|Home | About | Journals | Submit | Contact Us | Français|
Interleukin-35 (IL-35) belongs to the IL-12 family of heterodimeric cytokines but has a distinct functional profile. IL-35 suppresses T cell proliferation and converts naïve T cells into IL-35-producing iTr35. Here we show that IL-35 signals through a unique IL-12Rβ2:gp130 heterodimer or via homodimers. Conventional T cells are sensitive to IL-35-mediated suppression in the absence of one but not both receptor chains, whereas signaling through both chains is required for IL-35 expression and iTr35 conversion. IL35R signaling requires the transcription factors STAT1 and STAT4, which form a unique heterodimer that binds to distinct sites within the Il12a and Ebi3 promoters. This unconventional mode of signaling, which is distinct from other members of the IL-12 family, may broaden the spectrum and specificity of IL-35-mediated suppression.
Lethal autoimmune diseases can develop in the absence of effective immunosuppressive mediators. Regulatory T (Treg) cells are necessary for immune homeostasis and to prevent autoimmunity, yet also limit anti-tumor immunity1,2. Treg cells utilize a broad range of mechanisms to suppress immunity including suppressive cytokines such as IL-35, IL-10 and TGF-β. IL-35, belongs to the IL-12 cytokine family, which also includes IL-12, IL-23, and IL-27. The latter are heterodimeric proteins composed of shared α chains (p19 [Il23a], p28 [Il27] and p35 [Il12a]) and β chains (p40 [Il12b] and Ebi3 [Ebi3]), are produced by antigen producing cells, and have immunomodulatory activities. IL-12 and IL-23 are pro-inflammatory cytokines that can promote TH1 and TH17 differentiation and function3. IL-27 serves to initiate TH1 polarization but exhibits a host of immunomodulatory activities including limiting TH2 activity4, blocking TH17 differentiation and TGFβ-driven induced Treg (iTreg) formation, and IL-10 production5–7. IL-35 is distinct from other family members in that it is produced by Treg cell populations and is suppressive. In vitro and in vivo, IL-35 has two known biological effects; suppression of conventional T (Tconv) cell proliferation and conversion of naive Tconv cells into a strongly suppressive induced Treg cell population, termed iTr35, which function via IL-358–10. Thus a primary downstream target of IL-35 receptor (IL-35R) signaling is induction of Ebi3 and Il12a transcription.
Chain sharing extends beyond the cytokines to their receptors and signaling pathways. IL-12Rβ1 (Il12rb1) and IL-12Rβ2 (Il12rb2) form the IL-12 receptor11, while IL-23 signals through IL-12Rβ1 paired with the IL-23R (Il23r)12. In contrast, IL-27 utilizes gp130 (Il6st) and IL-27R, also known as WSX-1 or TCCR (Il27ra)13. Although IL-12Rβ1, IL-12Rβ2, IL-23R and IL-27R usage is restricted to the IL-12 cytokine receptor family, gp130 is ubiquitously expressed and is a component of receptors for many cytokines including IL-6, IL-11, leukocyte inhibitory factor (LIF), oncostatin M (OSM), ciliary neurotrophic factor (CNTF), cardiotrophin 1 (CT-1) and cardiotrophin-like cytokine (CLC)14.
Following receptor engagement, signaling is initiated by activation of members of the Janus kinase (Jak) family followed by phosphorylation and nuclear localization of members of the signal transducer and activator of transcription (STAT) family12, 15. A major conundrum in cytokine signaling is explaining how so many cytokines (over 50) can mediate such diverse functional events via only 7 STAT proteins16. While multiple STAT proteins can be phosphorylated following receptor engagement, usually only one or two are essential. For instance, although STAT1 (Stat1), STAT3 (Stat3), STAT4, (Stat4) and STAT5 (Stat5) can all be activated by IL-12 signaling, only STAT4 is required to mediate IL-12’s effects15, 17. Although STAT phosphorylation following IL-23 signaling is similar to that of IL-12, STAT3, but not STAT4, appears to be the critical signal transducer for IL-2312. Similarly, IL-27 receptor engagement activates STAT1, STAT2, STAT3 and STAT5 to varying degrees but STAT1 and STAT3 seem most important for complete IL-27 bioactivity18, 19.
While most receptor chains can contribute to both cytokine binding and downstream signaling, for some cytokine receptors one chain predominates in mediating signaling. For instance, Il12rb2−/− T cells are defective in their ability to phosphorylate STAT4, generate IFNγ and promote TH1 differentiation despite the presence of high affinity IL-12 binding sites on IL-12Rβ120. The relative contributions of gp130 and WSX1 in IL-27 signaling are more obscure. Although both receptor chains are necessary for STAT phosphorylation, IL-27 function can be abrogated with anti-gp130 mAb13.
Given that IL-35 is comprised of chains shared with IL-12 and IL-27, we hypothesized that the IL-35R chains and the STATs that mediate its signal transduction might also be components used to mediate IL-12 and IL-27 signaling. However, as IL-35 mediates distinct biological events from other members of the IL-12 family, the use of alternate receptors and signaling pathways could not be ruled out. In this study we addressed four important questions: (1) what receptor chains comprise the IL-35R, (2) how is IL-35R expression induced, (3) which STATs mediate IL-35R signaling, and (4) how does IL-35R signaling uniquely induce IL-35 secretion and iTr35 induction?
