|Home | About | Journals | Submit | Contact Us | Français|
We have used the public sequencing and annotation of the mouse genome to delimit the previously resolved type 1 diabetes (T1D) Idd18 interval to a region on chromosome 3 that includes the immunologically relevant candidate gene, Vav3. To test the candidacy of Vav3, we developed a novel congenic strain which enabled the resolution of Idd18 to a 604 kb interval, designated Idd18.1, which contains only two annotated genes: the complete sequence of Vav3, and the last exon of the gene encoding NETRIN G1, Ntng1. Targeted sequencing of Idd18.1 in the NOD mouse strain revealed that allelic variation between NOD and C57BL/6J (B6) occurs in non-coding regions with 138 single nucleotide polymorphisms (SNPs) concentrated in the introns between exons 20 and 27, and immediately after the 3′ UTR. We observed differential expression of VAV3 RNA transcripts in thymocytes when comparing congenic mouse strains with B6 or NOD alleles at Idd18.1. The T1D protection associated with B6 alleles of Idd18.1/Vav3 requires the presence of B6 protective alleles at Idd3, which are correlated with increased IL-2 production and regulatory T cell function. In the absence of B6 protective alleles at Idd3, we detected a second T1D protective B6 locus, Idd18.3, which is closely linked to, but distinct from, Idd18.1. Therefore, genetic mapping, sequencing, and gene expression evidence indicate that alteration of VAV3 expression is an etiological factor in the development of autoimmune beta-cell destruction in NOD mice. This study also demonstrates that a congenic strain mapping approach can isolate closely linked susceptibility genes.
Type 1 diabetes (T1D) is a multifactorial autoimmune disease, and both environmental factors and genetic loci spread throughout the genome govern T1D onset. The NOD mouse model of T1D (1) shares several features of etiology and genetics with human T1D, and has been instrumental in the identification of genes involved in T1D susceptibility and how variants of these genes influence the immune system (2-7). Our laboratory has previously utilized a congenic strain mapping technique in the NOD mouse model, in which genome segments from a T1D resistant mouse strain (B6 or C57BL/10J) are introgressed into the susceptible NOD background, to localize T1D susceptibility loci (2, 8). Four of these loci, insulin-dependent diabetes (Idd) 3, Idd10, Idd17, and Idd18, have been identified on mouse chromosome three, and only Idd3 and Idd10 have proposed candidate genes (9-12). The T1D-protection associated with the 650 kb Idd3 region is most likely accounted for by the differential expression between the NOD and B6 haplotypes of the IL-2 gene (4), and the proposed candidate gene for the 950 kb Idd10 region is Cd101 (12).
The initial mapping of Idd18, which is closely linked and distal to Idd10, was performed by sequentially truncating the Idd18 interval of a single homozygous B6-derived congenic segment that spanned Idd10 and Idd18. The resultant congenic strains that had lost the B6 protective alleles at Idd18 had higher T1D frequencies compared to the non-truncated congenic strains, but were more protected from T1D than NOD mice as B6 alleles at Idd10 provided protection (13). The determination of the recombination points of these novel congenic strains localized Idd18 to a genetic distance of 5.1 cM (13). Subsequently, a second congenic strain mapping strategy to positionally clone Idd18 was utilized, which exploited the enhanced T1D protection observed when B6-derived alleles are present at both the Idd10/18 and Idd3 intervals (14). This second strategy sequentially truncated the Idd18 interval of a bicongenic strain homozygous for two B6-derived introgressed segments: one spanning Idd10 and Idd18, and a second spanning Idd3. Notably, the protection associated with Idd10 is not observed in the context of B6 alleles at Idd3 (10); therefore, when B6 alleles at Idd18 were eliminated in these congenic strains, the protection observed was equal to that provided by Idd3 alone. T1D frequencies of novel congenic strains produced from this mapping effort reduced the Idd18 interval to a genetic distance of 2.04 cM (10). However, at that time (2001), the gene content of the 2.04 cM Idd18 interval was unknown.
