|Home | About | Journals | Submit | Contact Us | Français|
Toll-like receptors (TLRs) shape innate and adaptive immunity to microorganisms. The enzyme IRAK1 transduces signals from TLRs, but its activation and regulation mechanisms remain unknown. We show that TLR7 and TLR9 activated the isomerase Pin1, which then bound to IRAK1, resulting in IRAK1 activation and facilitating its release from the receptor complex to activate the transcription factor IRF7 and induce type I interferons. Consequently, Pin1-null cells and mice failed to mount TLR-mediated, interferon-dependent innate and adaptive immune responses. Given the critical role of aberrant IRAK1 activation and type I interferons in various immune diseases, controlling IRAK1 activation via Pin1 inhibition may represent a useful therapeutic approach.
Toll like-receptors (TLRs) recognize distinct pathogen-associated molecular patterns and elicit signaling cascades, leading to upregulation of type I interferons (IFNs) and proinflammatory cytokines to orchestrate innate and adaptive immunity against infection1-3. In particular, the intracellular TLRs are primarily involved in the recognition of viral (TLR3, TLR7) as well as viral and bacterial (TLR9) antigens and are especially adept at triggering type I IFN immune responses4-7. While TLR3 ligation induces IRF3 activation and subsequent interferon-β (IFN-β) secretion independent of the adaptor MyD88 but dependent on the adaptor TRIF, the TLR9 subfamily members TLR7 and TLR9 exclusively utilize MyD88, leading to robust type I IFN production mainly from plasmacytoid dendritic cells (pDCs)2, 8. Type I IFNs are essential for antiviral immune responses as they induce various IFN-inducible genes, promote natural killer cell cytotoxicity, dendritic cell (DC) maturation, and differentiation of virus-specific cytotoxic T lymphocytes, thereby linking innate and adaptive immunity1-3, 9. However, TLR activation is a ‘double-edged sword’, as inappropriate activation can participate in pathologic inflammation and diseases10, 11.
An essential molecule transducing signals from the upstream receptor complex for most TLRs to downstream transcription factors is the enzyme IRAK1 (IL-1 receptor associated kinase-1)12-14. Upon TLR ligation, IRAK1 is recruited to the TLR complex containing MyD88, IRAK4 and the adaptor TRAF6, where IRAK4 phosphorylates IRAK1 promoting IRAK1 autophosphorylation, leading to a characteristic ~20 kDa mobility shift in SDS-containing gels, a hallmark of IRAK activation. Activated IRAK1 dissociates from the receptor complex to activate IRF7 for induction of type I IFNs12-19. IRAK1-null mice show a blunted inflammatory response to TLR2 and TLR4 ligation20, but perhaps their most significant phenotype is the complete loss of IFN-α secretion following TLR7 or TLR9 ligation14. Although IRAK1 autophosphorylation is implicated as a key event in TLR signaling12, 13, 17, little is known about how such phosphorylation leads to IRAK1 activation and whether there is further regulation following phosphorylation.
Protein phosphorylation on serine or threonine residues preceding proline (Ser/Thr-Pro) is a central signaling mechanism in many cellular processes under both physiological and pathological conditions21. pSer/Thr-Pro motifs exist in two distinct cis and trans conformations, whose inter-conversion rate is markedly slowed down upon phosphorylation, but specifically accelerated by the unique prolyl isomerase Pin122. Pin1 has two structurally and functionally distinct domains, a pSer/Thr-Pro-binding WW domain and pSer/Thr-Pro-isomerizing catalytic domain. Pin1 controls the function of many key regulators in various cellular processes and its deregulation contributes to the pathogenesis of a growing number of diseases21.
Pin1 acts on IRF3 to affect IFN-β production upon TLR3 or activation of the helicase RIG-I 23. However, unlike IRF3- or TLR3-deficient mice, IRF7 or IRAK1-deficient mice completely fail to mount a type I IFN antiviral responses due to particular loss of type I IFN secretion from pDCs19, 24. These findings compelled us to examine the role of Pin1 in other TLR signal pathways. We show that Pin1 is a regulator of IRAK1 activation in TLR signaling and type I IFN-mediated innate and adaptive immunity. Pin1 inhibitors, which are under active development21, 25, may be useful in allowing selective inhibition of the type I IFN response while leaving proinflammatory cytokine production unaffected.
