Cellular FLIP long isoform (cFLIP_L)–IKKα interactions inhibit IRF7 activation, representing a new cellular strategy to inhibit IFNα expression

Abstract

Interferon α (IFNα) is important for antiviral and anticancer defenses. However, overproduction is associated with autoimmune disorders. Thus, the cell must precisely up- and down-regulate IFNα to achieve immune system homeostasis. The cellular FLICE-like inhibitory protein (cFLIP) is reported to inhibit IFNα production. However, the mechanism for this antagonism remained unknown. The goal here was to identify this mechanism. Here we examined the signal transduction events that occur during TLR9-induced IRF7 activation. The cFLIP long isoform (cFLIP_L) inhibited the expression of IRF7-controlled natural or synthetic genes in several cell lines, including those with abundant IRF7 protein levels (e.g. dendritic cells). cFLIP_L inhibited IRF7 phosphorylation; however, cFLIP_L-IRF7 interactions were not detectable, implying that cFLIP_L acted upstream of IRF7 dimerization. Interestingly, cFLIP_L co-immunoprecipitated with IKKα, and these interactions correlated with a loss of IKKα–IRF7 interactions. Thus, cFLIP appears to bind to IKKα to prevent IKKα from phosphorylating and activating IRF7. To the best of our knowledge, this is the first report of a cellular protein that uses this approach to inhibit IRF7 activation. Perhaps this cFLIP property could be engineered to minimize the deleterious effects of IFNα expression that occur during certain autoimmune disorders.

Introduction

Type I interferons (IFNs)² are comprised of IFNα and IFNβ, and their production is the first line of defense against virus infection (1). IFNα represents a group of cytokines (e.g. IFNα4 and IFNα6) that are predominately regulated by the interferon regulatory factor 7 (IRF7) transcription factor (2–4). In most cell types, IRF7 is expressed at low levels. However, IRF7 is expressed at high levels in hematopoietic cells like plasmacytoid dendritic cells (pDCs) (5, 6). IFNα production is increased in a variety of autoimmune diseases, including systemic lupus erythematosus, Sjögren's syndrome (7), type I diabetes (8), rheumatoid arthritis (9), and others (10, 11). This exemplifies that the precise up- and down-regulation of IFNα production is critical for proper immune system homeostasis.

IRF7 activation is required for robust IFNα expression (3). IRF7 activation occurs via the engagement of endosomal nucleic acid sensors (e.g. TLR7, TLR8, and TLR9). TLR9 homodimers are activated upon binding of viral (12) or bacterial unmethylated CpG motifs (e.g. CpG-A) (13) or DNAs involved in autoreactive immune complexes (14, 15). In all cases, the MyD88 protein is recruited to the cytoplasmic portion of these TLRs (16), acting as a critical signal adaptor molecule. Next is the assembly of a dynamic complex including at least IRAK1, IRAK4 (17), and TRAF6 (16). IKKα is subsequently recruited and activated, either by IRAK1 (18) or an unknown kinase (2, 19). Regardless, IKKα goes on to phosphorylate IRF7, whereas TRAF6 Lys-63–linked polyubiquitinates IRF7(16,17). Phospho-IRF7 then homodimerizes (20) and translocates to the nucleus, where it drives expression of IFNα genes as well as other interferon-stimulated genes (2).

Because IFNα has powerful pro-inflammatory properties, cells have mechanisms to down-regulate IFNα production in the absence of virus infection. For example, RTA-associated ubiquitin ligase (RAUL) is an E3 ligase that promotes IRF7 Lys-48–linked polyubiquitination and degradation (21). PP2A is a dephosphorylase that inactivates IRF7 (22). In contrast, 4E-BP1/2 inhibits IRF7 translation (23). The cellular aryl hydrocarbon receptor–interacting protein (AIP) inhibits IRF7 action downstream of IRF7 phosphorylation; it inhibits nuclear translocation of IRF7 homodimers (24).

