|Home | About | Journals | Submit | Contact Us | Français|
Uveitis is a complex multifactorial autoimmune disease of the eye characterized by inflammation of the uvea and retina, degeneration of the retina, and blindness in genetically predisposed patients. Using the rat model of experimental autoimmune uveitis (EAU), we previously identified three quantitative trait loci (QTL) associated with EAU on rat chromosomes 4, 12, and 10 (Eau1, Eau2, and Eau3). The primary goal of the current study is to delineate additional non-MHC chromosomal regions that control susceptibility to EAU, and to identify any QTLs that overlap with the QTLs of other autoimmune diseases. Using a set of informative microsatellite markers and F2 generations of resistant and susceptible MHC class II-matched rat strains (F344 and LEW), we have identified several new significant or suggestive QTLs on rat chromosomes 2, 3, 7, 10, and 19 that control susceptibility to EAU. A protective allele was identified in the susceptible LEW strain in the Eau5 locus at D7Wox18, and epistatic interactions between QTLs were found to influence the severity of disease. The newly identified regions (Eau4 through Eau9) colocalize with the genetic determinants of other autoimmune disease models, and to disease-regulating syntenic regions identified in autoimmune patients on human chromosomes 4q21-31, 5q31-33, 16q22-24, 17p11-q12, 20q11-13, and 22q12-13. Our results suggest that uveitis shares some of the pathogenic mechanisms associated with other autoimmune diseases, and lends support to the “common gene, common pathway” hypothesis for autoimmune disorders.
Autoimmune uveitis (1) is a major cause of visual loss in humans. The inflammatory form of uveitis can occur either as an isolated disease, or as part of a systemic inflammatory syndrome such as Behcet's disease, Vogt-Koyanagi-Harada syndrome, juvenile idiopathic arthritis, multiple sclerosis, spondyloarthropathies, and inflammatory bowel disease. Studies in humans and in animal models suggest that the disease is regulated by endogenous immune mechanisms (2) and is influenced by environmental and genetic factors (3, 4). Most forms of uveitis occur within families but do not follow a classic Mendelian pattern of inheritance. The inheritance pattern is complex and is influenced by multiple genetic factors (4). Because of the inflammatory nature of the disease and the allelic variations in the MHC genes that influence the immune response, susceptibility to uveitis has been associated with the major genes that are located within the MHC (HLA in human) class I (5–10) or class II (11–16) loci. In addition to the MHC class I and class II genes, genetic associations have also been reported for several non-MHC genes like TNF-α (17, 18), MCP-1/CCL2 promoter region (19), CCL2, CCL5 (20), CCR5 (21), the IL-1 gene cluster (22), ICAM-1 (23), eNOS (24), and CyP4501A1 (25). A detailed review of the genetic involvement in uveitis had been provided elsewhere (3, 4, 26).
Experimental autoimmune uveitis (EAU)4 in animals, induced by immunization with retinal proteins or their fragments, has been extensively used as an animal model for human uveitis. EAU is a T cell-mediated autoimmune disease whose expression is controlled by MHC as well as by non-MHC genes (27, 28). The autoimmune nature of human noninfectious uveitis, and the validity of EAU as its model are based on several lines of evidence, namely 1) uveitis patients exhibit responses to retinal proteins that are uveitogenic in animals, such as interphotoreceptor retinol-binding protein (IRBP), retinal soluble Ag, rhodopsin, phosducin, recoverin, melanin proteins, retinal pigment epithelium membrane proteins, and myelin basic protein (29). 2) The pathogenesis of human uveitis is mediated by a T cell-driven cellular immune response (30–33) similar to rheumatoid arthritis, type 1 diabetes mellitus, and multiple sclerosis. 3) Uveitis is one of the diseases found to occur in families in which clustering of other autoimmune diseases has been observed (multiplex families) (34–37), and finally, 4) genome-wide scans of uveitis patient families and our earlier study of EAU in the rat model have localized uveitis susceptibility loci to chromosomal regions that also harbor susceptibility loci for other autoimmune disorders (37–40).
