IL7Rα Expression and Upregulation by IFNβ in Dendritic Cell Subsets Is Haplotype-Dependent

The IL7Rα gene is unequivocally associated with susceptibility to multiple sclerosis (MS). Haplotype 2 (Hap 2) confers protection from MS, and T cells and dendritic cells (DCs) of Hap 2 exhibit reduced splicing of exon 6, resulting in production of relatively less soluble receptor, and potentially more response to ligand. We have previously shown in CD4 T cells that IL7Rα haplotypes 1 and 2, but not 4, respond to interferon beta (IFNβ), the most commonly used immunomodulatory drug in MS, and that haplotype 4 (Hap 4) homozygotes have the highest risk of developing MS. We now show that IL7R expression increases in myeloid cells in response to IFNβ, but that the response is haplotype-dependent, with cells from homozygotes for Hap 4 again showing no response. This was shown using freshly derived monocytes, in vitro cultured immature and mature monocyte-derived dendritic cells, and by comparing homozygotes for the common haplotypes, and relative expression of alleles in heterozygotes (Hap 4 vs not Hap 4). As for T cells, in all myeloid cell subsets examined, Hap 2 homozygotes showed a trend for reduced splicing of exon 6 compared to the other haplotypes, significantly so in most conditions. These data are consistent with increased signaling being protective from MS, constitutively and in response to IFNβ. We also demonstrate significant regulation of immune response, chemokine activity and cytokine biosynthesis pathways by IL7Rα signaling in IFNβ -treated myeloid subsets. IFNβ-responsive genes are over-represented amongst genes associated with MS susceptibility. IL7Rα haplotype may contribute to MS susceptibility through reduced capacity for IL7Rα signalling in myeloid cells, especially in the presence of IFNβ, and is currently under investigation as a predictor of therapeutic response.


Introduction
Multiple Sclerosis (MS) is the most common chronic neurological disease in young adults. Both genetic and environmental factors are implicated and interact in mediating pathogenesis [1][2][3][4][5][6]. The common polymorphisms mediating susceptibility to MS reside predominantly within, and neighboring, the genes affecting immune function, including the interferon response genes [2,7]. The challenge of the next phase of genetic research in MS is to identify the functional impact of these newly identified variants, possibly including a differential response to environmental susceptibility factors such as reduced vitamin D and viral infection [3,5].
We and others found an association of the interleukin-7 receptor a chain (IL7Ra) with MS [2,[8][9][10] and have identified key functional differences between IL7Ra haplotypes that are likely to affect MS susceptibility [10][11][12][13]. IL7Ra is a subunit of receptors for IL-7 and thymic stromal lymphopoietin (TSLP) and is expressed on T cells, dendritic cells (DCs) and other myeloid cells [14]. IL-7 is a critical and nonredundant cytokine mediating survival and differentiation of T cells and their progenitors [15,16]; while TSLP mediates T cell proliferation and the development of Th2 and regulatory T cells (Tregs) primarily through effects on dendritic cells [17][18][19][20][21][22]. IL7Ra has four major haplotypes [23]. Haplotype 2 (Hap 2) is protective against MS and cells exhibit reduced splicing of exon 6 from the transcript, producing less of a soluble isoform (sIL7Ra) [10,12]. However, we found that haplotype differences in splicing are greatly magnified in DCs compared to CD4 T cell subsets and PBMCs [12]. Thus, in the present study, we have examined in detail the effect of IL7Ra haplotype on myeloid cell subsets as potential drivers of the associated MS susceptible/protective phenotypes.
IL7Ra signaling exerts powerful effects on myeloid cell function, dependent on the microenvironment and on differentiation status. In monocytes, IL-7 induces inflammatory chemokines, IL-8 and MIP1b, and both TSLP and IL-7 induce the expression of the Th2-attracting chemokines, CCL17 and CCL22, [18]. TSLP skews peripheral DCs to induce an inflammatory Th2 phenotype in T cells however, in the tolerising environment of the gut, TSLP acts on intestinal DCs to induce a non-inflammatory Th2 phenotype characterized by production of the anti-inflammatory cytokine IL-10, and has been suggested to be involved in promoting tolerance to microflora and food antigens [24,25]. Importantly, TSLP acts on thymic DCs to stimulate the production of Tregs from CD4+CD25-single-positive thymocytes [21,26].