The cellular FLICE-inhibitory protein (cFLIP) was originally identified as an inhibitor of extrinsic apoptosis (25). There are two major isoforms of cFLIP, the long isoform (cFLIP_L) and a shorter splice variant (cFLIP_S), and both are members of the FLIP family (26). Our group recently identified cFLIP_L as an IRF3 antagonist; cFLIP_L binds to IRF3 to prevent enhanceosome formation (27). IRF3 demonstrates considerable sequence homology to IRF7 (28), begging the question whether cFLIP_L may bind to and antagonize IRF7 to control IFNα production. In support of this hypothesis is one report showing that overexpression of cFLIP_S correlates with a decrease in IFNα protein expression (29). To answer this question, we examined the effect of cFLIP on different steps of the TLR9-induced IRF7 activation pathway, using CpG-A to specifically trigger IRF7 dimerization. Several lines of evidence shown here suggest that cFLIP is a bona fide inhibitor of IRF7 activation and that it disrupts IKKα–IRF7 interactions as its antagonistic function.

Results

cFLIP_L inhibits IRF7-induced luciferase activity independent of IRF3 and IRF5

We showed previously that cFLIP_L inhibits IRF3-driven transcription by interrupting IRF3–CBP–DNA interactions (27). Because of the sequence and structural similarities of IRF3, IRF5, and IRF7 (28, 30), it was queried whether cFLIP_L could antagonize IRF5 or IRF7.

Luciferase reporter assays have been developed to specifically detect IRF5 or IRF7 activation and were used as a first step toward answering this question (31, 32). HEK293T (293T) cells were used because of their high transfection efficiency and their common use for luciferase reporter assays. Here the il12p40 promoter was fused to a luciferase gene to assess IRF5 activation (33) (Fig. 1A). Alternatively, the infa6 promoter was fused to a luciferase gene to assess IRF7 activation (34) (Fig. 1, B–E). Fig. 1A shows the specificity of the il12p40-luc plasmid for IRF5 activation; only cells transiently expressing IRF5 and TRAF6 stimulated luciferase gene expression robustly. Note that TRAF6 must be co-expressed with IRF5 for IRF5 homodimerization and subsequent IRF5 activation (35). Overexpression of a constitutively active IRF3 (IRF3CA) or IRF7 did not stimulate luciferase gene expression significantly above levels of pCI-transfected cells, as expected. Under these conditions, cFLIP_L had no effect on luciferase activity, suggesting that cFLIP_L did not antagonize IRF5 activation. A control for this assay was cells expressing Vpx, an HIV protein that is known to inhibit IRF5 activation (32).

Figure 1.

cFLIP_L inhibits IRF7-induced gene expression independent of IRF3 and IRF5. A and B, 293T cells were co-transfected with 450 ng of pil12p40-luc (A) or 450 ng of pifna6-luc and 50 ng of pRL-TK, 1000 ng of pCI, pIRF3CA, or pIRF7 or 500 ng each of pIRF5 and pTRAF6 (B). Cells were also co-transfected with either 1000 ng of pCI, pcFLIP_L, or pVpx (A) or pAIP (B). Cells were incubated for 24 h post-transfection. C, 293T cells were co-transfected with 450 ng of pifna6-luc and 50 ng of pRL-TK, 1000 ng of pCI, or 500 ng each of pIRF7 and pMyD88. Cells were also co-transfected with either 1000 ng of pCI, pcFLIP_L, or pIRF7DN. Cells were incubated for 24 h post-transfection. D, 293T cells were co-transfected with 450 ng of pifna6-luc and 50 ng of pRL-TK, 1000 ng of pCI, or 250 ng each of pIRF7, pMyD88, pIKKα, and pTRAF6. Cells were also co-transfected with either 1000 ng of pCI, pcFLIP_L, or pIRF7DN. Cells were incubated for 24 h post-transfection. E, HeLa cells were co-transfected with 450 ng of pifna6-luc, 50 ng of pRL-TK, 250 ng of pIFNα, and 1000 ng of pCI, pcFLIP_L, or pAIP. 24 h post-transfection, cells were incubated in medium lacking or containing 3 μm CpG-A for 3 h. For all experiments, cellular lysates were examined for luciferase activities. Results are shown as -fold induction of luciferase activity relative to pCI-transfected cells. A portion of each lysate was additionally examined for protein expression by using IB to detect FLAG-tagged cFLIP_L, myc-tagged Vpx, or myc-tagged AIP. Luciferase assays are representative of three technical replicates, and all luciferase assays were performed at least three times. Data are expressed as the mean ± S.D. Statistically significant differences in experimental samples versus unstimulated, pCI-transfected cells are denoted (*, p < 0.05).

