Sjögren's syndrome: studying the disease in mice

Sjögren's syndrome (SS), a systemic autoimmune disease, is characterized by inflammation of exocrine tissues accompanied by a significant loss of their secretory function. Clinical symptoms develop late and there are no diagnostic tests enabling early diagnosis of SS. Thus, particularly to study these covert stages, researchers turn to studying animal models where mice provide great freedom for genetic manipulation and testing the effect of experimental intervention. The present review summarizes current literature pertaining to both spontaneous and extrinsic-factor induced SS-like diseases in mouse models, discussing advantages and disadvantages related to the use of murine models in SS research.


Introduction
Assuming that studying a model organism will provide us with relevant information about the organism of our primary interest, investigation of nonhuman animals represents an important pillar in today's biomedical research. Over the past decades, the most popular experimental model to emerge is the common house mouse, irrespective of diff erent living environments, the evolutionary distance and some well-recognized dis crep an cies in innate and adaptive immune responses between mice and men. Despite such concerns, researchers generally accept these limitations in order to circum navigate techno logical and ethical issues related to research conducted in humans. Indeed, immunology has embraced the study of mice as a model organism and has accumulated tremendous insight into the intricacies of the human immune system and its involvement in both preventing and eff ecting disease.
In the present article, the murine models for Sjögren's syndrome (SS) are presented along the lines of spontaneous and extrinsic-factor induced models of SS-like disease and are discussed with special focus on disease phenotype and alterations induced in association with genetic modifi cation and experimental intervention. We also highlight common biological themes reported in context with both the etiology and the underlying pathogenic mechanisms of experimental SS and address their potential relevance for SS in humans.

Sjögren's syndrome: a summary
SS is a chronic autoimmune disease, which mainly aff ects the exocrine glands. Nearly all patients complain of a persistent feeling of dry mouth (xerostomia) and dry eyes (keratoconjunctivitis sicca) [1,2]. Th ese symptoms can be confi rmed by multiple objective tests indicating significant functional impairment of the salivary and lacrimal glands. Histological evaluation of minor salivary glands obtained from patients with SS usually shows large and persistent focal infi ltrates of mononuclear cells, often referred to as lymphocytic foci. Th ese foci consist of mainly T cells, fewer B cells and smaller numbers of natural killer cells and dendritic cells. Often associated with such infl ammation are acinar epithelial cell atrophy, progressive fi brosis and the presence of adipocytes in the salivary glands. Since approximately 60 to 80% of patients with SS produce anti-Ro antibodies and 40 to 60% produce anti-La autoantibodies [2], SS diagnosis also relies on the presence of these two biomarkers [3].
Aff ecting approximately 0.1 to 0.6% of the total population, SS is considered a relatively common rheumatic disease. In addition, SS is estimated to be 10 times more common in women compared with men. SS may extend from an autoimmune exocrinopathy to eff ect diverse extraglandular manifestations in the musculoskeletal, pulmonary, gastrointestinal, hepatobiliary, hematologic, vascular, dermatologic, renal and nervous systems. In contrast to systemic lupus erythematosus (SLE), where increased mortality has been reported as a consequence of the disease, the overall mortality in SS is comparable with the rate in the general population [4]. Th e risk of developing non-Hodgkin's lymphoma, however, is reported to be increased 16-fold in patients with SS compared with a control population [5]. Unfortunately, all therapies tested to date have proven ineff ective in reversing the course of SS. Regrettably, relatively few studies in the fi eld of rheuma tology address SS specifi cally.
With the possible exceptions of a few autoimmune diseases (for example, rheumatic heart disease), the etiology of most autoimmune diseases remains a mystery. Th e latter is also true for SS, despite multiple attempts to identify factors that might trigger the onset of a pathogenic immune response specifi cally directed against the exocrine glands. With a subset of SS patients exhibiting strong type 1 and type 2 interferon signatures [6,7], there is reason to believe that a viral agent is involved -but why some individuals are susceptible and others are not most probably resides in the individual's genetic background. An activated type 1 interferon system has also been des cribed in other autoimmune diseases (for example, SLE) [8]. Much is known about the exogenous and endo ge nous inducers of type 1 interferons and the molecular path ways that may mediate a continuous interferon produc tion involving Toll-like receptordependent ampli fi cation and propagation of the immune response [9]. Less is known, however, about and the functional role of specifi c gene variants in the regulation of the type 1 interferon system. Complicating this picture, unfortu nately, is the fact that an important share of patients with SS suff ers from secondary SS, defi ned as SS manifested in individuals diagnosed with other autoimmune diseases such as SLE, rheumatoid arthritis or scleroderma.
Another confounding factor in SS that has surfaced is the increasing recognition that the severity of secretory dysfunction does not necessarily correlate with the degree of leukocytic infi ltration or the loss of acinar tissue. Th is raises the distinct possibility that immune process-related alterations within the glandular tissues, disturbing saliva production and/or secretion, are involved in the impairment of exocrine gland secretion observed in patients with SS [10,11].
Th e aim of the present review is to provide the reader with an overview and specifi c information about murine strains that have been proposed as models of SS. Th e review also highlights fi ndings and hypotheses regarding the etiology and pathogenesis of SS that arose from research conducted in animal models [12,13].

