IBD-Associated TL1A Gene (TNFSF15) Haplotypes Determine Increased Expression of TL1A Protein

Background The recently identified member of the TNF superfamily TL1A (TNFSF15) increases IFN-γ production by T cells in peripheral and mucosal CCR9+ T cells. TL1A and its receptor DR3 are up-regulated during chronic intestinal inflammation in ulcerative colitis and Crohn's disease (CD). TL1A gene haplotypes increase CD susceptibility in Japanese, European, and US cohorts. Methodology and Principal Findings Here we report that the presence of TL1A gene haplotype B increases risk in Jewish CD patients with antibody titers for the E. coli outer membrane porin C (OmpC+) (Haplotype B frequency in Jewish CD patients: 24.9% for OmpC negative and 41.9% for OmpC positive patients, respectively, P≤0.001). CD14+ monocytes isolated from Jewish OmpC+ patients homozygous for TL1A gene haplotype B express higher levels of TL1A in response to FcγR stimulation, a known inducing pathway of TL1A, as measured by ELISA. Furthermore, the membrane expression of TL1A is increased on peripheral monocytes from Jewish but not non-Jewish CD patients with the risk haplotype. Conclusions and Significance These findings suggest that TL1A gene variation exacerbates induction of TL1A in response to FcγR stimulation in Jewish CD patients and this may lead to chronic intestinal inflammation via overwhelming T cell responses. Thus, TL1A may provide an important target for therapeutic intervention in this subgroup of IBD patients.


Introduction
TL1A, a recently identified member of the TNF superfamily, increases IL-2 response by anti-CD3/CD28-stimulated T cells [1]. Furthermore, we and others have shown that TL1A synergizes with IL-12 and IL-18 to augment IFN-c release in human T and NK cells and biases T cell differentiation towards a T H 1 phenotype [2,3,4]. TL1A expression is increased in inflamed tissue of colon and small bowel of CD patients and colocalizes to macrophages and T cells [2,5]. In particular, lamina propria, but also peripheral CD4+CCR9+ T cells, constitutively express membrane TL1A and are especially sensitive to TL1A stimulation [3,4]. In murine models of ileitis, TL1A is mainly expressed on lamina propria dendritic cells [6].
We have recently demonstrated that TL1A is produced by antigen-presenting cells, e.g. monocytes and dendritic cells, in response to FccR signaling but not in response to Toll-like receptor agonists or pro-inflammatory cytokines [7]. Stimulation with Immune Complexes (IC) leads to the expression of both membrane and secreted TL1A [1,7]. Neutralizing TL1A antibodies prevent and treat colitis in a murine model of chronic colitis by affecting both T H 1 and T H 17 responses, suggesting that TL1A is a central regulator of intestinal inflammation during colitis [8]. In addition, it has been demonstrated recently that TL1A also plays an important role in the pathogenesis of other inflammatory diseases, such as Experimental Autoimmune Encephalomyelitis (EAE) and allergic lung inflammation [9,10,11].
The first genome-wide association study of CD provided evidence that variation in TNFSF15, the TL1A gene, contribute to CD in Japanese and both CD and ulcerative colitis in the British population [12,13]. Haplotypes composed of 5 TNFSF15 SNPs were observed to confer significant CD risk (haplotype A) and protection (haplotype B), using both case/control and family-based study designs [12,13,14]. We have also observed a protective association between CD and the same haplotype B in a Los Angeles based cohort [15]. Stratification on Ashkenazi Jewish ethnicity suggested that haplotype B may have a different effect on CD susceptibility in the Jewish and non-Jewish populations. In contrast to the protective association seen in non-Jews, the opposite trend towards a risk association with haplotype B was observed in Ashkenazi Jews [15]. Similar observation of differential genetic risk association in diverse ethnic groups have been made in CD, in ulcerative colitis and other gentically complex diseases including schizophrenia and asthma [16,17,18,19,20,21,22]. Jewish CD patients carrying the TNFSF15 haplotype B were more likely to have more severe CD, as evidenced by a higher rate of surgery [15] and by the expression of antibody responses to microbial antigens, including the E. coli outer membrane porin C (OmpC+) [23,24]. To date, no functional basis for the relationship between TNFSF15 variation and disease severity in CD patients has been shown.
In order to determine the functional consequences of TNFSF15 genetic variation, we have identified subjects for immunological studies based on TNFSF15 haplotypes. Here we show that, in Jewish CD patients with seropositivity for OmpC, the TL1A risk haplotype TNFSF15 haplotype B is associated with higher TL1A expression upon stimulation of FccR. Furthermore, Jewish but not non-Jewish CD patients with the TNFSF15 risk haplotype B have a higher baseline expression of TL1A on peripheral monocytes, suggesting a higher baseline capacity for T cell stimulation. Collectively, our data define a role for TNFSF15 genetic variation in determining disease severity in Jewish CD patients, and support the concept that TL1A is a novel interventional target, at least for the subgroup of Jewish, OmpC+, CD patients.