IL-35 mediates two distinct functions in T cells: suppression of proliferation and conversion to IL-35-producing iTr35 cells8, 10. Thus, two anticipated functional phenotypes of genetic manipulation of IL-35R expression are resistance to IL-35-mediated suppression and prevention of iTr35 conversion. We first asked whether IL-35 could suppress the proliferation of CD4+CD45RBhiCD25− naïve CD4+ Tconv cells that lacked expression of each of the IL-12 family receptor chains. For these experiments we either used HEK293T-derived recombinant IL-35 (rIL-35) or, to remove any potential effects of 293T supernatant, Protein G beads coupled with either IL-35 conjugated via a non-neutralizing anti-p35 specific monoclonal antibody (mAb) or isotype control mAb8, 10. The ability of IL-35 to suppress Il27ra−/− and Il12rb1−/− Tconv cells was comparable to that seen with wild-type Tconv cells (Supplementary Fig. 1a). In contrast, Tconv cells that lacked expression of either IL-12Rβ2 (Il12rb2−/−) or gp130 (CD4cre × Il6stfl/fl referred to herein as Il6stΔT) were partially resistant to IL-35-mediated suppression (Fig. 1a and Supplementary Fig. 2a). However, T cells that lacked expression of both gp130 and IL-12Rβ2 (Il12rb2−/− × CD4cre × Il6stfl/fl mice, referred to herein as IL-35RΔT) were completely resistant to either soluble or bead-bound IL-35.
Next, we asked if IL-35R-deficient Tconv cells are also resistant to suppression mediated by regulatory T cell populations expressing a natural source of IL-35 (iTr35). The induced Treg population, iTr35 suppresses proliferation of Tconv cells exclusively via IL-35 8. Both exogenously added IL-35 and IL-35-producing Treg cells induced conversion of Tconv cells to iTr35 in vitro and in vivo under inflammatory conditions. As expected, neutralizing IL-35 activity with anti-IL-35, but not isotype control, blocked suppression of wild-type Tconv cells by iTr35 (Fig. 1b). Consistent with the rIL-35 observations, Tconv cells that lacked expression of IL-12Rβ2 or gp130 alone also appeared partially resistant to iTr35-mediated suppression. However, loss of both chains of the IL-35R (IL-35RΔT) rendered cells completely resistant to iTr35-mediated suppression (Fig. 1b).
Natural Treg cells that lack IL-35 expression (Ebi3−/− or Il12a−/−) have partially defective function in vitro and in vivo10. As anticipated, Il12rb2−/−, Il6stΔT and IL-35RΔT Tconv cells were partially resistant to natural Treg (nTreg)-mediated suppression (Supplementary Fig. 1b). Nevertheless, this defect is limited due to the multitude of alternative regulatory mechanisms available to nTreg cells1. However, nTreg cells activated in the presence of Tconv cells in the upper chamber of a Transwell plate can mediate potent IL-35-dependent, contact-independent suppression of Tconv cells in the bottom well9. Whereas wild-type Tconv cells were potently suppressed by nTreg-Tconv co-cultures, IL-35RΔT Tconv cells were completely resistant (Supplementary Fig. 1c). Taken together these data suggest that Tconv cells lacking both gp130 and IL-12Rβ2, but not IL-12Rβ1 and WSX1, are resistant to IL-35-mediated suppression regardless of source. Surprisingly, Tconv cells lacking just one chain of the putative IL-35R were partially resistant in in vitro suppression assays suggesting that signaling can occur with only one chain of the IL-35R expressed.
In addition to suppressing T cell proliferation, IL-35 also can induce Ebi1 and Il12a expression, and the conversion of Tconv cells into IL-35-producing iTr35 cells. We determined if loss of one or both chains of the putative IL-35R could impact the ability of receptor chain deficient Tconv cells to convert to iTr35. Although stimulation of naive wild-type Tconv cells in the presence of IL-35 resulted in the strong upregulation of Ebi3 and Il12a mRNA, deficiency in either receptor chain blocked IL-35 upregulation (Fig. 2a). Further, while wild-type cells differentiated into iTr35 were much more suppressive than cells differentiated to iTrCon cells, Il12rb2−/− and Il6stΔT cells differentiated in the presence of IL-35 failed to develop a suppressive iTr35 phenotype (Fig. 2b). Despite the fact that Il12rb2−/− and Il6stΔT Tconv cells are only partially resistant to IL-35-mediated suppression in vitro, they failed to convert into suppressive iTr35 cells. These data suggest that single-chain signaling is sufficient for partial IL-35-mediated suppression, but a complete IL-35R (IL-12Rβ2 + gp130) is required to induce IL-35 production.
Given that iTr35 mediate suppression entirely via IL-358, we utilized two different models to assess the requirement of IL-12Rβ2 and gp130 in IL-35-mediated suppression in vivo. First, iTr35 cells regulate the homeostatic expansion of Tconv cells in the lymphopenic environment of recombination activating gene-1 (Rag1)−/− mice. Purified wild-type, Il6stΔT, Il12rb2−/− or IL-35RΔT Thy1.2+ Tconv cells, either alone or in the presence of Thy1.1+ iTr35 cells, were adoptively transferred into Rag1−/− mice. Seven days after transfer, suppression of Tconv cell expansion was monitored by determining Thy1.2+ Tconv cell numbers. Whereas iTr35 cells significantly limited the proliferation of wild-type, Il6stΔT and Il12rb2−/− Thy1.2+ Tconv cells, they failed to block the expansion of IL-35RΔT Thy1.2+ Tconv cells (Fig. 3a).
