Common genetic variants of IL-7 receptor α (IL-7Rα) have recently been shown to affect susceptibility to multiple sclerosis (MS) and type 1 diabetes, and survival following bone marrow transplantation. Transcription of the gene produces two dominant isoforms, with or without exon 6, which code for membrane-bound or soluble IL-7Rα, respectively. The haplotypes produce different isoform ratios. We have tested IL-7Rα mRNA expression in cell subsets and in models of T cell homeostasis, activation, tolerance, and differentiation into regulatory T cell/Th1/Th2/Th17, memory, and dendritic cells (DCs) under the hypothesis that the conditions in which haplotype differences are maximal are those likely to be the basis for their association with disease pathogenesis. Maximal differences between haplotypes were found in DCs, where the ligand is mainly thymic stromal lymphopoietin (TSLP). The MS-protective haplotype produces a much lower ratio of soluble to membrane-bound receptor, and so potentially, DCs of this haplotype are more responsive to TSLP. The TSLP/IL-7Rα interaction on DCs is known to be critical for production of thymic regulatory T cells, and reduced production of these cells in MS susceptibility haplotypes may be a basis for its association with this disease. IL-7Rα mRNA expression varies greatly through cell differentiation so that it may be a useful marker for cell states. We also show that serum levels of soluble receptor are much higher for the MS susceptibility haplotype (p = 4 × 10−13). Because signaling through IL-7Rα controls T cell regulation, this haplotype difference is likely to affect the immunophenotype and disease pathogenesis.
Multiple sclerosis (MS) is the most common chronic neurologic disease of young adults, affecting >2 million people worldwide. Extensive studies of population genetics have shown a major heritable component to the disease, with an increasing risk with genetic closeness to an index case (1). Until 2007, only one inherited risk factor had been validated across multiple populations, the MHC class II DRB1*1501 allele, which confers a 3- to 4-fold relative risk, and implicates Ag presentation to CD4 T cells in MS pathogenesis. Several publications in 2007 and subsequently confirmed our earlier reports (2) of the gene encoding the IL-7 receptor α-chain (IL-7Rα, also known as CD127) as the first non-HLA gene influencing the risk of MS in studies involving >20,000 patients from Australia, the United States, the U.K., and Europe (3). Recently published genome-wide association studies by us (4) and others (5) have identified several further genes affecting MS susceptibility.
With gene identification progressing rapidly, determination of the functional effects of haplotypes influencing disease susceptibility represents the next major challenge in diseases of complex genetics. Haplotypes contain genetic variants that potentially affect regulation of transcription and mRNA processing, especially splicing. This is the first level at which genetic variance can produce functional differences for genes and the level where there is the least noise from other processes. For IL-7Rα, four common haplotypes have been described previously (6). Haplotype 2 is protective against the disease and is tagged by a T allele in exon 6 (rs6897932). This single nucleotide polymorphism (SNP) is associated with reduced exon splicing and the production of less of the gene encoding soluble IL-7Rα in PBMCs of healthy control individuals (7, 8). The putative proximal promoter contains several SNPs, and we have found evidence of haplotypic differences in expression and response to IFN-β (E. Hoe, F.C. McKay, S. Schibeci, R. Heard, G.J. Stewart, and D.R. Booth, submitted for publication). The haplotypes are also tagged by codon-changing SNPs whose functional effects, if any, are yet to be determined (6). The haplotypes have also been associated with other diseases: haplotype 2 decreases risk of type 1 diabetes (9), and haplotype 1 is associated with an increased risk of death following matched unrelated bone marrow transplantation (10).
IL-7Rα is expressed in multiple immune cell types, as a subunit of the heterodimeric receptors for IL-7 (with the common cytokine γ-chain) and for thymic stromal lymphopoietin (TSLP; with the TSLP receptor). IL-7 plays a key and nonredundant role in T lymphocyte differentiation, survival, and proliferation (11), and its heterodomeric receptor is predominantly expressed on T cells. TSLP promotes Th2 differentiation in the periphery and the development of regulatory T cells (Tregs) in the thymus (11). Its heterodomeric receptor is expressed predominantly on dendritic cells (DCs). Inherited loss of function of IL-7Rα in humans leads to a form of SCID (12), underlying its key role.