Human subjects
We collected peripheral blood from randomly selected patients attending the IBD center at Cedars-Sinai Medical Center who had previously been diagnosed with CD according to standard clinical, endoscopic, radiological, and histological findings. Written informed consent was obtained from all patients. Procedures were approved by the Institutional Review Board of Cedars-Sinai Medical Center (IRB number 3358 and 2673). The patient's demographics, diagnoses and medications at time of sample collection are provided in Table 1. The medications were equivalent in the different groups. Jewish ethnicity was defined as previously described by one or more grandparents of Ashkenazi Jewish descent [25,26]. Controls were matched for ethnicity and were usually spouses of CD patients.

TNFSF15 genotyping and haplotype assignment
Single nucleotide polymorphisms (SNPs) rs3810936, rs6478108, rs6478109, rs7848647, and rs7869487 were genotyped using either Illumina Golden Gate technology [27,28] or ABI TaqMan MGB technology [29,30] following the manufacturer's protocols (Illumina, San Diego, CA; ABI, Foster City, CA). Assays for these SNPs are available to other researchers through ABI as follows: rs3810936, hCV363308; rs6478108, hCV170492; rs6478109, hCV1305297; rs7848647, hCV11277159; and rs7869487, hCV11277149. Consistency between the two methods has been confirmed in our laboratory by genotyping over 200 samples with both methods. Details of the positions of these SNPs with respect to the TNFSF15 gene and the haplotypes from the individual SNPs are shown in Figure 1. While redundancy in the 5 SNPs may be apparent from Figure 1, all 5 SNPs are retained in this study in order to (1) distinguish the major haplotypes from minor haplotypes with low frequencies (data not shown) and (2) retain the haplotype terminology of the previous reports [12,13,15]. Haplotypes were assigned from the individual SNP data using Phase v2 and the probability of assignment was greater than 95%. Association of haplotypes with the presence of antibody per subject was tested by chi-square and the significance was estimated by permutation (Jewish OmpC2: N = 304, Jewish OmpC+: N = 196, non-Jewish OmpC2: N = 434, non-Jewish OmpC+: N = 234).

Serological Analysis
Sera were analyzed for expression of anti-OmpC antibodies by ELISA as previously described [31].
Isolation of CD14 + monocytes from patient's blood Blood was obtained from patients after informed consent in accordance with procedures established by the Cedars-Sinai Institutional Review Board. Monocytes were isolated from PBMC as described previously [7].

Cell culture and stimulation
Plate-bound, cross-linked human IgG (IC) was prepared as described previously [7]. Monocytes were incubated with IC for the indicated time-points.

ELISA
TL1A was quantified in undiluted supernatants from stimulated cells using an ELISA and Ab developed at Teva Pharmaceutical USA as described elsewhere [7].

