Sjögren's syndrome is a chronic illness manifested characteristically by immune injury to the salivary and lacrimal glands, resulting in dry mouth/eyes. Anti-Ro [Sjögren's syndrome antigen A (SSA)] and anti-La [Sjögren's syndrome antigen B (SSB)] autoantibodies are found frequently in Sjögren's subjects as well as in individuals who will go on to develop the disease. Immunization of BALB/c mice with Ro60 peptides results in epitope spreading with anti-Ro and anti-La along with lymphocyte infiltration of salivary glands similar to human Sjögren's. In addition, these animals have poor salivary function/low saliva volume. In this study, we examined whether Ro-peptide immunization produces a Sjögren's-like illness in other strains of mice. BALB/c, DBA-2, PL/J, SJL/J and C57BL/6 mice were immunized with Ro60 peptide-274. Sera from these mice were studied by immunoblot and enzyme-linked immunosorbent assay for autoantibodies. Timed salivary flow was determined after pharmacological stimulation, and salivary glands were examined pathologically. We found that SJL/J mice had no immune response to the peptide from Ro60, while C57BL/6 mice produced antibodies that bound the peptide but had no epitope spreading. PL/J mice had epitope spreading to other structures of Ro60 as well as to La, but like C57BL/6 and SJL/J had no salivary gland lymphocytic infiltration and no decrement of salivary function. DBA-2 and BALB/c mice had infiltration but only BALB/c had decreased salivary function. The immunological processes leading to a Sjögren's-like illness after Ro-peptide immunization were interrupted in a stepwise fashion in these differing mice strains. These data suggest that this is a model of preclinical disease with genetic control for epitope spreading, lymphocytic infiltration and glandular dysfunction.
Diseases are considered autoimmune on the basis of circulating antibodies that bind self as well as immune cell infiltration in affected organs. Autoimmunity, in the form of circulating autoantibodies, precedes autoimmune disease in both humans and animal models in every disease that has been studied prior to the onset of clinical illness [1, 2]. In addition, inflammatory infiltration may occur prior to clinical illness. For example, in humans with type 1 diabetes mellitus, as well as the non-obese diabetic (NOD) mouse, the pancreatic islets have infiltrating lymphocytes prior to the loss of beta cell function [3, 4]. The process and mechanism by which preclinical, perhaps benign, autoimmunity progresses to a harmful stage in which loss of function leads eventually to the onset of illness is not well explained. However, a failed series of immune check-points is considered a viable hypothesis and suggests a stepwise progression from initiation of benign autoimmunity to disease .
Sjögren's syndrome is a common, chronic autoimmune disorder in which exocrine organs such as the salivary and lacrimal glands are the targets of immune injury. Dry eye and dry mouth are the common clinical symptoms that lead to the diagnosis, but there are several other manifestations of Sjögren's syndrome, including lung and kidney disease as well as vasculitis . Criteria for classification for research purposes have been agreed upon , while disease activity and damage indices are not available . This illness may be one of the most common of the rheumatic autoimmune diseases . Anti-Ro [or Sjögren's syndrome antigen A (SSA)] and anti-La [or Sjögren's syndrome antigen B (SSB)] are autoantibodies found commonly in the circulation of patients with Sjögren's syndrome . In addition, these antibodies appear in the serum of mothers who give birth to babies with congenital complete heart block . Many of these mothers are healthy but go on to develop Sjögren's syndrome, in some cases many years later [11-13]. Thus, there is a preclinical period of autoimmunity in human Sjögren's syndrome.