It is not uncommon for several autoimmune disorders to coexist in a patient (41, 42). Additionally, there is familial clustering of different autoimmune disorders with a very high sibling relative risk value to siblings of autoimmune disorder patients (43). Furthermore, quantitative trait loci (QTL) regions for different autoimmune diseases overlap in humans (44–47) and in animal models (48–51), and there are common conserved gene expression profiles in patients with different autoimmune disorders (52, 53). These findings led to the hypothesis that there are common genetic mechanisms and common pathways that drive the susceptibility and outcomes of these autoimmune disorders. This hypothesis, together with our earlier identification of several EAU QTLs that colocalized with other autoimmune disorders (38), led us to search for additional genetic regions that might be shared between uveitis and other autoimmune diseases.
In the current study, we examine EAU-susceptible and EAU-resistant animals in an F2 population of rats generated from phenotypically different, resistant Fischer 344 (F344) and susceptible Lewis (LEW) strains, in which MHC class Ia and class II are matched (RT1l) and MHC class Ib is related (RT1l and RT1lvl). Using a panel of polymorphic rat microsatellite markers that are at the peak of the QTL regions of other autoimmune disease models, we were able to identify several additional loci that control susceptibility to EAU, and that colocalize with other autoimmune disease QTLs. Identification of these additional QTLs will allow us to better understand the autoimmune nature of uveitis.
Fischer 344 (F344/N) and Lewis (LEW/N) rats were obtained from Harlan Sprague Dawley. These strains share the MHC class II haplotype (RT1l). F2 intercross offspring were generated from (F344 × LEW) and (LEW × F344) F1 progeny using the convention of (female × male) to indicate strain parentage (38). The F2 population is the same as was used previously (38) to identify EAU QTLs. The animals were maintained under specific pathogen-free conditions. All the procedures conformed to the guidelines of the National Institutes of Health and to the resolution of Association for Research in Vision and Ophthalmology on use of animals in research.
EAU was induced as per protocols described previously (38). Briefly, the animals were immunized s.c. with 30 μg of the peptide encoding the major uveitogenic epitope of bovine IRBP (peptide R16, residues 1177–1191, sequence ADGSSWEGVGVVPDV), emulsified in CFA containing 2.5 mg/ml Mycobacterium tuberculosis. After the animals were euthanized on day 16–18 postimmunization, eyes were collected for histopathological evaluation, and disease was graded on a scale of 0–4 based on previously described criteria (27). The animals were ranked for their phenotype based on the average score of both eyes.
We have previously identified EAU QTL regions (38) including Eau1 (D4Arb8, D4Arb38, and D4Arb7) on chromosome 4, Eau2 (D12Arb2, D12Rat37, and D12Rat39) on chromosome 12, and Eau3 (D10Mit3, D10Mgh8, D10Arb5, D10Wox13, D10Mit8, and D10Mgh7) on chromosome 10. To further resolve these QTL regions, closely spaced microsatellite markers from chromosomes 2 (suggestive evidence around a site of strong segregation distortion at D2Mgh15), 4 (Eau1), 10 (Eau3), and 12 (Eau2), which are polymorphic between F344 and LEW strains, were selected (supplementary table Ia).5 To identify colocalization of EAU QTLs with loci identified for other autoimmune disorders, the peak markers for the QTLs of experimental autoimmune encephalomyelitis (EAE) models— insulin-dependent diabetes mellitus (IDDM) models, and various arthritis models such as collagen-induced arthritis (CIA), pristane-induced arthritis (PIA), adjuvant-induced arthritis (AIA), oil-induced arthritis (OIA)—were chosen (supplementary table Ia) based on information obtained from the Rat Genome Database (RGD) collated by the Medical College of Wisconsin (http://rgd.mcw.edu/objectSearch/qtlSubmit.jsp?species = Rat&go = Submit and Rat Map maintained by Göteborg University, Göteborg, Sweden (http://ratmap.org/qtler/)). If the reported peak marker(s) for any of these QTLs was noninformative in the F344 and LEW cross, they were replaced by polymorphic markers located within the peak and in close proximity to the reported marker (as marked in supplementary table Ia).