Flow cytometry
Monocytes stimulated with IC or bacteria for 16 h were stained with the TL1A Ab as described previously [7]. Cells were analyzed on a CyAn TM ADP flow cytometer (Dako Cytomation, Carpinteria, CA) and analyzed with the Summit 4.1 software package (Dako).

Statistics
Statistical significance between groups was determined by Student's t test. A value of p#0.05 was considered to be statistically significant.

Results
Association of TNFSF15 (TL1A) haplotype B and the serum expression of anti-OmpC in Jewish Crohn's disease patients Since the expression of anti-OmpC in CD patients is a familial trait and has been associated with severe CD [23,24,32], we tested the association between TNFSF15 haplotype B and the serum expression of anti-OmpC in Ashkenazi Jewish and non-Jewish CD patients ( Figure 1A-D). TNFSF15 haplotypes were assigned from individual SNP data using PHASE v2 with a probability greater than 95% (see Figure 1A-C for details of SNPs and haplotypes). In non-Jewish CD subjects, there was no difference in the frequency of haplotype B between OmpC positive and OmpC negative subjects (OmpC negative 38.6%, OmpC positive 38.3%; Figure 1D). The frequency of haplotype B in either group was similar to that previously observed for all CD in non-Jewish subjects [15]. In contrast, for Jewish CD subjects the frequency of haplotype B was significantly higher in OmpC positive subjects compared with OmpC negative subjects (Figure 1 D) (24.9% OmpC negative, 41.9% OmpC positive, P#0.001). The frequency of haplotype B in OmpC negative subjects was lower and in OmpC positive subjects was higher than that previously observed for all CD in Jewish subjects (32%) [15] ( Figure 1D). In addition, the level of anti-OmpC expression was higher in TNFSF15 (TL1A) haplotype B Jewish subjects (the median level of anti-OmpC was 23.9 ELISA units in Jewish CD patients with haplotype B compared with 14.9 in Jewish CD patients without haplotype B, P wilcoxon test #0.001). The frequency of OmpC seropositivity in Jewish CD patients with haplotype B is 52%, and 36% in non-Jewish CD patients with haplotype B. We did not observe an association between other CD-associated antibodies to microbial antigens (e.g. anti-Cbir1 antibodies, anti-I2 antibodies, anti-ASCA antibodies) and differences in haplotype B frequency in Jewish or non-Jewish CD patients (data not shown). To determine if the different TNFSF15 (TL1A) haplotypes associated with CD in Jewish patients correlate with functional consequences in TL1A expression, we isolated CD14+ monocytes from Jewish and non-Jewish CD patients with haplotypes A or B, and stimulated them with immune complexes (IC), a known inducer of TL1A gene expression [7]. We observed strong induction of TL1A secretion in response to IC in Jewish haplotype B carriers at 6 h of stimulation. Compared to Jewish haplotype A carriers, the secretion was significantly increased at 6 h in haplotype B carriers (Figure 2A). At 16 h we also observed an increase, although not significant (p = 0.22), in secretion of TL1A in haplotype B compared to haplotype A Jewish patients ( Figure 2A). In non-Jewish CD patients we also observed a significant difference between TL1A secretion in haplotypes A and B at 6 h of IC stimulation ( Figure 2B). Haplotype B carriers secreted significantly more TL1A at 6 h compared to haplotype A carriers. At 16 h the difference in TL1A secretion between the haplotypes were not significant (p = 0.29), although we observed a higher secretion in haplotype B non-Jewish patients. To determine if there are ethnic differences in expression of TL1A we compared Jewish and non-Jewish haplotype B carriers. Jewish haplotype B carriers secreted significant more TL1A at 6 h compared to non-Jewish haplotype B (p = 0.002). At 16 h we did not observe any differences in the ICinduced secretion of TL1A (p = 0.49). Next, we compared the TL1A secretion in Jewish and non-Jewish haplotype A carriers. There were no differences in the IC-induced secretion of TL1A in Jewish or non-Jewish CD haplotype A patients at 6 or 16 h (p = 0.07 and 0.09, respectively).