There are several animal models of Sjögren's syndrome . These include the NOD mouse as well as the MRL-lpr/lpr mouse. In the former, low salivary flow can be transferred with immunoglobulin . The latter also has a lupus-like illness but has no salivary gland dysfunction . Several other animal models of Sjögren's syndrome have been reported, including an acute sialoadenitis occurring in a graft-versus-host transplant model , mice transgenic for the envelope protein of the hepatitis C , NFS/sld mice bearing a mutation resulting in sublingual gland differentiation arrest , mice homozygous for the almphoplasia (aly)  and the congenic C57BL/6.NOD-Aec1Aec2 mouse strain NOD model of autoimmune exocrinopathy . Fleck and colleagues reported that infection of C57BL/6-lpr/lpr mice with murine cytomegalovirus infection have a resultant acute and chronic sialoadenitis . This inflammation persisted after clearance of the virus. High levels of anti-Ro (or SSA), anti-La (or SSB), rheumatoid factor and anti-dsDNA were produced. A subsequent study of these mice indicated that introduction of the Fas ligand ameliorates diseases . Another infection-related model is that of NZM2338 mice infected with murine cytomegalovirus (CMV) .
Immunization of mice with components of the Ro/La particle has been performed by several groups (reviewed in ) with reports of epitope spreading . We have immunized mice with short peptides from the sequence of the 60 kD Ro antigen and found that BALB/c mice develop an illness similar to human Sjögren's syndrome . Mice so immunized develop autoantibodies such that the entire Ro ribonucleoprotein particle becomes the target of the immune system , and also develop salivary gland lymphocytic infiltrates and salivary gland dysfunction . We have previously reported epitope spreading, or lack of this finding, in various strains of inbred mice immunized in this manner, but did not assess the development of a Sjögren's-like illness by determining salivary gland pathology and function in this previous study .
We undertook the present study to determine whether Ro60 peptide immunization of different mouse strains produces differences in development of the Sjögren's-like illness. We find that when immunized with Ro60 peptides there are stages at which the development of the disease is interrupted, ranging from immune response to the peptide, epitope spreading and systemic autoimmunity and lymphocytic infiltration of the salivary glands to dysfunction of the gland. Evaluation of the pathogenesis and genetics at each of these check-points in the different strains may illuminate pathogenesis, including genetic pathways important for the progression from preclinical autoimmunity to overt illness.
Mice were purchased from the Jackson Laboratories (Bar Harbor, ME, USA). BALB/c mice were raised in our local colony, the founders of which were acquired from Jackson. All animals were maintained in a specific pathogen-free environment at an accredited animal care facility at Oklahoma Medical Research Foundation (OMRF). The protocol was approved by the OMRF and the University of Oklahoma Health Sciences Center (OUHSC) Institutional Animal Care and Use Committee (IACUC). For muscarinic acetylcholine M3 receptor functional studies, male BALB/c mice (8–10 weeks of age) were sourced from the Flinders University Animal Facility. All functional assay protocols were approved by the Flinders University Animal Welfare Committee.
Fourteen mice, each belonging to BALB/c, DBA-2, SJL/J, C57Bl/6 and PL/J strains, were used in this experiment. Seven mice from each strain were immunized (day 1) with 50 μg of a synthetic linear peptide (representing amino acid residues 273–289 of the 60 kD Ro antigen) emulsified in Freund's complete adjuvant (FCA) and seven mice from each strain were administered saline in FCA, as reported previously [28, 29]. This peptide has the sequence LQEMPLTALLRNLGKMT, and is identical between human and mouse. Briefly, 50 μg of the peptide was emulsified in FCA for the initial immunization and in incomplete Freund's adjuvant for all subsequent immunization. The immunization procedure was carried out on days 1, 14, 36, 63 and 119 of the protocol, and blood was drawn on days 21, 42, 56, 70, 84, 98, 105 and 126. Sera were stored in aliquots at −80°C until use in immunoassays. Animals were killed by exsanguination on day 147 of the protocol. Control animals were immunized with Freund's adjuvant alone.
The Ro60 273–289 peptide with sequence LQEMPLTALLRNLGKMT was synthesized in the linear and multiple antigenic peptide (MAP) format at the University of Oklahoma Health Sciences Center Molecular Biology Facility using standard Fmoc chemistry [30, 31]. Peptides were purified to apparent homogeneity by high-performance liquid chromatography (HPLC). Purity was determined by mass spectroscopy. Multiple antigenic peptides (MAPs) representing the previously described B cell epitopes of 60-kDa Ro were also synthesized by the same facility.