To further characterize the functional effects of MS susceptibility haplotypes, we have taken two approaches: first, to determine the contexts in which the IL-7Rα haplotype may affect immune function, we selected two major cell types of potential importance in MS pathogenesis, namely T cells and DCs (13), and examined regulation of IL-7Rα under various conditions to determine whether the haplotypes are differentially regulated in response to varying stimuli or cellular transcriptional programs. We found a significant reduction in splicing out of exon 6 in the IL-7Rα MS-protective haplotype mRNA in whole blood, and this difference was much greater in DCs. Second, we examined the effect of haplotype on the concentration of circulating soluble receptor and demonstrate that all peripheral leukocytes are potentially exposed to lower effective signaling through the receptor in all individuals carrying the MS-susceptibility haplotypes.
Materials and Methods
In vitro generation of cell subsets
Peripheral blood was collected in EDTA from healthy control subjects of known IL-7Rα haplotype and mononuclear cells (PBMCs) isolated by Ficoll density gradient separation. Naive T helper cells (CD4CD45RA) were purified from PBMCs by magnetic separation (Miltenyi Biotec, Auburn, CA). For the activation model, CD4CD45RA cells were coincubated with anti-CD3 and anti-CD28 beads (Dynabeads; Invitrogen, Carlsbad, CA) for 48 h (14) at a ratio of 0.2 beads/cell. Th1 and Th2 subsets were generated by activation as above in the presence of IL-12 (30 ng/ml) and anti–IL-4 Ab (0.5 μg/ml) for Th1 cells, or IL-4 (20 ng/ml) and anti–IFN-γ Ab (2.5 μg/ml) for Th2 cells. IL-2 (200 IU/ml) was added to the culture on day 2, and all cytokines and Abs were replenished every 2 d until cell harvest on day 8 (15). For Th17 polarization, T cells were activated for 3 d in the presence of IL-1β (10 ng/ml), IL-6 (50 ng/ml), with anti–IL-4 and anti–IFN-γ Abs as above. On day 3, IL-2 (20 IU/ml) and IL-23 (20 ng/ml) were added, and all cytokines and Abs replenished every 2–3 d until cell harvest on day 12 (16). Tregs were generated by activation of T cells in the presence of TGF-β (10 μg/ml) for 4 d (17). Phenotypes were confirmed by flow cytometry (CD45RO+ for cellular activation; CD25highFoxP3high for Tregs) or RT-PCR (upregulation of T-bet, GATA-3, and RORγt for Th1, Th2, and Th17, respectively).
To deliver an activation signal intermediate between media alone and the activation model (as assessed by downregulation of IL-7Rα gene expression; “signal strength” model), anti-CD3 and anti-CD28 beads were used at a ratio of 1 bead per 40 cells. Conditions of peripheral homeostasis were modeled by culture of CD4CD45RA cells with IL-7 (10 ng/ml) for 7 d with cytokine replenishment every 2–3 d (18, 19), and IL-7–mediated upregulation of IL-2Rα (20) was confirmed by RT-PCR (see below). To model T cell tolerance, CD4CD45RA cells were cultured for 4 d with 2.5 anti-biotin microbeads (Miltenyi Biotec) loaded with 0.5 pg anti-CD3 beads (UCHT1 from eBioscience, San Diego, CA) per cell, rested for 2 d, then restimulated in wells coated with anti-CD3 (OKT3) at a concentration of 0.1 μg/ml for an additional 4 d (21), determined in titration experiments to induce relatively low levels of T cell proliferation by [3H]thymidine incorporation.