Second, iTr35 can also block the anti-tumor CD8+ T cell response against B16 melanoma, thereby facilitating tumor growth. Wild-type, Il6stΔT, Il12rb2−/− or IL-35RΔT CD4+ and CD8+ T cells, either alone or together with iTr35 cells, were adoptively transferred into Rag1−/− mice. The following day, mice were inoculated intradermally with B16 melanoma cells and tumor size determined after 14 days. In the absence of iTr35, tumor burden was similar between mice receiving CD4+ and CD8+ T cells, regardless of genotype (Fig. 3b). When present, iTr35 blocked anti-tumor immunity in wild-type, Il6stΔT or Il12rb2−/− CD4+ and CD8+ T cell recipients, which consequently developed large tumors. In contrast, IL-35RΔT CD4+ and CD8+ T cell recipients were resistant to iTr35-mediated suppression, resulting in limited tumor growth. Recipient antigen presenting cells (APCs) contribute to tumor immunity and may express the IL35R. However, the impact of IL-35 on APCs has not been examined here as only the transferred CD4+ and CD8+ T cells lack receptor expression. Also, tumor growth is equally determined by IL-35RΔT CD4+ or CD8+ T cells in the presence of iTr35 suggesting that an effect of IL-35 on APCs in this setting is at most limited. Collectively, these data demonstrate that IL-35R-deficient T cells are resistant to IL-35-mediated suppression in vivo. In both Rag1−/− and tumor models T cells lacking just one of the IL-35R chains (IL-12Rβ2 or gp130) could still be suppressed by IL-35 in a manner indistinguishable from wild-type cells, further suggesting that IL-35 can mediate signaling via a single receptor chain in vivo, contrasting observations showing that in vivo sensitivity to IL-12 and IL-27 are completely abolished following loss of one of the chains of their respective receptors20–22.
In order to obtain direct physical evidence that IL-35 binds to gp130 and IL-12Rβ2, we first employed laser scanning spinning disc confocal microscopy combined with total internal reflection fluorescence (TIRF) illumination and fluorescence resonance energy transfer (FRET)-based detection. mCherry-tagged IL-12Rβ2 plus either (GFP-tagged IL-12Rβ1 or GFP-tagged gp130 were transiently expressed on HEK293T cells. As expected, a significant increase in the normalized FRET (nFRET) signal was observed with cells expressing the IL-12R (IL-12Rβ2+IL-12Rβ1) following addition of IL-12 but not IL-35 (Fig. 4a). Similarly, a significant increase in nFRET was observed with cells expressing the putative IL-35R (IL-12Rβ2+gp130) following addition of IL-35 but not a negative control (Fig. 4b). The most likely mechanism of IL-35-induced signaling via a single chain is homodimerization. However, we could not detect gp130 or IL-12Rβ2 homodimerization using this approach (i.e. with gp130-GFP + gp130-mCherry; data not shown). It is possible that homodimeric receptors associate with reduced affinity and/or kinetics or an altered mode of assembly (distance and/or rotation) which may limit FRET efficiency.
As an alternate approach, we used the fluorescent protein-tagged receptor chains in co-immunoprecipitation experiments. Various combinations of receptor constructs were co-transfected into HEK293T cells and treated with either 293T-derived IL-12 or IL-35. Surface receptors were cross-linked and subjected to immunoprecipitation using mCherry (DsRed) -specific antibodies, followed by GFP detection by immunoblotting. As expected, IL-12 induced heterodimerization of IL-12R β1 and IL-12Rβ2, but none of the other receptor chain combinations tested (Fig. 4c,d). Consistent with our FRET observations, IL-35 induced the homodimerization and heterodimerization of gp130 and IL-12Rβ2 (Fig. 4c,d). These results suggest that IL-35 can bind to three receptor combinations, providing mechanistic insight into the partial resistance to IL-35-mediated suppression observed following single-chain deletion. It should be noted that while we use the term dimerization here as the minimal likely mode of receptor association, we cannot rule out the possibility of high order assembly structures being induced by IL-35 as has been suggested for IL-623. Taken together with the functional analysis above, these data confirm that IL-12Rβ2 and gp130 constitute the most plausible components of the IL-35R. However, additional molecules might facilitate cytokine binding or downstream signaling, although this would be unprecedented within the IL-6R and IL-12R families14, 24.
We next assessed the pattern of IL-35R expression and how this might correlate with and predict cellular sensitivity to IL-35. Il6st and, thus, gp130 surface expression is fairly ubiquitous on resting T cells, changing modestly following activation25, 26. Optimal responsiveness of Tconv cells to IL-12 requires activation in the presence of IL-2 or IL-27, which up-regulate expression of the IL-12β2 receptor chain27–29. Consistent with these previous observations, Tconv cell activation in the presence of increasing doses of IL-2 or IL-27 induced increased expression of Il12rb2 (Fig. 5a). Addition of neutralizing IL-2 antibodies resulted in little to no Il12rb2 upregulation upon T cell receptor (TCR) stimulation, suggesting that autocrine IL-2 was required for increased receptor expression upon T cell activation. In our system, Il12rb1 and Il6st levels did not change appreciably upon activation with either of these cytokines (Fig. 5a).
To examine the biological consequences of receptor upregulation, Tconv cells were activated in the presence of IL-2, IL-27, or neutralizing antibodies to IL-2, rested, and these previously activated T cells (Tact) restimulated in the presence of titered amounts of IL-35, with fresh Tconv cells included for comparison. Tconv cells activated in the presence of IL-2 or IL-27 were significantly more sensitive to IL-35 treatment than Tconv cells activated in their absence or fresh Tconv cells (Fig. 5b, c).