Ro peptide and La ELISA
The Ro60 273–289 MAP as well as other MAPs from 60 kD Ro and bovine La (Immunovision, Springdale, AK, USA) were used in an enzyme-linked immunosorbent assay (ELISA), as we have described previously [30-32].
Muscarinic acetylcholine M3 receptor peptides
The second extracellular loop peptide (amino acids 213–228) of the muscarinic acetylcholine M3 receptor with the sequence KRTVPPGECFIQFLSE and the third extracellular loop peptide (amino acids 514–527) of the same receptor with the sequence NTFCDSCIPKTFWN were synthesized as multiple antigenic peptides by GenScript USA, Inc. (Piscataway, NJ, USA).
Ro Western immunoblot
Purified bovine 60 kD Ro (Immunovision) was used for sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblot, as described previously .
Anti-nuclear antibodies (ANA)
ANA were determined with indirect inmmuofluorescence using HEp-2 cells as substrate and following instructions supplied by the manufacturer (Nova Diagnostics, San Diego, CA, USA), as described previously . ANA slides were read by an experienced laboratory technician who was blinded to other results. Sera were studied at a 1:40 dilution.
Saliva volume was measured over 10 min after pilocarpine injection, as reported previously , and expressed per 100 g mouse body weight.
Contractile responses of mouse bladder SM strips to electrical field stimulation (EFS) and carbachol were recorded and analysed as reported previously [33, 34]. Briefly, mice were euthanized and the bladder excised. The luminal epithelium was removed, and strips (10 mm in length) of detrusor SM prepared and suspended in Krebs' solution gassed with 95% O2/5% CO2, pH 7·4 at 37°C. At the beginning of each experiment, preparations were desensitized to capsaicin (10 μM; Sigma, St Louis, MO, USA) to inactivate sensory neurones, and hexamethonium (100 μM; Sigma) and guanethidine (0·3 μM; Sigma) were added to block nicotinic ganglionic neurotransmission and sympathetic neurones, respectively. EFS was performed at a pulse width of 0·3 ms at 60 V, using a Grass 588 stimulator (Grass Technologies, West Warwick, RI, USA) for the duration of 3 s at a frequency of between 1 and 50 Hz. SM carbachol (CCh) contraction–response curves were generated by the cumulative addition to the preparations of 0·1 to 30 uM of CCh. Contractile responses to EFS and CCh were measured using isometric transducers (Letica, Barcelona, Spain) connected to a PowerLab/8s data acquisition system (AD Instruments, Sydney, Australia).
Experimental protocol and data analysis
Following a 30-min equilibration period, baseline SM contractions induced by EFS or CCh were recorded. Serum (0·25% by volume) collected from BALB/c mice immunized with Ro60 273–289 or saline (day 42 post-immunization) was each added to two individual detrusor strip preparations and incubated for 30 min. Serum positive for anti-muscarinic 3 receptor antibodies was obtained from an insulin-dependent NOD mouse  (gift from Dr S. Grey). SM contractions induced by EFS or CCh in the presence of the serum were then recorded. The amplitudes of pre- and post-serum SM contractions were calculated using the peak parameters – peak amplitude function of the DataPad on Chart version 4.2 software (AD Instruments) and compared by two-way analysis of variance using GraphPad Prism (version 3.0a for Macintosh; GraphPad Software, San Diego, CA, USA). P-values of less than 0·05 were considered significant.
Formalin-fixed, paraffin-embedded salivary glands were stained with haematoxylin and eosin after sectioning. Lymphocytic infiltrates (defined as collections of greater than 50 mononuclear cells) were counted in five tissue sections from different areas of the gland. Examiners were blinded to control or peptide-immunized status.
Saliva flow did not have a Gaussian distribution and was therefore analysed by non-parametric, rank-order statistics. Student's t-test was used to compare ELISA results between groups.