Central memory T cells were generated by 5-d coculture of CD4CD45RA cells with mature autologous monocyte-derived DCs (described below; at a ratio of 10 T cells to 1 DC) and a nonspecific Ag, staphylococcal exotoxin A (1 ng/ml). IL-2 (20 IU/ml) was added for an additional 9 d with replenishment every 2–3 d (22), and CD45RO+CCR7+ phenotype was confirmed by flow cytometry. DCs were generated from CD14 monocytes (purified from PBMCs by positive selection; Miltenyi Biotec) by culture with IL-4 (17 ng/ml) and GM-CSF (67 ng/ml) over 5 d, with or without maturation with LPS (1 μg/ml) for an additional 2 d (23). The mature DC phenotype was confirmed by flow cytometry (CD83+ and HLA-DR+) (24). For each model, cells were washed briefly in ice-cold PBS and harvested immediately in Cells-to-signal lysis buffer (Ambion, Austin, TX) to stabilize labile mRNAs and stored at −80°C until RNA extraction (RNeasy; Qiagen, Valencia, CA).
Genotyping and gene expression
IL-7Rα haplotype was determined from DNA by RFLP as described previously (6). Total gene expression was measured by quantitative RT-PCR. Relative IL-7Rα haplotype expression was measured in heterozygotes by SNaPSHOT analysis (2, 25, 26) where the relative abundance of two codon region haplotype-tagging SNPs is quantitated in cDNA. The ratio of full-length to soluble isoforms was determined from specifically amplified IL-7Rα gene products (7) using an Agilent Bioanalyzer (Waldbronn, Germany). IL-2Rα mRNA expression was measured by using TaqMan Universal PCR Mastermix and commercially available TaqMan primers (TaqMan gene expression assay, accession number Hs 00907778; Applied Biosystems, Foster City, CA).
Soluble IL-7Rα ELISA
Sera were collected from people with MS (n = 115) and healthy control subjects (n = 100), each of whom had been genotyped for IL-7Rα, and soluble IL-7Rα levels were determined by sandwich ELISA using a capture Ab (5 μg/ml, MAB306) and a biotinylated detection Ab (50 μg/ml, BAF306; both from R&D Systems, Minneapolis, MN) incubated with HRP-conjugated streptavidin and tetramethylbenzidine as substrate. Standard curves were generated from a recombinant soluble IL-7Rα-Ig conjugate (Apollo Biosciences, Sydney, Australia).
Cell surface IL-7Rα expression
Whole blood was collected in EDTA from 18 people with MS and 20 age- and gender-matched healthy control subjects. MS subjects had not been treated with immunomodulatory therapy in the previous 3 mo. Blood was collected between 8 am and 2 pm, and samples were processed exactly 2 h postcollection. Buffy coat was stained either with CD45RA-FITC, CrTh2-PE, CD4-PerCP, IL-7Rα-APC, CCR7-PECy7, and CD25-APC-Cy7 or CD4-PerCP with isotype controls for FITC, PE, APC, PECy7, and APC-Cy7. Lymphocytes were gated on forward and side scatter parameters, and CD4 T cells were gated as a subset of lymphocytes from a plot of CD4 against side scatter. Mean fluorescence intensity for IL-7Rα-APC on all CD4 T cells is presented, corrected by subtraction of IgG-APC mean fluorescence intensity of CD4 T cells. Only the results for the CD4 subset are presented in this study; other Ab details are included for accurate reporting of the Ab panel from which the data were collected. All Abs were from BD Biosciences (San Jose, CA), except APC-conjugated Abs (from R&D Systems). Red cells were lysed with BD-Lyse (BD Biosciences), and the leukocytes were fixed, permeabilized, and stained with FoxP3-Pacific Blue (eBioscience) using the manufacturer’s buffer set and staining protocol. Samples were analyzed using an LSRII (BD Biosciences) on the day of staining, and all experiments were performed within a 1-mo window. Automatic compensation generated using FACSDiva (BD Biosciences) on the first day of staining was applied to all subsequent data files.