To determine if the observed differences in TL1A secretion also relate to differences in surface TL1A expression, we determined the expression of membrane TL1A as measured by the percentage of TL1A + monocytes and mean fluorescence intensity (MFI) after IC stimulation (Figure 3). We did not observe any significant differences in the percentage of control or IC-stimulated TL1A + monocytes in Jewish haplotype B carriers compared to haplotype A carriers ( Figure 3A, p = 0.19 and p = 0.49 for control and IC stimulation, respectively). Furthermore, we also did not observe any significant differences in the percentage of control or ICstimulated TL1A + monocytes in non-Jewish haplotype B carriers compared to haplotype A carriers ( Figure 3B, p = 0.83 and p = 0.72 for control and IC stimulation, respectively). When we analyzed MFI of TL1A staining we observed an increase in MFI in Jewish haplotype B compared to haplotype A carriers, although this increase did not reach significance ( Figure 3C, p = 0.66 and p = 0.52 for control and IC stimulation, respectively). We did not observe any differences in MFI in non-Jewish haplotype B compared to haplotype A carriers ( Figure 3D, p = 0.56 and p = 0.91 for control and IC stimulation, respectively). In summary, Jewish and non-Jewish haplotype B carriers secrete significantly more TL1A in response to IC compared to haplotype A carriers at 6 h (Table 2). Furthermore, Jewish haplotype B carriers secrete significantly more TL1A compared to non-Jewish haplotype B carriers ( Figure 2C). In contrast, we did not observe any significant differences in TL1A surface expression in patients with different haplotypes in either Jewish or non-Jewish CD patients (Table 2). TNFSF15 (TL1A) haplotypes in Jewish OmpC+ IBD patients result in differential TL1A expression in monocytes stimulated with IC Since we observed a risk association for TL1A haplotype B in Jewish CD patients, and this association was accentuated in Jewish OmpC+ patients, we subdivided the Jewish patients into OmpC+ and OmpC2 patients and analyzed TL1A secretion and membrane expression for haplotypes A and B. We observed a significant increase in TL1A secretion in patients carrying the B haplotype compared to A carriers in Jewish OmpC+ patients in response to IC ( Figure 4A). Differences in TL1A secretion were most prominent at 6 h post stimulation compared to 16 h ( Figure 4A, p = 0.34). We observed a TL1A response in Jewish OmpC+ patients heterozygous for haplotype A and B that was intermediate between that observed for patients carrying the A and  Figure 4C, p = 0.14 for 6 h IC stimulation), suggesting different kinetics in the TL1A secretion in Jewish vs. non-Jewish CD patients in response to FccR signaling. We did not observe any differences in TL1A secretion in non-Jewish OmpC2 patients with the B haplotype compared to patients with haplotype A ( Figure 4D, p = 0.21 for 16 h IC stimulation). Next, we analyzed the surface expression of TL1A in Jewish OmpC+ patients with different haplotypes by flow cytometry ( Figure 5). We observed a significant increase in the proportion of TL1A + monocytes in untreated monocytes from Jewish patients carrying the B haplotype compared to patients with the A haplotype in Jewish patients regardless of their OmpC status ( Figure 5A, B). Similar differences were not seen in non-Jewish patients ( Figure 5C, D, p = 0.37 and p = 0.43 for non-Jewish OmpC+ and non-Jewish OmpC2 patients, respectively). Surprisingly, we could not observe any differences in the percentage of monocytes expressing TL1A stimulated with IC between different haplotypes ( Figure 5A, B, p = 0.31). However, the mean fluorescence intensity (MFI) was significantly increased in IC stimulated monocytes in Jewish OmpC+ patients carrying the B haplotype compared to A carriers ( Figure 5A) but not in Jewish OmpC2 patients ( Figure 5B, p = 0.35). We did not observe any significant differences in surface expression of TL1A in untreated or IC stimulated monocytes between A and B carriers in non-Jewish patients ( Figure 5C, D,  OmpC2 haplotype B carriers secrete significantly more TL1A in response to IC compared to haplotype A carriers (Table 3). Furthermore, the proportion of TL1A + monocytes in untreated monocytes is significantly increased in Jewish OmpC+ and OmpC2 haplotype B carriers compared to Jewish OmpC+ and OmpC2 haplotype A carriers. Additionally, the MFI of TL1A membrane expression was significantly increased in IC stimulated monocytes in Jewish OmpC+ patients with the B haplotype compared to patients with the A haplotype (Table 3).