In MS, a defect in thymic output of natural Tregs has been observed, and correlated with reduced suppressive capacity [27][28][29]. In addition, a defect in inducible Tregs in MS has been suggested by impaired IL-10 production by Tr1 cells induced in vitro by CD46 costimulation [30,31]. Th2 markers are reduced in the peripheral blood of untreated MS patients [32], and immunomodulatory therapies are associated with skewing towards a Th2 profile [33]. We hypothesise that IL7Ra haplotype may influence susceptibility to MS by altering IL7Ra pathways in myeloid cells, such as those controlling thymic Treg output, and recruitment and differentiation of Th2/Tregs in the periphery.
While Hap 2 is protective against MS, Hap 4 homozygotes carry the highest risk of MS (odds ratio 1.35; p = 0.0011) [11]. We hypothesized that there are mechanisms in addition to altered exon 6 splicing by which IL7Ra haplotype affects MS susceptibility. In support of this, we previously demonstrated that IL7Ra is upregulated by IFNb in CD4 T cells, except in Hap 4 homozygotes [11]. IFNb exerts complex effects on innate and adaptive immunity and plays an important role in immune defense against viral infection. Both pro-and anti-inflammatory actions have been described [34,35]. Two recent reports have demonstrated that the anti-inflammatory effect of endogenous IFNa/b is mediated, not through direct effects on T cells, but via the IFN/b receptor on myeloid cells in the murine model of MS, experimental autoimmune encephalomyelitis [36,37]. IFNb is also an important first line immunomodulatory therapeutic for MS. IFNb therapy increases thymic output of Tregs [29] and skews the T effector profile to Th2 [33,38], mediated, at least in part, by effects on monocytes and DCs [39]. These data support a model in which IFNb upregulation of IL7Ra enhances TSLP/IL7 signaling in myeloid cells.
To investigate potential pathways affecting MS susceptibility, we determined the effects of haplotype on IL7Ra expression and splicing in myeloid cells ex vivo, and in vitro. We found effects of Hap 2 and Hap 4 on IL7Ra expression in all myeloid subsets examined and demonstrated distinct gene expression signatures in response to IL7Ra signalling depending on ligand, differentiation status of the responder cell and IFNbstimulation. The results suggest that the MS-protective haplotype is associated with increased IL7Ra expression, especially of the membrane-bound isoform, in monocytes and dendritic cells (Hap 2), that the cytokine IFNb increases IL7Ra expression in these cells, and that absence of this IFNb-responsiveness in dendritic cells is associated with the highest MS risk (Hap 4 homozygotes). These studies highlight the complex effects of IL7Ra haplotype in monocytes and DC subsets, and expand our understanding of pathways controlled by protective and susceptibility variants in MS.

Results
We had previously shown that functional differences between the MS-protective and -susceptible IL7Ra haplotypes are greatly magnified in monocyte-derived DCs compared to T cells [12]. In addition, the genotype with the highest risk of MS is unresponsive to IFNb in T cells, but this response has not been characterized in DCs or other myeloid cells. In this study we examine IL7Ra expression and function in myeloid cell subsets in detail, with the aim of identifying contexts potentially relevant to MS pathogenesis and response to IFNb therapy.
We first compared expression of IL7Ra and a major splice form in monocytes, ex vivo DC subsets (myeloid DCs and plasmacytoid DCs) and in vitro-generated DCs at two stages of differentiation (immature and maturing). We also determined whether expression and splicing change in response to IFNb stimulation. Secondly, we determined whether expression, splicing and response to IFNb differ between the IL-7R haplotypes at various stages of differentiation. Thirdly, to test the possible functional relevance of haplotype differences in IFNb response, we characterized gene profiles and immune pathways regulated by IL-7 and TSLP in IFNb-treated myeloid subsets. In addition, we induced a tolerogenic phenotype in dendritic cells using IFNb, and investigated whether TSLP in this context induces inflammatory or anti-inflammatory effects on T cell proliferation.