To examine the role of Pin1 in TLR signaling, we first compared cytokine production in response to various TLR ligands using DC subsets derived from Pin1 wild-type (WT) and Pin1 knockout (KO) mice. When stimulated with LPS (TLR4 ligand), Pam3 CSK4 (TLR2), R-848 (TLR7) or CpG DNA (TLR9), Pin1 KO myeloid DCs (mDCs) produced moderately less proinflammatory cytokines than Pin1 WT controls (Fig. 1a-c). Consistently, reduced proinflammatory cytokine secretion was also detected in Pin1 KO macrophages following stimulation of Pin1 WT and KO cells with each of the TLR ligands (Supplementary Fig. 1a, b). Stimulation of splenic plasmacytoid DCs (pDC) or Flt3 ligand–induced bone marrow pDCs with purified TLR7 or TLR9 ligand or with the influenza A virus (H1N1, a TLR7 stimulator) or mouse cytomegalovirus (MCMV, a TLR9 stimulator) showed robust IFN-α secretion in Pin1 WT cells (Fig. 1d-g), as shown6, 26. However, Pin1 KO cells almost completely failed to produce type I IFNs as measured by ELISA and quantitative RT-PCR analyses (Fig. 1d-h) . These effects of Pin1 deficiency on IFN-α production were highly specific because Pin1 KO neither affected the population of immune cells nor their TLR expression (Supplementary Fig. 2). Moreover, Pin1 enzymatic activity, although not protein amounts, was significantly elevated in R-848- or CpG-stimulated human peripheral blood mononuclear cells (Fig. 1i), which is consistent with the findings that Pin1 is kept inactivated until cellular cues are engaged21, 27. Thus, Pin1 plays a moderate role in proinflammatory cytokine production in mDCs in response to various TLR ligands, but is essential and specific for the type I IFN response in pDCs following TLR7 and TLR9 ligation.
To elucidate the molecular mechanism underlying the impact of Pin1 on type I IFN secretion, we used a proteomic approach to identify Pin1 substrates using a GST-Pin1 affinity purification procedure under high-salt and -detergent conditions22, 28, a procedure that has been used to identify almost all known Pin1 substrates21. We used R-848-stimulated human THP1 cells, a monocytic cell line that has a functional IFN response to TLR7 and TLR9 ligands and can be cultured in sufficient volumes. Following SDS-PAGE and mass spectrometry, one prominent and reproducible Pin1-binding protein at 100 kDa was identified to be IRAK1 (Fig. 2a, Supplementary Fig. 3a). The Pin1 KO phenotypes were strikingly similar to those observed in Irak1 KO cells and mice14, 20, 29. Furthermore, similar to Irak1 KO14, no obvious effect of Pin1 deficiency on IL-6 and IL-12p40 protein could be detected following pDCs stimulated with R-848 or CpG (Supplementary Fig. 1c, d). No difference was observed for IFN-β secretion from Pin1 WT and KO mDCs stimulated with CpG (Supplementary Fig. 1e), which is consistent with the previous results that CpG stimulation of mDCs induces IFN-β production in a MyD88- and IKKα-dependent but IRAK1-independent manner29. These results prompted us to examine the role of Pin1 in regulating IRAK1 function in TLR signaling. We confirmed the TLR7 and TLR9-dependent interaction between IRAK1 and Pin1 in THP1 monocytes and RAW264.7 macrophages by GST-Pin1 binding assay (Fig. 2b) or co-immunoprecipitation (Co-IP) (Fig. 2c). Pin1 predominantly bound to the activated form of IRAK1, which displayed a characteristic mobility shift on SDS gels after TLR ligation (Fig. 2b, c, solid arrows), suggesting that Pin1 might bind specifically to phosphorylated IRAK1. Indeed, this binding was mediated by the Pin1 WW domain (Supplementary Fig. 3b, c), a known pSer/Thr-Pro-binding module28, but was abolished either by IRAK1 dephosphorylation prior to Pin1 binding assay (Fig. 2d, Supplementary Fig. 3d) or mutating a key functional residue in the WW domain28(Supplementary Fig. 3b, e). In addition, Pin1 did not bind to the related kinases IRAK2 and IRAK4 following TLR7 and TLR9 stimulation (Supplementary Fig. 3f). Thus, following TLR7 or TLR9 ligation, Pin1 is activated and IRAK1 is phosphorylated, which allows Pin1 to interact specifically with IRAK1.
The binding of Pin1 to IRAK1 was somewhat surprising because Pin1 interacts only with specific pSer/Thr-Pro motifs21, 28 and there is little known about Pro-directed phosphorylation of IRAK1 in TLR signaling17. Consequently, we decided to define the Pin1 binding region and site(s) in IRAK1. Structurally, IRAK1 consists of an N-terminal death domain, a ProST-rich undetermined domain (UD) and a central kinase domain, with a C-terminal tail17(Supplementary Fig. 4a). To avoid interference of endogenous IRAK1, we expressed FLAG-IRAK1 or its mutants in IRAK1 null (I1A) 293 cells30, followed by a Pin1 binding assay. Overexpression of IRAK1, but not its K239S kinase-dead (KD) mutant, resulted in its auto-activation independently of TLR stimulation, as indicated by the characteristic mobility shift (Fig. 2e-gSupplementary Fig. 4b), as shown previously12, 13, 17. IRAK1, but not its KD mutant, interacted with Pin1 (Fig. 2e-h, Supplementary Fig. 4b). Furthermore, deletion of the UD abolished Pin1 binding (Supplementary Fig. 4b). These data suggest that Pin1 binds to kinase-active IRAK1, possibly through autophosphorylation sites in the UD.
