a factor in several autoimmune and inflammatory conditions, such as rheumatoid arthritis (RA) (3) and atherosclerosis (4). Hsp60 has also been identified as the crucial component in bacillus Calmette-Guérin–mediated inhibition of type 1 diabetes mellitus (DM) in NOD mice (5). As a consequence, a peptide vaccine (DiaPep277, amino acids [aa] 437–460), based on eukaryotic Hsp60, has been developed for the treatment of type 1 DM and is currently being tested in phase II clinical trials (6). Previous characterization of the immune response induced by aa 437–460 showed that it closely mimicked the effects of whole Hsp60 regarding prevention of type 1 DM, induction of T cell proliferation, and behavior in adaptive transfer experiments and other functional analyses, whereas use of other peptides from Hsp60 has resulted in outcomes very similar to those obtained by vaccination without peptide (7, 8).

Therapeutic and preventive effects, irrespective of the specific autoimmune condition studied, have been attributed to the capacity of Hsp60 to trigger antiinflammatory and regulatory mechanisms. Through a process dependent on Toll-like receptor 2 (TLR-2), Hsp60 and aa 437–460 have been shown to alter inflammatory chemotaxis and down-regulate T cell migration in vitro (8, 9). Treg cells seem to be innately responsive to Hsp60, and to be more effective in down-regulating CD4 and CD8 effector responses after Hsp60 and aa 437–460 engagement (10). In addition, Hsp60 may regulate Th1/Th2-related transcription factors and cytokines (11). Research efforts have also expanded the focus from self-Hsp60–specific T cells to antigen-presenting cells, which may directly interact with endogenous Hsp60 through identified and unidentified cell-specific surface receptors (12).

Sjögren's syndrome (SS) is a systemic autoimmune disease, which affects ∼0.3–0.6% of the total population and is manifested by severe impairment of exocrine gland function and focal mononuclear cell infiltrates within the salivary and lacrimal glands (13, 14). Treatments used today provide merely marginal symptomatic relief (15). Previous studies have shown that anti–type 3 muscarinic acetylcholine receptor (anti-M3R) autoantibodies may potentially interfere with acinar cell innervation, assigning for the first time a defined pathogenetic role to an autoantibody in SS (16). The disease can involve organs other than the exocrine glands, and up to 5% of patients develop lymphoid malignancies. As in other rheumatic diseases, anti-Hsp60 antibodies have been found to be elevated in patients with SS (17). Nevertheless, HSPs (18) and regulatory mechanisms, such as Treg cells (19), have not been extensively investigated in SS. Importantly, however, it seems that the protective effects of HSPs are independent of their antigenic relationship with the (unfortunately often unknown) disease-causing antigen (20).

The NOD mouse is the best-characterized model of SS (21–23). It spontaneously manifests SS-like histopathologic features and hyposalivation, following a specific time course for the onset of the different SS-related disease manifestations (24). Although some genetic loci related to diabetes have been found to contribute to inflammatory changes in the exocrine glands, diabetes and SS-like disease can develop independently of each other (25). Most knowledge regarding immunostimulatory interventions in type 1 DM has been accumulated through study of the NOD strain (1). Whether potential therapeutic agents may, however, modify SS disease manifestations has not yet been addressed. Nevertheless, some SS-related findings were described in a previous report characterizing NOD mice transgenic for Hsp60 (26). Despite abundant thymic expression of Hsp60, resulting in diminished susceptibility to type 1 DM, salivary gland inflammation was found to be aggravated, and interestingly, T cell responses to Hsp60 were not abolished.

The aim of the present study was to investigate the immunomodulatory potential of Hsp60 and its peptide aa 437–460 in SS-like disease in nondiabetic NOD mice. In addition, comprehensive biomarker profiles were analyzed for systemic and local alterations related to the treatment received and treatment efficacy in preventing the onset of hyposalivation.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Animals.

Female NOD/LtJ mice (stock no. 001976) and BALB/cJ mice (stock no. 000651) were purchased from The Jackson Laboratory (Bar Harbor, ME) and maintained in sterilized, individually ventilated cages at the animal facility of the University of Bergen Department of Physiology. Mice were fed with autoclaved RM1 pellets (Special Diet Service, Witham, UK). All procedures were carried out in a laminar flow hood. The study was approved by the Committee for Research on Animals/Forsøksdyrutvalget (project 12-05/BBB). At 7 weeks of age, 32 female NOD mice (16 per group) were immunized subcutaneously with 50 μl of one of the following agents emulsified 1:1 in Freund's incomplete adjuvant (IFA): 50 μg low-endotoxin (<0.05 endotoxin units/μg) human recombinant Hsp60 (no. ESP-540G; Stressgen, Victoria, British Columbia, Canada) (27) or 100 μg of aa 437–460 (VLGGGVALLRVIPALDSLTPANED) synthesized using 9H-fluoren-9-ylmethoxycarbonyl–based solid-phase peptide synthesis (Invitrogen, San Diego, CA) with >95% purity (determined by high-performance liquid chromatography). As controls, 22 NOD mice and 20 BALB/c mice were immunized with phosphate buffered saline (PBS) emulsified in IFA.

Assessment of diabetes.