IL7Ra mRNA Expression and Response to IFNb in Monocytes and DC Subsets
Total IL7Ra relative expression and response to IFNb were measured in ex vivo and in vitro generated myeloid cell subsets from healthy controls (Figure 1; Hap2/Hap4 heterozygotes; n = 3). In the absence of IFNb, there were significant differences between the subsets in IL7Ra expression (monocytes . maturing DC . immature DC; all p,0.05). In the presence of IFNb, myeloid DCs expressed significantly higher levels of IL7Ra than all other subsets (all p,0.05). In response to IFNb, mean IL7Ra expression was increased in monocytes (2.5 fold, not significant (ns)), maturing DCs (1.6 fold, ns), immature DCs (12 fold; p = 0.006) and myeloid DCs (10 fold; p = 0.005). There was a trend towards downregulation of IL7Ra by IFNb in plasmacytoid dendritic cells (pDCs) (10 fold; p = 0.071). (pDCs were conditioned with a higher IFNb dose (2,000 IU/ml) since lower doses of IFNb result in spontaneous apoptosis [40]). These represent larger changes in response to IFNbthan we had previously measured for CD4 T cells (1.2 fold) and PBMCs (2 fold) [11].

Differential Expression of IL7Ra Haplotypes in Myeloid Subsets
Relative haplotype expression compared in homozygous cells. Of the 3 haplotypes, IL7Ra Hap 2 was consistently expressed at the highest level in all IFNb-treated myeloid subsets examined (1.3-6.6 fold higher), as well as in untreated dendritic cell subsets (1.7-6.5 fold higher) (mean values, ns; Figure 2). Similar to our previous findings in CD4 T cells, IL7Ra Hap 4 was not upregulated in response to IFNb in any of the subsets ( Figure 2). In contrast, Hap 1 was upregulated in all subsets in response to IFNb (mono, p = 0.027; iDC, p = 0.046; matDC, p = 0.026), and mean Hap 2 IL7Ra levels were increased in all subsets (ns). As a result, Hap 4 was consistently expressed at the lowest level of all the haplotypes in all IFNb-treated cells (means), significantly lower than Hap 1 (p = 0.021) and Hap 2 (p = 0.046) in maturing DCs.
Relative haplotype expression measured in heterozygous cells. Relative expression of haplotypes can be measured under identical cellular conditions by quantitating the relative abundance of message containing haplotype-tagging SNPs in heterozygous cells. In response to IFNb, IL7Ra Hap 4 transcript (rs3194051 'G') was expressed at a relatively lower level compared to Hap 1 or Hap 2 (both rs3194051 'A') than in the absence of IFNb. Specifically, this was evident in monocytes (p = 0.025), immature DCs (p = 0.0003) and myeloid DCs (p = 0.012), with a trend in maturing DCs (p = 0.052) ( Figure 3A). These results are consistent with relative expression data in homozygotes (Figure 2), where the Hap 4 alleles are less responsive to IFNb than Hap 1 and Hap 2. However in plasmacytoid DCs, which downregulate IL7Ra on IFNb stimulation, the response (reduction) to IFNb is greater in Hap 4 (p = 0.022). Relative expression of Hap 4 was measured in monocytes, immature DCs and maturing DCs purified from previously frozen PBMCs from a second cohort of healthy controls ( Figure 3B) and from MS patients ( Figure 3C). A significant proportion of the combined cohort (n = 10; healthy control + MS) decreased relative expression of Hap4 in response to IFNb when the 3 cell subsets (mono, iDC, mDC) were included in the analysis (p = 0.043, sign test); this did not reach significance for individual groups or cell subsets, The trend was the same in MS and controls.
Haplotype differences in IL7Ra splicing and isoform expression. We had previously shown lower production of a soluble isoform, capable of inhibiting IL7Ra signaling in T cells and DCs of Hap 2 compared to the other haplotypes [12]. Here we determined the relative expression of this soluble isoform of IL7Ra in myeloid subsets, and whether this changes in response to IFNb ( Figure 4). The ratio of transcript encoding membranebound isoform to this soluble isoform (MB:Sol), generated by splicing out of exon 6, was determined as previously described [13] ( Figure 4). Hap 2 expressed a higher ratio of membrane-bound to soluble isoform than Hap 1 in all IFNb-treated myeloid subsets examined, and also in monocytes and immature DCs in the absence of IFNb( Figure 4). The mean ratio for Hap 4 was intermediate in all subsets and conditions (significantly lower than Hap 2 in IFNb-treated monocytes, and significantly higher than Hap 1 in IFNb-treated maturing DCs). IFNb treatment increased the relative expression of the membrane-bound isoform only in Hap 2 (significant in immature DCs) ( Figure 4). This is consistent with our finding that Hap 4 heterozygous cells showed no change in relative expression of the membrane-bound and soluble isoforms in response to IFNb ( Figure S1), including a second smaller and less abundant soluble isoform generated by splicing out of exon 5 and 6 [41] either in healthy controls or MS. We also found no difference between healthy controls and MS in relative expression of membrane-bound and soluble isoforms, either in the presence or absence of IFNb.