To demonstrate that Pin1 binds to autophosphorylated IRAK1, we co-expressed FLAG tagged KD-IRAK1 with or without WT-IRAK1 in IRAK1-deficient cells, followed by analyzing Pin1 binding specifically to KD-IRAK1. As shown previously12, KD-IRAK1 did not show the characteristic mobility shift and failed to interact with Pin1 when it was expressed alone (Fig. 2e left). However, when co-expressed with WT-IRAK1, KD-IRAK1 showed the mobility shift and also bound to Pin1 (Fig. 2e right), suggesting that Pin1 binds to autophosphorylated IRAK1. To confirm that Pin1 directly binds to IRAK1, we next performed Far-Western blotting analysis using WT and KD IRAK1 and GST-Pin1 WW domain. Indeed, Pin1 bound only to the slower mobility-shifted and presumably activated form of WT IRAK1, but there was no binding between Pin1 and KD IRAK1 (Fig. 2f). Finally, to confirm the binding of Pin1 to the active form of IRAK1, we performed a Pin1 binding assay using mouse embryonic fibroblasts (MEFs) stably transfected with WT and KD IRAK1 in the presence or absence of TLR7 activation. Pin1 bound to the active form of WT IRAK1, but not KD IRAK1 confirming that Pin1 predominately binds to activated IRAK1 (Fig. 2g). Taken together, these results indicate that upon TLR ligation, IRAK1 is activated by receptor recruitment and autophosphorylates on the pSer-Pro motifs, which in turn recruits Pin1 to act on IRAK1.
To identify the IRAK1 phosphorylation site(s) responsible for Pin1 binding, we mutated each of the six possible Pin1 binding pSer/Thr-Pro motifs in the UD of IRAK1 to Ala, and assessed their binding to Pin1 from retrovirally transfected cells. Although the mutation of S110, S163 or S196 had little effect on Pin1 binding, the mutation of S131, S144 or S173 alone to Ala considerably reduced IRAK1 activation and Pin1 binding, which was further reduced when all three sites were mutated together (Fig. 2h), indicating that phosphorylation of these sites participate in regulating IRAK1 activation and Pin1 binding. To confirm the phosphorylation status of these three sites, we used a two-step purification procedure to isolate IRAK1 using sequential FLAG immunoprecipitation and GST-Pin1 pulldown, followed by liquid chromatography–tandem mass spectrometry (LC-MS/MS) analysis. Both S131 and S144 were indeed phosphorylated (Supplementary Fig 5a, b). However, both trypsin and chymotrypsin digestions repeatedly failed to produce any peptides covering the region surrounding S173 (Supplementary Fig 5c), possibly due to the numerous proline and hydrophobic residues in this region. This prompted us to generate phospho-specific antibodies against phosphorylated S173 of IRAK1. The resulting anti-pS173 antibodies specifically recognized activated WT IRAK1, but not its S173A point mutant, even when it was highly overexpressed and activated using transient transfection, confirming that S173 in the UD of IRAK1 is indeed phosphorylated (Supplementary Fig. 6a). Moreover, S173 phosphorylation of IRAK1 was significantly induced in R-848- or CpG-stimulated human peripheral blood mononuclear and THP1 monocytes, as determined by flow cytometric (Fig. 2i) and immunoblotting (Supplementary Fig. 6b) analyses using pS173-specific IRAK1 antibodies. Taken together, these results indicate that upon TLR activation, S131-, 144-, and 173-Pro motifs in the UD of IRAK1 are not only phosphorylated in cells, but also are largely responsible for Pin1 binding.
The Pin1 WW domain and isomerase (PPIase) domains have been shown to bind and isomerize specific pSer/Thr-Pro motifs in its substrates, respectively22, 28, 31. To measure the Pin1 interaction with each of the implicated pSer-Pro motifs in IRAK1, we employed two-dimensional (2D) NMR spectroscopy to monitor the changes in the 15N-WW domain induced by titration with phosphopeptide ligands. In a 2D 15N-1H HSQC spectrum of a protein (Fig. 3a), each backbone NH group is represented by a peak, whose position reflects the chemical environment of that NH bond. Ligand binding to the protein is detected by changes in peak positions (fast exchange) or by the appearance of new peaks (slow exchange) as ligand is added. WW binding at each distinct IRAK1 site was measured using phosphopeptides containing residues 126-136 (pSer131-P132), 140-150 (pSer144-P145), and 157-180 (pSer173-P174) of IRAK1. The WW domain bound to each phosphopeptide and exhibited fast exchange kinetics, as demonstrated by changes in peak position, in each of the three titration experiments (Fig. 3a). Quantitative analysis of the change in chemical shift as a function of peptide concentration (Fig. 3b) yielded dissociation constants (KD) of 220 ± 15 μM, 120 ± 12 μM, and 260 ± 75 μM for the 126-136 (pSer131-P132), 140-150 (pSer144-P145), and 157-180 (pSer173-P174) phosphopeptides, respectively. The IRAK1-Pin1 interaction in the cell occurs as part of a multi-protein membrane-associated complex12-14, suggesting the potential for significant binding enhancement due to avidity.