Fig. 1B shows the specificity of the infa6-luc plasmid for IRF7 activation; luciferase activity was robust only in cells overexpressing IRF7 proteins. Additionally, overexpression of a IRF3CA or co-expression of IRF5 and TRAF6 did not stimulate the ifna6-controlled luciferase reporter gene significantly above levels of pCI-transfected cells, again showing specificity of ifna6-luc for IRF7. It is not fully clear how overexpressing WT IRF7 in 293T cells activates the ifna6-luc reporter, but this phenomenon has been seen in several publications (36–39). The most likely explanation is that the transfection process of plasmids mimics viral infection or CpG stimulation of TLR9 (40). cFLIP_L inhibited IR7-controlled luciferase activity, suggesting that cFLIP_L may act at one or more stages of the IRF7 signal transduction pathway. Note that luciferase activity is lower in cells transfected with IRF7 alone (15-fold, Fig. 1B) compared with cells co-overexpressing IRF7 and MyD88 (36-fold, Fig. 1C) or co-overexpressing IRF7, MyD88, IKKα, and TRAF6 (20-fold, Fig. 1D). The purpose of co-expressing IRF7, MyD88, TRAF6, and MyD88 (Fig. 1D) was to simulate formation of the myddosome (41). cFLIP_L inhibition of luciferase activity (Fig. 1, C and D) suggested that cFLIP_L antagonized one or more of these molecules or an event occurring downstream of myddosome formation. A dominant-negative mutant IRF7 (pIRF7DN) significantly inhibited ifna6-luc activity in all of these systems, as would be expected (Fig. 1, C and D). Interestingly, cFLIP_L inhibited IRF7-induced infa6-luc activity to a similar extent as AIP, a cellular protein known to inhibit IRF7 activation (42) (Fig. 1B).

The experiments shown in Fig. 1B overexpressed IRF7 to stimulate IRF7 activation because 293T cells do not express sufficient levels of IRF7 to drive promoter activity (42). In contrast, HeLa cells express IRF7, and IRF7 protein levels are increased when cells are transfected with a plasmid encoding IFNα (43, 44). Using this approach, incubation of HeLa cells with CpG-A stimulates the TLR9-induced IRF7 signal transduction pathway (45). Using this system, CpG-A activated IRF7 in vector-transfected cells, similar to another published report (Fig. 1E) (45). cFLIP_L significantly inhibited CpG-A-induced luciferase activity, and the extent of this inhibition was similar to the inhibition observed with AIP (Fig. 1E). Thus, cFLIP_L inhibited IRF7 activity in two separate experimental systems.

cFLIP_L does not associate with IRF7

We showed previously that cFLIP_L binds to an IRF3–CBP complex to prevent enhanceosome formation (27). Because IRF3 and IRF7 are similar, one possibility was that cFLIP_L would also interact with and inhibit IRF7.

293T cells were initially used to test this hypothesis because these cells have high rates of transfection efficiency and are used routinely to detect protein–protein interactions (27). Epitope-tagged versions of IRF7 were expressed in 293T cells because 293T cells have very low levels of endogenous IRF7 (46). As shown in Fig. 2A, despite the abundance of IRF7 in these cells, a FLAG-tagged cFLIP_L was not detectable in IRF7 immunoprecipitates. It was unlikely that this lack of detection was due to suboptimal conditions for protein–protein interactions because we detected IRF7 interacting with a known binding partner (AIP) (Fig. 2A, left panel) (24). Also, we detected cFLIP_L interacting with IRF3, a known cFLIP_L binding partner (Fig. 2A, right panel) (27).

Figure 2.

cFLIP_L does not co-immunoprecipitate with IRF7. A, 293T cells were co-transfected with 500 ng of pIRF7 or pIRF3 and 1000 ng of pCI, pcFLIP_L, or pAIP. 24 h post-transfection, cells were lysed. A portion of each lysate was incubated with anti-IRF7 or anti-IRF3 antibodies or nonspecific IgG. IB analysis of IP samples was performed to detect FLAG-tagged cFLIP_L and myc-tagged AIP, IRF7, or IRF3. B, HeLa cells were transfected with 1000 ng of pcFLIP_L or pAIP. 24 h post-transfection, cells were lysed, and a portion of each lysate was incubated with anti-IRF7 or nonspecific IgG antibodies. IB analysis of IP samples was performed to detect FLAG-tagged cFLIP_L and myc-tagged AIP and IRF7. For all IPs, a portion of each whole-cell lysate was also examined by immunoblotting to confirm expression of the proteins of interest.