Functional Significance of IL7Ra Signaling in IFNb-treated DCs
Gene expression. Having demonstrated that the protective haplotype is most highly expressed, and the risk haplotype is expressed at lowest levels in IFNb-treated myeloid subsets; we determined the immune genes and pathways regulated by IL7Ra signaling and thus potentially affecting MS risk in these subsets. IL-7 and TSLP significantly altered gene expression patterns in IFNb-treated monocytes, immature DCs and maturing DCs (Table 1). Of the differentially regulated genes, IL-7 upregulated a larger proportion in monocytes and immature DCs, but downregulated a larger proportion in maturing DCs. In contrast, TSLP up-and down-regulated approximately equal proportions of genes in each subset. Importantly, the number of genes differentially regulated by either ligand was related to the level of IL7Ra expression, particularly so for TSLP ( Figure 5). This underscores the functional significance of differences in IL7R expression, suggesting that lower-expressing haplotypes and lack of response to IFNb may confer reduced response to IL-7 and TSLP. IL-7 upregulated chemokine activity and immune system response pathways in IFNb-treated immature DC, including chemokines of the CXC (IL8) and C-C chemokine-ligand/-like families (CCL4L1, CCL4L2) and a matrix metalloproteinase (MMP9), but downregulated these pathways (and these four pathway members) in IFNb-treated mature DCs (Table 1). However overall, individual pathway members upregulated by IL-7 in these immature DCs mostly did not overlap with members downregulated by IL-7 in maturing DCs. Of interest, IL-7 upregulated the chemokine genes CCL17 and CCL22 in IFNbtreated immature DCs, and downregulated class II MHC genes in IFNb-treated maturing DCs (HLA-DRA, HLA-DPA1, HLA-DRB6). IL-7 also upregulated the cytokine biosynthetic process pathway in IFNb-treated maturing DCs. While no immune pathways were identified as regulated by IL-7 in IFNb-treated monocytes, IL-7 TSLP downregulated the immune response pathway in IFNbtreated monocytes, including a costimulatory molecule (CD86), antigen presentation (CD74, HLADQA1, HLA-C) complement (C1QB, C1QC) and chemokine receptor (CXCR4) genes; but upregulated IL8, CCL2, and CCL22. No immune pathways were identified as regulated by TSLP in IFNb-treated DC subsets, although in these maturing DCs, TSLP downregulated antigen processing molecules (CD74, HLADRB4) and, similar to IL-7, downregulated MMP9 and CCL8.

Effects on T cell proliferation
TSLP can induce the generation of tolerogenic DCs in the thymus and the periphery, in certain circumstances, but induces an inflammatory phenotype in others. We investigated the function of TSLP in a tolerogenic environment induced by IFNb,to determine whetherits actions are anti-inflammatory or proinflammatory in this context. We tested (1) if IFNb treatment induced a tolerogenic DC profile, and (2) if TSLP signaling in this context would stimulate or inhibit T cell proliferation in subsequent co-culture.