Urine glucose levels of >50 mg/dl (determined using Keto-Diabur-Test strips; Roche, Indianapolis, IN) on 2 consecutive measurements were considered to be evidence of onset of overt diabetes. This was confirmed by blood glucose levels of >300 mg/dl (Ascensia Microfill; Bayer Healthcare, Mishawaka, IN). At week 20 and week 21, all mice were screened for hyperglycemia.

Salivary secretion capacity.

Prior to salivary flow measurements, mice were fasted for a minimum of 5 hours with water ad libitum and subsequently anesthetized through intramuscular injection of Ketalar/Domitor (0.01 ml/gm body weight). Salivary flow, induced by intraperitoneal injection of 0.5 μg pilocarpine/gm body weight (P6503; Sigma, St. Louis, MO), was measured for 10 minutes. Preweighed tubes were weighed again after collection to determine the amount of saliva (1 μg = 1 μl). Protease inhibitor cocktail (P8340; Sigma) was added at a concentration of 1:500, and samples were kept at −80°C until analyzed. The salivary secretion rate is presented as microliters of saliva secreted per minute per gram of body weight.

Blood sampling, organ collection, and histopathologic analysis.

Blood, collected by heart puncture, was allowed to clot and centrifuged to obtain serum. Organs were fixed in 4% formalin prior to embedding in paraffin, sectioning, and staining with hematoxylin and eosin (H&E).

To determine the insulitis score, an average of 46 islets per mouse were scored, as described by Leiter (28), by an evaluator who was blinded with regard to treatment group.

In salivary glands, the focus score describing the frequency of inflammation (number of foci containing ≥50 cells/mm2 of glandular tissue) and the ratio index representing the loss of glandular epithelial tissue (square millimeter of inflamed area per square millimeter of glandular tissue) were determined as follows. Three independent H&E-stained sections of each salivary gland were qualitatively evaluated, and the section displaying the highest degree of inflammation was recorded as a multiple image composite, displaying the whole surface of the gland (magnification × 80). Total glandular area and the individual size of each focus were morphometrically analyzed. Sections of the kidneys, thyroid gland, thymus, heart, lung, liver, stomach, small and large intestine, appendix, and skin were also stained with H&E for basic histopathologic evaluation.

Multianalyte profiles.

Multianalyte profiles from serum (82 biomarkers) and saliva (75 biomarkers) were generated for each of the 36 nondiabetic NOD mice assessed for SS, and for 12 BALB/c mice, which were randomly selected from the original cohort of 20 PBS/IFA-injected BALB/c mice. For a list of the analytes used, see Supplementary Table 1, available on the Arthritis & Rheumatism Web site at http://www.mrw.interscience.wiley.com/suppmat/0004-3591/suppmat/. Results were obtained using a multiplex sandwich immunofluorescence assay based on color-coded and antibody-coated beads, carried out at Rules-Based Medicine (Austin, TX). Using an automated system for liquid handling, each sample was introduced into the capture microsphere multiplex of the multianalyte profile assay. After incubation, multiplexed cocktails of biotinylated reporter antibodies were added, developed with a streptavidin–phycoerythrin solution, and analyzed using a Luminex 100 instrument. For each multiplex, 8-point calibrators and 3-level controls were included. Detection antibodies recognizing autoantibodies were directed against all isotypes.

Quantification of anti-M3R antibodies.

M3R-transfected Chinese hamster ovary cells (pcDNA5/FRT/V5-His MsM3R-Flp-In cells [29]) were incubated with 10 μl of serum before addition of one of the following fluorescein isothiocyanate–conjugated goat anti-mouse antibodies (Southern Biotechnology, Birmingham, AL): isotype control, goat IgG (no. 0110-02), IgG (heavy and light) (no. 1031-02), IgG1 F(ab′)2 (no. 1072-02), IgG2b F(ab′)2 (no. 1092-02), IgG2c F(ab′)2 (no. 1079-02), and IgG3 F(ab′)2 (no. 1102-02). The cells were analyzed with a FACSCalibur flow cytometer using CellQuest software (BD Biosciences, San Jose, CA) and Flow Jo (Tree Star, San Carlos, CA) to compare the median fluorescence of the different populations.

Statistical analysis.

Means were compared with the specific reference group using one-way analysis of variance and Dunnett's post-test (2-tailed) to account for multiple group comparison. Since all pairs were compared, mean salivary flow was analyzed using the Bonferroni posttest. Prior to original and cross-validated (leave-one-out) group prediction, the discriminant potential of the respective variables was computed using discriminant analysis. Variables were entered simultaneously or in a forward stepwise manner, using Wiki's lambda as the inclusion/exclusion criterion. Relative risk (RR) and incidence were compared using Fisher's exact test (2-tailed). Statistical analyses were performed with SPSS 13 and Prism 4.0 for Mac OSX. Results are presented as the mean ± SEM.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Reduction of diabetes incidence and insulitis in nondiabetic NOD mice by immunization with Hsp60 or aa 437–460.