Susceptible (n = 117, EAU score ≥1.75) and resistant (n = 40, EAU score = 0) F2 rats were genotyped using a panel of markers (supplementary table Ia). DNA was extracted from snap-frozen liver tissue using the DNeasy Tissue kit (Qiagen). Fluorescently labeled (6-FAM or HEX or TET) and unlabeled rat microsatellite markers were custom synthesized by Applied Biosystems or were purchased from Research Genetics according to the sequence information obtained from RGD. Genotypes were determined by PCR amplification of simple sequence repeats as per the conditions suggested by Research Genetics. The polymorphic variants were resolved by electrophoresis using 2%Nusieve 3:1 agarose (Cambrex) gels in 1×TBE or using an ABI 3100 genetic analyzer (Applied Biosystems) and were analyzed using GeneScan 3.1 software (Applied Biosystems). Genetic markers were arranged according to their position in Mb (million base pairs) identified using the BLAST-like alignment tool (BLAT) search (http://genome.ucsc.edu/cgi-bin/hgBlat) of their primer sequences in the rat genome.
Statistical significance in the phenotypic difference between LEW and F344 strains was analyzed using the Mann-Whitney U test. The genotype vs phenotype associations at individual marker loci were analyzed by traditional association statistics for QTL mapping including the χ2 test with and without Yates correction, the G test, Fisher's exact test, odds ratio (OR), and relative risk using Exemplar version 4.0.4. Corrections were not made for multiple markers because markers used were selected for being high risk for autoimmune disease or they were not independent of each other, falling into a smaller set of regions with high linkage disequilibrium values. The linkage disequilibrium (D') between markers was analyzed based on their allelic frequencies in the F2 population using Exemplar software (data not shown). Models with interacting loci were predicted using a genetic algorithm learned from training sets using the software Exemplar for Genotyping version 4.0.4′ (www.sapiosciences.com/). Colocalizing autoimmune disease QTLs in human and mouse genomes were obtained from www.grc.nia.nih.gov/branches/rrb/dna/maps.htm and www.informatics.jax.org/searches/marker_form.shtml, respectively.
The susceptible LEW and the resistant F344 rats share common MHC class Ia and class II genes. This would suggest that the difference in the severity to IRBP-induced EAU in these two strains was under the control of the non-MHC genes. We have previously identified three uveitis-regulating regions on chromosomes 4 (Eau1), 12 (Eau2), and 10 (Eau3) using a genome-wide scan of 1,287 F2 rats derived from LEW and F344 strains that were immunized with a uveitogenic protocol of R16 peptide (residues 1177–1191 of IRBP, sequence ADGSSWEGVGVVPDV) (38). In the present study, we used the same samples derived from those 1287 animals to identify additional chromosomal regions. The severity and incidence of the disease in the F2 population had been reported previously (38). LEW rats are susceptible and F344 rats are resistant to EAU induction (Fig. 1). Intermediate scores of disease were observed in the F1 progeny with an incidence of 84.6% (n = 11/13) whereas 58.2% (n = 750/1287) of the F2 rats developed EAU (38), suggesting a complex inheritance pattern for susceptibility. Overall, there was no statistically significant (p = 0.0796) influence of sex noted in the susceptibility to EAU among the affected group (n = 750) of F2 population (399 females and 351 males). However, among the top 117 most severely affected F2 rats (EAU score ≥1.5), 63.2% were females (n = 74, average EAU score = 2.61) and only 36.7% were males (n = 43, average EAU score = 2.31), possibly suggesting that females are more prone to severe ocular inflammation than males in this model of uveitis (female vs male p = 0.0078 for the incidence and p = 0.00015 for the EAU severity).
With the objective of identifying common genetic determinants for multiple autoimmune diseases, we selected microsatellite markers that were 1) closely spaced on chromosomes 2, 4, 10, and 12, for fine resolution of the earlier identified EAU susceptibility loci Eau1, Eau2, and Eau3, or 2) in the peak regions of the QTLs of other autoimmune diseases or inflammatory disorders mapped in different strains of rats (supplementary Table Ia). The biallelic genotype data of 117 F2 rats that were highly susceptible to EAU and 40 F2 rats that were resistant to from EAU were analyzed for genetic association (p ≤ 0.05) based on a χ2 test as shown in Table I. As expected, homozygosity of LEW alleles from the susceptible strain was significantly associated with uveitis susceptibility at majority of these loci except at D7Wox18, D10Mit9, D19Mit6, and D2Mgh12 (Table I). D10Mit9 and D2Mgh12 both show decreased susceptibility in animals homozygous for the resistant F allele and a suggestion of increased susceptibility of animals homozygous for the susceptible L allele, while D19Mit6 shows decreased susceptibility of individuals heterozygous for the F and L allele with suggestion of increased susceptibility of individuals homozygous for the L allele (Table I). These newly identified regions were designated as Eau4 to Eau9 (Table II) in the order of their χ2 p value for the overall marker association as given in Table I. New QTL designations were not assigned to the loci that are in linkage disequilibrium with any of the loci included in the previously reported QTL regions or with any of the loci that have higher significance within the newly identified QTL regions.