Model organisms: a summary
In principle, due to the common descent of all living organi sms, discoveries made in one species might provide scientists with valuable information about another species. For this promise to be fulfi lled, researchers depend on a critical level of conservation between the species studied and the species of primary interest. As stated earlier in the manuscript, the mouse is the organism of choice for the majority of immunologists and has also become an integral element of the bed-tobedside drug development strategy. As species, humans and mice diverged approximately 70 million years ago; and from an immunologist's point of view of special importance, they evolved in two diff erent ecological niches. Nevertheless, the major paradigms about the working principles of the immune system appear to translate particularly well between the two species [14]. Th ere is a need, however, for further delineation of species-specifi c diff erences in order to increase predictability of how fi ndings from a murine strain may translate to a human population [14].
Direct comparison of the human and mouse genomes has confi rmed the close relationship of these two mammalian species, as there are only about 300 genes that are unique to either humans or mice [15]. Because of diff erences in the development and lifespan between humans and mice, one may certainly argue that significant diff erences exist in the timing of gene expressions, but the basic workings of molecular and biological pathways have been shown to be similar, if not identical. One must remember, however, the aspects of very distinct diff erences in innate and adaptive immunity that exist between mice and men [14]. Nonetheless, one can only be impressed by how relevant information from mouse studies is to humans. To researchers, perhaps the most compelling feature in the context of research using animal models is the mouse lifespan and fertility. In addition, most societies grant scientists considerable liberties in testing new hypotheses in mice by allowing genetic manipulation and the strict control of an animal's living environment. As a conse quence, resources related to research in mice became highly accessible -including thousands of inbred and genetically modifi ed strains, detailed experimental protocols, elaborated researchrelated reagents and data bases containing extensive data collections [16].

Experimental research and Sjögren's syndrome
Several aspects either directly related to the nature of SS or associated with current technical limitations underscore the necessity for research involving animal models. Prior to the onset of overt SS, physiological and structural changes are thought to take place in the exocrine glands, but due to the covert nature of the early stages of autoimmune diseases, studying these events in patients is virtually impossible. Similarly, collection of human specimens that represent a time course of the disease is diffi cult.
Ideally, a mouse model of SS mimics several clinical, histopathological and immunological features of the human disease combined with a high incidence of disease. In any cohort of SS patients, however, individuals are genetically diverse and the exhibited disease profi les are heterogeneous. Th ere is thus signifi cant reason to develop multiple murine strains, which manifest, to diff erent extents, SS-like disease manifestations. Th is is also important as, obviously, each mouse only represents one genetic back ground. Unsuccessful clinical trials, based on seemingly promising results of treating autoimmune diseases in mice, might be due to the fact that the trial's design relied too heavily on a single mouse model; for example, non obese diabetic (NOD) mice for the study of type 1 diabetes (T1D) [17].
Regarding mechanisms underlying the diff erent aspects of the pathogenesis of SS, studying genetically altered strains allows the testing of more specifi c hypotheses with regard to, for example, a specifi c protein, cell type or func tional pathway. Nevertheless, the overall functioning principles of a system can often not be predicted by studying the properties of its isolated parts as its state may strongly rely on the interaction of multiple components.

Murine models of spontaneous disease
In outbred populations of higher organisms such as rodents, cats, dogs and horses, the occurrence of infl ammatory conditions reminiscent of autoimmune diseases in humans is not uncommon. Also comparable with the situation in humans is heterogeneity in terms of genetic backgrounds, disease activity and clinical manifestations displayed under outbred conditions. A potential explanation for this phenomenon might be related to a possible null correlation between some autoimmune diseases and reproductive success. Hence, model organisms that develop a SS-like disease spontaneously might well, as a group, represent the assumed multifactorial etiology and complex pathogenesis of SS in humans. As mostly inbred strains are used for research purposes, the conclusions drawn from an experimental study may translate well to a sub population of patients while being only partly valid or invalid for other groups of patients with SS.
Th e earliest murine models of SS identifi ed were strains that develop SS-like disease manifestations spon ta neously. A pertinent fact is that these models mimic in part as a group the complex genetics and diverse disease phenotypes found in patients with SS. Th e disease phenotypes of these strains are discussed below (Tables 1  and 2).
Although other SS-related disease manifestations are not very pronounced in (NZB/NZW)F 1 mice, a more recent study demonstrated that an unspecifi c infl am matory stimulus, evoked by Freund's incomplete adjuvant, can trigger a signifi cant drop in salivary gland function already in an early phase of the disease, while this intervention protocol aff ected anti-Ro levels at a latter disease stage [22]. Th ought to alter the sizes of T-cell subsets, administration of anti-CD25 monoclonal anti bodies shortly after birth also exacerbates sialoadenitis and autoantibody production in this strain [23]. Lastly, Toll-like receptor 3 engagement through polyinosinic:polycytidylic acid has been studied in an attempt to recapitulate the eff ect of a dsRNA virus infection on the SS-like disease manifested in these mice [24]. As a result, infl ammatory mediators downstream of Toll-like receptor 3, such as type-1 interferon, were transcribed and a concomitant transient loss in salivary gland secretory function was observed [24].

MRL and MRL/lpr mice
In 1982 the MRL strain, at the time already established as a model of SLE, was reported to develop periductal lymphoid infi ltrates in the salivary glands [25]. MRL/lpr mice diff er from MRL mice with respect to a mutation involving the Fas gene [26]; however, negative selection in the thymus does not seem to be impaired in either strain [27]. In addition, irrespective of the lpr mutation in the Fas gene, MRL/lpr mice express a detectable amount of apoptosis-related FAS protein on lymphoid cells [28]. Nevertheless, defective apoptosis associated with the lpr mutation results in increased susceptibility and severity of the disease, most probably through acceleration of the disease course [26,28].
Immunohistochemical analyses of the organs targeted by the infl ammation show the presence of activated T cells [29,30], whose importance was further confi rmed in T-cell transfer experiments [31]. Infl ammatory lesions in the salivary glands of MRL/lpr mice contain B cells producing IgA and IgM rheumatoid factor [32] and were, in addition, identifi ed to be sites of IFNγ production [30]. Of potential concern, despite female predominance and the rare occurrence of anti-Ro autoantibodies, the clinical hallmarks of SS -hyposalivation and keratoconjunctivitis sicca -are absent in this model.