At 21 weeks of age, 10 (45.5%) of 22 NOD mice in the PBS/IFA-injected control group had diabetes, compared with 4 (25%) of 16 in the Hsp60-injected group and 3 (20%) of 15 in the aa 437–460–immunized group (Figure 1A). Due to the non-SS–related impact of hyperglycemia on the physiologic process of saliva secretion and possible distorting effects on the multianalyte profile, diabetic mice were excluded from all subsequent analyses. In addition, 1 aa 437–460–treated mouse was euthanized at 17 weeks of age due to poor health, and was found to have malignant Burkitt's lymphoma. The lymphoma had spread to all organs investigated and to the salivary glands (results are available online at http://www.uib.no/Broegelmann/docs/files/DelaleuN2008.pdf).

thumbnail image

Figure 1. Effect of treatment with phosphate buffered saline (PBS)/Freund's incomplete adjuvant (IFA), Hsp60, and amino acids 437–460 (aa 437–460) on A, incidence of diabetes in mice ages 10–21 weeks and B, the degree of pancreatic inflammation, measured by the insulitis score, in nondiabetic mice at 21 weeks. Bars in B show the mean and SEM. ∗∗ = P < 0.01. deSF = mice treated with either Hsp60 or aa 437–460 that had decreased salivary flow; reSF = mice treated with either Hsp60 or aa 437–460 that had retained salivary flow.

Download figure to PowerPoint

Consistent with the lower incidence of diabetes in both treated groups, the insulitis score in nondiabetic NOD mice was significantly decreased in the group immunized with Hsp60 (mean ± SEM 0.285 ± 0.033; n = 12) (P < 0.01) and in mice immunized with aa 437–460 (0.267 ± 0.027; n = 12) (P < 0.01), compared with PBS/IFA-injected NOD controls (0.454 ± 0.052; n = 12) (Figure 1B). These beneficial effects, previously described for type 1 DM (5), proved the general effectiveness of the treatment in our study per se, before the effect of Hsp60 and aa 437–460 on SS disease manifestations was assessed. BALB/c mice did not show any sign of insulitis, glucosuria, or hyperglycemia.

Overall decrease of salivary gland inflammation and partial retention of secretory function in mice treated with Hsp60 or aa 437–460.

All NOD mice developed focal mononuclear cell infiltrates in the salivary glands, whereas BALB/c mice were free of any sign of glandular inflammation. The mean ± SEM focus score in PBS/IFA-injected NOD controls was 1.007 ± 0.087 foci/mm2. In comparison, Hsp60 treatment led to a significant reduction in focus score (mean ± SEM 0.611 ± 0.085) (P < 0.01). Similarly, aa 437–460–immunized mice had significantly lower focus scores than did PBS/IFA-injected NOD control mice (0.708 ± 0.074) (P < 0.05) (Figure 2A). The ratio index in both treatment groups followed the same trend, but the differences did not reach statistical significance (Figure 2A).

thumbnail image

Figure 2. Effect of treatment with PBS/IFA, Hsp60, and aa 437–460 on A, salivary gland inflammation, measured by focus score (solid and hatched black bars) and ratio index (solid and hatched gray bars), B, salivary secretion capacity, and C, salivary flow rates shown according to treatment and prevention of hyposalivation, in NOD mice. BALB/c mice were used as controls. Bars in A show the mean and SEM. In B and C, values in 12 mice per group are shown; the horizontal and vertical lines show the mean ± SEM. The dotted line in C represents the threshold for retained salivary flow. ∗ = P < 0.05; ∗∗ = P < 0.01; ∗∗∗ = P < 0.001. NS = not significant (see Figure 1 for other definitions).

Download figure to PowerPoint

At 21 weeks of age, salivary secretion in PBS/IFA-injected NOD controls (n = 12) was decreased by 42% (mean ± SEM 0.367 ± 0.026 μl/minute/gm) compared with BALB/c mice (0.637 ± 0.024 μl/minute/gm; n = 12) (P < 0.001). Mean salivary flow was higher in Hsp60-immunized mice (0.429 ± 0.046 μl/minute/gm; n = 12) and in aa 437–460–injected mice (0.369 ± 0.039 μl/minute/gm; n = 12) than in NOD controls (Figure 2B). Nevertheless, we observed an obvious dichotomy regarding salivary secretion capacity in both treated groups. Mice within each group could be separated by using the highest salivary flow rate measured in PBS/IFA-injected NOD controls (0.459 μl/minute/gm) as the threshold (Figure 2C). Notably, mice immunized with Hsp60 or aa 437–460 that had salivary flow rates above the threshold did not have significantly lower salivary flow rates than BALB/c mice (P > 0.05) (Figure 2C).

The 6 Hsp60-treated NOD mice and the 4 aa 437–460–treated NOD mice with salivary flow rates above the threshold were referred to as treated mice with retained salivary flow. The 6 Hsp60-treated NOD mice and the 8 aa 437–460–treated NOD mice with salivary flow rates below the threshold were referred to as treated mice with decreased salivary flow. Importantly, treated mice with retained salivary flow and treated mice with decreased salivary flow were very similar with respect to insulitis score, focus score, and ratio index (Figures 1B and 2A).

Treatment with Hsp60 or aa 437–460 does not induce significant amelioration or aggravation of extraglandular disease manifestations.

SS disease may involve organs other than the exocrine glands. Representative sections of the thyroid gland, thymus, heart, liver, gastrointestinal tract, and skin exhibited no apparent changes reminiscent of SS extraglandular disease manifestations or other autoimmune diseases. The observed slight increase in average aortic wall thickness in NOD mice compared with BALB/c mice was less pronounced in the groups immunized with Hsp60 or aa 437–460. (Data available online at http://www.uib.no/Broegelmann/docs/files/DelaleuN2008.pdf.) Signs of vasculitis or atherosclerotic plaque were not found in any mouse.