Analysis of TH1, TH2 and TH17 cells revealed variable IL-35R chain expression (Supplementary Fig. 2). Notably, TH1 cells express significantly higher levels of Il12rb2 than Tconv cells, TH2 or TH17 cells, while Il6st is more highly expressed on TH1 and TH2 cells compared to Tconv cells and TH17. Tconv cells, TH1 and TH2 cells are all susceptible to IL-35-mediated suppression while TH17 cells are resistant (VC, LC, DAAV et al, unpublished results). Thus, while IL-35R chain expression is likely to influence susceptibility to IL-35-mediated suppression, other factors may also contribute. Nevertheless, our data implicate IL-2 and IL-27 as important inducers of IL-35R upregulation and thereby sensitization to IL-35-mediated suppression.
Given that the IL-35R utilizes IL-12Rβ2 and gp130, it seemed logical that signaling would be mediated via the STAT family of transcription factors. Consistent with previous reports27, 28, IL-12 treatment of Tconv cells resulted in intracellular phosphorylation of STAT4 while IL-27 signaling induced STAT1 and STAT3 phosphorylation (Fig. 6a). IL-35 treatment of wild-type Tconv cells resulted in phosphorylation of STAT1 and STAT4, but no activation of either STAT3 or STAT5 above background. Moreover, no induction of pSTAT1 or pSTAT4 was observed in Tconv cells that lacked the IL-35 receptor. Immunoblotting confirmed the induction of substantial pSTAT1, but somewhat weaker pSTAT4 activation, in Tconv cell blasts treated with IL-35 (Fig. 6b).
Given that T cells expressing only one IL-35R chain were still resistant to IL-35-mediated suppression, we asked which STATs associated with signaling via what receptor chain. Intracellular phospho-flow cytometry and immunoblot analysis showed that T cells that express IL-12Rβ2, but not gp130, only induce pSTAT4 in response to IL-35, while T cells that express gp130, but not IL-12Rβ2, only induce pSTAT1 (Fig. 6c,d). These data further confirm the ability of IL-35 to mediate signaling in the presence of only one of the IL-35R chains.
To determine which STATs were required to mediate IL-35 signaling, we examined the susceptibility of Tconv cells that lacked STAT1, STAT3 or STAT4 to IL-35-mediated suppression. Whereas IL-35 suppressed the proliferation of Stat3ΔT Tconv cells (CD4cre × Stat3fl/fl) to a degree comparable to wild-type Tconv cells, suppression of Stat1−/− and Stat4 −/− Tconv cells was significantly reduced (Fig. 6e). Similarly, nTreg-mediated suppression of Stat1−/− and Stat4−/−, but not Stat3ΔT, Tconv cells was also impaired (Supplementary Fig. 3). Importantly, Il12rb2 and Il6st receptor mRNA expression was comparable in Stat1−/−, Stat4 −/− and wild-type Tconv cells (data not shown). Collectively, these data suggest that both STAT1 and STAT4 are required for maximal IL-35-mediated suppression.
IL-35 can convert Tconv cells into hypo-responsive, strongly suppressive iTr358. Thus IL-35 signaling induces Ebi3 and Il12a expression, IL-35 production and the generation of iTr35. Mutants deficient in one chain of the IL-35R that were only able to activate one of the two required STAT proteins, failed to convert to iTr35 (Fig. 2). This suggested that both STAT1 and STAT4 signaling downstream of the IL-35R were required to induce transcription of Ebi3 and Il12a, a prerequisite for conversion of T cells into iTr35. IL-35 induced significantly less Ebi3 and Il12a transcription in STAT1- and STAT4-deficient Tconv cells, implicating a coordinated role for these transcription factors in mediating IL-35 production and iTr35 conversion (Supplementary Fig. 4a). Considering the rapid increase in Ebi3 and Il12a expression after stimulation with IL-35 (1–3h), it is unlikely that signaling is occurring through an intermediate step (Supplementary Fig. 4b).
Several cytokine receptors activate STAT1 and STAT4 to drive pro-inflammatory TH1-type responses, so it is unclear how their activation via the IL-35R could give rise to a very different outcome. Two well-known activators of STAT1 and STAT4 are IFN-γ and IL-12, respectively17, 30. Treatment of activated Tconv cells with IL-12 + IFN-γ or IL-35 alone induced STAT1 and STAT4 activation (Fig. 7a). However, Tconv cells stimulated with IL-12 + IFN-γ did not upregulate Ebi3 or Il12a, and thus would not secrete IL-35 and become iTr35 (Fig. 7b). Thus, the activation of STAT1 and STAT4 by IL-35 leads to a distinct outcome.
Although STAT proteins generally transduce cytokine signals by homodimerization, which promotes their translocation to the nucleus, STAT1:STAT2 and STAT1:STAT3 heterodimerization has been implicated in interferon and IL-6 receptor signaling, respectively31. Reciprocal co-immunoprecipitation and immunoblot analysis demonstrated that IL-35 induces the formation of a STAT1:STAT4 heterodimer 30 min after stimulation (Fig. 7c). In contrast, treatment of activated Tconv cells with IL-12 + IFN-γ failed to induce appreciable formation of STAT1:STAT4 heterodimers despite potent activation of both STAT proteins (Fig. 7a,c). These data suggest that IL-35 can mediate distinct signaling consequences compared with IFN-γ and IL-12 by inducing the formation of a STAT1:STAT4 heterodimer.