Here we have used B6 mouse genome sequencing data and Ensembl gene annotation to design novel polymorphic microsatellite markers and characterize the gene content of Idd18. This enabled us to refine the 2.04 cM interval, mapped in the context of B6 protective alleles at Idd3, to a physical distance of 1.27 Mb and identify the immunologically relevant gene, Vav3, as a candidate. To test the candidacy of Vav3, we developed a novel congenic strain homozygous for B6 alleles at Idd3 and an introgressed B6 segment spanning Idd10 and Idd18 truncated immediately upstream of Vav3. This new congenic strain mapping (in the context of B6 alleles at Idd3) has refined Idd18 to an interval of 604 kb, now designated Idd18.1, that contains only two annotated genes, Vav3 and the last exon of Ntng1; thereby highlighting the candidacy of Vav3. The likelihood that Vav3 is an Idd gene was strengthened by our demonstration of extensive sequence variation in Vav3 non-coding regions, and allele-dependent expression of VAV3 mRNA. We also tested the level of protection associated with Idd18.1 without protective B6 alleles at Idd3. Unexpectedly, in the absence of B6 alleles at Idd3, the protection associated with Idd18.1 was not observed, and a different Idd18-associated interval, designated Idd18.3, was found to be required for Idd18-mediated T1D protection. The Idd18.3 interval is 996 kb and sequence polymorphisms highlight Fam102b as a candidate gene.
Primer3 (15) was used to design primers for PCR and primer and probe sets for real-time quantitative RT-PCR (qPCR). These were synthesized by Sigma-Genosys (Haverhill, U.K.); the probes were dual labeled with TAMRA and FAM fluorescent dyes. Sequences of D3Nds and D3Mit microsatellite markers are available at http://www-gene.cimr.cam.ac.uk/todd/public_data/mouse/NDS/NDSMicrosTop.html and http://mouse.ensembl.org, respectively. All remaining primers and probes used in this study are available in Supplemental Tables 1, 2, and 3.
DNA for genotyping was extracted from tail biopsies using a salting out method (16). PCRs included 25 ng DNA, 0.5 U AmpliTaq Gold® (Applied Biosystems, Foster City, CA), 0.2 mM dNTPs, 1-4 mM MgCl2, 10% v/v glycerol, and 62.5 ng of each primer. Microsatellite markers that were not resolvable by 4% agarose gel electrophoresis were fluorescently labeled on the forward primer and PCR products were electrophoresed on an ABI 3100 Sequencer (Applied Biosystems, Foster City, CA) and sized using GeneScan® (version 3.5.1) and Genotyper® (version 3.6) software (Applied Biosystems, Foster City, CA). RFLP markers were digested with the appropriate restriction enzyme (NEB, Herts, UK) and were resolved on 2-3% agarose gels.
All mice were housed under specific pathogen-free conditions, and the appropriate institutional review committee approved experimental procedures. NOD/MrkTacfBR (from here on designated as NOD) mice were purchased from Taconic Inc. (Germantown, NY). The derivation of the following congenic strains has been described previously: line 1098, NOD.B6 Idd3R450 (also known as NOD.B67) (N12) (9, 17); line 1538, NOD.B6 Idd3 Idd10 Idd18R323 (N12) (10); line R135, NOD.B6 Idd3 Idd10 Idd18R135 (N12) (10); line 1101, NOD.B6 Idd10R2 (N8) (13); line R3, NOD.B6 Idd10R3 (N9) (13); and line R1, NOD.B6 Idd3 Idd10R1 (N12) (10). Line 1100, NOD.B6 Idd3 Idd10 (N12), was developed contemporaneously with the strains described in Lyons et al (10).
To develop the congenic mouse strains used to test the candidacy of Vav3 as Idd18, NOD.B6 Idd3 Idd10 Idd18 (line 1538) congenic mice were crossed with NOD mice. The progeny were intercrossed and tail DNA was genotyped to identify mice with a recombination event proximal to Vav3, which would produce a chromosome that was NOD at Vav3 and B6 for the remainder of the Idd18 interval. One mouse with a recombination event proximal to Vav3 was backcrossed to NOD and progeny heterozygous for the desired recombination event, and that had retained the B6 allele at Idd3, were intercrossed to produce homozygous mice for line 2399. To produce line 2412, mice heterozygous for the desired recombination event and that had an additional recombination event between Idd3 and Idd10, resulting in a heterozygous genotype at Idd10 and a NOD homozygous genotype at Idd3, were selected and intercrossed.
Subsequent to the completion of the studies presented in this manuscript, the genetic backgrounds of lines 1538 and 1101 were genotyped by ParAllele Biosciences (South San Francisco, CA) using their 5 K mouse SNP chip. The assay demonstrated a 3.6 Mb segment of B6 DNA on chromosome 18 in line 1101, in addition to the known congenic regions on chromosome 3; whereas line 1538 only contained the known congenic regions. Line 1101 was backcrossed to remove this contaminating B6 DNA segment and the chromosome 3 B6 DNA segment was fixed to homozygosity; this new congenic strain was designated line 7754. The T1D frequencies of lines 1101 and 7754 were equivalent when assessed contemporaneously (Supplemental Figure 1). Lines 1098, 7754, 1538, and 2412 are available from Taconic Inc. (Germantown, NY). Lines R1, R3, R135, 1101, and 2399 are no longer extant. Note that all the congenic mouse strains used in this study are homozygous for NOD alleles at the Idd17 region, which is located between Idd3 and Idd10 (11).