Relative risk of focal lymphoid infiltrates in the kidneys (>50 lymphocytes) was increased in Hsp60-treated mice (present in 9 of 12 mice) and in aa 437–460–treated mice (present in 9 of 12 mice), compared with PBS/IFA-injected NOD controls (present in 4 of 12 mice). Relative risk of the occurrence of hyaline casts was also increased in mice treated with Hsp60 (RR 2.25) or aa 437–460 (RR 1.29) (Data available online at http://www.uib.no/Broegelmann/docs/files/DelaleuN2008.pdf.). Of 12 PBS/IFA-injected NOD controls, 11 exhibited foamy macrophages in the lungs, a phenomenon that occurred less frequently in the mice treated with Hsp60 (9 of 12 mice) and those treated with aa 437–460 (8 of 12 mice). In contrast, the frequency of lymphoid cell infiltrates in the lungs was constant across all groups of NOD mice (present in 3 mice per group). However, none of these differences in frequency were statistically significant (P > 0.05 by Fisher's exact test).

Alterations in biomarker profiles classify NOD mice according to the specific treatment received.

The multianalyte profile had a strong focus on proteins involved in processes previously identified in in vitro experiments to be sensitive to modulation by Hsp60 and aa 437–460. In order to detect such effects in vivo, quantities of 87 mediators of inflammation in serum and 75 in saliva (see Supplementary Table 1, available on the Arthritis & Rheumatism Web site at http://www.mrw.interscience.wiley.com/suppmat/0004-3591/suppmat/) were measured individually in all mice assessed for SS (n = 12 mice per group). Circulating interferon-γ–inducible 10-kd protein (IP-10) and eotaxin were significantly lower in both treatment groups compared with PBS/IFA-injected NOD controls (Table 1). Serum monocyte chemotactic protein 1 (MCP-1), MCP-3, interleukin-10 (IL-10), and leptin, and salivary IL-2 followed the same trend in terms of findings in the Hsp60-treated group and in the aa 437–460 group relative to the PBS/IFA-injected NOD controls. Differences from controls, however, were statistically significant only in mice treated with aa 437–460. In contrast, levels of anti-M3R IgG1 and antimitochondrial antibodies in serum were significantly increased in mice immunized with Hsp60, compared with PBS/IFA-injected NOD controls. Circulating serum amyloid P was also significantly higher in Hsp60-immunized mice compared with PBS/IFA-injected NOD controls, whereas serum levels of growth hormone were significantly increased in aa 437–460–injected mice only.

Table 1. Significant alterations in biomarker profiles in mice treated with PBS/IFA, Hsp60, or aa 437–460, irrespective of treatment outcome regarding hyposalivation*
 PBS/IFAHsp60Paa 437–460P
  • *

    Values are the mean ± SEM (% of expression in mice injected with phosphate buffered saline [PBS]/Freund's incomplete adjuvant [IFA]). Biomarker quantities from control NOD mice injected with PBS/IFA (n = 12) were compared with the levels detected in mice treated with Hsp60 (n = 12) and mice treated with amino acids 437–460 (aa 437–460) (n = 12). Data were analyzed using one-way analysis of variance combined with 2-tailed Dunnett's post-test. IP-10 = interferon-γ–inducible 10-kd protein; MCP-1 = monocyte chemotactic protein 1; IL-10 = interleukin-10; SAP = serum amyloid P; anti-M3R = anti–type 3 muscarinic acetylcholine receptor.

  • Versus PBS/IFA-treated mice. Significance was set at P < 0.05.

Serum     
 IP-10 (CXCL10), pg/ml71.25 ± 10.344.67 ± 5.58 (63)0.02041.42 ± 2.79 (58)0.009
 MCP-1 (CCL2), pg/ml125.2 ± 13.095.33 ± 10.0 (76)0.07985.08 ± 5.94 (68)0.015
 MCP-3 (CCL7), pg/ml288.5 ± 34.0210.3 ± 22.9 (73)0.069190.2 ± 16.9 (66)0.020
 Eotaxin (CCL11), pg/ml947.4 ± 58.5768.6 ± 33.4 (81)0.027786.1 ± 51.8 (83)0.048
 IL-10, pg/ml441.8 ± 14.2407.6 ± 12.4 (92)0.139369.9 ± 13.2 (84)0.001
 Leptin, pg/ml607.4 ± 113421.4 ± 70.0 (69)0.227277.6 ± 64.4 (46)0.019
 SAP, μg/ml25.00 ± 0.9429.25 ± 1.21 (117)0.01927.33 ± 1.14 (109)0.246
 Growth hormone, ng/ml63.62 ± 9.3165.75 ± 6.94 (103)0.990107.8 ± 19.0 (169)0.038
 Anti-M3R IgG1, units4.435 ± 0.4813.80 ± 1.22 (311)<0.0014.234 ± 0.43 (95)0.977
 Antimitochondrial  antibody, units3.467 ± 0.607.833 ± 1.59 (226)0.0083.558 ± 0.35 (103)0.997
Saliva     
 IL-2, pg/ml14.61 ± 2.8522.92 ± 3.35 (157)0.14225.58 ± 3.54 (175)0.043

