3.1. Genome Scans and Linkage Analysis in Autoimmune Diseases
During the last 10 years, several groups have collected multiplex families of the various autoimmune diseases for genome screening with microsatellites and linkage analysis. The very first report of linkage analysis came from the field of type 1 diabetes (T1D) mellitus (Davies et al., 1994) followed by multiple sclerosis (MS) (Ebers etal., 1996), rheumatoid arthritis (RA) (Cornelis etal., 1998), and systemic lupus erythematosus (SLE) (Gaffney et al., 1998; Moser et al., 1998). I limit this review to these diseases, adding only Crohn's disease (CD) and ulcerative colitis (UC).
3.2. Autoimmune Diabetes (T1D)
As mentioned above, the first genome scan performed with multiplex families was on this disease. It identified 20 potential susceptibility regions of which the locus for insulin was already known (Davies et al., 1994). A second genome scan did not allow replication of most of the regions identified in the first scan (Concannon et al., 1998). This led to skepticism regarding genome scans, probably because of the inflated expectations promoted initially by some investigators.
Later on, through careful fine mapping, it was possible to pinpoint a variable number tandem repeat (VNTR) as the causative polymorphism within the insulin gene that represented the IDDM2 locus. The VNTR was classified in types
I-III depending on the size of the repeats. It was also found that class III alleles could associate with low expression levels of insulin despite the fact that these alleles have previously shown a protective effect. In Caucasians, the insulin gene VNTR alleles divide into two discrete size classes. Class I alleles (26-63 repeats) predispose in a recessive way to T1D, while class III alleles (140 to more than 200 repeats) are dominantly protective. The protective effect may be explained by higher levels of class III VNTR-associated insulin mRNA in thymus such that elevated levels of preproinsulin protein enhance immune tolerance to preproin-sulin, a key autoantigen in T1D pathogenesis. The paternal effect (preferential transmission of a given allele from the father) was observed only when the father's untransmitted allele was class III. Therefore, the hypothesis has been that in individuals carrying the class I VNTR alleles the low expression of proinsulin in the thymus leads to lack of induction of immune tolerance (Undlien et al., 1995; Bennett et al., 1996, 1997; Vafiadis et al., 1997; Ahmed et al., 1999).
A second important region of linkage identified by the genome scans was at 2q33. Recently, an extremely thorough analysis of the region, including the genes ICOS and CTLA4, was performed. A common allelic variation was found to be genetically associated not only with T1D but also with Graves' disease and autoimmune hyperthyroidism correlated with the presence of a molecule producing lower levels of a soluble alternative splice form of CTLA4 lacking the CD80/CD86 ligand-binding domain, providing an interesting mechanistic explanation of the genetic association with CTLA4 (Nistico et al., 1996; Hill et al., 2000; Ueda et al., 2003). In short, the insulin gene VNTR and the splice form of the CTLA4 gene have with certainty been identified as susceptibility genes for T1D. Various other IDDM loci were detected in the genome scans performed to date and the genes contributing to the effects are being identified.
Just as with T1D, full-genome scans have been performed for MS (Ebers et al., 1996). In the initial genome scan, possible loci were detected on chromosome 6p21 and 5p and evidence was later provided on 5p with 21 Finnish MS families (Kuokkanen et al., 1996). A full-genome scan also performed in the Finnish families identified a region in chromosome 17q22-24 (Kuokkanen et al., 1997) also revealed in a genome scan from the UK and in a limited scan in Scandinavian families and in a whole Nordic linkage analysis. However the 5p region has not been replicated (Oturai et al., 1999; Akesson et al., 2002). A new region was found in chromosome 19q13, in particular the 19q13.1 subregion when a larger set of Finnish families were stratified for HLA-DR15, known to be strongly associated with MS (Reunanen et al., 2002), as seen with HLA and several other autoimmune diseases (Ligers et al., 2001; Masterman and Hillert, 2002). An interaction between markers at HLA-DR15 and the CTLA4 gene has been described in MS (Alizadeh et al., 2003). The CTLA4 gene has been an interesting candidate as linkage has also been found to 2q33, where this gene lies. It appears that the identification of genes for MS is more difficult than for other autoimmune diseases. An interesting candidate gene, CD45 (PTPRC) was found to have a splice variant associated with MS in German individuals, but this association was not replicated in several other studies (Jacobsen et al., 2002; Gomez-Lira et al., 2003; Sabouri et al., 2003). In short, the most consistent genomic region containing a potential gene for MS susceptibility is the 17q22 region, but no gene has been identified to date though some interesting observations have been done (Nelissen et al., 2000; Saarela et al., 2002; Chen et al., 2004).