The new QTL Eau4 on chromosome 3 is marked by the locus D3Mgh10, which is also the peak locus for Iddm13, a QTL for another autoimmune disease, namely, type 1 diabetes. The association of this locus with uveitis is highly significant (p = 0.0011, Table I). F2 rats homozygous for LEW alleles at this locus showed significantly higher rate of disease incidence (p = 0.0003, OR = 17.72, Table I) and severity of uveitis (p = 0.0062, Table III) compared with other genotypes in this population suggesting that while the F/L genotype increases risk slightly, homozygosity for LEW alleles contributes strongly to the susceptibility to uveitis, resembling a recessive mode of inheritance (Fig. 2).
D2Arb24 at the distal end of rat chromosome 2 (Eau5) shows significant association (p = 0.0021) with EAU susceptibility (Table I). Interestingly, uveitis association at this locus was statistically significant with both F344 (p = 0.0039, OR = 0.39) and LEW (p = 0.0049, OR = 7.07) homozygotes (Table I). The protection conferred by the F344 allele, and the susceptibility rendered by the LEW allele revealed an allelic dose dependency (Fig. 2). A similar trend in allelic dose dependency with respect to disease severity was also observed at the D2Mgh12 locus (p = 0.0262, Table III) which is in linkage disequilibrium with the peak locus of Eau5, i.e., D2Arb24. This region of the rat chromosome 2 had been identified previously as the disease-controlling region for two other autoimmune diseases, collagen-induced arthritis (Cia10), and type 1 diabetes (Iddm3) (54) (supplementary table Ia).
The newly identified regulatory region on chromosome 7 (Eau6) at D7Wox18 (Table II) colocalizes with the QTL for a nonautoimmune disease model in rats, namely noninsulin-dependent diabetes mellitus Niddm14 at its peak locus (supplementary table Ia). Eau6 also overlaps with the QTL Niddm19. It is interesting to note that although LEW is the susceptible parental strain, homozygosity of the LEW alleles at this locus significantly reduces the severity of uveitis in the F2 rats, suggesting a protective role for the LEW allele(s) of candidate gene(s) within this QTL (QTL p value = 0.068) (Fig. 2b). The association of the homozygous LEW genotype with the uveitis phenotype is highly significant (p = 0.0009, OR = 0.28) compared with the other two genotypes of this locus (OR for F/F = 2.96, OR for F/L = 1.93, Table I).
With the use of dense polymorphic markers on rat chromosome 10, we were able to reveal two additional disease-controlling regions on this chromosome (Eau7 and Eau9) that are independent of the Eau3 QTL originally reported. The Eau7 region has a peak at the D10Mit9 locus (p = 0.0049, Table I), mapping proximal to the Eau3 QTL region. Two adjacent markers, D10Arb3 and D10Mgh10, that are in linkage disequilibrium with D10Mit9 also show significant association with EAU (p = 0.0155 and 0.0188, respectively, Table I). Genotypic associations at these three marker loci within the Eau7 region reveal significant association of both homozygous genotypes (F/F and L/L) to the phenotype of uveitis (Table I), one with protection and the other with susceptibility (Table I and Fig. 2). Alleles at none of the markers in the Eau7 locus were in linkage disequilibrium with the alleles of markers in Eau3 region (data not shown). The peak locus for Eau7 (D10Mit9) corresponds to the peak locus of the Eae3 QTL for the EAE model for multiple sclerosis and the Pia15 QTL for the PIA model for rheumatoid arthritis in rats (Table II) and overlaps with the QTL regions of Pia10 and Aia5 QTLs for autoimmune arthritis models.