This study was approved by the Sydney West Area Health Service Human Research Ethics Committee HREC2002/9/3.6(1425), and all participants gave written informed consent.
We used in vitro modeling of CD4 and DC subsets to define contexts in which the MS-susceptibility haplotypes of IL-7Rα may be differentially regulated and thus potentially relevant in MS pathogenesis. We used different conditions designed to model T cell activation, differentiation, tolerance, and homeostasis, and DC maturation and measured total and relative haplotype expression in these models. We also determined the effect of haplotype on soluble and cell surface IL-7Rα protein expression.
IL-7Rα total expression in CD4 T cells and DCs
We first measured total expression of IL-7Rα mRNA in each of the cellular models. IL-7Rα mRNA was expressed at a high level in resting naive CD4 T cells and was downregulated over time of activation by CD3 and CD28 stimulation, with reciprocal upregulation of the IL-2Rα (Fig. 1A, 1B). Downregulation of IL-7Rα was observed in all models of CD4 T cell differentiation (Th1, Th2, Th17, Treg, central memory), during strong (activation) and intermediate (signal strength) levels of CD3/CD28 stimulation, and in response to consumption of IL-7 (homeostasis) (Fig. 1C). Weak CD3 stimulation in the absence of CD28 costimulation, mimicking T cell tolerance to low-affinity Ag (e.g., self-Ag) did not downregulate IL-7Rα (Fig. 1C). IL-7Rα was expressed at a very low level on immature, monocyte-derived DCs; it was significantly upregulated upon maturation with LPS but only to levels ∼20-fold lower than those expressed on naive CD4 T cells (Fig. 1D).
IL-7Rα relative haplotype expression in CD4 T cells and DCs
To determine whether particular IL-7Rα haplotypes are preferentially expressed in certain cellular contexts, we measured expression of each haplotype in heterozygote cells under different conditions. Heterozygotes are particularly useful for this purpose, because both haplotypes are in identical conditions (the same cells) but can respond differently if their regulatory sequences are differentially controlled. For each condition, we examined at least four individuals. The MS-protective haplotype, haplotype 2, was never preferentially expressed in CD4 T cells: comparisons in Hap 1/Hap 2 and Hap 4/Hap 2 heterozygotes demonstrate that haplotype 2 is expressed in equimolar proportions, with the MS-susceptibility haplotypes in CD4 T cells, including following Th1 or Th17 differentiation or IL-7 consumption (homeostasis) for haplotype 1, and following differentiation to a Th2 or central memory phenotype for haplotype 4 (Fig. 2A). In mature DCs, the converse was observed: the MS-protective haplotype was preferentially expressed (at ∼30% higher level) over the MS-susceptibility alleles. Relative expression was examined over time in the activation model, and equal expression of haplotypes 2 and 4 in resting naive T cells was maintained over 48 h of activation-induced downregulation (Fig. 2B).
IL-7Rα splicing and protein expression of receptor isoforms
We and others have found previously that the MS-protective haplotype is associated with reduced splicing of exon 6 and the production of less of the gene encoding soluble IL-7Rα in PBMCs of healthy control individuals (7, 8) and MS subjects (E. Hoe, F.C. McKay, S. Schibeci, R. Heard, G.J. Stewart, and D.R. Booth, submitted for publication). However, in whole blood, where the majority of the IL-7Rα transcript is T cell derived, the soluble isoform represented only a minor component of the total gene RNA in all haplotypes.
To determine whether exon 6 splicing is similar for the haplotypes in other cell types, full-length and soluble isoforms were compared in whole blood (Fig. 3A) and mature monocyte-derived DCs (Fig. 3B). Lower expression of the soluble isoform was observed in whole-blood RNA from haplotype 2 than haplotype 1, with haplotype 1/2 heterozygotes showing intermediate levels of exon splicing. As for whole-blood RNA, the MS-protective haplotype showed reduced splicing of exon 6 in mature monocyte-derived DCs. However, the transcript encoding soluble IL-7Rα isoform was expressed at a much higher level relative to the full-length isoform in DCs, such that full-length and soluble transcripts are equimolar for the MS-susceptibility haplotype. DCs have both increased expression (Fig. 2A) and a higher proportion of the isoform encoding the membrane-bound receptor for the MS-protective haplotype (Fig. 3).