NFS/sld mice
Th e NFS/sld mouse provides a model in which aberrant immune responses against α-fodrin are elicited [33]. A defect in salivary gland development leads to aberrant enzymatic proteolysis of the structural protein fodrin by caspase [33]. Indeed, some patients with SS produce antibodies specifi c to the 125 kDa subunit of α-fodrin [34]. However, the association between antibodies to αfodrin and SS does not seem to be as strong as originally thought [35]. Th ymectomy performed in NFS/sld mice 3 days after birth results in development of T-cell dominated infi ltrates in the salivary and lacrimal glands, and -secondary to the SS-like disease -the NFS/sld mice undergoing thymectomy 3 days after birth also tend to develop infl ammatory lesions in other organs [36].

IQI/Jic mic e
Th e IQI/Jic strain was developed from the same stock that gave rise to the NOD mouse. Selection, however, was for mice that exhibited a SS-like disease comparable with NOD mice but in the absence of T1D. IQI/Jic mice develop focal infl ammation in the salivary and lacrimal glands, accompanied by parenchymal destruction [37]. Sialoadenitis progresses over time and becomes more prominent in females compared with males. IQI/Jic mice also develop infl ammatory lesions in several other organs, including the lung, pancreas and kidneys [38].
Interestingly, kallikrein-13 has recently been suggested to play a role in the etiology of the SS-like disease manifested in IQI/Jic mice [39]. Kallikreins, together with other proteases, were found to be part of the salivary proteome characteristic for patients with SS [40].

Nonobese diabetic mice and related strains
Th e NOD strain descends from a cataract-prone strain of outbred Jcl/ICR mice and is today the most extensively characterized model of SS and T1D. Although some genetic loci related to diabetes (idd s loci) contribute to the infl ammatory changes in the exocrine glands, it seems that diabetes and SS develop independently of each other [41][42][43]. T1D in NOD mice is restricted to the expression of the class II major histocompatibility MRL SG and LG infl . SLE-like disease [29] lpr  SG and LG infl . lpr exacerbates the MRL disease phenotype [30] NFS/sld None Aberrant proteolysis of α-fodrin [33] 3d-Tx SG and LG infl . Infl ammatory lesions in multiple organs [34] IQI/Jic SG and LG infl . Infl ammatory lesions in multiple organs [37] NOD SG and LG infl .,  SG and LG function T1D, multiple immune system-related alterations [12] H2 b Similar to original NOD strain No T1D [42] C57BL/6 May develop SG infl . at an old age Widely used recipient strain NOD-Aec1Aec2 SG and LG infl .,  SG and LG function Increased applicability compared with NOD mice [41] Il2 -/-SG and LG infl .,  SG function Generalized lymphoproliferative disease [91] Il2Rα -/-SG and LG infl .,  SG function Generalized lymphoproliferative disease [91] Foxp3 -/-None More susceptible to SS-like disease [94] Il10 Tg SG and LG infl .,  SG and LG function  apoptosis in the SG [95] Il14α Tg SG infl .,  SG function High incidence of CD5 + lymphoma, nephritis [101] Baff Tg SG and LG infl .,  SG function MZ B-cell dominated infl ., SLE-like disease [106] Tgfβ1 -/-SG and LG infl .,  SG function Poor viability, mixed infl ammatory cell response [112] Tgfβ1 Tg  SG function Impaired SG architecture, SG tissue fi brosis [113] Thbs1 -/-LG infl ., anti-Ro and anti-La Detailed assessment of the eye component [116] Ar -/-SG infl .
Estrogen-dependent disease [119] RbAp48 Tg SG and LG infl .,  SG function, anti-Ro and anti-La Assessment of the infl ammatory milieu in the SG [122] Id3 -/-SG and LG infl .,  SG function, anti-Ro and anti-La Exocrine gland dysfunction precedes SG and LG infl .