In addition, a region distal to the previously reported Eau3 QTL on chromosome 10 showed significant association with EAU susceptibility. Two adjacent loci in this new QTL (Eau9), D10Rat153 (p = 0.0169) and D10Arb27 (p = 0.0186), were found not to be in linkage disequilibrium with any of the loci in Eau3 QTL (D10Mgh8, D10Arb5, D10Wox13, D10Mit8, and D10Mgh7), suggesting that these two regions are independently segregating EAU regulatory regions. A number of autoimmune disease-QTLs including Eae18, Aia5, Ciaa2, and Pia10 have been mapped to this region of the rat chromosome 10.
Because of the inclusion of several closely spaced polymorphic microsatellite markers on chromosome 10 in the current study as compared with the genome-wide scan reported earlier, we were able to redefine the peak marker of the originally described Eau3 QTL. The new peak locus is now assigned to D10Wox13 with an overall marker association of p = 0.0012 compared with the previously reported peak at D10Mit3 (p = 0.0246, Table I).
The phenotypes of all the three possible genotypes (F/F, F/L, and L/L) at Eau3 (D10Wox13), Eau7 (D10Mit9), and Eau9 (D10Rat153) on rat chromosome 10 regulatory regions were similar (Fig. 2a) where the heterozygous animals expressed intermediate phenotype when compared with the severity of uveitis in F344 or LEW homozygous animals. One allele from either the F344 or LEW strain at these three QTLs could significantly confer either protection or susceptibility to EAU (Table III, Fig. 2).
A region on chromosome 19, including the D19Mit6 locus (Eau8), that influences CIA (Cia14) in rats was also shown to regulate the EAU phenotype significantly (p = 0.0163, Table I). Although comparison of the severity of uveitis among different genotypes at this locus did not achieve statistical significance (QTL p value = 0.0716), it was interesting to note that coexistence of both F344 and LEW alleles in the heterozygous animals showed suggestions of reduced severity of uveitis (Fig. 2). This is supported by the statistically significant reduction in the odds ratio (0.33, 95% confidence interval = 0.15– 0.72) and significant association of the F/L genotype with protection from uveitis (p = 0.0046) compared with homozygotes.
Although the population analyzed in this and the previous study (38) were the same, and all the markers that showed significant disease association in the previous study at Eau1 (chromosome 4), Eau2 (chromosome 12), and Eau3 (chromosome 10) loci were included in the current study, we did not find Eau1 on chromosome 4 to be a significant chromosomal region controlling the EAU phenotype because the association statistics used here for biallelic loci considered both the highly susceptible and the resistant groups of rats. However, the Eau1 region (at D4Arb7 and D4Arb8) was found to interact with other chromosomal regions in controlling EAU susceptibility as described in the next section. Using the current statistical analysis, the Eau2 QTL (D12Rat37) on chromosome 12, reported earlier, was found to have somewhat lower significance of association (p = 0.0324) when compared with rest of the QTLs (Eau3–Eau9). The Eau3 region on chromosome 10 remained the same with a shift in the peak locus from D10Mit3 (41.35-Mb position) in the genome-wide scan study to D10Wox13 (55.86-Mb position) in the present study.
Additive allelic effects (incomplete dominance) were noticed at markers within Eau3, Eau5, Eau7, and Eau9 (Fig. 2a) with two copies of F344 or LEW alleles conferring protection or susceptibility, respectively. In the case of Eau4 and Eau6, a significant difference in phenotype (Table III) was observed only when two copies of LEW alleles (homozygous recessive) were present (Fig. 2b). Interestingly, heterozygous individuals had a significantly lower susceptibility for uveitis at locus Eau8 (D19Mit6, Fig. 2c).
Because the rat genome is not as well annotated as the human and the mouse genomes, we extrapolated rat QTL regions to the homologous regions of human and mouse chromosomes to identify putative candidate genes involved in susceptibility to EAU (Table II). This extrapolation may help to delineate the interactions between putative genes harboring the rat EAU QTLs in the pathogenesis of uveitis and to predict possible risk associated factors for human uveitis.
The syntenic human chromosomal regions of these newly identified rat EAU QTLs have been reported to be associated with and/or linked to several autoimmune diseases in patients suffering from systemic lupus erythematosus, multiple sclerosis, rheumatic arthritis, IDDM, Crohn's disease, ulcerative colitis, Graves' disease, psoriasis, ankylosing spondylitis, and allergic rhinitis (Table II).