A similar co-immunoprecipitation was performed in HeLa cells (Fig. 2B). Endogenous IRF7 protein levels were detected in HeLa cells, allowing us to examine whether cFLIP_L interacted with endogenous IRF7. Similar to Fig. 2A, cFLIP_L was not detected in IRF7 immunoprecipitates. Again, IRF7–AIP interactions remained detectable, showing that conditions were optimal for detecting IRF7 binding partners. Thus, it appeared that cFLIP_L did not exert its antagonistic effects via interacting with IRF7.

cFLIP_L inhibits IRF7 phosphorylation

One critical step in the TLR9-induced IRF7 activation pathway is IRF7 phosphorylation at Ser-477 and Ser-479 (36). After IRF7 is phosphorylated, IRF7 changes conformation, exposing the interferon association domain to allow IRF7 homodimerization, nuclear translocation, recruitment of critical co-factors such as CBP (47), and DNA binding (20).

Because cFLIP_L did not co-immunoprecipitate with IRF7 (Fig. 2), we asked whether cFLIP_L prevented IRF7 phosphorylation. To test this, HeLa cells were transfected with pIFNα to increase endogenous IRF7 expression and then stimulated with CpG-A, resulting in IRF7 phosphorylation (Fig. 3A). Phospho-IRF7 was also observed when AIP was expressed in cells, and this was expected because AIP inhibits IRF7 activation downstream of IRF7 phosphorylation (24). In contrast, IRF7 phosphorylation was not detected in cFLIP_L-expressing cells (Fig. 3A). This suggested that cFLIP_L targeted a signaling event upstream of IRF7 phosphorylation. The data in Fig. 3B further supported this concept. In this luciferase reporter assay, IRF7CA was overexpressed. IRF7CA is sufficient to stimulate infa6-luc activity because phosphomimetic amino acid substitutions (Ser-477 and Ser-479 to Asp) yield an IRF7 protein that is constitutively active without the need for a kinase (36, 48). cFLIP_L did not inhibit the activity of a constitutively active IRF7 mutant, suggesting that it works upstream of phosphorylation. AIP blocked IRF7-controlled luciferase activity, and this was expected because AIP prevents nuclear translocation of IRF7 (24) (Fig. 3B).

Figure 3.

cFLIP_L inhibits IRF7 phosphorylation. A, HeLa cells were transfected with 1.5 μg of pIFNα and 6 μg of pCI, pcFLIP_L or pAIP. 24 h post-transfection, cells were incubated with medium lacking or containing 3 μm CpG-A for 3 h. Cells were lysed, and immunoblotting was performed to detect phospho-IRF7, IRF7, FLAG-tagged cFLIP_L, myc-tagged AIP, and β-actin proteins. B, 293T cells were co-transfected with 450 ng of pifna6-luc; 50 ng of pRL-TK; 1000 ng of pCI, cFLIP_L, or pAIP; and 500 ng of pCI or pIRF7CA. 24 post-transfection, cells were lysed, and luciferase activities were quantified. Luciferase assays are representative of three technical replicates, and all luciferase assays were performed at least three times. Results are shown as -fold induction of luciferase activity relative to unstimulated, pCI-transfected cells. Immunoblot analysis of whole-cell lysates also was performed to detect FLAG-tagged cFLIP_L and myc-tagged AIP. C, 293T cells were transfected with 500 ng of IRF7, 500 ng of pCI or pMAVS, and1000 ng of pCI, pcFLIP_L, or pnsp11. 24 h post-transfection, cells were lysed, and immunoblotting was performed to detect phospho-IRF7, IRF7, FLAG-tagged cFLIP_L, FLAG-tagged nsp11, and β-actin proteins. The experiments shown here are representative of experiments performed at least three times. Data are expressed as the mean ± S.D. Statistically significant differences in experimental samples compared with cells transfected with empty vector are denoted (*, p < 0.05).