Table 2. Alterations in disease phenotype observed in association with genetic modifi cation and experimental intervention
LG function, SG infl . and SG function not assessed LG function, SG infl . and SG function not assessed LG function, SG infl . and SG function not assessed LG function, SG infl . and SG function not assessed [ com plex (MHC) haplotype H2 g7 [44]. Whereas NOD.B10-H2 b mice are resistant to the onset of overt T1D, they still exhibit the main disease manifestations of SS [42]. Th e exact extent and cellular composition of the glandular infl ammation in NOD.B10-H2 b mice, however, remains to be defi ned. NOD mice in which the original MHC H2 g7 haplotype was replaced with an H2 q or H2 p haplotype were also investigated. In summary, while the diff erence in H2 haplo type did not seem to aff ect the frequency of sialoadenitis, the disease severity varied among these strains [43]. Interestingly, introduction of the H2 q haplotype directed the autoimmune response towards the production of SLE-associated autoantibodies and a higher incidence of kidney pathology [43].
Autoimmune manifestations in NOD mice represent a complex disease involving genetics, sensitivity to exogenous factors and defects in central and peripheral tolerance [44]. Th ese factors have also been reported to contribute to the susceptibility of the strain to develop autoimmune thyroiditis [45], SLE [46], myasthenia gravis [47] and autoimmune encephalomyelitis [44] subsequent to specifi c intervention.
In NOD mice, focal infl ammation in the submandibular salivary glands and the lacrimal glands develops from approximately 8 weeks of age onwards. Th e foci appear comparable in structure and cellular composition with infi ltrates found in human salivary glands ( Figure 1) [48,49], and gender-related diff erences in the degree of exocrine gland infl ammation have also been reported in this strain [50]. As in patients with SS, in NOD mice the relationship between histopathological changes and hyposalivation is not always obvious -which indicates a certain autonomy of the autoimmune manifestations of SS ( Figure 2) [51]. Exocrine gland infl ammation in NOD mice appears to precede the onset of hyposalivation by a considerable amount of time [52]. Interestingly, transition to an overt disease does not necessarily need to be associated with a signifi cantly higher degree of glandular infl ammation [52], but hyposalivation and reduction in lacrimation were rather correlated with the occurrence of B-cell response-related gene transcripts in the exocrine glands [53,54].
Supporting the notion of a certain independence between degree of infl ammation and glandular hypofunc tion, introduction of an NZW-derived interval of chromosome 7 (annotated Ssial3) into NOD mice moderated sialoadenitis without ameliorating salivary gland function [55]. Analyses of dozens of infl ammatory mediators in serum and saliva obtained from NOD mice, furthermore, only revealed a minimal number of biomarkers corre lating with several SS-related disease manifes tations in an association network [56]. In addition, successful prevention of hyposalivation -through administration of 60 kDa heat-shock protein and of 60 kDA heat-shock protein-derived peptide amino acids 437 to 460 -did not coincide with a corresponding decrease in salivary gland infl ammation [57]. In contrast, biomarker signatures generated from saliva, indicating qualitative changes in salivary gland infl ammation, predicted treatment outcome and salivary gland function with high accuracy [57]. Several lines of evidence indicate that as T1D progresses from early insulitis to overt diabetes there is a loss of immune cell subsets, such as regulatory T cells (T regs ) and invariant natural killer T cells within the islets [17]. Unfortunately, little is still known about the role of these cell subsets in the progression of SS. Nevertheless, NOD mice defi cient for E2F transcription factor 1 -a regulator of T-cell proliferation, diff erentiation, and apoptosis -have a pronounced decrease in CD4 + CD25 + T regs and seem to be highly predisposed not only to T1D but also to SS [58]. In order to investigate the eff ects of E2F transcription factor 1 defi ciency prior to the involvement of the adaptive immune system, the SS disease profi le was later assessed in NOD-E2f1 -/mice, which, in addition, carried the severe combined immunodefi ciency (scid) mutation. Interestingly, this strain's saliva secretion capacity was found impaired [59] irrespective of the severe defi ciencies in adaptive immunity and the absence of exocrine gland infl ammation reminiscent of SS mediated by the scid mutation [60].
Another possible connection between SS and T1D in NOD mice might involve common autoantigens. Disruption of the islet-cell autoantigen 69 kDa gene in NOD mice, a self-antigen associated with diabetes that is expressed not only in the pancreas but also in the exocrine glands, reduced SS-related histopathology and glandular hypofunction [61]. A study investigating a large cohort of patients with SS could not, however, confi rm a role or true frequency of islet-cell autoantigen 69 kDa autoimmunity in patients with SS [62]. Studying the role of autoimmune regulator defi ciency and central tolerance in the context of SS in NOD and Balb/c mice identifi ed odorant binding protein 1a as a potential autoantigen involved in the etiology of autoimmune-mediated lacrimal gland pathogenesis [63].
To determine whether B cells contribute to the SS-like disease, experiments were carried out in NOD-Igμ null mice, which lack mature B cells [64]. Th e results indicate that in SS, in contrast to T1D, B cells do not signifi cantly participate in the initiation phase of the disease [44,64]. However, B-cell activity appears to be critical in the transition to an overt disease stage in these mice, since, despite the presence of T cells in the salivary glands, NOD-Igμ null mice fail to develop hyposalivation [64]. Subsequent studies also documented the concomitant lack of hyposalivation and anti-muscarinic acetylcholine type-3 receptor (M3R) autoantibodies of the IgG 1 isotype in IL-4-defi cient and signal transducer and activator of transcription 6-defi cient NOD strains [65,66]. In connection with possible non-infl ammatory mechanisms underlying the onset of hyposalivation, an altered aquaporin 5 distri bution -similar to the patterns observed in human specimens -has also been described in exocrine glands obtained from NOD mice [67,68].
Protection from T1D in NOD mice has been associated with a shift from a Th 1 to a Th 2 cytokine expression profi le in autoreactive T cells [17]. Results obtained in subsequent studies, however, indicated that compartment ali zation into disease-promoting Th 1 and protective Th 2 cytokines cannot be applied to the overall pathogenesis manifested in NOD mice [69]. Th e emergence of novel immune-cell subsets such as T regs and Th 17 cells further questions the validity of such models [70,71].
Cytokine expression in the exocrine glands obtained from NOD mice has been analyzed [72,73]. In a later study, taking advantage of recent technological developments, more comprehensive sets of infl ammatory mediators were analyzed in serum and saliva obtained from NOD mice [56]. Furthermore, blocking of either lymphotoxin βR or TNFR1 signaling has given insight into the implication of these two TNF family members in the development of the SS-like disease in NOD mice [48,74]. Whereas lymphotoxin βR signaling appears to aff ect the degree and cellular composition of salivary gland infl ammation [48], inhibition of TNFR1 engagement has been suggested to exacerbate the manifestation of hyposalivation [74]. In an earlier study, however, transgenic overexpression of TNFR1 inhibited exocrine gland infl ammation [75]. Investigation of antibodymediated inhibition of lymphocyte migration as a potential treatment strategy demonstrated that α 4 β 1 -integrin, leuko cyte selectin and leukocyte function-asso ciated  antigen 1 expression on lymphocytes and that vascular cell adhesion molecule 1 expression and peripheral node addressin on endothelial cells are required for lympho cyte homing to the lacrimal gland of NOD mice [76].
Th e functional roles of Th 1 and Th 2 cytokines in the pathogenesis of SS have been assessed in some detail by comparing a set of gene knockout mice: NOD-Il4 -/- [77], NOD.B10-H2 b -Il4 -/- [65], NOD.B10-H2 b -C.Stat6 -/- [66], NOD-Ifnγ -/and NOD-Ifnγr -/- [78] mice. Il4 -/-NOD mice and Stat6 -/-NOD mice retain salivary secretion rates similar to Balb/c mice despite the fact that they continued to present with exocrine gland infl ammation [65,66]. NOD-Ifnγ -/mice and NOD-Ifnγr -/mice were found to develop neither sialoadenitis or hyposalivation nor to present the signs of delayed salivary gland organogenesis present in the salivary gland of the parental NOD strain [60,78]. Of note, the mononuclear cell infi ltrates within the lacrimal glands persisted in these two latter strains [78]. Results regarding the more recently described Th 17-cell subset suggest that the Th 17/IL-23 system is activated in a NOD-derived strain during the overt state of the disease [79]. Interestingly, local IL-17A expression as a result of adenovirus vector-associated Il17a delivery to the salivary gland of SS nonsusceptible C57BL/6 mice recapitulated to a large extent the SS-like disease phenotype described in the NOD strain [80]. Subsequent investigation of IL-17 as a therapeutic target at diff erent disease stages showed that gene therapyinduced inhibition of IL-17, through expression of its receptor in the salivary gland, had the capacity to significantly reduce several important features of the SS-like disease, including salivary gland infl ammation and severity of hyposalivation [81].
To investigate the importance of specifi c gene regions with regard to SS-like disease manifestations, NODspecifi c genetic loci were introduced into either a C57BL/6 background [41] or a C57BL/10 background [82]. For both strains, gene expression of the salivary gland tissues was compared with their respective parental strain [54,82,83]. Unfortunately, the C57BL/10-based model termed B10.Q-Nss1/Idd5 has not been assessed for salivary gland hypofunction [82].
Th e principal aim for the development of the C57BL/6based model named C57BL/6.NOD-Aec1Aec2 was primarily to circumnavigate three problems associated with its parental NOD strain: the known impact of overt T1D on the physiological process of saliva and tear secretion as well as the possible interference of T1D, overt or asymptomatic, with biological readouts obtained from the NOD strain; the fact that there is no appropriate comparative nondiseased control strain for NOD mice; and the presence of a multitude of immune systemassociated defects in the NOD strain [44].
Th e genes within the genetic regions designated Aec1 (Idd5 on chromosome 1) and Aec2 (Idd3 on chromosome 3) appear suffi cient for the manifestation of a SS-like disease phenotype comparable with the one manifested in NOD mice [41]. First steps towards fi ne-mapping of Aec2 were undertaken with the purpose of identifying candidate genes potentially regulating SS-associated autoimmunity [84]. Nevertheless, although considered nonsusceptible to the development of a SS-like disease, the genomes of C57BL/6J or C57BL/10 might still contribute to the congenic strain's disease phenotype by enhancing the primary eff ects introduced by the congenic regions [85]. Such phenomena render it more diffi cult to discriminate between disease-causing and diseasepromoting gene segments. In addition, the two recipient strains may develop spontaneous sialoadenitis as they age [82,86]. Th e improved applicability of the C57BL/6.NOD-Aec1Aec2 strain compared with the original NOD mice, however, facilitated the study of proteases in the initiation phase of the disease [87], a more distinct delineation of the salivary and lacrimal gland transcriptome prior to and during the onset of the SS-like disease [53,54], as well as assessment of a potential role of complement 3 in SS [88].