Initially, we wished to determine whether or not there were differences in the immune response to the immunizing peptide among the strains employed. We found that SJL/J animals had virtually no circulating antibodies binding the peptide as measured in ELISA. Meanwhile, C57BL/6 had some but minimal anti-Ro273–289. The other strains, BALB/c, DBA-2 and PL/J, had high titre antibodies binding the peptide (Fig. 1a). Control animals immunized with Freund's adjuvant only had no demonstrable antibody binding to the peptide (Fig. 1b). Thus, three of the five strains studied mounted a vigorous immune response to immunization with the short peptide from 60 kD Ro, while two strains did not.
In order to determine binding to the whole 60 kD Ro molecule, we subjected intact 60 kD Ro to SDS-PAGE and transfer to nitrocellulose for immunblotting. These studies showed that of the three strains binding the peptide of immunization, the majority of BALB/c and PL/J animals also bound the intact molecule (Fig. 2). Only one DBA-2 mouse had convincing, but low-level binding to 60 kD Ro. Previous results show that binding 60 kD Ro in this assay is evidence of B cell epitope spreading [27-29]. The two strains, SJL/J and C57/BL/6, that either did not bind or bound the Ro 273–289 peptide poorly had no evidence of binding to the parent molecule. These data demonstrate that of the three strains of mice with an initial immune response to the peptide, two went on to have B cell epitope spreading with resultant antibodies that bind 60 kD Ro.
We next determined whether antibody binding additional epitopes of 60 kD Ro was found in the sera of the Ro peptide-immunized animals. Based on previous data, we hypothesized that animals binding intact 60 kD Ro would, in fact, have B cell epitope spreading such that other regions of the molecule were bound by antibody. We found that strains binding intact 60 kD Ro did have antibody binding other peptide epitopes of this molecule, including peptide Ro480 and Ro355 (data not shown). Our previous data showed that the Ro60 reaction in an animal upon immunization with a peptide from Ro60 is due in fact to the induction of multiple antibodies and not to antibodies against the immunizing peptide . Affinity absorption used in our study demonstrated that some autoantibody was present that bound regions of Ro60 other than that used for immunization . In that study, we immunodepleted serum containing antibodies against the immunogen by passing the serum through a Sepharose 4B column coupled with the immunizing Ro peptide. The depleted serum was found to bind to Ro60, confirming that epitope spreading had occurred .
In addition, we determined binding of the La protein by sera from these mice. Similar to intact 60 kD Ro binding, we found that BALB/c and PL/J mice had anti-La as determined by a sensitive ELISA, although the titre of anti-La in the PL/J was low but statistically elevated compared to same-strain controls (Fig. 3). We also found that mice from some strains had a positive ANA (Fig. 4). These included BALB/c and DBA-2, while SJL/J animals tended to have fluorescence after immunization with Freund's adjuvant alone (Fig. 4). Thus, BALB/c, DBA-2 and PL/J strains had evidence of autoantibodies binding structures other than the peptide of immunization.
The presence of a Sjögren's-like illness was assessed first by measuring salivary flow. We found that these experiments replicated our previous work  in that timed, stimulated saliva volume was reduced in BALB/c mice immunized with the 60 kD Ro 274 peptide (P < 0·017; Kruskal–Wallis rank order testing) (Fig. 5). Meanwhile, there was no reduction of salivary flow in the other strains (Fig. 5). Thus, only one of the strains developed clinical organ failure associated with Sjögren's syndrome after immunization with a short peptide from 60 kD Ro.
The second aspect of a clinical Sjögren's-like illness to be studied was pathology of the salivary gland after Ro peptide immunization. Again, as reported previously, BALB/c mice immunized with the Ro peptide had lymphocytic infiltration of the salivary gland that was highly similar to the findings in human Sjögren's syndrome patients, while control mice did not (Fig. 6). DBA-2 mice also had similar findings, with most Ro peptide-immunized mice developing lymphocytic infiltration while same-strain control mice did not have appreciable immune infiltration (Fig. 6). Thus, similar to MRL-lpr/lpr mice, this strain had lymphocytic infiltration but no salivary gland dysfunction. Both SJL/J and PL/J mice had lymphocytic infiltration in equal numbers of mice from the experimental and control groups. Finally, there were no lymphocytic infiltrates in the either control or immunized C57BL/6 mice (Fig. 6).