To increase understanding of the observed alterations in the multianalyte profiles described above, discriminant analyses were performed to predict treatment group membership based on a linear combination of variables. Discriminant analyses including all variables that were significantly altered as a consequence of treatment classified 97.2% of all cases according to the treatment received. However, using cross-validated prediction, which is based on all cases except the given case and is thought to give a better estimate of the classification accuracy in the population than an original nonvalidated classification, the model was not found to be very reliable in predicting treatment group membership (cross-validated hit rate 75%). Simplification of the model by stepwise forward discriminant analyses, assessing the importance of each variable, led to a more reliable model, which included only IL-10, anti-M3R IgG1, and leptin as predictors of treatment group membership (original and cross-validated hit rates 86.1%) (Figure 3).

thumbnail image

Figure 3. Potential of interleukin-10 (IL-10), anti–type 3 muscarinic acetylcholine receptor (anti-M3R) IgG1, and leptin measured in serum to predict treatment group (IFA, Hsp60, or aa 437–460). All variables that were significantly altered in NOD mice after treatment with either Hsp60 or aa 437–460 were entered in a forward stepwise manner into the discriminant analysis. Wiki's lambda was used as the inclusion/exclusion criterion, resulting in a model consisting of IL-10, anti-M3R IgG1, and leptin as the best predictors of treatment group (original and cross-validated hit rates 86.1%). The x-, y-, and z-axes show the discriminant score. Circles represent individual mice. See Figure 1 for other definitions.

Download figure to PowerPoint

Distinct biomarker signatures in treated mice with retained salivary flow.

The observed dichotomy within the treatment groups regarding prevention of hyposalivation enabled the delineation of a biomarker signature strictly related to salivary secretion capacity. For this purpose, biomarker profiles from treated mice with retained salivary flow (n = 6 Hsp60-treated mice and 4 aa 437–460–treated mice) were compared with treated mice with decreased salivary flow (n = 6 Hsp60-treated mice and 8 aa 437–460–treated mice). Levels of 36 proteins were significantly altered between these 2 groups (Table 2). (For additional data, see Supplementary Figure 1, available on the Arthritis & Rheumatism Web site at http://www.mrw.interscience.wiley.com/suppmat/0004-3591/suppmat/.)

Table 2. Significant alterations in biomarker profiles associated with the prevention of hyposalivation in treated mice with retained salivary flow, treated mice with decreased salivary flow, and control BALB/c mice*
 Retained salivary flowDecreased salivary flowPBALB/cP
  • *

    Values are the mean ± SEM (% of expression in treated mice with retained salivary flow). Treated mice with retained salivary flow were mice treated with either Hsp60 or aa 437–460 that had salivary flow rates >0.459 μl/minute/gm. Treated mice with decreased salivary flow were mice treated with either Hsp60 or aa 437–460 that had salivary flow rates <0.459 μl/minute/gm. Biomarker levels in treated mice with retained salivary flow (n = 10) were compared with the levels measured in treated mice with decreased salivary flow (n = 14) and control BALB/c mice (n = 12). Data were analyzed using one-way analysis of variance combined with Dunnett's post-test. SGOT = serum glutamic oxaloacetic transaminase; VCAM = vascular cell adhesion molecule 1; Apo A-I = apolipoprotein A-I; MGSA = melanoma growth-stimulatory activity protein; GCP-2 = granulocyte chemotactic protein 2; MIP-2 = macrophage inflammatory protein 2; MDC = macrophage-derived chemokine; TNFα = tumor necrosis factor α; M-CSF = macrophage colony-stimulating factor; MMP-9 = matrix metalloproteinase 9; TIMP-1 = tissue inhibitor of metalloproteinases 1; ND = not detectable; MPO = myeloperoxidase; Lptn = lymphotactin; OSM = oncostatin M; FGF-9 = fibroblast growth factor 9; TPO = thrombopoietin; VEGF-A = vascular endothelial growth factor A (see Table 1 for other definitions).

  • Versus treated mice with retained salivary flow. Significance was set at P < 0.05.