Given that Ebi3-Il12a transcription is an exclusive direct downstream consequence of IL-35R signaling, we sought to determine if STAT1:STAT4 heterodimers possessed a unique binding pattern within these promoters. In silico analysis of the Il12a and Ebi3 promoters using a proposed consensus STAT motif (TTC-N1–5-AGA, allowing for mismatch) identified nine potential binding sites (Fig. 7d)32, 33. We designed primers to interrogate six of these sites by chromatin immunoprecipitation (ChIP) analysis using antibodies to STAT1 and STAT4. As controls we also included known STAT1 and STAT4 binding sites within the Irf1 and Il18ra promoters, respectively34, 35. T cells, four days post-activation, were stimulated with IL-12, IFN-γ or IL-35 and assayed for STAT1 and STAT4 binding by ChIP. As expected, IL-12 induced STAT4 binding to Il18ra-750 and IFN-γ induced STAT1 binding to Irf1-250 (Fig. 7e,f). However, IL-35 did not induce STAT binding to either site. Although IL-12 or IFN-γ induced minimal STAT1 and STAT4 binding to the Il12a and Ebi3 promoters, IL-35 induced STAT binding to multiple sites. The pattern of STAT1 and STAT4 binding following IL-35 treatment was very similar implicating the use of a heterodimer. To directly address this possibility, we conducted ChIP-reChIP analysis in which the anti-STAT4 ChIP product was subjected to a second ChIP with anti-STAT1, and vice versa. T cells stimulated with IL-12 + IFN-γ induced minimal STAT1:STAT4 interaction with the Il12a and Ebi3 promoter STAT consensus sites. However, T cells stimulated with IL-35 showed strong enrichment of STAT1 and STAT4 binding to Il12a-250 and Ebi3-500, as well as other sites in the promoters. Although the amplicons used contain only one putative STAT binding consensus sequence, we cannot rule out the possibly that STAT 1 and STAT4 might bind at adjacent sites that are either within this amplicon or in the fragments generated during DNA shearing. These data implicate a STAT1:STAT4 heterodimer as the unique biochemical effector of IL-35 signaling, which appears to specifically bind to multiple sites within Il12a and Ebi3 promoters.
The composition of the IL-35R presents a conundrum regarding the mode of IL-35:IL-35R interaction and assembly when considered in light of the site 1, 2 and 3 architectural paradigm originally established for IL-636. In this blueprint, site 3 of the cytokine engages the N-terminal Ig-domain (IgD) of the signaling receptor, thus the presence of an N-terminal IgD is required for it to act as a site 3 binder (Fig. 8). Absence of the N-terminal IgD as seen, for example, in WSX-1 and IL-12Rβ1 would appear to relegate them to site 2 binding. While there are no known crystal structures of IL-12, IL-27 or IL-23 receptor signaling complexes, current models of the IL-12:IL-12R interaction suggest that the IL-12Rβ2 IgD domain binds to site 3 on p35, while also interacting with the p40 subunit via its N-terminal domain20, 21, 24, 37 (Fig. 8b). However, in the IL-27:IL-27R complex gp130 is also predicted to bind to site 3 in the p35 homologue p2724. This presents a conundrum as IL-12Rβ2 and gp130 both contain N-terminal IgD that could serve as the site 3 interacting domain, yet clearly both cannot be binding to the same site and suggests differences in the formation of the IL-35:IL-35R complex compared with the IL-12 and IL-27 receptor complexes. There are three possibilities (Fig. 8c). First, gp130 binds to site 2 in the IL-6, CNTF and LIF complexes24 and thus could do so in the IL-35:IL-35R complex, leaving IL-12Rβ2 to bind to site 3. Second, the IL-35:IL-35R complex could form symmetric homo-hexameric assemblies (2:2:2) analogous to the IL-6R complex23, thus allowing IL-12Rβ2 and gp130 to each bind to site 3 on p35. Third, gp130 and IL-12Rβ2 could both be capable of site 2 and site 3 binding, and therefore exist in an interchanging equlibrium of hetero-tetrameric complexes (1:1:1:1, gp130: IL-12Rβ2:EBI3:P35) (Fig. 8c - two ‘hetero’ models). Indeed, studies on the IL-6 complex suggest that it may be able to signal as tetrameric or hexameric assembly23. Finally, a completely novel mode of binding may exist that utilizes the site 1/2/3 model in a manner not predicted from the existing structural information on IL-6, LIF, and CNTF24, 36, 38. While a definitive answer will require detailed IL-35:IL-35R structural analysis, it is interesting to note that the IL-35 neutralizing antibodies we have described, which are specific for Ebi3, can immunoprecipitate IL-27 and IL-35 but only neutralize IL-35, while those specific for p35 can immunoprecipitate IL-12 and IL-35 but only neutralize IL-35. This further supports the notion that IL-35:IL-35R complex formation is distinct and differs from the IL-12 and IL-27 receptor complexes.
The family of gp130-related shared cytokine receptors has been the most actively growing sub-group of the Type I cytokine family, with the most recent addition being IL-35. Important similarities and differences between IL-12, IL-27 and IL-35 signaling have been revealed in this study. Given that IL-35 is part of the IL-12 cytokine family, it was reasonable to presume that the IL-35R might share usage of receptor chains and utilize common signaling molecules. However, since the biological activity of IL-35 contrasts with that of its cytokine family members, commonality in signaling could not be assumed. We found that the IL-35 receptor and signaling pathway does overlap with IL-12, by using IL-12Rβ2 which is part of the IL-12R, and IL-27, by using gp130 which is part of the IL-27R. However, in contrast to IL-12 and IL-27, IL-35 can signal through a single receptor chain with gp130 alone inducing pSTAT1 and IL-12Rβ2 inducing pSTAT4. Indeed, in the absence of either one of the IL-35 receptor chains in vivo, the suppressive effects of IL-35 remain intact. To completely abrogate IL-35 activity in vivo, loss of both IL-12Rβ2 and gp130 was required. Given our current understanding of receptor signaling and data presented here, IL-35 appears to mediate the homodimerization of either IL-12Rβ2 or gp130, as well as IL-12Rβ2:gp130 heterodimerization. However, we cannot entirely rule out the possibility that the various forms of the IL-35R may form higher order multimers of these receptor chains. gp130 can signal as either a homodimer (e.g. with IL-6:sIL6Rα or viral IL-6) or a heterodimer with LIF-R (e.g. CNTF and LIF). This is also a formal possibility for IL-12Rβ2, which so far is only known in a heterodimeric context in the IL-12 receptor complex. Whether signaling induced by IL-35 can be productively mediated via a single IL-35R chain in all physiological conditions and disease scenarios remains to be determined. However the complete IL-12Rβ2:gp130 receptor is required for maximal suppression, IL-35 induction and iTr35 conversion.