In order to determine whether Pin1 catalysis occurs at each of these sites, homonuclear 2D ROESY NMR was used as previously reported32. In the presence of a catalytic amount of Pin1, exchange crosspeaks between the cis and trans isomers of the pSer-Pro peptide bond were clearly observed for each peptide (Fig. 3c, top panels). Conversely, in the absence of Pin1, exchange crosspeaks were missing (Fig. 3c, bottom panels). These results demonstrate that Pin1 accelerates the cis-trans isomerization at each pSer-Pro motif, thereby confirming these sites as Pin1 substrates.
Given that Pin1 binds to and isomerizes multiple pSer-Pro motifs in IRAK1 upon TLR activation, a key question is whether Pin1 regulates IRAK1 function in TLR signaling. Therefore, we examined the effects of Pin1 KO on IRAK1 activation in response to activation of various TLRs using Pin1 WT and KO MEFs and pDCs. Although TLR7 and TLR9 ligation activated IRAK1 in a time-dependent fashion in both Pin1 WT cells (Fig. 4a), as indicated by the mobility shift and increased kinase activity (Fig 4b), as previously described12, 13, 17, there was no evidence for IRAK1 activation in either assay in Pin1 KO MEFs or pDCs (Fig. 4a, b) or in Pin1-silenced THP1 cells using RNAi (Fig. 4c). Moreover, Pin1 KO also completely abolished IRAK1 activation in response to ligation of other TLRs including TLR2 and TLR4 (Supplementary Fig. 7a, b). These effects were highly specific because Pin1 KO did not affect activation of the IRAK1 upstream kinase IRAK4 (Fig. 4b), or mitogen-activated protein kinases (MAPK) including ERKs, JNKs and p38 MAPKs upon TLR activation (Supplementary Fig. 8). Similar observations were made following LPS stimulation of macrophages (Supplementary Fig. 9a). We also assessed the effects of Pin1 deficiency on IκB degradation following pDC stimulation with R-848 and CpG or treatment of macrophages with LPS and did not see any obvious difference between Pin1 WT and KO cells (Supplementary Fig. 9b, c). To further confirm this effect of Pin1 on IRAK1 activation, we developed an assay to measure the kinase activity of IRAK1 in cells utilizing the fact that IRAK1 can phosphorylate the N-terminal 220 amino acid IRAK1 fragment containing the UD in trans, as shown by the characteristic mobility shift after co-expression with WT IRAK1 (Fig. 4d), as demonstrated previously12, 13, 17. As expected, exogenously-expressed IRAK1 in Pin1 WT MEFs efficiently phosphorylated the IRAK1 N-terminal fragment, inducing the characteristic mobility shift (Fig 4d). However, like KD Irak1, WT Irak1 in Pin1 KO MEFs completely failed to induce any mobility shift of the N-terminal IRAK1 (Fig. 4d). These results together indicate that Pin1 is required for IRAK1 activation.
To further demonstrate the importance of Pin1 for the time-dependent activation of IRAK1 following TLR ligation, we overexpressed WT Irak1 and KD Irak1 in Pin1 WT and KO MEFs using a retroviral expression system. Under overexpression conditions, WT-IRAK1 was partially activated, which was further activated upon TLR7 ligation in Pin1 WT cells, as shown by the characteristic mobility shift (Fig. 4e), consistent with the findings that IRAK1 activation is sensitive to the amount of IRAK protein13, 17. However, no IRAK1 activation was observed in Pin1 KO cells, even after stimulation (Fig. 4e), further confirming the role of Pin1 in IRAK1 activation. Importantly, KD IRAK1 did not show any evidence of activation following TLR ligation both in Pin1 WT and KO cells (Fig. 4e). These results indicated that IRAK1 failed to be activated in Pin1 KO cells. To confirm that defective IRAK1 activation in Pin1 KO cells is specifically due to loss of Pin1 and to examine the importance of Pin1 binding and isomerase activities for IRAK1 activation, we performed rescue experiments by re-expressing WT Pin1 or its point mutants, W34A mutant (in the WW domain) or K63A mutant (in the catalytic domain), which fail to bind to or isomerase Pin1 substrates, respectively22, 28. Re-expression of Pin1, but neither of its WW domain (W34A) nor catalytic domain (K63A) point mutant, completely restored IRAK1 activation in Pin1 KO cells expressing IRAK1 (Fig. 4f), reminiscent of IRAK1 activation found in Pin1 WT cells (Fig. 4e). Taken together, these results demonstrate an essential role for Pin1 in IRAK1 activation during TLR signaling.
Given that Pin1 was required for activation of IRAK1 we wondered whether Pin1 regulates IRAK1 mediated downstream signaling. Following TLR activation, IRAK1 is recruited to the receptor complex via MyD88 and IRAK4, where it is activated and released from the receptor complex. This allows transcription factors such as IRF7, the master regulator of IFN-α, to translocate into the nucleus and activate IFN-α transcription, making IRAK1 activation a key step in the TLR7 and TLR9 signaling cascade12-19. Therefore, we examined whether Pin1 KO affects the ability of IRAK1 to transduce TLR signals.