Serum     
 IL-1α, pg/ml746.9 ± 23.6576.5 ± 37.5 (77)0.007959.1 ± 45.2 (128)0.001
 Haptoglobin, μg/ml23.10 ± 1.0820.14 ± 0.59 (87)0.03118.92 ± 0.87 (82)0.003
 SGOT, μg/ml28.70 ± 2.0820.86 ± 1.55 (73)0.00215.39 ± 1.00 (54)<0.001
 VCAM-1, ng/ml1,482 ± 35.01,359 ± 38.7 (92)0.043965.9 ± 32.6 (65)<0.001
 Apo A-1, μg/ml70.70 ± 3.0560.50 ± 2.52 (86)0.01152.58 ± 1.55 (74)<0.001
 Clusterin, μg/ml227.0 ± 5.75204.4 ± 4.72 (90)0.008202.1 ± 5.07 (89)0.005
Saliva     
 MGSA (CXCL1), pg/ml66.30 ± 10.9146.2 ± 29.2 (221)0.02433.50 ± 9.54 (51)0.474
 GCP-2 (CXCL5), pg/ml280.0 ± 34.0588.5 ± 57.7 (210)<0.001271.3 ± 27.6 (97)0.986
 MIP-2 (CXCL2), pg/ml7.800 ± 1.4218.81 ± 2.41 (241)<0.0016.442 ± 1.07 (83)0.836
 MCP-1 (CCL2), pg/ml4.510 ± 1.2610.31 ± 1.65 (229)0.0081.067 ± 0.58 (24)0.148
 MDC (CCL22), pg/ml13.73 ± 2.9633.79 ± 3.73 (246)<0.0017.617 ± 2.44 (55)0.338
 MIP-3β (CCL19), pg/ml9.400 ± 4.8536.90 ± 4.69 (393)0.0018.600 ± 5.02 (91)0.990
 IL-1α, pg/ml185.5 ± 14.5358.6 ± 29.4 (193)0.003298.3 ± 47.7 (161)0.063
 IL-1β, pg/ml62.00 ± 32.0186.7 ± 44.7 (301)0.04850.00 ± 27.6 (81)0.965
 IL-10, pg/ml172.7 ± 7.54261.1 ± 21.3 (151)0.004153.3 ± 19.6 (89)0.699
 IL-11, pg/ml10.80 ± 3.2025.07 ± 4.00 (232)0.0060.917 ± 0.92 (8)0.075
 TNFα, pg/ml1.300 ± 1.3017.20 ± 5.40 (1,323)0.010ND0.958
 CD40, pg/ml4.410 ± 1.0510.61 ± 1.97 (241)0.0101.292 ± 0.71 (29)0.246
 M-CSF (CSF-1), pg/ml57.00 ± 9.16111.1 ± 13.4 (195)0.00452.80 ± 9.04 (93)0.952
 VCAM-1, pg/ml138.0 ± 28.7309.2 ± 29.9 (224)0.001146.8 ± 28.1 (106)0.968
 MMP-9, ng/ml41.30 ± 10.186.54 ± 11.4 (210)0.00640.50 ± 7.02 (98)0.998
 TIMP-1, pg/ml120.5 ± 14.4238.6 ± 27.1 (198)0.00199.40 ± 14.2 (82)0.721
 Leptin, pg/mlND5.300 ± 2.010.020ND1.000
 MPO, ng/ml39.20 ± 6.31151.6 ± 44.3 (387)0.02139.92 ± 8.43 (102)1.000
 Anti–Scl-70, units0.926 ± 0.031.193 ± 0.16 (129)0.0330.869 ± 0.01 (94)0.765
 IP-10 (CXCL10), pg/ml40.70 ± 4.8568.00 ± 7.90 (167)0.0068.808 ± 1.93 (22)0.002
 Lptn (XCL1), pg/ml29.46 ± 5.4644.87 ± 5.08 (152)0.04711.32 ± 2.85 (38)0.023
 MCP-3 (CCL7), pg/ml13.51 ± 2.2921.94 ± 2.30 (162)0.0114.075 ± 1.12 (30)0.006
 MIP-1β (CCL4), pg/ml56.80 ± 10.999.64 ± 11.5 (175)0.01017.75 ± 6.21 (31)0.024
 IL-7, pg/ml53.40 ± 10.288.20 ± 10.3 (165)0.02418.80 ± 6.30 (35)0.030
 OSM, pg/ml43.50 ± 9.6379.60 ± 11.5 (183)0.0198.300 ± 3.39 (19)0.027
 FGF-9, ng/ml2.070 ± 0.302.936 ± 0.27 (142)0.0370.603 ± 0.15 (29)0.001
 TPO, ng/ml1.690 ± 0.252.761 ± 0.33 (163)0.0180.398 ± 0.18 (24)0.006
 VEGF-A, ng/ml0.725 ± 0.061.413 ± 0.09 (195)0.0051.500 ± 0.22 (207)0.002
 Haptoglobin, μg/ml2.810 ± 0.301.628 ± 0.22 (58)0.0121.495 ± 0.32 (53)0.005
 Myoglobin, pg/mlND140.0 ± 28.30.003135.0 ± 35.70.005

Saliva from treated mice with retained salivary flow contained significantly lower quantities of multiple chemokines (such as melanoma growth-stimulatory activity protein, granulocyte chemotactic protein 2, macrophage inflammatory protein 2 [MIP-2], MCP-1, macrophage-derived chemokine [MDC], MIP-3β, IP-10, lymphotactin, MCP-3, and MIP-1β), cytokines and cytokine receptors (such as IL-1α, IL-1β, IL-10, IL-11, tumor necrosis factor α [TNFα], CD40, IL-7, and oncostatin M [OSM]), growth factors (such as macrophage colony-stimulating factor [M-CSF], fibroblast growth factor 9 [FGF-9], thrombopoietin [TPO], and vascular endothelial growth factor A [VEGF-A]), and proteins belonging to other protein families (such as vascular cell adhesion molecule 1 [VCAM-1], matrix metalloproteinase 9 [MMP-9], tissue inhibitor of metalloproteinases 1 [TIMP-1], leptin, myeloperoxidase [MPO], anti–Scl-70, and myoglobin). Importantly, levels of 19 of these analytes detected in lower quantities in treated mice with retained salivary flow were not significantly different from the levels detected in BALB/c mice (P > 0.05), indicating a normalization of the quantity of these inflammation mediators in parallel with protection from hyposalivation (Table 2 and Supplementary Figure 1A).