All diabetes cumulative frequency studies were conducted using female mice. The presence of T1D was tested every 10 to 14 days beginning at approximately 80 days of age by the detection of urinary glucose >500 mg/dL using Diastix (Miles, Elkhart, IN). Studies were terminated at 214 days of age. Kaplan-Meier survival curves were plotted for each mouse strain, and these were compared using the logrank test (Prism®4 software).
The resequencing of Idd18.1 in the NOD mouse strain involved aligning the bacterial artificial chromosome (BAC) clone end sequences of the NOD library against the B6 mouse genome sequence (18). From this, five NOD BAC clones that formed a minimal sequencing tile path spanning the 604 kb Idd18.1 interval were selected and sequenced at WTSI, and deposited at EMBL (http://www.ebi.ac.uk/embl/) (clone DN-315E9, accession number CR936848; DN-164L21, CR938729; DN-127N6, CR936838; DN-250A21, CR936849; and DN-250O19, CT010461). To identify polymorphisms between NOD and B6 in the Idd18.1 interval, the NOD BAC clone sequences spanning the Idd18.1 interval were aligned to the B6 mouse genome sequence (NCBIm build 36) using SSAHA (19), and SNPs, microsatellites, and other insertion/deletion polymorphisms were determined computationally and confirmed manually. The polymorphisms were entered into T1DBase (20, 21) and displayed graphically using GBrowse (22). The SNP density plots were generated by counting the number of SNPs in 10 kb windows, sliding 2 kb at a time, and plotting the count at the midpoint of each window. All the annotation of the Idd18.1 and Idd18.3 intervals can be viewed at http://www.t1dbase.org.
To view B6 and NOD/LtJ variation across the Idd18.3 interval, the National Institute of Environmental Health Sciences (NIEHS) sponsored resequencing data, which was generated through the resequencing of 15 inbred strains of mouse by Perlegen Sciences (23), was downloaded, converted into gff format, and loaded into GBrowse.
Whole spleen, kidney, and brain were harvested from 9-week old female mice and immediately homogenized in TRIzol® using a polytron. Spleen and thymus were obtained from 3-week old male mice and single cell suspensions prepared. Erythrocytes were removed from the splenocyte samples. Total RNA was extracted from the cells and homogenized tissues using TRIzol® following the manufacturer’s instructions. 1 μg of total RNA was used as the template for cDNA synthesis using SuperScript™ II reverse transcriptase (Invitrogen Ltd. Paisley, U.K.) following the manufacturer’s instructions.
To verify the VAV3 genetic structure, the VAV3 and VAV3.1 mRNA sequences, NM_020505 and NM_146139, were aligned to the B6 genomic sequence spanning Idd18.1 using est2genome from the EMBOSS suite of programs (24, 25). NM_020505 was searched against the mouse expressed sequence tag (EST) database at NCBI (26) using BLAST (27). Five ESTs (BG060768, CA870586, CA872533, BI966969, BY752269, and BY565884) were identified and aligned against the B6 genome sequence spanning Idd18.1 using est2genome. Primers for RT-PCR were designed to span the unique exon-exon boundaries present, and were used to test for expression in whole spleen and kidney cDNA.
The expression levels of VAV3 and NETRIN G1 transcripts were measured relative to the expression of beta-2 microglobulin using qPCR in an ABI Prism 7300 sequence detector (Applied Biosystems). All reactions were performed using TaqMan® Universal Master Mix. The expression levels are given as dCT, which are calculated by subtracting the cycle threshold value (the cycle number at which message is first detected) for beta-2 microglobulin from the cycle threshold value for each transcript. Lower dCT values indicate higher RNA levels. As the cycle thresholds were in the exponential phase of amplification, a 1 dCT difference is equivalent to a 2-fold change, and a 4 dCT difference to a 16-fold change in RNA levels.
The strategy to map Idd18 that utilized the interaction between Idd3 and Idd10/18 (14), resolved the Idd18 interval to 2.04 cM between the boundaries of congenic strains R1 and R135 (10), which were both homozygous for B6 alleles at Idd3 and Idd10, but only R135 was homozygous for B6 alleles at Idd18 (Fig. 1A). R135 was shown to have similar levels of T1D as the control congenic strain, which was homozygous for B6 alleles at Idd3, Idd10, and Idd18, whereas R1 had lost the protection associated with Idd18 and had similar levels of T1D as the Idd3 congenic strain tested (10). To refine and determine the physical length of Idd18 in the context of B6 Idd3 alleles, the recombination points of lines R1 and R135 were resolved using microsatellite markers designed using the public mouse genome sequence (28). The proximal recombination point of Idd18 is now defined as a 147 kb region between markers AL683823_12 and AL671917_10, and the distal recombination point as a 3.6 kb region between markers AL683824_5 and AL683824_5_1. Therefore, from mouse Ensembl (build 36), the Idd18 interval, in the context of B6 Idd3 alleles, is a physical distance of 1.27 Mb located between, but not including, the microsatellite markers AL683823_12 and AL683824_5_1 (Fig. 1B).