In order to determine the possible reason for the differential effects observed in the various strains of mice immunized with the Ro60 peptide, we investigated the presence of antibodies directed against the muscarinic acetylcholine M3 receptor in these mice. We found that none of the five strains of mice immunized with the Ro60 peptide had significant levels of antibodies directed against the second and third extracellular loops of the muscarinic acetylcholine M3 receptor. We also carried out a functional muscarinic M3 receptor assay using bladder SM strips as described in Materials and methods using pooled sera obtained from Ro 273–289 peptide or Freund's immunized BALB/c mice. Serum from a NOD mouse was used as a positive control. The NOD mouse serum inhibited both EFS and CCH responses (P > 0·05). However, there was no inhibition with the Ro peptide-immunized BALB/c mice sera.
We have shown in a previous study that both B and T lymphocyte subtypes infiltrate into the salivary glands of Ro60 peptide-immunized mice. The infiltrates consisted of 45% CD4+ T cells, 18% CD8+ T cells and 35% CD19 B cells, a finding similar to that seen in humans with Sjögren's syndrome .
The strains used in this study are given in Table 1. BALB/c mice had antibodies binding the peptide of immunization, epitope spreading with binding of intact 60 kD Ro as well as La, lymphocytic infiltration of the salivary gland and salivary gland dysfunction, as reported previously . DBA-2 mice developed all these features except that there was no glandular failure, in that salivary flow was intact in these animals despite the presence of infiltrates. PL/J mice had antibodies binding the peptide of immunization as well as evidence of epitope spreading, as sera from these animals contained antibodies against intact 60 kD Ro and La, but there was no evidence of an immune attack directed against the salivary glands. C57BL/6 mice mounted an immune response against the 60 kD Ro 273–289 peptide but had no evidence of epitope spreading, immune attack against the salivary glands or salivary gland dysfunction. Lastly, SJL/J mice had no detectable antibodies binding the immunizing peptide or other features of the Sjögren's-like illness.
Table 1. Manifestations, both clinical and immunological, that developed after Ro60 273–289 peptide immunizations in the strains of mice tested.
†Strains designated as positive had circulating antibodies binding the immunizing peptide, Ro60 Ro273–289. ‡Strains designated as positive had circulating antibodies binding other epitopes of Ro60 as well as antibodies binding whole Ro60. §Strains designated as positive had lymphocytic infiltration of the salivary glands meeting the criteria given in the Methods. ¶Strains designated as positive had statistically low stimulated salivary flow compared to sham-immunized same-strain controls.
We have found that upon immunization with a short peptide from 60 kD Ro, there were marked differences in the immune response and its outcome among these strains of mice. As we and now others  have reported previously, BALB/c developed a Sjögren's syndrome-like illness after immunization with the Ro60 273–289 peptide with development of epitope spreading, lymphocytic infiltration of the salivary gland and low salivary production. As shown in Table 1, the other strains of mice all had different points to which the immune response and its consequences proceeded along the putative pathway to the Sjögren's-like illness. Thus, these other strains are arrested in various stages of the development of the disease, such that each can be construed to have developed preclinical autoimmunity.
SJL/J mice did not mount an immune response to the peptide, at least not one manifested by the presence of antibodies in the peripheral blood. We hypothesize that this lack of response is related to major histocompatibility complex (MHC) differences such that do not allow an immune response to this short peptide. This could lie at the level of the T cell or B cell compartment. We have shown in previous work that the peptide used herein is both a B and T cell epitope in BALB/c mice . SJL/J mice have H-2s, and are the only strain among those tested that have this MHC type.