To address whether IRAK1 is still recruited to the TLR receptor complex in Pin1 KO cells, we transfected HA-MyD88 into both Pin1 WT and KO cells retrovirally expressing FLAG-IRAK1, followed by immunoprecipitation with anti-HA antibodies and then immunoblotting with anti-FLAG antibodies. As shown previously17, the activated form of IRAK1 in Pin1 WT cells was not readily found in the MyD88 immune complexes (Fig 5a). However, IRAK1 in Pin1 KO cells formed a stable interaction with HA-MyD88 (Fig 5a), presumably due to the fact that IRAK1 is not fully activated in these cells (Fig. 4a, ,5a5a). Thus it appears that IRAK1 in Pin1 KO cells is unable to dissociate from the receptor complex due to its lack of autophosphorylation13, 17, presumably retaining IRAK1 at the receptor complex.
Given that Pin1 is required for IRAK1 activation and dissociation from the receptor complex, we examined whether Pin1 affects IRF7 activation using Pin1 knockdown and knockout. Pin1 knockdown in THP1 monocytes using Pin1-RNAi not only abolished the IRF7 and TRAF6 interaction, as shown by Co-IP experiments (Fig. 5b), but also blocked IRF7 nuclear translocation in response to TLR7 and TLR9 activation, as determined by subcellular fractionation followed by immunoblotting analysis (Fig. 5c, d). To further confirm these results, we immunostained for IRF7 in primary Pin1 WT and Pin1 KO pDCs after TLR7 and TLR9 ligation. Upon TLR activation, IRF7 translocated to the nucleus in Pin1 WT but not Pin1 KO pDCs (Fig. 5e). These results suggest that Pin1 activates IRAK1 to cause IRF7 nuclear translocation in response to TLR7 or TLR9 stimulation.
This suggestion was further supported by our findings from IRAK1-mediated IRF7 functional assays. Specifically, Pin1 KO abolished IRF7 reporter activity following TLR7 or TLR9 stimulation (Fig. 6a, b), and these defects were fully rescued by Pin1, but not its binding-inactive- or isomerase-defective mutant, as measured by IRF7 reporter activity and IFN-α production (Fig. 6c, d). To further investigate the role of Pin1 and IRAK1 kinase activity in IRF7 activation, we co-expressed MyD88, a Gal4-IRF7 reporter construct and various amounts of KD IRAK1 in Pin1 WT and KO MEFs. IRF7 activation in WT cells decreased as the amount of transfected KD IRAK1 was increased. In contrast, IRF7 activation was consistently lower in Pin1 KO cells and unaffected by the amount of KD IRAK1 transfected (Fig. 6e). These results demonstrate that both Pin1 and IRAK1 kinase activity are necessary for activation of IRF7. These findings are consistent with the previous findings 1) that IRAK1, but not its KD mutant, phosphorylates IRF714, 2) that IRAK1 kinase activity is necessary for the transcriptional activity of IRF7, but not NF-κB14, 3) that KD IRAK1 inhibits MyD88-induced IRF7 activation in a dominant-negative manner14, and 4) that inhibition of IRAK kinase activity with a synthetic inhibitor prevents IRF7 phosphorylation, but not NF-κB phosphorylation in CpG stimulated pDCs33.
IRAK1 mutations that prevented Pin1 binding in retrovirally transfected MEFs including S131A, S144A and S173A alone or together also decreased IRF7 promoter activation and IFN-α secretion similar to kinase-inactivating IRAK1 mutation or Pin1 KO (Fig. 6f, g). To confirm the importance of Pin1 in IRAK1- and IFN-α– mediated antiviral activity, we performed plaque formation assays using GFP-expressing vesicular stomatitis virus (VSV). Specifically, L929 cells were infected with GFP-VSV and incubated with supernatants from Pin1 WT and KO MEFs expressing IRF7 and IRAK1 or its mutants, followed by GFP-positive plaque quantification. While supernatants from Pin1 WT MEFs expressing WT IRAK1 had potent antiviral activity, those from Pin1 WT MEF expressing Pin1 binding IRAK1 mutants or KD IRAK1 had little activity, similar to Pin1 KO MEFs (Fig. 6h, i), consistent with IRF7 activity and IFNα production in these cells (Fig. 6f, g). Thus, disrupting the IRAK1 activation by inhibiting Pin1 or by preventing IRAK1 from acting as a Pin1 substrate drastically abrogates IRF7 activation, subsequent IFN-α production and antiviral response in vitro.