The annotation present in mouse Ensembl identified 16 genes and 1 pseudogene within the Idd18 interval: 1700013F07Rik, Tmem167b, Taf13, Wdr47, Clcc1, Gpsm2, 4921525H12Rik, Stxbp3a, Fndc7, Prpf38b, 4921515j06Rik, Fam102b, 4930443g12Rik, Slc25a24, Zpbp2_pseudogene, Vav3, and Ntng1 (listed in order from the centromere). Only Vav3 has functions known to be associated with the immune system and was, therefore, the most likely gene to mediate the effects of Idd18 in T1D.
To test the hypothesis that NOD alleles at Vav3 mediate the T1D susceptibility of Idd18, a new congenic mouse strain, designated line 2399, was developed from line 1538. Line 2399 has B6 alleles at Idd3, Idd10, and the 14 proximal genes in the 1.27 Mb Idd18 region (as defined by lines R1 and R135), and NOD alleles at Vav3 and Ntng1 (Fig. 1A, B). Assuming Vav3 is the causative gene of Idd18, and from the previous congenic strain mapping of Idd18 (10), we expected the novel line 2399 congenic mice to lose the protection associated with the B6 alleles of Vav3 and have a diabetes frequency similar to NOD.B6 Idd3 congenic mice.
Diabetes frequency studies were conducted on cohorts of female mice from line 2399 and several control congenic strains. Line 2399 has a higher frequency of T1D at seven months of age than line 1538, which is NOD.B6 Idd3 Idd10 Idd18, and a similar T1D frequency to lines 1098 and 1100, which are NOD.B6 Idd3 and NOD.B6 Idd3 Idd10, respectively, indicating that line 2399 has lost the protection associated with the B6 alleles of the Idd18 interval (Fig. 1C). Analysis of the survival curves for line 2399 and the control strain, line 1538, indicates that this difference is statistically significant (P = 0.003). As lines 1538 and 2399 differ by the presence of B6 and NOD alleles at Vav3, respectively, these results confirm that Vav3 is a candidate gene for Idd18. From here on, we designate the Idd18 interval refined by line 2399 in the context of B6 alleles at Idd3 as Idd18.1.
Based on the different T1D frequencies of lines 1538 and 2399, the distal recombination point in line 2399 defines the proximal boundary of Idd18.1 to 59 kb between the microsatellite markers AL845310_13 and AL845310_10 (Fig. 1B). The distal boundary of Idd18.1 is defined by the distal recombination point in line R135 (Fig. 1B). Therefore, the Idd18.1 interval is a maximum of 603,824 nucleotides between, but not including, the markers AL845310_13 and AL683824_5_1 (Fig.2).
The only annotated genes in the Idd18.1 interval are Vav3 and the last exon of Ntng1 (Fig. 2). Zpbp2_pseudogene is also within the Idd18.1 interval; however, this processed pseudogene of Zona pellucida binding protein 2 does not have an open reading frame. There are no other RNA-coding sequences annotated in the region. From mouse Ensembl, Idd18.1 also contains several Genscan predictions; however, the predictions do not have EST evidence and do not have orthologs in human or rat suggesting that they are chance open reading frames in the genomic sequence. NETRIN G1 is important neurologically (29-31) and mRNA expression is only detectable in brain and eye tissues (32). In addition, there is no published function of NETRIN G1 in immune cell types whereas VAV3 has many published functions in the immune system (33, 34). Although Ntng1 is unlikely to be the Idd18.1 candidate gene, Ntng1 and Vav3 were both included in further sequence and expression analyses.
To identify any sequence polymorphisms in the VAV3 and NETRIN G1 genes between NOD and B6, five NOD BAC clones spanning the complete Idd18.1 interval (Fig. 2) were selected and sequenced at the WTSI. In the mouse, Vav3 spans a genomic distance of 345 kb and consists of 27 exons; all the donor and acceptor splice site sequences of Vav3 are in agreement with the canonical GT-AG splice site (35). This is very similar to the gene structure and size of human VAV3 (data not shown). Polymorphisms between NOD and B6 are not present in the coding sequence or splice sites of Vav3, nor in the last exon of Ntng1. However, 218 SNPs, 182 microsatellites, and 23 other insertion/deletion polymorphisms were identified throughout the Idd18.1 interval. Two peaks of high SNP frequency are observed in Idd18.1: the first spans intron 20 to 2.5 kb after the 3′ UTR of Vav3 and contains 138 SNPs in 61 kb of sequence; the second overlaps the distal boundary of Idd18.1, spanning the final intron of Ntng1 to 13 kb downstream of Ntng1 and contains 90 SNPs in 40 kb. 20 kb of the second peak, containing 46 SNPs, is located in the Idd18.1 region (Fig. 2).