Discussion
Our findings demonstrate a direct association between genetic variation in the TL1A gene TNFSF15, and the induction of TL1A in Jewish, and particularly in OmpC+ Jewish CD patients. Patients carrying the TNFSF15 (TL1A) risk haplotype B have a significantly increased expression of membrane and soluble TL1A in response to IC. In the current era of finding genetic associations with disease using genome-wide association studies, our report also illustrates that detailed molecular phenotypic studies of patients with particular haplotypes will lead to the discovery of important functional differences that occur as a result of the different haplotypes.
We have previously reported that TNFSF15 (TL1A) haplotype B is associated with a more severe disease phenotype in Jewish patients, as these individuals have a higher frequency of small bowel surgery [15]. Furthermore, TNFSF15 haplotype B is associated with increased risk of CD in Ashkenazi Jews (32% in    CD vs. 26% in controls) but with a decreased risk in non-Jewish CD patients (39% in CD vs. 50% in controls) [15]. Here, we extend our findings and demonstrate that Jewish OmpC+ CD patients (a group characterized by more severe disease) have a further increased frequency of haplotype B (also a group characterized by more severe disease) (42% CD OmpC+ vs. 25% CD OmpC2) suggesting that an antibody titer for the E. coli outer membrane porin C (OmpC) increases the risk of CD in Jewish patients carrying the B haplotype. We observed that this hierarchical relationship of TL1A risk haplotype B and OmpC titers in Jewish CD patients also applies to the functional response, e.g. the immune responses to FccR signaling in Jewish CD patients. In both Jewish and non-Jewish CD we observed a significant increase of TL1A secretion in haplotype B compared to haplotype A. However, we observed an earlier onset of TL1A secretion and an overall greater amount of TL1A secretion in Jewish patients with haplotype B compared to non-Jewish haplotype B carriers, suggesting different kinetics in the induction of TL1A expression in Jewish vs. non-Jewish patients carrying haplotype B (see Table 2 for data summary). When we further sub-stratified Jewish and non-Jewish patients into patient groups with or without positive serology for OmpC antibodies, this association of TL1A haplotype B and increased TL1A expression was further accentuated in Jewish OmpC+ patients. Jewish OmpC+ patients with haplotype B have increased secretion of soluble and enhanced expression of membrane TL1A in response to IC, as well as increased baseline membrane TL1A expression, when compared to haplotype A (see Table 3 for data summary). On the other hand Jewish OmpC2 patients with haplotype B have increased secretion of soluble TL1A in response to IC, and increased baseline membrane TL1A expression, but TL1A and membrane TL1A in response to IC are not significantly different when compared to haplotype A. In non-Jewish OmpC+ patients with haplotype B, we only observed an increased secretion of soluble TL1A in response to IC, but with a distinct and delayed kinetics compared to Jewish OmpC+ and OmpC2 patients of the same haplotype. The higher baseline expression of membrane TL1A on monocytes that we observed in Jewish but not non-Jewish CD patients carrying haplotype B might suggest a higher capacity of TL1A dependent T cell activation in these patients [5,7]. Interestingly, we observed that the percentages of monocytes with TL1A surface expression being approximately twice as high in OmpC2 Jewish patients compared to OmpC+ patients for either haplotype when stimulated with IC ( Figures 5A, B). We analyzed the TL1A surface expression of monocytes in response to IC stimulation at a 16 h time-point. We have demonstrated that monocytes from Jewish OmpC+ and OmpC2 patients have a faster induction of TL1A secretion compared to non-Jewish patients (Figures 2, 4). We observed maximal induction of TL1A secretion in Jewish OmpC+ patients at 6 h and a decline of TL1A secretion at 16 h. However, in Jewish OmpC2 patients we observed a further increase of TL1A secretion at 16 h (Figure 4 A  and B). The mechanism for the differential induction in Jewish OmpC+ vs. Jewish OmpC2 and Jewish vs. non-Jewish patients is currently unknown. The differences in surface expression in Jewish OmpC+ vs. OmpC2 patients might therefore just reflect the differences in cleavage-kinetics of TL1A from the membrane due to higher turn-over of TL1A in Jewish OmpC+ patients.