Discussion
Our driving hypothesis is that IL7Ra haplotypes are associated with MS because they alter expression of the gene. Under this hypothesis, where there are no haplotype-specific effects, the cell subset/microenvironment does not contribute to the association of the genotype with MS; and conversely, where there are large haplotype effects, the cell subset/microenvironment is likely to contribute to pathogenesis. In the absence of IFNb, large haplotype differences were found in immature and LPS-stimulated maturing DCs, with the protective haplotype having the highest expression. With IFNb stimulation, monocytes, immature DCs and maturing DCs displayed the same trend with highest expression by the protective haplotype, and additionally, lowest expression by the highest risk haplotype. Under all conditions, in all subsets, the protective haplotype preferentially expressed the membrane-bound isoform. Using microarray analysis, we found that cellular response to IL-7 and TSLP was related to the level of IL7Ra expression in myeloid cells. Taken together, these results suggest that the MS-protective haplotype is associated with increased capacity of dendritic cells and monocytes to respond to IL-7 or TSLP in these contexts. IL-7 is produced by stromal cells at a fixed level, and cellular competition for this limited resource is determined by the level of IL7Ra expression [16], with the soluble IL7Ra isoform able to competitively inhibit cell-surface binding [49,50]. We had previously demonstrated that T cells and inflammatory DCs of the protective haplotype produce a lower ratio of the soluble isoform [12]; and here show that this is also the case in immature DCs and monocytes, including under IFNb stimulation. We also demonstrate here that splicing is not different between MS and controls in Hap 4 heterozygous carriers. T cells presumably contribute substantially to haplotype differences in serum/ extracellular fluid levels of sIL7Ra, by virtue of their abundance in peripheral blood. However, we hypothesise that any given concentration of sIL7Ra may have a more significant inhibitory effect on IL7Ra signaling in DCs/monocytes than in T cells, because of the lower cell-surface receptor density of the former (10-100-fold less than T cells) [12]. In addition, we found earlier that haplotype differences in splicing are magnified in DCs compared to CD4 T cell subsets and PBMCs, with almost equimolar ratios of membrane-bound and soluble receptor generated by DCs of the susceptibility haplotype in the microenvironment [12]. This may be of particular relevance where DCs themselves [51][52][53] or cells within the immediate microenvironment [54] produce IL-7 or TSLP. Overall, increased sIL7Ra generated by T cells and DCs of the susceptible haplotypes may have a greater inhibitory effect on IL7Ra signaling in DCs and monocytes.
The functional effects of IL7Ra signalling in myeloid cells are pleiotropic and microenvironment-dependent. Based on our finding that increased myeloid cell expression of IL7Ra is associated with the MS-protective haplotype, and current knowledge of IL7Ra biology in myeloid cells, we suggest that several pathways may be worthy of further investigation as potential contributors to MS susceptibility. Firstly, both IL-7 and TSLP induce expression of the Th2-attracting chemokines CCL17 and CCL22 in monocytes [18,19]. Here we also demonstrated induction of these chemokines by IL7/TSLP in IFNb-treated myeloid cells. The expression of these renders subsequently differentiated monocyte-derived DCs highly chemoattractive to regulatory T cells, and skews the balance of effectors to CD4 rather than CD8 [55]. Skewing of the T helper balance to Th2/ Treg recruited at sites of inflammation [56][57][58] is one potential mechanism by which higher IL7Ra signaling might protect against autoimmunity.
Secondly, TSLP also acts on DCs to tolerise T cells in at least two important contexts of potential relevance to MS: the gut and the thymus. There is accumulating evidence that the relationship between host immunity and gut microbiota may influence the development of autoimmunity (reviewed in [59]). TSLP is constitutively expressed in healthy human colon and is inducible in a human enterocyte cell line upon exposure to gut commensals, acting on intestinal DCs to induce a non-inflammatory Th2 phenotype characterized by IL-10 and IL-13 production [24,25]. Induction of a more tolerogenic gut DC phenotype [60] and increases in IL-10 and IL-13 [61] by manipulation of gut microbiota are associated with attenuation of subsequently induced experimental autoimmune encephalomyelitis. Reduced IL-17 and interferon gamma in both intestine and spinal cord in Relative expression of membrane-bound (full-length) and ex 6 soluble (sol) IL7Ra isoform was measured as previously described [13] and expressed as a ratio. IFNb increases the ratio of membrane-bound to ex 6 solIL7Ra receptor isoform expressed in myeloid cells of Hap 2 (Hap 2 iDC, p = 0.026; Hap 2 mono showed a trend, p = 0.056). Mean +/ 2 SEM is shown; *significantly different from media condition; bars represent significant differences between haplotypes in the same subset under the same condition by Student's t test (p,0.05). doi:10.1371/journal.pone.0077508.g004  these animals suggests that the intestinal microenvironment can influence neurologic inflammation [60]. A second mechanism by which higher myeloid expression of IL7Ra could potentially confer protection from MS is via increased tolerising effects of TSLP in the gut.
Thirdly, TSLP also acts on DCs in Hassall's corpuscles of the thymus to induce Treg production from CD4+CD25-thymocytes [20,26]. It has been suggested recently that minor perturbations in the number or function of Tregs may be permissive for the development of autoimmune disease, allowing differentiation and proliferation of low affinity self-reactive CD4 T cells that have escaped thymic deletion [62]. In MS, reduced thymic Treg production has been observed [27][28][29], while current first-line immunomodulatory therapies increase thymic Treg output [29]. Higher IL7Ra expression by the MS-protective haplotype could potentially improve thymic TSLP signaling and Treg output.