Since there was not a structural difference in VAV3 between NOD and B6, we hypothesized that a change in RNA expression or a change in splicing efficiency of VAV3 could cause susceptibility to T1D. As well as full-length VAV3, two alternate mRNA transcripts have been published for the VAV3 gene: one identified in both mice and humans, VAV3.1 (36), and one identified only in humans, VAV3β (37). These transcripts are predicted to encode partial VAV3 proteins that have alternative promoters in introns of the VAV3 gene; however, there is no evidence that these transcripts are translated into protein. VAV3.1 has a unique 5′ UTR located upstream of exon 18 of full-length VAV3, and contains eight novel amino acids upstream of the normal reading frame of exons 18 to 27. From our sequence data, polymorphisms are not present in the novel coding sequence of VAV3.1. We found no evidence of the novel VAV3β exon in mouse.
In order to fully investigate the expression of the mouse VAV3 gene, we searched for alternative VAV3 transcripts in the public EST database at NCBI. We identified six ESTs that gave evidence for novel splicing events not observed in the known VAV3 transcripts. To confirm that these splicing events were not EST artifacts, we used RT-PCR to amplify the novel spliced exons. This confirmed the expression of BG060768 (novel 3′ UTR in intron 17), CA870586 (novel 3′ UTR in intron 6), and CA872533 (novel 3′ UTR in intron 22). As our initial aim was to identify expression differences, we did not characterize these transcripts further.
Real-time quantitative RT-PCR (qPCR) was used to measure the relative expression of full-length VAV3 and the alternatively spliced VAV3 transcripts compared to beta-2 microglobulin. As it was not possible to specifically assess the expression of the full-length VAV3 transcript, since all the exon-exon boundaries of full-length VAV3 are present in alternatively spliced transcripts, we assessed the expression of full-length VAV3 at three different locations in the transcript (the exon 1-2 boundary, the exon 10-11-12 boundaries, and the exon 21-22 boundary). The expression levels were essentially identical for all three qPCR assays; therefore, we proceeded to assess expression levels of full-length VAV3 using only the exon 1-2 boundary qPCR assay. Differential expression of full-length VAV3 was observed when comparing thymocytes from 3-week old male congenic mice from lines 2399 and 1538, which have NOD and B6 alleles at Idd18.1, respectively (Fig. 3A). Line 2399 has 1.84-fold higher expression of full-length VAV3 mRNA compared to line 1538 indicating that the NOD alleles at Idd18.1 are expressed higher than the B6 alleles. This differential expression is not due to differences in thymocyte cell populations, as the congenic strains tested have a similar proportion of CD4 and CD8 negative, single positive, and double positive cells (data not shown). The expression levels of the alternatively spliced transcripts ranged from 64- to 256-fold lower compared to the exon 1-2 boundary in full-length VAV3 in 3-week old male thymocytes (data not shown). A similar trend in differential expression was observed in the alternatively spliced transcripts, although this was not statistically significant for all transcripts (data not shown).
The expression of full-length NETRIN G1 (NETRIN G1A) mRNA and spliced transcripts that contain the terminal exon present in the Idd18.1 interval and alternative penultimate exons as compared to full-length NETRIN G1 (NETRIN G1C, D, E, and G) (31, 38) were tested by qPCR. NETRIN G1A, C, D, and E were detected in whole brain (dCT 8, 18, 10, and 11, respectively), but were absent from whole spleen and thymocytes (data not shown). NETRIN G1G was not detected in the tissues tested (data not shown).