Given the essential role for Pin1 on IRAK1-dependent antiviral cellular responses in vitro, we next examined the effects of Pin1 KO in vivo using Pin1 WT and KO mice. Following injection with R-848 or CpG, robust IFN-α production could be observed in Pin1 WT mice (Fig. 7a, b), as shown14. In contrast, serum IFN-α concentrations in Pin1 KO littermates were reduced (Fig. 7a, b). When injecting mice with LPS or R-848, the serum concentrations of IL-6 and IL-12p40 were lower in Pin1 KO mice, compared to WT controls, albeit not as dramatically as IFN-α (Supplementary Fig. 10a-c). As the MyD88-IRF7 pathway has been shown to be essential for IFN-α production during MCMV infection34-37, we next examined the effects of Pin1 KO on systemic MCMV infection. Whereas IFN-α concentration in Pin1 WT animals peaked after 36 hours following MCMV infection, IFN-α induction was almost entirely suppressed in Pin1 KO mice (Fig. 7c). Moreover, Pin1 KO mice were much more vulnerable to systemic MCMV infection than their WT littermates, resulting in increased weight loss (Fig. 7d) and morbidity (Fig. 7e). These phenotypes are similar to those observed in Irf7 or Myd88 KO mice34-37 and further highlight the contribution of Pin1 to the antiviral immune response in vivo.
Co-stimulation of TLR9 and CD40 induces CD8+ T cell expansion in a pDC, IRF7- and IFN-α-dependent manner, thereby playing a major role in regulation of adaptive immune responses19, 38. To study the effects of Pin1 deficiency on adaptive immunity, we next investigated the effects of Pin1 KO on the induction of antigen-specific CD8+ T-cell responses. As reported19, 38, treatment with ovalbumin and anti-CD40 alone did not induce specific CD8+ T cell expansion, whereas co-inoculation of CpG-A complexed to DOTAP, a CD40 agonistic antibody and ovalbumin induced a strong expansion of antigen-specific CD8+ T cells in Pin1 WT mice (Fig. 7f, left). In contrast, the ovalbumin-specific CD8+ T-cell response was greatly impaired in Pin1 KO mice (Fig. 7f, right).
These results together indicate that Pin1 is essential for IRAK1 activation in response to TLR7 and TLR9 stimulation. Upon activation, Pin1 activity is upregulated and IRAK1 is autophosphorylated in the UD, allowing Pin1 to bind to and isomerize phosphorylated IRAK1. Such Pin1-catalyzed conformational change facilitates the dissociation of IRAK1 from the receptor complex and recruitment of TRAF6, which combines with IRAK1 to activate IRF7 by promoting nuclear translocation for the induction of type I IFNs to mediate innate and adaptive immunity (Supplementary Fig. 11a). Pin1 deficiency specifically prevents IRAK1 activation and release from the receptor complex, preventing TRAF6 recruitment and IRF7 activation, leading to loss of type I IFN production and its mediated innate and adaptive immunity (Supplementary Fig. 11b).
TLRs play an integral role in host defense against a broad range of microorganisms1-3, with TLR7 and TLR9 being particularly important components of the antiviral machinery4-7. IRAK1 has been shown to be essential for TLR7 and TLR9 mediated IFN-α induction12-14. However, the mechanisms underlying IRAK1 activation and regulation are not well understood. Our results demonstrate for the first time that Pin1 is a regulator of IRAK1, playing an essential role in TLR signaling and type I IFN-mediated innate and adaptive immunity. Given the major role of aberrant IRAK1 activation and type I IFN overproduction in various immune diseases10, 11, these results suggest that Pin1 inhibitors, which are under active development21, 25, may offer a useful therapeutic approach.
We found that Pin1 activity was increased during TLR signaling and that Pin1 genetic deletion modestly inhibited TLR7- and TLR9-dependent, proinflammatory cytokine production in mDCs, but completely abrogated type I IFN production from pDCs. Consistent with these phenotypes, Pin1 specifically acted on IRAK1 autophosphorylation sites in a TLR-dependent manner. Mechanistic evaluation of the consequences of mutations of the Pin1 binding and isomerizing sites in IRAK1 or genetic deletion of Pin1 demonstrated that the role of these phosphorylation sites is to promote a conformational change, leading to IRAK1 activation. Such Pin1-catalyzed conformational change facilitates the dissociation of IRAK1 from the receptor complex to activate downstream transcription factors for the induction of type I interferon. Specifically, Pin1 KO did not affect proximal IRAK1 signaling as its recruitment to the TLR complex and IRAK4 activation were normal. Furthermore, Pin1 KO did not affect other TLR activated kinases such as the MAPKs. However, Pin1 KO prevented IRAK1 activation and release from the receptor complex, leading to loss of recruitment of TRAF6 to the complex, resulting in failure of nuclear translocation of IRF7. As a result, Pin1-deficient cells and mice failed to mount a robust systemic type I IFN response following R-848 and CpG treatment or MCMV inoculation and became highly susceptible to viral infection. These results demonstrate an essential role for Pin1 in IFN-α mediated innate immunity.