In the presence of IFNb, IL7Ra was upregulated in all myeloid subsets examined, and expressed at the highest level in myeloid DCs (10-fold upregulated), suggesting these cells may be the most responsive to IL7Ra signaling under conditions of IFNb stimulation. Both IL-7 and TSLP induced significant changes in expression of immune pathways in IFNb-treated monocytes and monocyte-derived DCs, indicating that the upregulation may be of functional significance. IFNb stimulation also unmasked a functional difference in the haplotype conferring highest MS risk: lack of upregulation and consequent lowest expression of IL7Ra in IFNb-treated dendritic cells. IFNb is specifically induced at sites of viral and other microbial infections [35,63] and autoimmune inflammation [64] including the acute MS lesion [65], but is also constitutively expressed at high levels by thymic epithelial cells under non-inflammatory conditions, as shown recently using a reporter mouse model [63]. Attenuated responses of DCs, particularly myeloid DCs, to IL-7 or TSLP in the presence of IFNb, such as during microbial infection, central nervous system inflammation or in the thymus, may contribute to increased MS risk in Hap 4 homozygotes.
Whether upregulation of IL7Ra is involved in the clinical effect of IFNb is not yet known. IFNb treatment increases Treg output from the thymus with concomitant improvements in Treg suppressive function [29], and skewing of the T cell phenotype to Th2 [33,39] has been associated with a long-term favorable clinical response [38]. These effects could potentially result from enhanced TSLP signaling. We are currently investigating the effect of IL7Ra haplotype on immune and clinical response to IFNb in MS patients. To further characterize the effects of TSLP in a tolerogenic environment induced by IFNb, we examined whether subsequent TSLP treatment of IFNb-stimulated DCs would exert an additive suppressive capacity on TCR-mediated T cell proliferation. TSLP partially reversed the suppressive effect of IFNb in this context. These in vitro conditions might not mimic a microenvironment in which improved IL7Ra signaling reduces an autoimmune inflammatory process.
Interestingly, IL7Ra was downregulated by IFNb in DCs of the lymphoid lineage, the plasmacytoid DCs. Plasmacytoid DC precursors are the primary producers of type I IFN in response to viruses such as HSV [66] and induction of IFNb by viral infection has been demonstrated recently using a reporter mouse model [67]. However the amount of type I IFN produced in response to virus is tightly controlled by other cytokines, with IL-7 increasing production [66]. We hypothesise that the downregulation of IL7Ra we observed in response to IFNb represents a negative feedback loop once IL-7-primed pDCs have secreted type I IFN.
The immunomodulatory effects of IFNb are diverse. Interferon responsive genes are over-represented in the list of genes associated with MS susceptibility. Of the 55 genes identified as associated with MS [2], approximately one-third are identified as interferon responsive in the interferome, the database of interferon-regulated genes (n = 18 genes) [68]. Only one-tenth genes of the 20,000 or so genes now described from the human genome are listed as interferon responsive (n = 1996 genes) [68], so the excess of interferon-responsive genes associated with MS is remarkable Table 1. Gene expression in response to IL7Ra signaling in IFNb-treated DC subsets 1 .

Monocytes
All genes 2 : 143 q, 67 Q All genes: 215 q, 248 Q Pathways 3 : none identified Pathways: Q immune response 9 Immature DCs All genes: 79 q, 57 Q All genes: 5 q, 5 Q Pathways: q chemokine activity 4 q immune system response 5 Pathways: none identified Maturing DCs All genes: 16 q, 57 Q All genes: 68 q, 61 Q Pathways: q up-regulation of cytokine biosynthetic process 6 Q chemokine activity 7 Q immune system response 8 Pathways: none identified 1 freshly purified monocytes or monocyte-derived DCs were preincubated for 48 h with IFNb (1000 IU/ml; +LPS for maturing) before 24 h stimulation with IL7 or TSLP (10 ng/ml) and gene expression measured by microarray. 2 total number of genes upregulated (q) or downregulated (Q) at least 1.5-fold by IL-7 or TSLP. This data is also presented as a sum of total genes regulated in Figure 5. (p,E 28 , Fishers exact test). Our findings raise the possibility that IL7Ra and IFNb may work together in myeloid cells to affect susceptibility to MS.