Similarly, homologous mouse chromosomal regions of these rat EAU QTLs have been identified as the disease controlling regions for many autoimmune disease models such as EAE (Eae), type 1 diabetes (Idd), proteoglycan-induced arthritis (Pgia), autoimmune orchitis (Orch), cytokine deficiency induced colitis susceptibility (Cdcs), and Borrelia burgdorferi-associated arthritis (Bbaa) and various immune response regulating traits like T cell phenotype modifier (Tcpm), TCR-induced activation (Tria), lymphoproliferation modifier (lprm), immunoregulation (Im), and cytokine-induced activation (Cinda) identified in crosses of different inbred mouse strains (Table II).
EAU is a polygenic disease in which any one gene/locus is neither necessary nor sufficient to confer susceptibility. Using the genetic algorithm included in the Exemplar program package, we predicted models of interacting loci that determine EAU susceptibility. This is a machine learning approach to find combinations of alleles at multiple loci correlated with the disease phenotype. Permutation testing was then applied to assess the reliability of the models. These models thus predict the contributions from different loci in determining susceptibility to uveitis. We identified genotypes of several minor loci that significantly influenced the severity of EAU in rats when coexisting in combination (used three loci) with certain genotypes of other major or minor loci using the AND model (Table IV), and genotypes of several loci that interacted in a substitutive fashion using the AND/OR model (Table V). All the models were iterated for 10,000 permutations.
The results from the AND model of interacting loci are given in Table IV. The models had higher significance for uveitis association when compared with the independent association as individual marker loci (Table IV and supplementary table Ib). A total of 23 models could be predicted with χ2 p < 0.05, of which four had permutation p values >0.05. From these models five new chromosomal regions could be identified that indirectly modified the contributions of independent EAU QTLs to disease susceptibility. These interacting loci are: 1) the region of chromosome 12 (models 1–6, Table IV) marked by four closely linked loci D12Rat37, D12Rat38, D12Wox5 and D12Mit1 interacting with the D14Rat3 locus on chromosome 14 and the D1Rat90 locus on chromosome 1. The chromosome 12 region has previously been identified as the Eau2 QTL; 2) the region of chromosome 2 spanning the 30–35 Mb position marked by linked loci D2Rat154, D2Rat116, D2Rat11, and D2Rat10 (models 7–12, Table IV). This region interacts with Eau9 on chromosome 10 (D10Arb27, D10Mgh5, and D10Wox20). 3) Locus D7Rat26 on chromosome 7 interacts with multiple regions to modify EAU susceptibility (models 16–20, Table IV), one of which is Eau1 (D4Arb7 and D4Arb8) that was originally identified. 4) The chromosome 1 locus D1Rat90 interacts with multiple regions (models 13–16, Table IV), and 5) the chromosome 5 locus D5Mit10, which is the peak marker for Eae16 (supplementary table Ia), which interacts with the regions on chromosomes 2, 7, and 10. These models show that heterozygous genotypes at these interacting loci modulate the EAU susceptibility that is associated with EAU QTLs on the same or different chromosomes. The AND model also predicted the interaction between multiple loci in Eau3 (D10Mit8 and D10Wox13) and a locus in Eau9 (D10Arb27) on chromosome 10 (model 23 in Table IV).
Using the AND/OR model predictions we identified several complex interacting loci that significantly affected the susceptibility to uveitis (Table V). The significance of these models to uveitis association is much higher than the independent loci involved in the interaction (Table V and supplementary table Ib). Using this strategy, it is evident that one locus can substitute for another in determining susceptibility when it occurs in combination with another locus or loci. Because many of the loci involved in the AND/OR model did not show significant association independently, the genetic contribution of each of these loci to the overall expression of EAU phenotype in the population is probably minimal. However, as can be seen in Table V, in an animal with the right genetic background, they can have a dramatic effect. Perhaps more importantly, as seen in the AND model set, they can modify the susceptibility and severity imparted by one of the independently identified loci. Homozygous F344 genotypes at loci D1Wox25, D2Rat154, D10Rat117, D10Mit1, D12Mit4, and D19Mit6 (Table V) were involved in these interactions, suggesting probability of disease promoting alleles from the resistant F344 strain.