To confirm that the inhibition of phospho-IRF7 by cFLIP_L was not indirectly due to CpG-A activation of TBK1–IKKϵ–mediated IRF7 phosphorylation (49), we performed an IRF7 phosphorylation assay in 293T cells expressing either empty vector, cFLIP_L, or nsp11, a porcine respiratory virus protein known to inhibit IRF7 phosphorylation (50). To stimulate TBK1–IKKϵ-mediated IRF7 phosphorylation, we overexpressed the upstream signaling molecule MAVS (51). Here cFLIP_L did not inhibit IRF7 phosphorylation, in contrast to nsp11 (Fig. 3C). IRF7 protein levels were greatly reduced in Nsp11-expressing cells because NSP11 is an endoribonuclease (50).These data suggest that cFLIP_L does not antagonize the TBK1–IKKϵ kinase complex. This is further supported by the finding that cFLIP_L does not inhibit TBK1-induced IRF3 phosphorylation (27).

The N-terminal DED-containing region of cFLIP is necessary to inhibit IRF7 phosphorylation and activation

Fig. 4A shows that cFLIP_L is comprised of two death effector domains (DEDs) and a C terminus containing a caspase-like domain (CLD). In contrast, cFLIP_S lacks the CLD. We showed previously that the CLD of cFLIP_L is sufficient to inhibit the IRF3 activation pathway (27). Thus, the DEDs were dispensable for cFLIP_L inhibition of IRF3 activity. We were curious whether the CLD also provided IRF7 inhibition. We used the same IRF7-specific luciferase reporter assay as shown in Fig. 2D to map the cFLIP_L domain(s) required for inhibition. As shown in Fig. 4B, cFLIP_L and cFLIP_S each significantly inhibited CpG-A–induced ifna6-luc activity, suggesting that one or more DEDs possess the inhibitory function. These data also agree with the finding that cFLIP_S inhibits IFNα production (29). However, the CLD did not antagonize IRF7 activation (Fig. 4B). Consistent with luciferase assay results, cFLIP_L and cFLIP_S, but not the CLD, inhibited IRF7 phosphorylation triggered by either CpG-A treatment of cells (Fig. 4C) or when IRF7 and IKKα were overexpressed (Fig. 4D). Thus, the DED regions of cFLIP_L and cFLIP_S are important for IRF7 antagonism. Equally important, Fig. 4D showed that IKKα overexpression resulted in IRF7 activation in a manner presumed to be independent of TBK1 and IKKϵ. Thus, cFLIP_L inhibition of IKKα-induced IRF7 phosphorylation continues to suggest that cFLIP_L does not act on the TBK1–IKKϵ complex to inhibit IRF7 activation.

Figure 4.

The N-terminal DEDs of cFLIP inhibit IRF7 activation. A, schematic of the wildtype and mutant (CLD) cFLIP_L protein and the alternative splice variant cFLIP_S. cFLIP_L and cFLIP_S each contain tandem DEDs, whereas only cFLIP_L possesses a CLD. B, HeLa cells were co-transfected with 450 ng of pifna6-luc, 50 ng of pRL-TK, 250 ng of pIFNα, and 1000 ng of pCI, pcFLIP_L, pcFLIP_S, or pCLD. 24 h post-transfection, cells were incubated in medium lacking or containing 3 μM CpG-A for 3 h. Cells were lysed, and luciferase activities were quantified. Results are shown as -fold induction of luciferase activity relative to untreated pCI-transfected cells. Immunoblot analysis of lysates was also performed. Luciferase assays are representative of three technical replicates, and all luciferase assays were performed at least three times. Data are expressed as the mean ± S.D. Statistically significant differences in experimental samples compared with cells transfected with empty vector are denoted (*, p < 0.05). C, HeLa cells seeded in 10-cm dishes were transfected with 1.5 μg of pIFNα and 6 μg of pCI, pcFLIP_L, pcFLIP_S, or pCLD. 24 h post-transfection, cells were incubated with medium lacking or containing 3 μm CpG-A for 3 h. Cells were lysed in 100 μl to concentrate protein. D, 293T cells were transfected with 1000 ng of pCI or 500 ng of pIRF7 and pIKKα and 1000 ng of pCI, pcFLIP_L, pcFLIP_S, or pCLD. 24 h post-transfection, cells were lysed. Immunoblotting was performed to detect phospho-IRF7, total IRF7, each FLAG-tagged FLIP, or β-actin.