Several genome scans have been produced for RA in France (mainly south European) (Cornelis et al., 1998) and the UK (MacKay et al., 2002), and two screens from the NARAC Consortium in the US focused on European Americans and erosive arthritis (Jawaheer et al., 2001, 2003). As for the other autoimmune diseases, the HLA region at 6p21 showed the strongest linkage signal in all studies performed to date. The results from the various scans were rather consistent for non-HLA regions in spite of the apparent nonreplication observed initially. The most outstanding novel region that was found in the French genome screen was located at 18q22 as well as at 3q13. When the first NARAC genome scan appeared the results were quite different from those identified in the French screen. The main regions identified were at chromosomes 1, 4, 12, 16, and 17, but the second NARAC screen identified chromosome 18q21 and new loci at 9p22 and 10q21. The combined analysis supported regions from both scans. Regions most influenced by using the HLA-DRp alleles as covariates (shared epitope alleles such as DRp1*0401) were in chromosomes 1p, 1q, and 18q (Gregersen et al., 1987; Jawaheer et al, 2002).
The results of the UK genome scan were rather disappointing as the only region reaching nominal significance was the HLA while all non-HLA regions did not. However, the study results got support from the previous NARAC and French genome screens in six regions. To date, no gene has been identified through the direct use of the multiplex families, but the NARAC material has been used in alternative approaches, as will be described further below.
Work performed by Japanese groups has used a completely different approach and has identified at least three new interesting and potential genes involved in RA, PADI4, RUNX1, and SLC22A4 (OCTN1). The approach consisted in the analysis of a dense map of SNPs in regions selected for having been found in human linkage studies or containing interesting candidate genes. The two regions were in 1p36 and in 5q31, the latter being that of the cytokine gene family. The former study identified the PADI4 gene (peptidylarginine deiminase type 4), an enzyme involved in citrullination, a process thought to be defective in RA where antibodies against citrullinated peptides are major biomarkers of the disease (Suzuki et al., 2003). Whether PADI4 is the gene for RA is as yet uncertain as the results have not been replicated (Barton et al., 2004).
The second gene identified by the Japanese groups was after the analysis of the 5q31 region with a very dense map of SNPs (Tokuhiro et al., 2003). This analysis identified the SLC22A4 gene, an organic cation transporter member of the OCTN family that was recently found also associated with CD (Peltekova et al., 2004). The mutation found within the first intron of this gene results in the disruption of a binding site for the runt-domain family of regulatory proteins known as RUNX, in particular RUNX1. A similar disruption was previously identified in SLE (described below). Thus the Japanese approach led to the discovery of PADI4, SLC22A4 (OCTN1), and preliminarily RUNX1.
Interestingly, another candidate gene, TNFRII, has been found associated in familial RA in Europeans but not in sporadic cases (Dieude et al., 2002). However, clinically, no differences have been described between sporadic and familial cases of RA. The possibility exists that several genes are responsible for disease susceptibility in this cytogenetic region.
Most recently, the genotyping of 87 SNPs that were considered functional (showing amino acid changes) was performed in a case-control design in RA. This analysis led to the identification of an amino acid change in codon 620 of the protein tyrosine phosphatase PTPN22, also known as LYP (Begovich et al., 2004), an enzyme that dephosphorylates and inactivates antigen-induced T cell activation. The change converted an arginine to tryptophan and disrupted the binding of the SH3 domain of LYP to the kinase Csk. This variant was found previously to be associated with T1D (Bottini et al., 2004).
3.5. Crohn's Disease and Ulcerative Colitis
As for the other autoimmune diseases, genome scans have been performed for both CD and UC (Rioux et al., 2000). The best results have been obtained with the former while linkages remain elusive for UC. Some of the most important results were found for chromosomes 5q31-33, 3p, and 6p (Rioux et al., 2001) as well as for chromosome 12 in a separate scan and in chromosome 16 in a French study (Lesage et al., 2000). Importantly, the first breakthrough in gene identification in autoimmune diseases and complex diseases came from identification of mutations in the CARD15 gene in CD (Hugot et al., 2001; Ogura et al., 2001). CARD15, or NOD2, is a leucine-rich domain-containing protein that has an inhibitory role and acts as a receptor for intracellular pathogens. Three main variants were identified in three independent low-frequency haplotypes, thus challenging the common disease-common variant hypothesis. Weak association of CARD15 has been found with UC and epistasis involving 5q31 and CARD15.
The next region to be studied, 5q31, was recently fine-mapped and as a surprise the results pinned down to two related genes, SLC22A4 (OCTN1) and SLC22A5 (OCTN2) (Peltekova et al., 2004). As noted previously, variants of the SLC22A4 gene were identified in susceptibility to RA. Two mutations were identified in these genes in CD and it remains uncertain if the main contributor to susceptibility is SLC22A4 or SLC22A5. Recently, it has been observed that OCTN1 is importantly expressed in lymphoid tissues, while OCTN2 is mainly expressed in kidney and is a proven carnitine transporter involved in carnitine deficiency (Xuan et al., 2003; Yamada et al., 2004). The mutation found in OCTN1 was a missense substitution, while that identified in OCTN2 was a G-C transversion in the promoter, leading to changes in gene expression (Peltekova et al., 2004). However, o gene has been identified for UC.