We also examined haplotype differences at the level of protein expression. DCs of the MS-protective haplotype secreted ∼2-fold lower levels of soluble IL-7Rα protein upon maturation with LPS in vitro, compared with the MS-susceptibility haplotype (Fig. 4). Ex vivo, soluble receptor was also measured in peripheral blood (sera) from a large cohort of people with MS (n = 115) and healthy controls (n = 100), including individuals who do not carry the MS-protective haplotype (not Hap 2; n = 132) and individuals who carry either one (n = 69) or two (n = 14) copies of this haplotype (Hap 2). Soluble receptor was significantly lower in sera from individuals with the MS-protective haplotype (p < 4 × 10−13; Fig. 5A), as well as in the MS (p < 2 × 10−7) and healthy control cohorts (p < 6 × 10−7) analyzed separately. A significant dose effect of the haplotype 2 allele was observed. There was no difference in serum IL-7Rα (sIL-7Rα) between MS and controls. There was a trend toward higher sIL-7Rα in the MS group compared with healthy controls (45.9 [95% CI: 41.50–50.37] and 41.7 [37.8–45.6] ng/ml, respectively; p = 0.303), as would be predicted from their expected lower frequency of the protective haplotype. However, haplotype 2, the protective haplotype, was as common in MS as controls in this cohort (frequency 0.23 in both MS and controls). A much larger study would be needed for a genotype odds ratio of 1.2 to produce statistically significant differences in haplotype frequency and its effect on serum sIL7Rα between the two cohorts. We also analyzed cell-surface expression of IL-7Rα on CD4 T cells of healthy controls (n = 20) and people with MS (n = 18) and found a trend toward higher cell-surface expression of IL-7Rα for the MS-protective haplotype (n = 17) compared with those who do not carry haplotype 2 (n = 21) (Fig. 5B), but no differences between MS and healthy controls. We found no evidence to suggest that membrane-bound IL7Rα is different in cells of people with MS, or that serum (soluble) IL-7Rα is different in disease; nor to expect that mRNA splicing may be different in MS.
These data demonstrate IL-7Rα mRNA expression varies >20-fold during CD4 T cell and DC differentiation (Fig. 1). The IL-7Rα isoform ratio and level of expression is affected by haplotype, especially in DCs (Fig. 3). However, the relative reduction or increase in mRNA expression for each haplotype was not different throughout cell differentiation and in response to stimulation, as measured by haplotype expression in heterozygotes in T cell and DC models (Fig. 2). In DCs, the haplotypes appear to be under different regulation from each other, more so than in T cells. The MS-protective haplotype has previously been reported to produce significantly less soluble IL-7Rα mRNA in PBMCs and cell lines (7, 8), and we demonstrate in this study that this is also so in CD4 T cell and DC subsets. We confirm this haplotype difference extends to the protein level by demonstrating that in sera there is a highly significant difference in circulating sIL-7Rα, and a difference in cell surface IL-7Rα associated with the haplotype 2 genotype (Fig. 5).