cFLIP_L associates with IKKα and prevents IKKα–IRF7 interactions

The data above showed that, although cFLIP_L inhibited IRF7 phosphorylation, it did not bind to IRF7. Two kinases (IRAK1 and IKKα) are reported to promote IRF7 phosphorylation during TLR9 stimulation (17, 19). The current dogma is that the IRAK1–IKKα kinase cascade leads to phosphorylation of IRF7 (19, 52). Thus, we queried whether cFLIP_L disrupts members of the signaling complex that are critical for IRF7 phosphorylation. To test this, we performed IRF7 co-immunoprecipitation, where IKKα, IRF7, cFLIP_L, and TRAF6 were ectopically expressed. We observed that IKKα–IRF7 interactions were greatly diminished when cFLIP_L was present (Fig. 5A), implying that cFLIP_L inhibited IRF7–IKKα interactions. As expected, cFLIP_L–IRF7 interactions were not detected, similar to the observations shown in Fig. 2. IRF7 activation by IKKα is preceded by its ubiquitination by TRAF6 (17, 53). Interestingly, overexpression of cFLIP_L did not prevent IRF7–TRAF6 interactions (Fig. 5A). This suggested that cFLIP_L acted downstream of the formation of the MyD88-based complex containing TRAF6.

Figure 5.

cFLIP_L associates with IKKα and prevents IKKα–IRF7 interactions. A, 293T cells were co-transfected with 500 ng of pIRF7, 1000 ng of pCI or pcFLIP_L, 500 ng of pIKKα, and 500 ng of pTRAF6 as indicated. 24 h post-transfection, cells were lysed, and a portion of each lysate was immunoprecipitated with anti-IRF7 or nonspecific IgG antibodies. A separate portion of each lysate was used to examine protein expression levels. IB analysis of co-immunoprecipitated samples was performed to detect IKKα, myc-tagged TRAF6, FLAG-tagged cFLIP_L, or IRF7 proteins. B, 293T cells were co-transfected with 1000 ng of pIKKα, pCI, pcFLIP_L, and pIRF7 as indicated. 24 h post-transfection, cells were lysed, and a portion of each lysate was immunoprecipitated with anti-IKKα or nonspecific IgG antibodies. A separate portion of each lysate was used to examine protein expression levels. IB analysis of co-immunoprecipitated samples was performed to detect FLAG-tagged cFLIP_L, IKKα, or IRF7. C, 293T cells were co-transfected with 500 ng of pIKKα and 1000 ng of pCI, pcFLIP_L, pcFLIP_S, or pCLD. 24 h post-transfection, cells were lysed, and a portion of each lysate was incubated with anti-IKKα or nonspecific IgG antibodies. IB analysis of IP samples was performed to detect FLAG-tagged cFLIP constructs and IKKα. For all lysates, immunoblot analysis of whole-cell lysates was also performed. The asterisk denotes the heavy chain.

We next wanted to ask whether cFLIP_L disrupted IRF7–IKKα interactions by competitive inhibition. Co-immunoprecipitations were performed to examine interactions between IKKα and cFLIP_L (Fig. 5B). For this experiment, epitope-tagged IKKα and cFLIP_L were co-overexpressed in 293T cells. Fig. 5B shows that cFLIP_L indeed co-immunoprecipitated with IKKα. This was not unexpected given that a variant of cFLIP_L (p43) was reported to bind to IKKα (54). As a control, we were also able to detect IKKα–IRF7 interactions in cells ectopically expressing IKKα and IRF7 (note that the thick band representing the heavy chain has a slightly different mobility than the IRF7-containing band) (Fig. 5B). Fig. 4 suggested that the DEDs of cFLIP were critical for IRF7 inhibition, whereas the CLD is dispensable. We performed co-immunoprecipitation to identify the cFLIP region that associated with IKKα. Indeed, cFLIP_L and cFLIP_S co-immunoprecipitated with IRF7 whereas the CLD did not (Fig. 5C), further supporting the model that IRF7–cFLIP interactions are critical for the inhibitory mechanism of cFLIP_L.