Gene knockout and transgenic models
Genetic modifi cations have been shown to trigger diff erent aspects of the SS-like disease in murine models. Th e fact that silencing or overexpression of a single gene can result in a disease profi le reminiscent of SS points to pathways downstream of this particular gene. Th ese pathways are commonly associated with either regulating the immune response, governing developmental processes or contributing to exocrine gland homeostasis. A discussion of gene knockout and transgenic models of SS follows (Tables 1 to 3).

IL-2-defi cient, IL-2Rα-defi cient and forkhead box P3-defi cient mice
Th e fi rst indication that IL-2 activities are diminished in NOD mice was fi rst reported in 1993 [89] and was later found to be associated with Idd3 [90]. Today, IL-2 is recognized as a critical factor in promoting diff erentiation and activation of T regs . Concordantly, inhibition of circulat ing IL-2 led to aggravation of diverse autoimmune manifestations in NOD mice [45], and both IL-2-defi cient and IL-2Rα-defi cient C57BL/6 mice present with exocrine gland infl ammation and hyposalivation [91,92]. Th e SS-like disease in the two strains develops, however, secondary to a generalized lymphoproliferative disease character ized by autoimmune hemolytic anemia and infl ammatory bowel disease [93]. Nevertheless, data collected on IL-2 in a SS-related context indicate that in conditions with decreased regulatory cell populations the salivary glands are prone to exhibit autoimmune manifestations. Incon sistently, however, mice that carry the forkhead box P3 (Foxp3) sf mutant gene and are therefore defi cient for Foxp3 + -positive T regs remain free of glandular infl amma tion as long as they are not exposed to lipopolysaccharide [94]. Nonetheless, cells isolated from their lymph nodes had the capacity to induce sialoadenitis in immuno compromised, recombination activating gene-1-defi cient recipient mice [94].

IL-10 transgenic mice
Overexpression of IL-10 in C57BL/6J mice provokes progressive histopathology and hyposalivation suggestive of SS [95]. Il10 transfer into NOD mice, however, partially suppressed the appearance of SS-like features [96] indicating a dual role of IL-10 in SS, most probably dependent on temporal or site-specifi c expression patterns of IL-10.

IL-12 transgenic mice
Th e infl uence of IL-12 is considerably clearer. Both CBA [97] and SJL [98] mice transgenic for Il12 exhibit focal infl ammation within their exocrine glands, with the latter strain showing an additional array of SS-related mani festations, including hyposalivation and modest increases in autoantibody levels upon aging [98]. In this context, it should be noted that SJL mice are generally susceptible to pathogenic autoimmunity and are also highly prone to develop B-cell lymphoma [99]. In an earlier study, concurrent intraperitoneal injection of IL-12 and IL-18 triggered severe atrophy of the salivary and lacrimal glands of various strains [100]. Th ese degenerative changes in glandular tissues, however, occurred without lymphocytes concomitantly infi ltrating the aff ected glands [100].