IL-27 activates a heterogeneous mixture of STAT proteins depending on the cell type in question, although in T cells STAT1 and STAT3 dominate22, 27, 28, 39. Like IL-27, IL-35 signals through two different STAT proteins, STAT1 and STAT4. However, STAT3 appears dispensable for IL-35 signaling, an interesting observation given its importance for IL-27 signaling, which is downstream of gp130 engagement. IL-12 family cytokine signaling involves phosphorylation of many different STAT proteins, which can homodimerize, thereby acting alone to influence gene transcription, or can heterodimerize to work in concert to elicit an alternate response. The intensity as well as the kinetics of STAT activation has previously been shown to influence downstream signaling events, gene expression and the nature of the biological response. For instance, IFNα signaling in myeloid cells can either promote or inhibit immunity through phosphorylation of STAT1, STAT2, and STAT331. A primarily STAT1:STAT2 response initiates the anti-viral, pro-inflammatory response. When STAT3 expression dominates, STAT1-dependent gene expression is inhibited, directly suppressing immunity. In addition, dominant STAT3 can indirectly alter the IFNα signal by dimerizing with STAT1, thereby inhibiting homodimerization of STAT1 and the biological activities that are induced in response to STAT1 homodimers31. By this mechanism, the IFNα signal is shifted from a pro-inflammatory to an anti-inflammatory response as a result of changing the relative expression and activation of STAT proteins as well as their dimerization capacity. In addition, STAT1:STAT3 heterodimers have been previously described, therefore dominance of one STAT protein over another may influence not only the activity of the STAT protein in question, but also its ability to homo- and heterodimerize to mediate distinct biological outcomes. We have shown here that IL-35 induces the formation of a STAT1:STAT4 heterodimer, promoting a distinct mode of Il12a and Ebi3 promoter binding and gene activation. It is possible that STAT heterodimers may be generally linked to anti-inflammatory programs.
Since IL-35 appears to utilize receptor chains and STATs that are similar to those used by other IL-12 family members, another important question is how a T cell can translate potentially similar signals into such distinct biological outcomes16. This enigma is certainly seen with IL-10 and IL-6. Although both cytokines signal primarily through STAT3, the outcome of their signaling and their biological activity is vastly different and has been a focus of intense research for more than a decade30. Thus, the utilization of STAT1:STAT4 heterodimers gives biochemical distinction to IL-35 signaling. However, it is still unclear why a STAT1:STAT4 heterodimer predominates upon IL-35 stimulation and not after simultaneous stimulation with STAT1 and STAT4 activators, such as IFN-γ and IL-12 respectively. It is clear that gp130 and IL12Rβ2 homodimers cannot induce both pSTAT1 and pSTAT4, suggesting that a heterodimer only forms after ‘canonical’ gp130:IL12Rβ2 receptor activation. It is possible that bringing these two receptor chains into close proximity favors STAT heterodimerization that may not occur following ligation of two spatially distinct receptors. Receptors that activate STAT1 or STAT4 may be spatially segregated, preventing the formation of an anti-inflammatory STAT1:STAT4 heterodimer. Indeed, the IFN-γR tracks readily to the immunologic synapse following TCR activation, while the IL-12R has a more diffuse localization40, 41. One might speculate that this is required to negate the close proximity of these receptors during T cell activation and thus prevent inadvertent STAT1:STAT4 heterodimer formation.
The expression pattern of the IL-35 receptor will provide insight into potential IL-35 target cell types. While gp130 is fairly ubiquitously expressed, IL-12Rβ2 is expressed mainly by activated T cells and natural killer cells21. In T cells, the expression of IL-12Rβ2 is most highly expressed by TH1 cells, and its expression correlates with responsiveness to IL-1242 and IL-35. Expression of IL-12Rβ2 has also been shown on other cell types, such as dendritic cells43, which may affect the spectrum of IL-35 bioactivity in the immune system. IL-12Rβ2 is undetectable on most resting T cells, but can be rapidly upregulated by exposure to IL-2, IL-12, IL-27, IFN-y, tumor-necrosis factor (TNF) and co-stimulation through CD28. Indeed, we have shown here that IL-2 or IL-27 pretreatment increases T cell sensitivity to IL-35 mediated suppression. Thus, IL-35 might have biological effects on a variety of cellular targets and under a variety of disease conditions. IL-27 is of particular interest given its capacity to exhibit inhibitory activity. Although it is thought that this is mediated, in part, via its induction of IL-1044, 45, it is also possible that IL-27-mediated induction of IL-12Rβ2 may sensitize cells to IL-35-mediated suppression and thus may contribute to the perceived inhibitory activity of IL-27. The capacity of IL-35 to mediate suppression via IL-12Rβ2:gp130 heterodimers and IL-12Rβ2 and gp130 homodimers could have profound implications for its range and breadth of regulatory control. This is especially relevant for gp130, which is broadly expressed21. Indeed, utilization of gp130 homodimers may provide a mechanism for IL-35 to suppress naïve T cells that do not express IL-12Rβ2.