In addition to their central role in innate immunity, TLRs have profound effects on the ensuing adaptive immune response. Type I IFNs not only limit viral replication, but also exert various immunoregulatory effects such as B and T cell activation39 as well as mDC maturation40. Importantly, type I IFNs, following viral infection, also enable cross-priming of CD8+ T cells, thereby allowing the presentation of exogenous antigens in the context of major histocompatability complex (MHC) class I molecules38. Pin1 KO mice were severely defective in triggering an antigen specific CD8+ T-cell response. These results demonstrate the critical role for Pin1 in regulating IRAK1, thereby representing an important new mechanism pivotally positioned at the interface of innate and adaptive immunity.
The tight regulation of IRAK1 is likely required to avoid inflammatory disease. For example, IRAK1 genetic changes are associated with pathological conditions such as SLE, leading to speculation that its increased expression or constitutive activation makes IRAK1 a disease susceptibility factor for IFN-driven autoimmune disorders41. Similarly, elevated type I IFN production has been intimately linked to many autoimmune diseases10, 11, 42.
In particular, aberrant type I IFN signaling now has a firmly demonstrated role in increased susceptibility to various viral infections, autoimmune disease and cancer. Important unifying themes of such diseases are the lack of treatments and the rather unusual prevalence of these diseases in females. For example, women progress to AIDS much faster than men who have the same viral load and this has been at least in part attributed to substantial sex differences in IFN-α secretion arising from HIV-1 stimulation of TLR7 in dendritic cells, which drives increased CD8 T cell activity43. Furthermore, the TLR-IRAK1-IRF7-Type I IFN pathway as a key contributor in systemic lupus erythematosus (SLE) pathogenesis, a disease that is particularly prominent in females44, 45. Specifically, in SLE patients increased concentrations of antibodies to self antigens such as nucleic acids activate TLRs to drive type I IFNs, which then perpetrate autoimmune tissue destruction and pathogenesis46. Furthermore, recent genetic evidence has definitively associated components of the TLR-IRAK-IRF pathway with SLE47. A single nucleotide polymorphism in IRAK1 has been associated with increased susceptibility to SLE and Irak1 KO effectively suppresses SLE in animal models41. These examples highlight the important role of IRAK1 and type I IFN in a range of diseases.
One of the challenges arising from the recent wealth of knowledge on TLR signaling is how to develop a strategy to inhibit specific arms of TLR-mediated immune regulation while leaving other critical defensive nodes untouched, a question no doubt asked by many biologists but one rarely answered. The data here reveal that Pin1 inhibition completely abrogates activation of IRAK1 kinase, and fully suppresses type I IFN production, but with only a moderate effects on pro-inflammatory cytokine production. These results suggest that inhibiting either Pin1 and/or IRAK1 kinase activity might allow pharmacological discrimination, which would allow selective inhibition of the type I IFN response while leaving proinflammatory cytokine production unaffected. Such a pharmacological approach might have advantages over conventional immunosuppressing strategies.
We thank Lewis Cantley and Tony Hunter for constructive advice and critical reading of the manuscript, and Laurent Brossay, Raymond Welsh, John D Hamilton and Sean Whelan for the viruses. A.T.K. is a fellow of the Swiss Foundation for grants in Biology and Medicine. The work was supported by NIH grants AG029385 to L.K.N. and K.P.L., DK066917 to M.A.E, and GM058556, American Asthma Foundation Senior Investigator Award and Target Identification in Lupus Grant from Alliance for Lupus Research to K.P.L.
Pin1 knockout mice have been backcrossed to C57L/B6 for 15 generations to obtain a pure genetic background, with Pin1-/- and Pin1+ littermates being used in all experiments48.
To prepare bone-marrow derived pDC, bone marrow cells were cultured with 100 ng/ml Flt3L (Peprotech) for 6 days. For mDC generation, bone marrow cells were cultured in the presence of 10 ng/ml GM-CSF (PeproTech) for 6 days. Splenic B220+CD11c+ pDC were sorted using a LSR II flow cytometer (BD Biosciences). IL-6, IL-12 p40, TNF-α in mDC supernatants following stimulation with the indicated ligands (all from Invivogen) were measured by ELISA (eBioscience). IFN-α in pDC following stimulation with R-848, CpG-A, MCMV (MOI=1) or influenza A virus A/PR/8/34 (H1N1) (MOI=1) was measured by ELISA (PBL InterferonSource) after 24 h.
PDCs were stimulated for 6 h with R-848 or CpG-A and then subjected to quantitative real-time PCR analysis using RotorGene and SYBR Green (Qiagen). Data were normalized to the level of GAPDH expression in each sample. Primers for β-actin, IFN-α1 and IFN-β have previously been described19.
The Pin1 PPIase assay was performed as previously described22.
Frequencies of B+ cells (CD19+,CD3-), CD4+ T cells (CD19-CD3+CD4+CD8-), CD8+ T cells (CD19-CD3+CD4-CD8+), myeloid DC (MHC-II+CD11c+) and pDC (MHC-II+B220+) were determined by staining with the corresponding antibodies (eBioscience). Cells were analyzed using a LSR II flow cytometer (BD Biosciences) and FlowJo software.