We previously published that TNFSF15 haplotype B is associated with increased risk of CD in Ashkenazi Jews but with a decreased risk in non-Jewish CD patients [15]. To our surprise, we did not observe a decrease in TL1A expression in haplotype B carriers in non-Jewish patients compared to haplotype A carriers as one would expect ( Figures 2B, 3B, 3D). However, from the data presented here we are concluding that Jewish CD patients demonstrate a faster TL1A induction than non-Jewish patients. We observed an increase of TL1A secretion at 16 h compared to 6 h post IC-stimulation in non-Jewish patients suggesting a maximal secretion at 16 h. However, we did not analyze the TL1A secretion beyond the 16 h time-point and it might be possible that in non-Jewish patients the maximal secretion of TL1A is reached beyond 16 h. Therefore, it cannot be excluded that a difference between TL1A secretion in haplotype A and B might occur at a later time-point in non-Jewish patients. Furthermore, based on our data we hypothesize that TL1A is a CD severity more importantly than susceptibility gene. We have previously demonstrated that TNFSF15 haplotype B is associated with increased risk of CD in Jews but with a decreased risk in non-Jewish CD patients [15]. However, when we analyzed haplotype association with disease severity, as measured by clinical parameters (fibrostenosis, internal perforation, perianal penetration, small bowel surgeries), no association was observed between haplotypes in non-Jewish patients but association was observed in Jewish CD patients carrying haplotype B [15]. These data are consistent with our findings of no significant differences in TL1A expression in non-Jewish patients.
The TNFSF15 genetic variation and TL1A functional changes observed here are of importance in the context of the recent demonstration that the IL23R gene is a major determinant of CD susceptibility [14,33,34,35,36,37,38,39]. In particular, we have been able to demonstrate that TL1A synergizes with IL-23 to induce IL-17 production by CD4+ T cells in a mouse model of chronic colitis [8]. Furthermore, these observations are consistent with the role of TL1A during the development of chronic colitis in mice. We have been able to demonstrate that administration of neutralizing TL1A antibodies not only prevents but also treats established chronic colitis, likely by directly blocking IFN-c production by T H 1 cells and IL-17 and IL-6 production from T H 17 cells, and also by blocking the enhancing effect of TL1A/IL-12 and TL1A/IL-23 on these CD4 + T cell subsets [8]. These data support the notion that TL1A expression is increased during chronic mucosal inflammation and that leads to further activation of TH1 and TH17 cells.
In this study, we demonstrated that there is a direct association between genetic variants of the TL1A gene and the induction of TL1A in Jewish CD patients. Since the induction of TL1A gene expression seems to be restricted to stimulation by FccR signaling and/or enteric bacteria [7,40,41], blockage of TL1A might have therapeutic benefit in IBD patients without markedly increasing the susceptibility to infection. Defining CD patients by ethnicity, serotyping, and genotyping, will allow us to identify patients with the potential for higher TL1A response to immune complexes and commensal bacteria and hence may well identify those CD patients that would respond best to therapeutic blockade of TL1A function.