Given the potent immunosuppressive capacity of IL-35, blocking its function may be therapeutically beneficial in a number of disease settings. We have previously shown that inhibiting IL-35 activity with a neutralizing anti-IL-35 mAb has significant in vivo efficacy8. The unique IL-35 receptor composition of IL-12Rβ2 and gp130 presents some interesting considerations and therapeutic opportunities. Animals lacking gp130 are not viable, likely due to the large number of cytokines that utilize gp130. For this reason it is difficult to infer a lack of IL-35 signaling as causative of this phenotype. In contrast, mice lacking IL-12Rβ2 have a mild phenotype, with reports suggesting that they are more susceptible to autoimmunity and B cell malignancies46. Interestingly, several autoimmune and inflammatory diseases have been linked to IL-12Rβ2 including atopic dermatitis, primary biliary cirrhosis, type 1 diabetes, and Behçet’s disease47–49. Taken together, these observations suggest that the clinical consequences of gp130 and IL-12Rβ2 dysfunction may affect immune regulation via the IL-35:IL-35R signaling axis, suggesting that further studies are required to assess the importance of the IL-35 signaling pathway in human disease.
Spleens and lymph nodes from Il12rb1−/− mice were provided D. Fairweather and J.A. Frisancho (Johns Hopkins University), CD4cre × Il6stfl/fl (Il6stΔT) mice by M. Karin and S. Grivennikov (University of California at San Diego), Il27ra−/− mice by C. Hunter and J. Stumhofer (University of Pennsylvania), Stat1−/− mice by A. Satoskar and P. Reville (Ohio State University), and Stat3ΔT (CD4 Cre × STAT3 fl/fl) mice by C. Drake and H.R. Yen (Johns Hopkins University). Il12rb2−/−, Stat4−/−, Rag1−/−, C57BL/6, B6.PL and Balb/c mice were purchased from Jackson Laboratory. Some experiments with CD4cre × Il6stfl/fl mice were performed with a colony established at St. Jude from mice obtained from Rodger McEver (University of Oklahoma Health Sciences Center). These were crossed with Il12rb2−/− mice to generate IL-35ΔT mice (CD4cre × Il6stfl/fl × Il12rb2−/−). All receptor and STAT deficient mice are on a C57BL/6 background, with the exception of Stat4−/− mice which are on a Balb/c background. In all experiments, wild type mice used were of the appropriate genetic background. Animal experiments were performed in American Association for the Accreditation of Laboratory Animal Care-accredited, specific-pathogen-free facilities in the St. Jude Animal Resource Center. Animal protocols were approved by the St Jude Animal Care and Use Committee.
Neutralizing IL-35 mAb was developed with recombinant murine Ebi3 protein, as described.8
IL-35 constructs were generated by recombinant PCR as described10. Constructs containing Ebi3 and Il12a linked by a 2A “self-cleaving” peptide or a flexible linker were used for IL-35 generation and an empty vector containing an IRES-GFP construct was used as a control. 1.0 × 106 293T cells were seeded into culture plates, then transfected using Trans-IT reagent (Mirus Biotechnologies). After 16 h media was changed and supernatant collected for 48 h. Supernatants were filtered, concentrated, and reconstituted into fresh RPMI. Each lot of supernatant is screened for its ability to inhibit T cell proliferation and validated extensively as previously described (see Ref. 8; Supplementary Fig. 2 and 4).
Naïve Tconv (CD4+CD25−CD45RBhi) and nTreg (CD4+CD25+CD45RBlo) cells were purified as previously described8, 10. Splenic and lymph node suspensions depleted of non-T cells by magnetic sorting were purified flow cytometrically. iTr35 cells were generated as previously described8.
Following 18h activation with anti-CD3 + anti-CD28 coated beads, cells were treated with 100ng/ml IL-12, IL-27 or IL-35. Cells were resuspended in lysis buffer and subjected to immunoblotting for pSTAT1 and pSTAT4 (Cell Signaling Technology and Santa Cruz Biotechnology). For immunoprecipitations, cells were resuspended in lysis buffer, precleared with Protein G beads, incubated with antibodies overnight, and immunoprecipitated using Protein G beads. Blots were developed using ECL (Amersham Biosciences).
IL-12 family receptor chains cloned into modified MSCV retroviral vectors yielded receptor chains tagged with either GFP (donor) or mCherry (acceptor) (IL-12Rβ1-GFP, IL-12Rβ2-mCherry, and GP130-GFP). Specific details regarding primers and constructs can be obtained from Kate Vignali (Kate.Vignali/at/st.jude.org). 293T cells were transfected 24 h prior to plating on Poly-L-lysine-coated 8-well tissue culture slides. Cells were analyzed using confocal microscopy and TIRF illumination (Intelligent Imaging Innovations). Images were acquired at 2 sec intervals for a total time of 30 sec to calculate basal FRET values, followed by addition of rIL-12 [25 ng/ml] or mIL-35 [25 ng/ml] and continued image acquisition at 2 sec intervals. 293T cells were transfected with single GFP or mCherry constructs for determination of donor bleedthrough fluorescence (average Fd/Dd = 0.0004) as well as direct excitation of the acceptor (average Fa/Aa = 0.003). Three color mathematical correction was applied to generate a FRET channel, where Fc = Transfer − Fa/Aa*Acceptor-Fd/Dd*Donor. Normalized FRET (nFRET) represents the ratio of corrected FRET signal detected prior to and following cytokine, with basal FRET signals arbitrarily set to 1.