cFLIP_L inhibits IRF7 in the THP-1 and CAL-1 cell lines

The above experiments showed that cFLIP_L inhibited IRF7 activation in HeLa and 293T cells. IRF7 is expressed at higher levels in hematopoietic cells like macrophages and pDCs (23, 55–57). If the cFLIP function identified in HeLa and 293T cells was relevant, then cFLIP_L should antagonize IRF7 activation in these professional antigen-presenting cells (APCs). There were two possible ways to test this mechanism in physiologically relevant cell lines. We could silence endogenous cFLIP_L and ask whether that results in an increase in IR7 activation and IFNα gene expression, However, this approach is technically difficult because cFLIP_L is required for macrophages (58) because of the anti-apoptosis properties of cFLIP_L (59, 60). In our hands, attempts at silencing cFLIP_L also resulted in cell death, making it difficult to collect sufficient amounts of cells for experimentation. An alternative strategy is to overexpress cFLIP_L and ask whether this correlates with a decrease in IRF7 phosphorylation and IFNα expression. This approach was feasible because cFLIP_L was not expressed at high levels in the THP-1 and CAL-1 cells (Figs. 6, A and C, note that cFLIP was not detected in cells transduced with the control (con) lentivirus and subsequently left untreated or treated with CpG-A). When the cFLIP_L gene (cflar) was stably introduced into the THP-1 human monocyte cell line via lentivirus transduction (61), cFLIP_L protein expression was detected (Fig. 6). We picked this cell line because PMA-treated THP-1 cells differentiate to macrophage-like cells (62). In this state, THP-1 cells respond to CpG-A stimulation and express high levels of IRF7-controlled IFNα and interferon-stimulated gene transcripts (63, 64). We also transduced the CAL-1 cell line with the same cFLIP_L-expressing lentivirus. The CAL-1 cell line was developed for use as a surrogate for primary pDCs to study type I IFN signaling and production (65). One benefit of using this cell line as opposed to primary human cells is that it avoids donor-to-donor variation. Although CAL-1 cells produce IFNα to a lesser extent than primary pDCs (65), the IRF7 signal transduction and activation pathway is maintained (66). As a control, a separate set of THP-1 and CAL-1 cells was transduced with lentiviruses that lacked the cFLIP_L gene (depicted as control in Fig. 6).

Figure 6.

cFLIP_L inhibits IRF7 activation and IKKα–IRF7 interactions in THP-1 and CAL-1 cell lines. A and B, THP-1 cells (A) and CAL-1 pDC cells (B) were transduced with a control lentivirus (con) or cFLIP_L-expressing lentivirus (FLIP). THP-1 cells were differentiated into macrophages by treatment with PMA (10 ng/ml for 16 h). Transduced cells were incubated with medium lacking or containing 10 μm CpG-A or CpG-B for 5 h. Cells were lysed, and total RNA was extracted. A portion of each lysate was also used to detect cFLIP_L protein expression. The levels of ifna4, ifna6, and il12p40 mRNA were quantified by using quantitative RT-PCR. C, transduced CAL-1 cells were incubated in medium lacking or containing 10 μm CpG-A for 5 h. Cells were then lysed, and a portion of lysate was immunoprecipitated with anti-IRF7 or nonspecific IgG antibodies. IB analysis of IP samples was performed to detect phospho-IRF7 or IRF7. A portion of each lysate prior to immunoprecipitation was also analyzed for expression of IRF7, cFLIP_L, or β-actin. The asterisk denotes the heavy chain. D, transduced CAL-1 cells were incubated in medium lacking or containing 10 μm CpG-A for 5 h. Cells were then lysed, and a portion of lysate was immunoprecipitated with anti-IKKα or nonspecific IgG antibodies. IB analysis of IP samples was performed to detect endogenous IKKα, cFLIP_L, or IRF7. A portion of each lysate prior to immunoprecipitation was also analyzed for protein expression.

Transduced THP-1 cells or CAL-1 cells were incubated with CpG-A to trigger IRF7 activation (57, 67, 68). The transcription of two genes known to be controlled by IRF7 homodimers (ifna4 and ifna6) was examined to assess the function of cFLIP_L inhibition in both cell lines (69). As shown in Fig. 6, A and B, CpG-A–induced ifna4 and infa6 mRNA expression was significantly inhibited in cFLIP_L-expressing THP-1 and CAL-1 cells, respectively, compared with cells transduced with a virus lacking the cFLIP_L gene. As a control, the transcription of a gene not controlled by IRF7, il12p40 (2), was examined to assess the specificity of cFLIP_L on TLR9-mediated, IRF7-driven transcription. CpG-B, but not CpG-A, will stimulate il12p40 expression (2). As shown in Fig. 6B, there was no significant difference in il12p40 mRNA levels in control or cFLIP_L-expressing CAL-1 cells during CpG-B stimulation. There was a slight increase in il12p40 mRNA levels in cFLIP_L-expressing cells versus control cells when CpG-A was used, and this may be due to the action of cFLIP_L as an NF-κB activator (70). This suggests that the inhibitory role of cFLIP_L is IRF7-specific, validating the luciferase results we observed in Fig. 1A.