Materials and Methods
The study was approved by the Western Sydney Local Health District Human Research Ethics Committee (HREC2004(1914)) and blood donors gave written informed consent.

Cell Separation and Culture
Peripheral blood mononucleocytes (PBMCs) were isolated from EDTA-treated whole blood of healthy controls using Ficoll Paque Plus (GE Healthcare, Sweden). The study was approved by the Western Sydney Local Health District Human Research Ethics Committee and blood donors gave informed consent.The age range of donors was 28-67 years, both males and females were included and gender was not significantly different between compared groups. Number of subjects used for each experiment is given in Figure legends. Freshly drawn blood was used for all experiments except Figures 3B, 3C and Suppl Fig 1, where monocytes were purified from thawed cryopreserved PBMCs, Monocytes were isolated by positive CD14+ selection using antibody-conjugated magnetic beads (Miltenyi Biotec, Germany) and the purity obtained was routinely 90%. Immature DCs (iDCs) were prepared from monocytes by incubation with IL-4 (20 ng/ml) Figure 6. TSLP treatment of IFNb-stimulated dendritic cells induces T cell proliferation in co-culture (A) IFNb induces expression of tolerogenic genes in dendritic cells. Expression of IL27p28, IDO and IL10 was measured in in vitro cultured immature (IL-4, GM.CSF; iDC) and maturing monocyte-derived dendritic cells (IL-4, GM.CSF, LPS; matDC) (each in triplicate). Expression was measured by RTPCR relative to GAPDH; mean +/2 SEM is shown; *significantly different from media condition by paired t test (IL27p28, p = 0.045; IDO, p = 0.016). (B) IFNb-treated DCs (IFNb) reduce proliferation of CD4 T cells compared to untreated DCs (media), but subsequent DC treatment with TSLP partially reverses this effect (IFNb + TSLP). CFDA-loaded naïve CD4 T cells and antiCD3/antiCD28 beads were added on day 8 to maturing DC culture previously conditioned +/2 IFNb (day 5) and +/2 TSLP (day 7). Proliferation of CD4+ cells was assessed on day 13 by reduction in CFDA median fluorescence intensity. The results are representative of two independent experiments. doi:10.1371/journal.pone.0077508.g006 IL7Ra Haplotype Expression in Dendritic Cells PLOS ONE | www.plosone.org and GM.CSF (70 ng/ml) on day 0, 2 and 4 of culture [69]. Maturing DCs (matDC) were induced by the addition of 1 mg/ml LPS on day 5 of culture for 24 h. Plasmacytoid dendritic cells (pDCs) and myeloid dendritic cells (mDCs) were purified as previously described by magnetic separation [70,71]. The purities of pDC and mDC subsets were routinely 99% and 80% respectively. IFNb-1a (1000 IU/ml Biogen Idec, Switzerland) was added to freshly purified monocytes, mDCs and to day 5-cultured iDCs (simultaneously with LPS for maturing DC culture) for 24 h. Freshly purified pDCs were cultured with 2000 IU/ml IFNb-1a for 24 h. All cells were cultured in triplicate at 10 6 /ml at 37uC in 5% CO 2 in X-Vivo 15 medium (Lonza, Switzerland), washed with phosphate-buffered saline and lysed using Cells-to-Signal Lysis buffer (Ambion, TX, USA). For microarray experiments, fresh monocytes, day 5 iDCs (no LPS) and day 5 maturing DCs (+LPS) were cultured with IFNb for 48 h before addition of TSLP or IL-7 (10 ng/ml; R&D Systems, MN, USA) for 24 h, followed by washing and lysing as above. For proliferation experiments, and day 5 maturing DCs (+LPS) were cultured with IFNbfor 48 h before addition of TSLP or IL-7 (10 ng/ml; R&D Systems, MN, USA) for 24 h, followed by addition of autologous T cells.