Anti-CD3- + anti-CD28-coated latex beads were generated as previously described8, 10. In some experiments, superparamagnetic anti-CD3 + anti-CD28 beads were used (Dynabeads, Life Technologies). To determine suppression, 1.25 × 104 or 2.5×104 cells were activated with anti-CD3- + anti-CD28-coated beads with IL-35 supernatants, IL-35 beads, or iTr35 for 72 h; proliferation was determined by [3H]-thymidine incorporation.
“% suppression” is calculated as: 100 – 100*[(CPM of “suppressed” Teff)/(average CPM of Teff alone)] to normalize data across experiments (CPM can vary significantly from experiment to experiment). Ranges of CPM values obtained with Tconv cells activated alone for each set of experiments are detailed in the figure legends (typically ranged from 20,500–91,000).
Homeostasis assays were performed as described10. Briefly, WT or knockout naive Thy1.2+ Tconv cells were purified and used as “responder” cells in adoptive transfer. Thy 1.1+ iTr35 were generated and used as “suppressor” cells. Splenic T cell numbers were determined by flow cytometry seven days after transfer.
Rag1−/− mice received indicated cells i.v. on day −1 of experiment. Wild type or receptor-deficient naïve CD4+CD25− (9×106/mouse) and CD8+ T cells (6×106/mouse) alone or in combination with iTr35 cells (106/mouse) were adoptively transferred. B16-F10 melanoma was a gift from Mary Jo Turk (Dartmouth College, Hanover, NH) and cultured as described8.
Naïve Tconv cells were stimulated with plate-bound anti-CD3 (3 μg/mL) and anti-CD28 (2 μg/mL) for 48 h in the presence of 100 U/mL of rhIL-2, then expanded 48h in IL-2 without TCR or CD28 stimulation. 25 × 106 cells were washed, rested 2 h on ice in serum-free media, then stimulated with cytokines (rIL-12: 25ng/mL, IFN-γ: 100ng/mL, rIL-35: 25% culture volume of 293T supernatant) for 90 min. Cells were crosslinked with 1% formaldehyde for 30 minutes, followed by a 15 minute quench with glycine. Chromatin was prepared using SDS lysis, and sheared to 250–750bp fragments with a Diagenode Bioruptor (Belgium) using 15 × 30s pulses. Chromatin was precleared using ssDNA-(Sigma) and BSA-blocked Protein G beads (GE Healthcare). STAT1 and STAT4 or isotype-matched IgG control antibodies (Cell Signaling, Danvers, MA) were added and rotated overnight. Protein-DNA complexes were immunoprecipitated using blocked beads, washed in low salt, high salt, LiCl, and TE buffers, and eluted with NaHCO3. For ChIP-reChIP, complexes were first eluted with 0.1M DTT and subjected to a second round of immunoprecipitation, washes, and elution with NaHCO3. Eluates were reverse-crosslinked, proteinase K and RNase A treated, and DNA purified using Qiaquick DNA Purification columns (QIAGEN). DNA was assayed by real-time PCR using SYBR Green (Applied Biosystems).
We are grateful to J.A. Frisancho, S. Grivennikov, M. Karin, J. Stumhofer, P. Reville and H.R. Yen for assistance with harvesting and shipping spleens and lymph nodes, J. Partridge and P. Brindle for assistance designing ChIP experiments, Karen Forbes, Ashley Castellaw and Amy Krause for maintenance, breeding and genotyping of mouse colonies, Richard Cross, Greig Lennon and Stephanie Morgan for FACS, the staff of the Shared Animal Resource Center at St Jude for the animal husbandry, and the Hartwell Center for Biotechnology and Bioinformatics at St Jude for real-time PCR primer/probe synthesis. This work was supported by the National Institutes of Health (R01 AI091977, to D.A.A.V.), a sponsored research agreement from NovoNordisk Inc. (to D.A.A.V.)[z1], an Individual NRSA (F32 AI072816, L.W.C.), NCI Comprehensive Cancer Center Support CORE grant (CA21765, D.A.A.V.), and the American Lebanese Syrian Associated Charities (ALSAC, D.A.A.V.).
AUTHOR CONTRIBUTIONSL.W.C. designed (with help from D.A.A.V) and executed a substantial proportion of the experiments, analyzed data and wrote the manuscript; G.M.D. performed STAT co-IP and ChIP/ChIP-reChIP experiments, cytokine-receptor co-IP experiments, some of the pSTAT analysis and functional assays and wrote the manuscript; C.Guy performed confocal microscopy/FRET experiments; K.M.V. generated all the constructs used in the project; V.C. performed TH1/TH2 polarization for receptor analysis; D.F., A.R.S., C.A.H. and C.G.D. provided mice; K.C.G performed structural modeling of cytokine receptor complexes; P.M. performed phospho-STAT Western blot; C.A.H., P.M., K.C.G, C.G.D. and K.M.V. commented on the manuscript, and D.A.A.V. conceptualized the research, directed the study, and wrote the manuscript.
COMPETING FINANCIAL INTERESTS
The authors declare competing financial interests. D.A.A.V., L.W.C. and K.M.V. have submitted patents that are pending and are entitled to a share in net income generated from licensing of these patent rights for commercial development. Furthermore, this work was supported in part by a sponsored research agreement with NovoNordisk to D.A.A.V.