Serum IL-1α levels in treated mice with retained salivary flow were closer to the levels found in BALB/c mice than were those found in treated mice with reduced salivary flow. In contrast, levels of haptoglobin in serum and saliva and circulating serum glutamic oxaloacetic transaminase (SGOT), as well serum VCAM-1, were significantly higher in treated mice with retained salivary flow than in treated mice with reduced salivary flow or in BALB/c mice. The same was true of the antiinflammatory apolipoproteins apolipoprotein A-I (Apo A-I) and clusterin (Table 2).

Discriminant analyses were performed to identify the biomarkers that could most accurately predict treatment effectiveness regarding prevention of hyposalivation. A low level of salivary VEGF-A was the best individual predictor of successful treatment, whereas high VEGF-A levels indicated onset of hyposalivation despite treatment with Hsp60 or aa 437–460 (original and cross-validated hit rates 87.5%). Expanding the model by including circulating Apo A-I and secreted M-CSF further improved prediction accuracy (original hit rate 91.7%; cross-validated hit rate 95.8%) (Figure 4A).

thumbnail image

Figure 4. Potential of biomarker signatures to predict treatment outcome regarding hyposalivation and normal or impaired salivary secretion capacity in BALB/c and NOD mice, irrespective of the treatment they received. All biomarkers found in significantly different levels in treated mice with retained salivary flow versus treated mice with decreased salivary flow were entered in a forward stepwise manner into the discriminant analysis. Wiki's lambda was used as the inclusion/exclusion criterion. A, Potential of vascular endothelial growth factor A (VEGF-A) and macrophage colony-stimulating factor (M-CSF) measured in saliva and apolipoprotein A-I (Apo A-I) measured in serum to predict prevention of hyposalivation onset in mice treated with Hsp60 and aa 437–460 (original hit rate 91.7%; cross-validated hit rate 95.8%). The x-, y-, and z-axes show the discriminant score. B, Potential of a model combining granulocyte chemotactic protein 2 measured in saliva, interleukin-1α (IL-1α) measured in serum, and myeloperoxidase (MPO), myoglobin, and macrophage inflammatory protein 3β (MIP-3β) measured in saliva to predict normal salivary secretion irrespective of treatment and mouse strain (original and cross-validated hit rates 93.8%). The x-axis shows the salivary flow rate in microliters per minute per gram; the y-axis shows the discriminant score of salivary granulocyte chemotactic protein 2, circulating IL-1α, and salivary MPO, myoglobin, and MIP-3β. Circles represent individual mice. See Figure 1 for other definitions.

Download figure to PowerPoint

Discriminant analyses were also performed to identify a set of biomarkers with the highest potential to discriminate between normal and impaired salivary secretion capacity, irrespective of treatment group and mouse strain. Granulocyte chemotactic protein 2 in saliva was identified as the most accurate single predictor (original and cross-validated hit rates 81.3%). A forward stepwise model combining salivary granulocyte chemotactic protein 2 with circulating IL-1α and salivary MPO, myoglobin, and MIP-3β resulted in accurate classification of 93.8% of the cases, using original and cross-validated predictions (Figure 4B).

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

The present report describes the first preventive antigen-specific intervention to successfully target experimental spontaneous SS. Although the antigenic relationships between Hsp60 and antigens specifically related to SS are unknown, our results indicate that immune responses to Hsp60 are relevant in the control of pathogenic autoimmunity. Consistent with the notion that Hsp60 and aa 437–460 treatment can alter chemotaxis of T cells in vitro (8, 9), decreased inflammation was associated with lower serum levels of chemoattractants for Th1 (IP-10) and Th2 cells (eotaxin) (30), even 14 weeks after immunization. In addition, levels of MCP-1 and MCP-3, to which monocytes and plasmacytoid dendritic cells are most responsive (30), were also reduced in serum after treatment.

Disease severity in RA, and more recently in juvenile idiopathic arthritis, has been shown to be related to the extent of Hsp60-induced propagation and activation of Treg cells (31, 32). It has been suggested that IL-10, secreted upon Hsp60 signaling through TLR-2, is a key molecule involved in this process (10). Our findings do not support the notion of increased IL-10 secretion by Treg cells in favor of immunosuppression. However, such alterations may be detectable only in specific microenvironments. Nevertheless, it should be noted that systemic overexpression of IL-10 has been shown to induce SS in mice (33). In contrast, in the present study, IL-2, which is crucial in maintaining immunologic self tolerance by promoting growth and function of Treg cells (34), was increased in saliva upon treatment with Hsp60 or aa 437–460.

Compared with IL-2, leptin exerts a reciprocal effect on effector and Treg cell populations (35). The observed modulation of leptin as a consequence of treatment therefore further supports the idea of strengthened regulatory pathways. Indeed, induction of regulatory processes might be the key to controlling pathogenic autoimmunity while avoiding long-term immunosuppression. In addition to Treg cells, IL-17 cells have received considerable attention in the study of autoimmune diseases (36). Nevertheless, our treatment regimens were not accompanied by detectable alterations in IL-17 in saliva or in serum (data not shown). In our study, induction of anti-M3R IgG1 production was strongly related to immunization with Hsp60. However, although IgG1 has previously been reported to be the anti-M3R isotype that can mediate hyposalivation (37), in our study the induction of the anti-M3R IgG1 isotype, generally considered to be an antiinflammatory isotype (38), did not seem to be related to manifestation of or protection against hyposalivation.