We next assessed the protection of Idd18.1 in the absence of protective B6 alleles at Idd3 by developing a congenic strain with NOD alleles at Idd3 and the same B6 introgressed segment spanning Idd10 and the 14 genes at the proximal end of Idd18 that is present in line 2399 (Fig. 4A). Based on the increased T1D frequency of line 2399 (NOD.B6 Idd3 Idd10) compared to line 1538 (NOD.B6 Idd3 Idd10 Idd18), due to the loss of B6 protective alleles at Idd18.1 (Fig. 1C), we expected line 2412 (NOD.B6 Idd10) to have an increased frequency of diabetes compared to line 1101 (NOD.B6 Idd10 Idd18) (Fig. 4A) since line 2412 does not have protective B6 alleles at Idd18.1 (Fig. 4A). However, the diabetes frequency of line 2412 was not different (P = 0.99) from line 1101 (Fig. 4C), and both lines were equally protected from T1D as compared to NOD mice (P = 6.0 × 10−5 and P = 5.0 × 10−5, respectively). Since the result of line 2412 was unexpected, we repeated the diabetes frequency study of lines 2412, 1101, and NOD (Fig. 4D). We observed the same result; thereby, confirming that the diabetes frequency of line 2412 was the same as line 1101.
Based upon the difference in diabetes frequency observed when comparing lines 2399 (NOD.B6 Idd3 Idd10) and 1538 (NOD.B6 Idd3 Idd10 Idd18) that differ at Idd18.1 and have B6 alleles at Idd3 and Idd10, and equal diabetes frequency of lines 2412 (NOD.B6 Idd10) and 1101 (NOD.B6 Idd10 Idd18) that also differ at Idd18.1 and have B6 alleles at only Idd10, we propose that the T1D protection associated with the B6 alleles of Idd18.1 is only observed in the context of protective B6 alleles at Idd3.
To determine whether the VAV3 expression difference, observed when comparing congenic mice from lines 1538 and 2399 (Fig. 3A), was dependent on B6 protective alleles at Idd3, we assessed the expression of full-length VAV3 (at the exon 1-2 boundary) in 3-week old thymocytes from male line 1101 (NOD.B6 Idd10 Idd18) and line 2412 (NOD.B6 Idd10) congenic mice that have NOD alleles at Idd3, B6 alleles at Idd10, and B6 or NOD alleles at Idd18.1, respectively (Fig. 3B). Line 2412 expressed 1.46-fold more full-length VAV3 compared to line 1101 (Fig. 3B) indicating that the differential expression of B6 and NOD Vav3 alleles was not dependent on the allelic status of Idd3.
The first Idd18 interval discovered was identified due to the difference in diabetes frequency between lines 1101 (NOD.B6 Idd10 Idd18) and R3 (NOD.B6 Idd10) (Fig. 4A). Line R3 had a higher frequency of diabetes than line 1101, and had, therefore, lost the protection associated with the B6 alleles of Idd18, which was defined as a 5.1 cM interval (Fig. 4A) (13). Notably, the protection associated with the 5.1 cM Idd18 interval was not dependent on B6 alleles at Idd3. Therefore, we propose that in addition to Idd18.1, which is dependent upon B6 alleles at Idd3, another Idd interval, which is not dependent upon B6 protective alleles at Idd3, is located in the original 5.1 cM Idd18 interval. We designate this Idd3 independent interval, Idd18.3. Note that Idd18.3 has not been designated Idd18.2 since another region has this gene name (see Discussion).
The distal recombination points of lines 1101 and R3 defining the original Idd18 interval were resolved to 17.1 Mb between, but not including, the microsatellite markers AC093365_1 and D3Mit370 (Fig. 4A). This region contains the Idd18.3 interval, the protection of which is not dependent upon B6 alleles at Idd3. As there is no difference in T1D frequency between lines 1101 and 2412 (Fig. 4, C and D), this excludes Idd18.3 from the region between the distal recombination points of lines 1101 and 2412, which is between, but not including, the microsatellite markers AL845310_13 and D3Mit370 (Fig. 4A). Therefore, we postulate that the Idd18.3 interval is located between the distal recombination points of lines 2412 and R3, which is between, but not including, the microsatellite markers AC093365_1 and AL845310_10 (Fig. 4B). The Idd18.3 interval is 996 kb and contains 22 genes and 1 pseudogene (Fig. 5). From the NIEHS resequencing data there are 55 SNPs polymorphic between NOD/LtJ and B6 in the Idd18.3 interval, none of which map to the coding regions of the 22 genes located in Idd18.3. However, this does not exclude the existence of NOD/B6 coding SNPs in these genes, as the NIEHS resequencing does not capture all the sequence variation due to technical and biological issues. The highest density of SNPs within Idd18.3 is located in and immediately upstream of Fam102b, with 8 and 24 SNPs in intron 1 and 22 kb upstream of the 5′ UTR, respectively (Fig. 5).
In this study we have identified that the original 5.1 cM Idd18 interval consists of two smaller subcongenic intervals, Idd18.1 and Idd18.3. Idd18.1 is 604 kb and contains only two annotated genes: the full sequence of Vav3 and the last exon of Ntng1. Interestingly, the protection associated with the B6 alleles of Idd18.1 is only observed when B6 protective alleles are also present at Idd3. In the absence of protective alleles at Idd3, we provide evidence for Idd18.3, a B6 protective interval that is 996 kb and contains 22 genes.