We hypothesize that the differences between the other strains that are responsible for different outcomes are also genetic, but these genetics are not MHC. Conversely, these genetics may be highly informative with regard to the pathogenesis of Sjögren's syndrome. For example, C57BL/6 mice bound the peptide but had no spreading, while PL/J mice had epitope spreading. Thus, exploring the genetic differences in these mice related to this finding may illuminate the initiation of Sjögren's and/or other autoimmune diseases. PL/J mice did not show infiltrating lymphocytes in the salivary gland but the DBA-2 strain did have these infiltrates, while both strains had epitope spreading. Thus, there may be genetics controlling lymphocyte phenotype, especially cell surface markers important for lymphocyte movement, which are critical to this difference. One can speculate that experiments using immunization of back-crosses or recombinant inbred strains could identify the genetics important for each of these steps.
It is interesting to note that the anti-Ro 274 antibody titre does not correlate with lymphocytic infiltration of salivary glands. For example, all the Ro274-immunized BALB/c mice have high anti-Ro274 antibodies (Fig. 1). However, only three of six BALB/c mice developed infiltrates (Table 2). A similar situation is seen in DBA and PL/J mice, which also have high titre anti-Ro274 antibodies (Fig. 1).
Table 2. Salivary gland infiltrates in mice immunized with Ro274 peptide or Freund's adjuvant.
BALB/c and DBA-2 mice differed in that while both had epitope spreading and lymphocytic infiltrates in the salivary glands, only BALB/c had decreased salivary volume. There are at least two possible explanations for this difference. In both human Sjögren's syndrome as well as in NOD mice, there is evidence that the glandular dysfunction may be due to autoantibodies binding the muscarinic receptor 3 , which mediate neurological control of the salivary glands. However, the Sjögren's-like illness observed in our mice does not appear to be mediated by anti-muscarinic acetylcholine M3 receptor antibodies, as none of the five strains of mice immunized with the Ro60 peptide had significant levels of antibodies directed against the second and third extracellular loop of the muscarinic acetylcholine M3 receptor (M3 receptor functional assay using bladder SM strips also confirmed this result in BALB/c mice). Alternatively, similar to the differences found between other strains at different points of the process, as described above, there may be genetic differences that mediate the differential response of salivary flow in these two mouse strains to Ro peptide immunization. Recent data have implicated the innate immune system in that treatment with Toll-like receptor 3 agonist gave a biphasic but accelerated Sjögren's-like disease in NZB/W F1 mice with follicular T helper cells within the glandular infiltrates . Thus, either innate or acquired immunity might be involved in the differences between these strains with and without glandular dysfunction after immunization with 60 kD Ro peptides.
If, in fact, there are quantum steps leading to eventual disease in this model of Sjögren's syndrome, then this is reminiscent of the situation in NZM2410 mice. Three susceptibility genetic intervals have been identified in this strain, which develops a lupus-like illness. Each one of these mediates a difference aspect of the autoimmune diathesis in these animals . However, the situation is complicated in NZM2410 mice, with the genetic intervals containing multiple susceptibility alleles and protective alleles [40, 41], as well as multiple gene interactions .
Of course, the ideas presented for our model are predicated on the assumption that there is a temporal relationship to the events seen. In particular, we have assumed that epitope spreading with development of a full-blown autoimmune response to the 60 kD Ro protein must occur prior to lymphocyte infiltration of the glands, and that lymphocytic infiltration of the gland is a prerequisite to failure of the salivary gland to secrete properly. In fact, we do not know the precise order of occurrence of the events, nor do we know whether there are cause-and-effect relationships between the different aspects of the Sjögren's-like illness.
Immunization of mice with a single, short peptide from the Ro60 autoantigen results in a Sjögren's-like illness. Using multiple strains of mice, we find that this process is interrupted at a specific point in the various strains. Thus, this animal model may allow dissection of the genetics of pathological features of preclinical autoimmunity, including epitope spreading, lymphocytic infiltration and glandular dysfunction.
This work was supported in part by NIH grants to R.H.S. (DE017561) as well as grants from the Oklahoma Center for the Advancement of Science and Technology (to R.H.S. and B.T.K.). We thank Lucy Ramon, Skyler P. Dillon, Yaser Dorri, Sima Asfa, Sherry Hubbell, Olga Yeliosof and Ali Khalili for technical assistance.
The authors have no competing interests to declare.