IL-14α transgenic mice
By promoting expansion and activation of specifi c B-cell subsets, Il14α transgenic C57BL/6 mice present with hyper gammaglobulinemia by 3 months of age, with exocrine gland infl ammation by 6 months of age [101,102]. In addition, this strain develops immune-complexmediated nephritis, as well as exhibiting a high incidence of CD5 + B-cell lymphoma [101]. Strengthening the relevance of this model, a recent study demonstrated a strong dependence of the SS-like disease manifestations on local expression of lympho toxin α, a molecule crucial for the maintenance of organized lymphoid microenviron ments in target tissues of autoimmune diseases [103]. Il14α transgenic mice defi cient for lymphotoxin α retained normal saliva secre tion and presented no signs of salivary gland infl am mation or secondary lymphoma development [104]. Th is strain also no longer presented the disproportionally large CD5 + B-cell compartment, characteristic for mice overexpressing Il14α [104].

B-cell-activating factor transgenic mice
B-cell-activating factor (BAFF), also known as B-lymphocyte stimulator, has emerged as a critical regulator of Bcell survival and maturation, demonstrating the need for an obligate survival signal for both maturing and fully diff erentiated B cells [105]. Excess BAFF-mediated survival signals are thought to aid autoreactive B cells to escape apoptosis, to expand and, subsequently, to exert their potentially pathogenic activities [105].
BAFF received considerable attention following develop ment of Baff transgenic strains, of which one was shown to develop features reminiscent of SS, including lymphoid infi ltrates in the salivary and lacrimal glands and hyposalivation [106]. Th e manifestation of the SSlike disease thereby critically depends on B cells with a marginal zone B-cell like phenotype, which are the dominant lymphocyte population infi ltrating the salivary glands from this strain [107]. Despite the high numbers of B cells, anti-Ro antibodies or anti-La antibodies were not detected. Strains overexpressing BAFF also develop severe autoimmune manifestations traditionally associated with SLE, including circulating immune complexes, anti-DNA antibodies and immunoglobulin deposition in the kidneys [108,109]. Disruption of Tnfα in Baff transgenic mice furthermore revealed a critical role of the anti-tumor activity of TNFα in this strain [110].

Transforming growth factor beta 1 transgenic and defi cient mice
Transforming growth factor (TGF) beta1 is a multifunctional molecule that has eff ects on many developmental, physio logical and immunological processes. Animals carry ing a mutated Tgfβ1 allele present a syndrome marked by mixed infl ammatory cell responses and tissue necrosis, in many cases leading to organ failure and death [111]. In surviving mice, the syndrome includes infl ammation of the exocrine glands in a large share of the animals that can, however, be prevented by systemic injections of synthetic fi bronectin peptides [112]. Mice overexpressing TGFβ1 in the secretory cells of both the mammary and salivary glands exhibit impaired salivary gland architecture concomitant with salivary gland hypofunction [113]. It is important to note that altered TGFβ1 expression in mice results in poor viability and surviving mice suff er from hyposalivation in association with infl ammation, acinar cell atrophy and fi brosis in the salivary glands [111][112][113].

Adaptor molecule Act-1-defi cient mice
As a negative regulator of BAFF and CD40, the adaptor molecule Act 1 (Act1) crucially modulates the survival of all B cells [114]. In Act1-defi cient mice, similar to mice transgenic for BAFF, marginal-zone-like B cells dominate the infl ammation in the exocrine gland that develops around 6 months of age [115]. At 8 months of age, levels of salivary secretion appear to be slightly diminished, while indications for dry eyes -such as scratching and skin lesions around the eyes -were observed during the breeding process as early as 3 weeks postpartum [115]. Another shared trait between the Act1-defi cient strain and Baff transgenic mice is the production of SLEassociated anti-DNA autoantibodies and the manifestation of glomerulonephritis [115]. In contrast with Baff transgenic mice [106], however, defi ciency in Act1 causes the production of autoantibodies specifi c for Ro and La [115]. Further investigation of this phenomenon revealed that, compared with the other autoantibody specifi cities found in these mice, the production of anti-Ro autoantibodies and anti-La autoantibodies critically depends on functional CD40. As Act1 was more recently also identifi ed as a critical signaling component of the IL-17 signaling pathway, this aspect needs to be addressed in further studies [114].

Thrombospondin-1-defi cient mice
A recent study showed that silencing the thrombospondin 1 gene (Th bs1) in C57BL/6 mice causes the development of a severe and remarkably complete SS-like disease with respect to the involvement of the eye [116]. Increased apoptosis in the lacrimal glands accompanies their progressive deterioration and, in addition, anti-Ro autoantibodies and anti-La autoantibodies were detected in this strain [116]. Some THBS1-defi cient mice also displayed external signs of dry eyes, although the tear volume secreted upon stimulation did not diff er between the genetically modifi ed mice and the wild-type strain [116]. Although a few parameters of lacrimal gland function signifi cantly decreased before the signifi cant infl ux of infl ammatory cells into the glands, the authors argue for a critical involvement of the immune system, in particu lar the Th 17 system, in the observed pathology [116]. THBS1 is capable of activating latent TGFβ and, as alluded to previously, dysregulation of the TGFβ system aff ects the immune system as well as multiple developmental processes. Although less devastating compared with direct deletion of Tgfβ1, THBS1-defi cient mice exhibit a similar infl ammatory condition and display histo logical abnormalities described for TGFβ-defi cient mice [117]. Being a multidomain matrix glycoprotein capable of interacting with multiple cell adhesion molecules and proteases involved in angiogenesis, it is not surprising that THBS1 plays important roles in the development of diverse tissues [118]. Taking into account these properties of THBS1, investigating immune system-unrelated alterations in the exocrine tissues of Th bs1-defi cient mice would further clarify the etiology of the disease they manifest.