The prototype of systemic autoimmune disease, SLE, has been extensively studied and four groups have performed genome screenings for various populations (Gaffney et al, 1998; Moser et al, 1998; Shai et al, 1999; Lindqvist et al, 2000). The disease has several features that make the genetic component even more important than for the other autoimmune diseases. There are animal models that develop the disease spontaneously. Conversely, the genetic component for SLE is not dominated by the HLA with the exception of some populations where HLA DR3 has been implicated. However, DR3 is in linkage disequilibrium with a null allele of C4, making interpretations difficult (Arnett, 1985; Howard et al., 1986). One of the genome scans in Caucasians detected the MHC region as an important susceptibility locus. A similar finding was observed in Caucasian Swedes, but not in other ethnic groups (Gaffney et al., 1998; Lindqvist et al., 2000).
The most consistent loci identified for SLE were found in chromosome 1 (1p36, 1q23, 1q31, and 1q42-44), but the most significant were detected in chromosomes 2q37, 4p13, 4p15, and 16q12 (Gaffney et al., 1998; Moser et al., 1998; Shai et al., 1999; Lindqvist et al., 2000). Other minor loci have been identified by several groups and replicated as well, but remain with minor effects (Gaffney et al., 2000; Tsao, 2000). Some of the strongest linkages were found in families of Icelandic and Swedish origins (Lindqvist et al., 2000). Recently, a genome screen in Finnish families identified and confirmed loci in 14q and 6q observed in previous scans (Koskenmies et al., 2004). Another locus identified recently in families of Argentine-European origin is on chromosome 17q12 (Johansson et al., 2004). As expected, high degree of heterogeneity has been found between populations.
The region 1q23 contains as major candidate genes the low-affinity receptors for the Fc portion of immunoglobulins or FcyR. Earlier studies had been performed on these genes, because their products are involved in immune complex clearance (Salmon et al., 1996). The main association has been found for FcyRIIIA, with a clearly functional polymorphism (valine to phenylalanine substitution) affecting immunoglobulin-binding affinity, but association to FcyRIIA has also been observed (Seligman et al., 2001; Zuniga et al., 2001). However, the functional polymorphism of FcyRIIA (histidine to arginine substitution) has also been associated. Recently it was proposed that variants in both genes are required for susceptibility and act functionally as if it would be a compound heterozygos-ity situation (Magnusson et al., 2004). In other populations, in particular Asians, the polymorphism found in the FcyRIIB gene that could affect the transmembrane region (Li et al., 2003) is associated with SLE.
Most of the loci on chromosome 1 have been found also to be quantitative trait loci in animal models of SLE although importance of the FcG receptors has not been confirmed in mice. Instead, these genes confer susceptibility in animals deficient for the FcyRIIB crossed with lupus susceptible mice (Bolland et al., 2002).
A breakthrough in SLE has been the identification of variation in the PDCD1 gene coding for the immunoreceptor PD-1, a 50-kDa protein expressed on the surface of various early lymphoid tissues and activated T and B cells (Prokunina et al., 2002). It is involved in inhibition of T cell activation and in thymic tolerance. It had been previously shown that a knockout model for PD-1 developed a lupus-like disease, and the gene became an important candidate when linkage was identified to 2q37.3 in sets of Nordic multiplex families with SLE, where the gene for PD-1, PDCD1, had been mapped (Magnusson et al., 2000). The variant identified to be associated segregated in a single haplotype and disrupted the binding site for the runt-related silencer RUNX1, the same identified later for RA, and even psoriasis (Helms et al., 2003; Tokuhiro et al., 2003). Several studies have now replicated the genetic association initially found in lupus and have also identified association of the same polymorphism with RA and T1D in Europeans (Nielsen et al., 2003; Lin et al., 2004, 2004; Prokunina et al., 2004a,b). Association with RA was identified with a second polymorphism in PDCD1 in the Chinese (Lin et al, 2004).
The overlap of clinical manifestations, the presence of similar autoantibody specificities in several diseases, and the common immunologic features suggest that genetic susceptibility may be shared among various autoimmune diseases as it was suspected in earlier studies. Now that the genes are beginning to be identified, it is indeed confirmed that this is the case. For example, a splicing mutation in CTLA4 has been found associated with T1D as well as with SLE. PDCD1 also has been associated with RA and T1D, and, as mentioned previously, SLC22A4 has been associated both with CD and with RA. The coding and functional variant of PTPN22 was first identified in T1D and then independently discovered in RA. Association was also recently identified with SLE (Kyogoku et al., 2004). However, the gene most commonly shared among autoimmune diseases has been the HLA DRß1 (Table 5.1). Does this gene modify the clinical picture of the autoimmune diseases depending on the variation? Why do we identify several genetic variants being shared among clinically distinct autoimmune diseases? "How does the combination of susceptibility variation lead to disease" is becoming a question with a higher probability to be answered.
Table 5.1. Genes Shared between Autoimmune Diseases
SLC22A4 RA/Crohn's disease
Was this article helpful?