Ntng1 and Vav3 are positional candidate genes for Idd18.1 (Fig.2). NETRIN G1, also known as LAMINET1, is a cell surface glycoprotein and forms part of the UNC-6/Netrin family. NETRIN G1 functions in nervous system plasticity by providing short-range cues to neurons to promote neurite outgrowth (39, 40), and recent studies have found an association between NETRIN G1 and genetic risk for schizophrenia (30, 41). The expression of NETRIN G1 is primarily restricted to the central nervous system (31, 38), and in our study we did not observe expression of NETRIN G1 transcripts in thymocytes or spleen. Based on the function, expression, and phenotypes associated with disruption of NETRIN G1, Ntng1 is unlikely to be a functional candidate gene for T1D.
The VAV proteins belong to the DBL family of Rho/Rac/Cdc42 guanine nucleotide exchange factors (GEFs). Three mammalian family members are known VAV1, VAV2, and VAV3 each with overlapping but distinct specificities for different GTPases and contribute to signaling pathways involved in cytoskeletal modulation and transcriptional alterations. VAV1 is expressed exclusively in immune cells whereas VAV2 and VAV3 are expressed more broadly. The VAV family as a whole is necessary for the adaptive immune system as triple VAV1/2/3 knockout mice have severe defects in T and B cell development and function (34). VAV3 deficiency has been shown to affect cytoskeletal plasticity in several cell types including fibroblasts (42) and platelets (43), and VAV3 is required for macrophage phagocytosis of apoptotic neutrophils (44), and B cell antigen receptor endocytosis and antigen presentation (45). Based on this evidence for a functional role of Vav3 in immune cells, we propose that Vav3 is the T1D etiological gene of Idd18.1.
In addition to its strong functional candidacy, we identified allele-dependent VAV3 mRNA expression differences (Fig.3). The NOD allele correlates with ~50% higher levels of VAV3 transcripts compared to the B6 allele; this relatively small expression difference is characteristic of common functional variants (4, 6, 46). The expression difference is likely due to one or more of the 138 SNPs or 95 microsatellite and other insertion/deletion polymorphisms present between exon 20 and 2.5 kb downstream of the 3′ UTR, which could modulate RNA stability, transcription (by altering an enhancer sequence), and/or efficiency of splicing (47, 48).
In this study, we have observed that the protection associated with Idd18.1 is dependent on B6 alleles at Idd3. This dependency is not due to altered VAV3 expression caused by the Idd3-determined change in IL-2 levels, since congenic strains with NOD alleles at Idd3 (lines 1101, 2412) and B6 alleles at Idd3 (lines 1538, 2399) were both observed to have VAV3 differential expression when comparing congenic strains with NOD and B6 alleles at Idd18.1 (Fig. 3). This suggests that the expression of VAV3 is dependent upon the Idd18.1 genotype alone, whereas the T1D protective phenotype of Idd18.1 is a result of the interaction between the immune pathway(s) involving Idd3/Il2 and Idd18.1/Vav3 within a cell type or between cell types. The B6 protective Idd3 alleles have been shown to produce 2-fold more IL-2 than the NOD susceptible alleles resulting in an increase in the function of CD4+ CD25+ T-regulatory cells, a T cell subset that can prevent the activation of effector T cells (4). The B6 allele at Idd3 has also been shown to influence the function of antigen presenting cells (49), and the deletion of islet-specific T cells (5). The lower VAV3 expression associated with B6 Idd18.1 alleles likely alters the function of one or more immune cell types to enhance the dampening of the autoimmune process established by the higher-expressing B6 alleles of IL-2. However, the lower VAV3 expression from B6 Idd18.1 alleles in an environment generated by NOD alleles at Idd3 and B6 alleles at Idd10 does not appear to dampen the autoimmune process, indicative of a functional epistasis between the effects of these loci and their haplotypes.
We have previously shown that protective B6 alleles at the Idd3 and Idd10/18 gene regions interact in a non-multiplicative manner to provide increased T1D protection (14). Although the T1D protection mediated by Idd18.1 depends on the allelic status of Idd3, other Idd loci in the Idd10/18 region could also contribute to the non-multiplicative interaction observed in the previous study. It is important to note that the dependency between Idd3 and Idd18.1 has been observed using congenic strains that have B6 introgressed DNA at Idd10, Idd18.3 and an unpublished (HIF and LSW) B6 susceptibility locus, Idd18.2, located between Idd10 and Idd18.3. A congenic strain with a small introgressed segment (<2 Mb) spanning only Idd18.1/Vav3 is needed to assess the protective effects of Idd18.1 alone and in conjunction with protective alleles at Idd3/Il2 without the effects of neighboring Idd loci.