Genotyping and qRTPCR
Donors were genotyped and IL7Ra haplotype determined as described [13]. For experiments comparing haplotype differences using homozygotes, Hap 1, 2, and 4 were compared, excluding Hap 3 due to lack of availability (low minor allele frequency; [23]). RNA was extracted from cell lysates using the RNeasy Mini RNA extraction Kit (Qiagen, Germany) and reverse-transcribed using Superscript III (Invitrogen, CA, USA) according to the manufacturer's instructions. IL-7Ra semi-quantitative RT-PCR was performed using SYBR Green (Applied Biosystems, CA, USA) with primers spanning intron 7 (Forward: 59-CTGGAA-CATCTTTGTAAGAAACCAAG-39; Reverse: 59-TAGCTT-GAATGTCATCCACCCT-39). Comparative measurement of mRNA from each haplotype in heterozygotes and determination of the ratio of IL-7Ra mRNA encoding the full length isoform (membrane-bound) to a soluble isoform (exon 6 spliced out) was performed as previously described [13]. Another isoform generated by splicing out of both exon 5 and 6 has recently been described and relative quantitation of this, membrane-bound, and soluble (exon 6 spliced out) IL7Ra was performed as described [41] in a second cohort. Semi-quantification of tolerogenic genes in dendritic cells was performed using SYBR green with primers for IL-27p28 (Forward: 59

Microarray Analysis
Total RNA was extracted from cell lysates using RNeasy Mini RNA extraction Kit (Qiagen, Germany). Total RNA quality was assessed using the Agilent RNA 6000 series II Nano kit (Agilent Technologies, CA, USA), and concentration was measured by NanoPhotometer TM Spectrophotometry (Implen GmbH, Germany). 180 ng total RNA from each sample was biotinylated and amplified using IlluminaH TotalPrep RNA Amplification Kit (Ambion, TX, USA). The complementary mRNA (cRNA) yield was then measured using the NanoPhotometer TM Spectrophotometer (Implen GmbH, Germany). 700 ng of cRNA samples were hybridized onto human HT-12_V3 expression beadchips (Illumina, Inc. CA, USA) profiling 48,804 transcripts per sample.
Raw data was processed using Beadstudio v3 (Illumina, Inc. CA, USA) which involved filtering of the probe sets to keep only those probes which were detected above background in at least one sample. Data were exported into BRB-ArrayTools version 3.8 and quantile normalization was applied to the data. Genes whose expression levels differed by more than 1.5 fold between groups, were deemed to be differentially expressed. Gene lists were exported into GeneGo TM Metacore (St Joseph, MI, USA) where tests were carried out for over-representation of the gene lists in curated biological pathways. A cutoff of 5% false discovery rate was used to determine significance of over-represented pathways.
Naïve CD4 T cell Proliferation Assay CD4+ cells were purified by negative selection, then further purified by positive selection for CD45RA+ cells using antibodyconjugated magnetic beads (Miltenyi Biotec, Germany). Cells were washed twice with PBS and incubated at 10 7 cells/ml in 10 mM carboxyfluorescein diacetate (CFDA)/PBS for 10 min at 37uC. Cells were washed twice with culture medium and resuspended at a final concentration of 4610 6 cells per ml in the DC culture medium. 50 ml of this suspension was added to each well in the presence of anti-CD3/anti-CD28 beads (Invitrogen, CA, USA) (DCs:naïve T cells:beads = 1:1:1). Proliferation was assessed five days later by assessing the reduction in CFDA median fluorescence intensity of CD4+ cells (anti-CD4 by flow cytometry gated on CD4 versus side scatter; LSRII, BD Biosciences, CA, USA).

Statistics
Comparisons were made between gene expression in the presence and absence of IFNb, using a paired t test, and between haplotypes or subsets, using the Student's t test assuming unequal variance. Direction of change in response to IFNb was assessed using the sign test. The comparisons were considered significant if p#0.05. P values were not corrected for multiple comparisons. Error bars in Figures represent standard error of the mean. Figure S1 Splicing of IL7Ra is unchanged in MS and upon IFNb stimulation in myeloid cells of heterozygous Hap 4 carriers. Cryopreserved PBMCs from healthy controls (n = 5) and MS patients (n = 5) were thawed, monocytes purified, and monocytes (A), in vitro cultured immature dendritic cells (IL-4, GM.CSF; iDC) (B) and maturing monocyte-derived dendritic cells (IL-4, GM.CSF, LPS; matDC) (C) were incubated +/2 IFNb (1000 IU/ml) for 24 h. Relative expression of membrane-bound (MB), exon 6 soluble (Sol(-Ex6)), and exon 5,6 soluble (Sol(-Ex5,6)) IL7Ra isoforms was measured as previously described [41] and expressed as a proportion of the total. Mean +/2 SEM is shown. There were no significant differences between controls and MS, or between media and IFNb; bars represent significant differences between isoforms in the same subset under the same condition by paired t test (p,0.05). (DOCX)