Aromatase-defi cient mice
Th e high female predominance and the late onset of SS in humans suggest a possible role of estrogen in the etiology of SS. Whereas neither estrogen receptor-alpha-defi cient nor estrogen receptor-beta-defi cient strains exhibit SS-like disease manifestations, another model for estrogen defi ciency -the aromatase knockout mouse -develops a lympho proliferative condition that in some aspects resembles the histopathological manifestation of SS in the salivary glands [119]. In parallel with sialoadenitis, B-celldominated infl ammation of the kidneys and enlargement of the spleen were also reported for this strain [119].

Retinoblastoma-associated protein 48 transgenic mice
Estrogen defi ciency caused by ovariectomy was shown to increase the number of epithelial cells undergoing apoptosis in the salivary glands as well as to have an eff ect on the cleavage of structural proteins [120]. Th e observation that retinoblastoma-associated protein 48 (RbAp48) contributes to the observed estrogen-dependent modulation of apoptosis exclusively in the salivary glands gave reason to investigate its role in a strain overexpressing RbAp48 under the control of a salivary gland specifi c promoter [121,122]. From 20 weeks of age onwards, RbAp48 transgenic mice exhibit infl ammatory lesions in the salivary and lacrimal glands, which comprise mostly T cells and fewer B cells. At 30 weeks of age, salivary gland function was shown to be impaired in the transgenic mice compared with the wild-type strain. Increased levels of anti-Ro autoantibodies and anti-La auto antibodies complemented the SS-like phenotype observed in RbAp48 transgenic mice [122]. Th e investigators also collected considerable data that suggest a dominant role of resident cells in the initiation and perpetuation of the disease, especially by contributing to MHC-II-dependent antigen presentation and modulation of the cytokine milieu [122].

Id3-defi cient mice
Inhibitors of DNA binding (ID) proteins are inhibitors of basic helix-loop-helix transcription factors and act as regulators of proliferation and diff erentiation of immune and non-immune cells [123]. Th e immune system of C57BL/6-Id3 -/mice is characterized by alteration in humoral immune reactions, marginal zone B-cell develop ment, B-cell precursor survival and both MHC-Irestricted and MHC-II-restricted positive and negative selection [124]. In these mice, T-cell-dominated focal infl ammation develops between 6 and 12 months of age, coinciding with anti-Ro autoantibody and anti-La autoantibody production. Th e same strain, however, shows severe exocrine gland dysfunction as early as between 6 and 18 weeks of age -long before the appearance of focal lymphocytic foci in their exocrine glands [125]. Nevertheless, the notion that disruption of certain genes triggers distinct pathological changes, almost exclusively limited to the salivary and lacrimal glands, may encourage further investigation of possible inter relationships between organ and lymphocyte development and the etiology of autoimmune diseases. In this context, investigation of a T-cell-targeted conditional Id3 knockout strain revealed the strong dependence of the original C57BL/6-Id3 -/strain's disease profi le on ID3-defi cient T cells in particular [126]. Interestingly, depletion of B cells ameliorated the SS-like disease in this strain [127]. Th e role of ID3 in SS is now under investigation in humans, but to date there are no SNPs in Id3 associated with SS in humans [128].

Phosphoinositide 3 kinase class-IA-defi cient mice
Th ere is an abundance of information that implicates phosphoinositide 3 kinase class IA in the development and onset of autoimmune diseases [129]. Mice with a Tcell-specifi c loss of phosphoinositide 3 kinase class IA (r1ΔT/r2n) develop an infl ammatory condition reminiscent of SS in the lacrimal glands that parallels the occurrence of infl ammatory lesions in the lungs, liver and intestines in these mice [130]. While exocrine gland function was not assessed in detail, the authors report a decrease in T regs in the periphery and increased anti-Ro antibodies and anti-La antibodies as a result of this specifi c genetic modifi cation.

Knockin mice with mutated κB enhancers in the IκBα promoter
Aberrant regulation of nuclear factor of kappa light polypeptide gene enhancer in B cells (NF-κB) has been associated with infl ammatory and autoimmune diseases since its crucial role in both innate and adaptive immunity was reported [131]. Among other autoimmune conditions, certain poly morphisms in the promoter of NF-κB inhibitor nuclear factor of kappa light polypeptide gene enhancer in B-cells inhibitor alpha (IκBα) might contribute to an individual's susceptibility to develop SS [132]. Removal of the feedback regulation of NF-κB by introducing κB enhancers in the IκBα (IκBα M/M ) promoter of C57BL/6 mice altered the expression of NF-κBassociated genes such as Il17 and genes involved in T-cell development [133]. Subsequently, infl ammation in the exocrine glands, concomitant with production of anti-Ro autoantibodies, anti-La autoantibodies and anti-DNA autoantibodies, were observed in these mice without, however, assessing exocrine gland function [133]. Th e overall phenotype of this strain, in addition, is characterized by involvement of various other organs, shortened lifespan and hypersensitivity to septic shock [133].

Extrinsic factor-induced models
For the strains described earlier in this manuscript, the etiology of SS-like disease manifestations is assumed to be, to a large extent, associated with the mouse's specifi c genetic background. Th e experimental models presented below are strains in which the development of a SS-like pathology requires administration of extrinsic factors, such as proteins and peptides or viruses (Table 3). Such protocols are based on the concept that injecting specifi c components emulsifi ed in an adjuvant can break immuno logical tolerance to certain organ-specifi c or organ-unspecifi c structures. Th is event might subsequently be followed by an immune-system-initiated pathogenesis. Th e components injected in such studies are mostly selected on the basis on their suspected role in the disease of interest. As alluded to previously, unfortunately, the current knowledge about disease-relevant autoantigens in SS is limited -which might be one reason why induced models are not of equal importance in SS as, for example, in studying multiple sclerosis [134] or rheumatoid arthritis [135].