Autoimmune diseases occur in 3–5% of the population and are manifested in multiple forms depending on the tissues or the organs involved. Despite these clinical differences, several autoimmune diseases appear to share common susceptibility loci, probably representing common factors in pathogenesis of the final disease. These colocalizing regions contain gene clusters which are differentially expressed during an autoimmune disease process (55) and are therefore of clinical interest as potential therapeutic targets. It is notable that genetic association studies for uveitis in patients suffering from ankylosing spondylitis (37) and Behcet's disease (40) also show overlapping genetic regions with those identified in patients with multiple sclerosis, Crohn's disease, rheumatoid arthritis, IDDM, and psoriasis. These findings strongly support the concept that common pathways may be involved in uveitis and other autoimmune diseases.
Studies to delineate specific uveitis-associated genes are limited in humans due to the wide clinical heterogeneity and low incidence of a particular disease entity. Inbred animals serve as informative models in such situations because genetic crosses can be made to segregate the phenotypes, and the influence of different genetic regions on the manifestations of the disease can be studied under controlled genetic and environmental conditions. In general, results obtained from animal models of other diseases have been found to corroborate with the association studies from patient data when extrapolated to the conserved syntenic regions in humans (47, 48).
The first systematic study on the genetic control of uveitis, reported by our group using the rat model, demonstrated that Eau1, Eau2, and Eau3 overlapped with the disease regulating regions of other autoimmune disease models including EAE, CIA, and type 1 diabetes (38). Since our previous study (38) was based on a whole genome-wide scanning approach using widely spaced microsatellite markers, it is highly likely that some important disease controlling regions may have been missed. By choosing polymorphic microsatellite marker loci located within the QTLs of other disease models, the potential for discovering new EAU control regions was enhanced. Using this approach, we have identified several new EAU QTLs Eau4–Eau9 (Table II). The overlapping autoimmune disease QTLs in rats and the list of candidate genes harboring the new EAU QTLs are given in Table II. Many of these genes have been reported to be associated with various autoimmune diseases in patients and in mouse models (Table II). Although we tested a total of 124 markers (supplementary table Ia), standard corrections for multiple testing were not included in our analysis. There are two basic reasons for this. First, each of these markers was selected for previously being associated with autoimmune diseases and had a relatively high a priori risk of being associated with EAU. Second, markers tested on chromosomes 2, 4, 10, and 12 show significant linkage disequilibrium, as some of them were selected to refine previously identified EAU-associated regions. Thus, any corrections made would be for a relatively small group of risk regions rather than on the basis of the number of markers tested. However, significance of individual and multiple model markers was corrected by permutation testing.
The two parental strains of rats selected for this study match in their MHC haplotype but differ drastically in their susceptibility to many autoimmune diseases and inflammatory responses. Because these strains are not selectively bred for EAU susceptibility or resistance, it may be possible to detect some segregating QTL alleles that are “protective” in the susceptible LEW strain and “disease promoting” in the resistant F344 strain. This is evident from the associations of the newly identified locus D7Wox18 in the Eau6 QTL where presence of two copies of LEW alleles resulted in decreased EAU severity (Fig. 2b). Similarly, homozygous the F344 genotype at D1Wox25, D2Rat154, D10Rat117, D10Mit1, D12Mit4, and D19Mit6 loci were found to be associated with disease susceptibility when in combination with the LEW alleles of the interacting loci in the AND/OR predicted models (Table V). Overdominance of the phenotype in heterozygous animals at Eau8 (Fig. 2c) is an interesting phenomenon. Compared with the hybrid vigor for survival fitness, this type of allelic interaction was also reported earlier at the suggestive locus for the arthritis model (PIA) on rat chromosome 14 (56) and at different MHC loci for most of the infectious (57) and autoimmune diseases (58).
Models predicting interacting loci provide some additional insights into the genetic risks for EAU. These interacting models (Tables (TablesIVIV and andV)V) are supportive of the polygenic mode of inheritance of EAU susceptibility. This has direct relevance in the immunopathogenesis of tissue specific autoimmune diseases where complex network of interacting proteins such as cytokines and chemokines, signal transduction molecules, transcription factors, neurotransmitters and their receptors, and proteins involved in oxidative stress leads to immune mediated tissue damage.