Prevention of hyposalivation, the other hallmark of SS, was achieved in a substantial number of mice for the duration of the experiment. Further studies are needed to determine whether NOD mice can be permanently protected against development of hyposalivation and whether such treatments might also be able to induce remission of SS. Refinement of treatment regimens may further increase the percentage of mice protected against hyposalivation, as has been shown recently in studies of diabetes using tandem repeats of aa 437–460 (39).

Interestingly, effective protection against hyposalivation was associated almost exclusively with changes in the quality of inflammation reflected by the multianalyte profiles. At the same time, we found hyposalivation to be unrelated to the quantitative degree of glandular inflammation. With focus on immune mediators, multianalyte profiles generated an overview, although it was not all-embracing, of the serum and salivary proteome. However, unlike in gene expression profiling, we could exclude distorting effects, such as RNA stability and poor correlation between messenger RNA and protein levels (40), from our analyses. Hu et al (40) used mass spectrometry to compare the salivary proteome from SS patients versus healthy controls. Not unexpectedly, of the 42 proteins reported in that study, none was included in our multianalyte profile, which shows the differential focus and strengths of the 2 methods (41, 42).

We identified low levels of VEGF-A, a key molecule mediating vascularization, as a primary predictor of prevention of hyposalivation, followed by low levels of M-CSF, which, like TPO and OSM, is an inducer of VEGF-A (43). Other molecules implicated in the process of neovascularization that were decreased in saliva in treated mice in which hyposalivation was prevented were MMP-9 and FGF-9 (44), as well as all CXCR2 ligands measured (granulocyte chemotactic protein 2, melanoma growth-stimulatory activity protein, and MIP-2) (30). These results support the notion of a strong interrelationship between pathogenic neovascularization and impaired secretory function. The importance of neovascularization in RA is recognized, and it has been explored as a target for therapeutic interventions (45). Unfortunately, the issue of pathogenic neovascularization in SS has not yet been addressed.

The salivary multianalyte profile associated with retained salivary flow in NOD mice resulted from the decrease and normalization of multiple chemokine levels. Such modulation was potentially promoted by the simultaneous decrease in levels of TNFα and IL-7, both key antagonists of Treg cell–induced immunoregulation (46). Down-regulation of CD40, IL-10, and IL-11 (47) may further suggest that Th2-associated responses were modulated by treatment. An effect of Hsp60 on B cells through innate signaling pathways has been reported (48) and may also contribute to the modulation of autoimmune diseases with a confirmed or potential B cell aspect. Nevertheless, in our study the levels of most autoantibodies remained unaffected. Similarly, we did not identify significant alterations in extraglandular disease manifestation as a consequence of the 2 treatments investigated. The development of malignancy in one mouse remains difficult to interpret, since lymphomas develop rather frequently in aged NOD mice (49). Nevertheless, antiinflammatory treatment should be critically reviewed regarding alterations induced, which might critically affect antitumor-related or host-defense–related immunity.

One other encouraging aspect of our results regards the capacity of granulocyte chemotactic protein 2, which, individually or combined with 4 other analytes, accurately predicted impaired salivary flow irrespective of treatment group and mouse strain. Validation of a biomarker-based diagnostic criterion in humans would indeed represent a significant advance in SS diagnosis and allow close followup of disease progression and the effects of therapeutic interventions.

In conclusion, our results indicated that Hsp60 vaccination may have a preventive effect in experimental SS. In addition, we found that the Hsp60 peptide aa 437–460 had, in general, the same beneficial effects as did whole Hsp60. Biomarker profiles indicated that down-regulation of inflammatory chemotaxis in parallel with strengthened regulatory and antiinflammatory mechanisms were major consequences of the treatment. Successful prevention of hyposalivation was related to a significant decrease in mediators of inflammation related to pathologic neovascularization, inflammatory chemotaxis, and cell activation. The multianalyte profile was also considerably successful in predicting treatment effectiveness. These biomarkers, in addition to the analytes indicative of normal and impaired salivary flow across the strains, should be explored further regarding their diagnostic significance in humans. In addition, the processes we identified to be involved in the onset of hyposalivation should be considered in the development of new therapeutic strategies.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Dr. Delaleu had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Delaleu, Jonsson.

Acquisition of data. Delaleu, Madureira, Immervoll.

Analysis and interpretation of data. Delaleu, Immervoll, Jonsson.

Manuscript preparation. Delaleu, Jonsson.

Statistical analysis. Delaleu.

PhD supervision. Jonsson.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

We thank Gry Bernes for excellent technical assistance, Associate Professor Ellen Berggreen for valuable advice, Dr. B. Delaleu-Justitz for careful revision of the manuscript, and Janet Cornelius and Dr. S. Küster for their support.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
art_23656_sm_Figure1.doc416KSupplementary Figure 1
art_23656_sm_Table1.doc54KSupplementary Table 1. Analytes included in the multi-analyte profile

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.