The outcome of signaling from both TSLP and IL-7 is known to vary according to microenvironment. The ratio of soluble to insoluble mRNA isoforms varies greatly between cells being maximal in DCs. These cells respond to TSLP. Studies on TSLP function have so far indicated that in the periphery it promotes Th2 polarization and, in the thymus, Treg differentiation (27). Decreased signaling because of increased production of the soluble IL-7R protein could result in fewer or less effective Tregs. In MS, thymic output of Tregs is reduced (28), and diminished suppressive capacity of Tregs has been described previously (reviewed in Ref. 29). This could also be the basis for the IL-7Rα association with type 1 diabetes, where IL-7Rα haplotype 2 is also protective (9). In other diseases, the clinical significance of the haplotype may have a different basis. In allogeneic bone marrow transplants from matched unrelated donors, when the donor carries haplotype 1, the prognosis is poor (10). The haplotype 1 donor T cells may be less likely to proliferate in transplants because of decreased relative availability of IL-7. They then fail to confer protection from infection, and so, such transplants have an increased mortality. Haplotype 2 carriers have better recovery from lymphopenia in therapy for HIV (R. Rajasuriar et al., personal communication), possibly as a result of improved availability of IL-7 to promote CD4 T cell proliferation.
Soluble cytokine receptors are well-known to regulate signaling (30). Goodwin et al. (31) found evidence this was also true for sIL-7Rα, when they demonstrated harvested supernatants from COS-7 cells expressing human sIL-7R inhibited the binding of 125I-IL-7 to an IL-7–dependent murine cell line. sIL-7Rα binding to TSLP has yet to be demonstrated. Although reduced signaling through the receptor may have predictable consequences in vitro on immediate exposure, for continuous exposure and in vivo, the outcome is far less predictable. Membrane-bound receptor is internalized on binding to ligand (32), and a feedback loop may exist such that cell surface expression is dependent on available ligand. Homeostatic correction through internalization of receptor may reduce the impact of an altered balance of sIL-7Rα and IL-7/TSLP in carriers of the different haplotypes.
A further complication is that ligand availability is not dependent solely on unbound protein. IL-7 binds extensively to fibronectin and heparan sulfate components of the extracellular matrix (33) and thus may be largely displayed in cell-associated form and consumed by cells at the sites of production (34). Soluble receptor may affect response to ligand both locally (around sites of production and cellular interactions) and systemically (because of binding to serum or lymph circulating ligand).
The effect of increased IL-7 availability may be inferred from its effect when added experimentally or therapeutically. IL-7 injection into mice increases T cell numbers, in thymic and nonthymic compartments (35). It also increases immune response when used as an adjuvant (36). In humans, it has been tested to improve recovery from lymphopenia induced by drug treatment in two clinical trials. In both trials the numbers of CD4 and CD8 cells, both naive and memory, increased, but the numbers of Tregs did not (37, 38). However, in HIV patients, a single dose of recombinant human IL-7 increased CD4 and CD8 counts but did not change Treg frequency (39). Such effects would not be expected to be useful in autoimmunity, where an increase in relative Tregs would be desirable. Relative T cell subset responses to IL-7 in these clinical settings were from lymphopenic individuals, the immunophenotype response to increased IL-7 (or TSLP) availability in other settings may be different. The net effect of differential haplotype effects on expression may need to be determined from correlations of genotype with immunophenotype. In MS there is evidence of an altered immunophenotype (7, 40), but as yet no correlation with IL-7Rα genotype has been described for these aberrant cell subsets. It is also possible that immunophenotype differences are dynamic, differing at the onset of disease and as it progresses, so that genotype effect on it are masked or difficult to dissect.
As the current genome-wide analysis studies identify additional genes, the approach we have used could be useful in identifying the functional role of haplotypes of genes associated with MS and other common autoimmune diseases. This is especially so for those genes expressed predominantly in T cells and DCs, such as IL-2Rα (CD25), CD6, CD40, CD58, IFN regulatory factor 8, and CD120a, now also confirmed as associated with MS (4, 40, 41).
We thank the people with MS and healthy control participants for donating blood for the study and Najwa Marmash for excellent technical assistance.
Disclosures The authors have no financial conflicts of interest.
This work was supported by an Australian Research Council Industry Linkage grant with Biogen Idec (to E.H., K.G., and F.C.M.).
Abbreviations used in this paper:
single nucleotide polymorphism
regulatory T cell
thymic stromal lymphopoietin.