Ro peptides
Repeated intraperitoneal injection of the Ro peptides -Ro amino acids 480 to 494 or Ro amino acids 274 to 290emulsifi ed in complete Freund's adjuvant and later in Freund's incomplete adjuvant has been shown to recapitulate some manifestations of SS in Balb/c mice [136]. Th ese mice present with hyposalivation, SS-like histopathology and production of anti-Ro antibodies and anti-La antibodies by 38 weeks of age [136]. Unfor tunately, the actual penetrance rate of the SS-like disease proved to be low, thereby limiting the potential value of the model [136]. Oral feeding of Ro or Ro peptides abolished the susceptibility of Balb/c mice to SS-like disease induction through the experimental procedure described above [137]. While these studies were designed to determine whether Ro, as an autoantigen, is impor tant in the etiology of SS, there is still a question as to how Ro might actually be presented to the immune system [138]. In light of a recent study indicating that Ro52 is a negative regulator of proinfl ammatory cytokine production [139], if and how these newly described pro perties of Ro52 contribute to SS remain to be investigated.

Muscarinic acetylcholine type-3 receptor peptides
As alluded to above, antibodies targeting the M3R may directly mediate the inhibition of exocrine gland secretion by inhibiting neuronal innervation of acinar cells. A recent study assessed the question further by vaccinating C57BL/6-M3r -/mice with a six-valent mixture of freeform extracellular peptides of M3R [140]. Indeed, the inoculation of splenocytes or CD3 + T cells into immunodefi cient C57BL/6-Rag1 -/mice triggered the development of marked mononuclear cell infl ammation in the exocrine glands accompanied by salivary gland hypofunction [140]. Th is study further supports the notion of a direct pathogenic role of anti-M3R immunity in SS [10].

Carbonic anhydrase
A subset of patients with autoimmune diseases, including patients with SS, produces autoantibodies against carbonic anhydrase II [141]. Studies carried out in mice revealed that experimental sialoadenitis can be induced through carbonic anhydrase II immunization of PL/J mice [142] as well as congenic strains of PL/J mice carrying the a H2 s or a H2 u haplotype [142]. Additional studies are required, however, to be able to estimate in more detail the resemblance of the disease manifested in this model with SS in humans.

Murine cytomegalovirus
Intraperitoneal injection of murine cytomegalovirus has been documented to lead to sialoadenitis and production of anti-Ro autoantibodies and anti-La autoantibodies in genetically modifi ed C57BL/6 mice [143]. Th e modifi cations, aff ecting either FAS-mediated or TNFR1-mediated apoptosis, resulted in an incomplete clearance of murine cytomegalovirus, suggesting that any defect in this response may evoke chronic infl ammation that resembles the histopathologlogical changes characteristic for SS [143]. In a subsequent study, C57BL/6-gld/gld mice, which are Fas ligand defi cient, were treated with an adenoviral viral vector inducing the overexpression of Fas ligand [144]. In light of high levels of Fas ligand expression following injection of the vector, fewer than 5% of ductal and acinar cells proved to be apoptotic. Nevertheless, the intervention caused signifi cant reductions in the number of infl ammatory foci and the degree of tissue destruction in the salivary glands [144].

Conclusions
SS is a complex autoimmune exocrinopathy that over time often progresses to a systemic disease. Interpatient heterogeneity is a major component of this rheumatic disease, as evidenced by the array of symptoms exhibited by patients at clinic visits. Although numerous mouse strains are being proposed as models of SS, it is not surprising that no single model can perfectly match the full spectrum of SS observed in a human population. In the present review, we describe how the genetic background of these models and intervention protocols modulated the disease profi le they project. Individual genetic alterations and their contribution to diff erent disease stages and specifi c manifestations of SS might one day be assembled to depict a more complete and integrated picture of SS.
Today researchers are presented with several alter natives regarding spontaneous and genetically modifi ed models of SS: yet, because of the complexity of SS, additional models will undoubtedly be required. Unfortunately, the limited knowledge about SS disease-relevant autoantigens and SS-related genetic risk factors continues to impede the development of extrinsic factorinduced models of SS.
In recent years there has been a positive trend towards testing hypotheses through genetic modifi cation or intervention protocols in established models of SS. Results from these studies have often yielded insight into mechanisms potentially associated with the pathology of SS. Th e identifi cation of B-cell-dependent mechanisms of pathogenesis in murine models, coupled with investigation of treatment strategies such as anti-CD20 antibodies targeting B cells in patients with SS, exemplify such translational advances. Nevertheless, it has become ever more challenging to keep pace with the developments in immunology and to be able to, at least partly, assess the importance of newly discovered components such as novel immune-cell subsets or regulatory pathways in a SS-related context. An assortment of wellcharacterized murine strains is needed in order to investigate possible roles of these components at the diff erent stages of SS. In the past few years, there has also been an increase in the number of studies assessing the role of possibly relevant and immune system-unrelated processes in the etiology of SS. Again depending greatly on animal models, such research initiatives are expected to yield an increasing number of relevant biomarkers, which may specify an individual's risk of developing SS or may indicate an early stage of disease.
It also has become clear that murine strains, in some cases, represent a stereotypic or incomplete picture of their human disease counterpart. To counteract this issue, in-depth characterization of individual models as well as reliance on results obtained in multiple models is, however, anticipated to increase the success rate of translational studies.
In conclusion, many advances in the fi eld of SS have their basis founded in discoveries initially made in animal models. Improved collaboration among scientists that develop animal models, researchers that apply animal models to investigate SS-related aims and clinicians that have access to well-defi ned SS patient cohorts should accelerate the discovery of novel disease mechanisms that lead to development of eff ective treatment regimens.

Competing interests
The authors declare that they have no competing interests.