For cytospins, pDCs were spun onto glass slides, air dried, fixed in methanol and thereafter stained with an anti-IRF7 antibody (Abcam), followed by an Alexa Fluor 488 conjugated secondary antibody. Slides were counterstained with DAPI.
GST pulldown, immunoprecipitation, and immunoblotting analyses were performed as described28. Antibodies against IRAK1, IRAK4, IκBα, IRF7, p-ERK, p-JNK, p-p38, and lamin antibodies were from Cell Signaling. In some experiments an IRAK1 antibody from Millipore was used. TRAF6 antibodies were from Santa Cruz, Tubulin antibody was from Sigma. MyD88 antibody was from Stressgen. Anti-Pin1 monoclonal antibody was made as described48.
Far-Western analysis was performed as described22 .
Coomassie-stained SDS-PAGE gel bands were excised, reduced with DTT, alkylation with iodoacetamide followed by in-gel digestion with trypsin overnight. Reversed-phase microcapillary liquid chromatography tandem mass spectrometry (LC-MS/MS) was performed..
Phosphorylated Ser173-specific IRAK1 antibodies were raised by immunizing rabbits with a KLH-coupled phosphorylated Ser173-containing IRAK1 peptide (Proteintech Group) and were affinity purified, as described48.
Recombinant Pin1 and its WW domain were expressed as N-terminal GST or His fusion proteins, followed by thrombin cut to remove the tag31. NMR experiments were performed at a temperature of 25°C on a Varian Inova 600 MHz spectrometer equipped with a (H,C,N) Z-axis gradient probe, as described32. Spectra were processed and analyzed using the software tools nmrPipe, nmrDraw and Sparky.
For retroviral transduction, viral supernatants were prepared using Phoenix HEK293 cells as per manufacturers instructions (Orbigen). Stable cell pools were checked for protein expression by immunoblotting analysis with various antibodies to confirm protein expression. Stable cell lines were maintained in culture using 1 μg/ml puromycin.
In vitro kinase assays were performed as previously described27.
Nuclear and cytosolic fractions were prepared with NE-PER (Pierce), as described previously49.
Pin1+ and Pin1-/- MEFs stably expressing IRAK1 and its mutants in 24-well plates were transfected with Gal4-IRF7 and 40 ng of the UAS(GAL)-luciferase reporter gene using Lipofectamine 2000 (Invitrogen), as described previously49. In some experiments, cells were simultaneously transfected with constructs for Myd88 (20 ng), Pin1 WT, Pin1 W34A and Pin1 K63A (each 100 ng/well) or TLR7 and TLR9 (each 10 ng/well). Renilla luciferase reporter gene (50 ng) was co-transfected as an internal control.
MEFs stably expressing IRAK1 and its mutants or empty vector were transiently transfected with mIRF7 (20 ng/well) (Invivogen) in 24-well plates. In some experiments, cells were co-transfected with Pin1 WT, Pin1 W34A or Pin1 K63A (100 ng/well). IFN-α concentrations in the supernatants were measured by ELISA after 24 h. To measure serum IFN-α levels by ELISA, mice were intravenously injected with R-848 (50 nmol), CpG-A (5 μg) complexed to DOTAP (Roche) or intraperitoneally with MCMV (105 PFU). Blood was drawn from the tail vein at various time points.
L929 cells cultured in 6 well plates were incubated over night with culture supernatants of MEFs stably expressing IRAK1 and its mutants or empty vector. L929 cells were then infected with VSV-GFP (a kind of Dr. Sean Whelan, Harvard Medical School) at a multiplicity of infection (MOI) of 0.02 for 4 h. Medium was then removed and the corresponding cell culture supernatants containing 0.5% agarose were added to each well. GFP positive plaques were counted after 24 h using a fluorescence microscope.
To assess the effect of MCMV infection on body weight, mice (n=6) were injected i.p. with 2.5× 104 PFU of the salivary gland derived MCMV clone RVG-102 (MCMV-GFP) (kindly provided by Dr. Laurent Brossay, Brown University) which is recombinant for GFP under the immediate-early gene 1 (Ie-1) promoter and has acute and latent infection characteristics similar to those of wild-type MCMV50. To measure MCMV-mediated morbidity, mice were i.p. injected with 105 PFU. Mice were sacrificed once they reached a moribund state. All experiments were performed according to a protocol approved by the Beth Israel Deaconess IACUC.
Expansion of ovalbumin-specific CD8+ T cells was measured using a MHC class I tetramer38. Briefly, mice were injected i.p with 30 μg CpG-A complexed to DOTAP, 50 μg anti-CD40 (clone FGK45; Alexis) and 0.5 mg ovalbumin (Sigma-Aldrich). After 6 d, ova-specific splenic CD 8+ T cells were detected with CD8a-APC and CD44-FITC antibodies and a PE-conjugated H-2Kb tetramer containing the SIINFEKL peptide. Cells were analyzed using a flow cytometer. Results are representative of 3 independent experiments.