Idd18.3 is a 996 kb interval immediately proximal to Idd18.1 that is pivotal for the protection observed in the context of NOD alleles at Idd3 (Fig. 4). Based on polymorphism frequency (Fig. 5), Fam102b is the most likely candidate for Idd18.3. Although Fam102b is expressed in many tissues, its function is not known. Notably, Fam102b has high expression in dendritic cells and macrophages (50, 51) including upregulated expression in bone-marrow derived dendritic cells upon stimulation with CpG oligonucleotides (52). Dendritic cells have been shown to be necessary for T cell-mediated T1D in NOD mice (53); therefore, Fam102b may increase T1D susceptibility via the activity of dendritic cells.
This article highlights the phenomenon of the clustering of Idd loci. The original linkage analysis peak identified on chromosome 3 (54, 55) is now known to consists of at least five Idd loci confirmed by congenic strain mapping, Idd3, Idd10, Idd17, Idd18.3, and Idd18.1. In addition, we are fine mapping loci between Idd10 and Idd18.3 and have identified at least one interval that modulates the frequency of T1D, Idd18.2 (HIF and LSW, unpublished). This clustering of loci is not unique to chromosome 3 or T1D: the Idd4 locus on chromosome 11 is composed of two subcongenic intervals, Idd4.1 and Idd4.2 (56), and the Sle1 interval that confers protection against systemic lupus erythematosus is made up of at least 3 subcongenic intervals (57). The dissection of such clustered disease-causing genes requires a congenic strain mapping strategy and is a complex task since in some cases the NOD alleles confer resistance to T1D, rather than the expected susceptibility, or the clustered genes may be involved in gene-gene interactions, defined as when alleles at one locus obscure (mask) or enhance (synergize) the phenotypes conferred by alleles of a linked or unlinked locus.
Gene-gene interactions are frequently encountered in congenic strain mapping studies and on the one hand complicate the mapping of disease susceptibility genes whilst on the other enable the detection of some loci. For example, as observed here (Fig. 1C), and previously (10), B6 Idd3 alleles mask the protection associated with B6 Idd10 alleles, and NOD Idd3 alleles mask the protection associated with B6 Idd18.1 alleles (Fig. 4). Notably, Idd18.1 would not have been identified if only the Idd10/18 congenic strain (line 1101, which has NOD alleles at Idd3) had been used to fine-map the region. Similarly, the T1D susceptibility associated with alleles at Idd5.4 is dependent on the presence of susceptibility alleles at Idd5.1/Ctla4, and is not observed if protective alleles are present at Idd5.1/Ctla4 (58). Congenic strain mapping enables specific gene-gene interactions to be evaluated in a fixed genetic environment, a situation not achievable in human studies. Although gene-gene interactions have been documented in human T1D (59, 60), it is likely that most instances of masking and synergy are obscured by the fact that hundreds of T1D variants are segregating simultaneously in the human population (61, 62).
There is no evidence to date for association of human VAV3 with T1D (61, 63) (www.t1dbase.org). However, regions synonymous with Vav3 have been identified by linkage analysis in the biobreeding diabetes-prone rat model of T1D: Iddm26 is 41 Mb, contains 450 genes, and overlaps Idd10, Idd18.2, Idd18.3, and Idd18.1 (64); and a second unnamed region is 26 Mb, contains 196 genes, and overlaps Idd18.2, Idd18.3 and Idd18.1 (65). Identification of Vav3 as a causal T1D gene in the NOD mouse and a possible T1D gene in the biobreeding diabetes-prone rat helps us prioritize analysis of human T1D gene candidates that are known to function in VAV3 dependent signaling pathways controlling cytoskeletal assembly and other immune response functions.
We would like to thank Dr. Jane Rogers for assistance with B6 BAC clone sequencing.
HIF was funded by a Wellcome Trust 4-year studentship. LSW and JAT are supported by a joint grant from the Juvenile Diabetes Research Foundation (JDRF) and the Wellcome Trust. Cambridge Institute for Medical Research (CIMR) is in receipt of a Wellcome Trust Strategic Award (079895). The resequencing of Idd18.1 in the NOD mouse strain was performed at WTSI and was funded by the Immune Tolerance Network (ITN) contract AI 15416, which was sponsored by the National Institute of Allergy and Infectious Diseases (NIAID), the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), and the Juvenile Diabetes Research Foundation International (JDRF). The availability of NOD congenic mice through the Taconic Emerging Models Program has been supported by grants from the Merck Genome Research Institute, NIAID, and the JDRF.