Various markers of a major immunoregulatory region on chromosome 12 (Table IV) show significant interaction with the D14Rat3 locus. This chromosome 14 region harbors genes involved in immune cell migration such as chemokine factors (Cxcl9, Cxcl10), vascular adhesion protein (Vap1) and proteins with cytokine functions like bone morphogenetic protein (Bmp3) and osteopontin (Spp1). The level of expression or function of these proteins may be regulated by the gene products of the interacting chromosome 12 region. The candidate genes from chromosome 12 QTL possibly taking part in this interaction are neutrophil cytosolic factor (Ncf1) that is involved in leukocyte transendothelial migration; erythropoietin (Epo) and platelet-derived growth factor (Pdgf) involved in cytokine-cytokine receptor interaction; and other immunologically relevant proteins like tripartite motif protein (Trim50), galectin-related interfiber protein (Grifin), the adaptor protein Aps, and the translation initiation factor Eif2ak1. Among these osteopontin (Spp1) on chromosome 14 (59, 60) and Ncf1 on chromosome 12 (61) have been previously shown to be associated with autoimmune disease models.
Among the predicted models (Table IV) the chromosome 2 region mainly interacts with the chromosome 10 EAU QTLs that harbor the cytokine gene cluster and several other immunologically relevant genes (Table II). The chromosome 2 region (Table IV) includes the serotonin receptor (Htr1a), and cyclin B (Ccnb1), which play an important role in inflammatory responses. Studies have shown that Htr1a binds to serotonin and increases the production of proinflammatory cytokines IL-6 and TNF-α (62).
The previously identified Eau1 (D4Arb7) locus interacts with a region on chromosome 7 (D7Rat26, Table IV) that includes genes encoding cytokine responsiveness factor cyclin-dependent kinase (cdk4) and inhibin-β (Inhbc), which is involved in TGF-β signaling and cytokine-cytokine receptor interaction pathways. The chromosome 5 locus at D5Mit10 interacts with many loci (Table IV) on different chromosomes. The gene products aquaporin 3 and 7 (Aqp3 and Aqp7) from this region have been shown to be associated with lymphocyte activation and dendritic cell maturation (63), which are critical for mounting effective immune response to self Ags like retinal proteins.
Almost all of the QTLs identified for EAU in this model harbor regulatory regions for other autoimmune disorders (Table II). However, overlapping QTL regions do not necessarily mean that identical genes control the outcome of all these diseases. A finer resolution of each of these loci and eventually identification of the specific genes and functional studies of their protein products will be necessary to identify single gene effects.
The current study only partially identified some of the genetic regions shared between EAU and other autoimmune disease models. This was largely due to the lack of genetic variability between F344 and LEW strains at some known autoimmune QTL regions. Such overlapping regions would have to be studied in a different susceptible/resistant rat strain combination, having identifiable polymorphisms in those regions. Also, the peak markers for other autoimmune disease QTLs may not precisely coincide with the peaks for EAU QTL within the same chromosomal interval due to the differences in the strains used in this and other studies. However, these experiments have implicated in EAU an initial set of genetic loci that will serve as candidate loci for uveitis in humans and other species. In addition, we have begun to unravel the complex interactions between genetic susceptibility loci, which presumably correlate to physiological interactions in the pathogenesis of uveitis. For practical reasons, we have not attempted to use all the polymorphic markers within the entire QTL interval of each of these autoimmune disease QTLs. Nevertheless, our current data strongly support the conclusion that uveitis is indeed an autoimmune disease that is regulated by a group of common genetic determinants that modulate other autoimmune diseases and that it shares common pathways and mechanisms with other autoimmune diseases, thereby suggesting shared methods for therapy.
The authors have no financial conflict of interest.
1This work was supported by The Intramural Research Program of the National Eye Institute, National Institutes of Health.
4Abbreviations used in this paper: EAU, experimental autoimmune uveitis; IRBP, interphotoreceptor retinol-binding protein; QTL, quantitative trait locus; EAE, experimental autoimmune encephalomyelitis; CIA, collagen-induced arthritis; IDDM, insulin-dependent diabetes mellitus; PIA, pristane-induced arthritis; AIA, adjuvant-induced arthritis; OR, odds ratio.
5The online version of this article contains supplemental material.