Focusing on just CAL-1 cells, we observed that CpG-A–mediated IRF7 phosphorylation was decreased in CAL-1 cells expressing cFLIP_L (Fig. 7). Fig. 7 shows cFLIP_L co-immunoprecipitated with IKKα in both unstimulated and stimulated cells. Additionally, IKKα–IRF7 interactions were greatly reduced in cFLIP_L-transduced cells versus cells transduced with an empty vector (Fig. 7). Thus, cFLIP_L inhibits IRF7 activation by interacting with IKKα in antigen-presenting cells (Fig. 7). We attempted to examine IKKα interactions with endogenous cFLIP_L but failed to reliably and consistently detect cFLIP_L.

Figure 7.

Proposed mechanism for cFLIP-mediated inhibition of IRF7-driven IFNα production. Activation of endosomal TLRs such as TLR7 by single-stranded RNAs and TLR9 by CpG motifs (e.g. CpG-A) leads to recruitment of the MyD88 protein. Next is the formation of a dynamic complex including at least IRAK4, IRAK1, and TRAF6. This complex triggers TRAF6-mediated Lys-63–linked ubiquitination of IRF7, followed by IRF7 phosphorylation. A current favored model proposes that IRAK4 phosphorylates IRAK1, leading to phosphorylation of IKKα. IKKα, in turn, activates IRF7. Phosphorylated IRF7 homodimerizes and translocates to the nucleus, where it drives expression of IFNα. The data shown here suggest that cFLIP binds to IKKα in a manner that prevents IKKα-mediated IRF7 phosphorylation and subsequent downstream IRF7 action.

Discussion

IRF7 is critical for IFNα gene expression (2–4). There is one previous report showing that cFLIP inhibits IFNα production (29). However, the antagonistic mechanism of cFLIP remained unknown. The goal here was to identify this function by examining the effect of cFLIP_L on well-known signal transduction events of the TLR9-induced IRF7 activation pathway. We observed that cFLIP_L prevented IRF7 phosphorylation. IKKα is one well-known IRF7 kinase (19). We performed co-immunoprecipitation assays and found that IRF7-IKKα interactions were abrogated by cFLIP_L, concomitant with cFLIP_L-IKKα interactions. Thus, we conclude that cFLIP_L disrupts IRF7–IKKα interactions, interactions that are otherwise required for IRF7 activation (Fig. 7).

To the best of our knowledge, this is the first report of a cellular protein that disrupts IKKα–IRF7 interactions as a strategy to antagonize IRF7 activation. Most cellular IRF7 antagonists target IRF7 itself. For example, AIP binds to IRF7, and this interaction prevents IRF7 nuclear translocation (24). RAUL inhibits IRF7 (and IRF3) by targeting these IRFs for proteasomal degradation (21). Other proteins act indirectly on IRF7. Namely, transforming growth factor β1 promotes Lys-63–linked ubiquitination of TRAF6, which correlates with a decrease in IRF7 phosphorylation through unknown mechanisms (45). The myriad cellular strategies to decrease IRF7 activation are a testament to how the host cell has evolved multiple mechanisms to achieve immune system homeostasis.

Within the family of IRF proteins, IRF3 and IRF7 are most closely related (28). There are several lines of evidence showing that cFLIP_L antagonizes IRF7 using a mechanism distinct from its strategy to antagonize IRF3. For example, cFLIP_L inhibited CpG-A–induced IRF7 activation, a signaling pathway that does not activate IRF3 (2, 71, 72). Second, the domain of cFLIP_L required for IRF7 activation (tandem DEDs) is distinct from the region required for IRF3 inhibition (CLD) (27). Third, cFLIP_L co-immunoprecipitates with IRF3 but not IRF7 (27). Thus, cFLIP_L has at least two separate mechanisms to antagonize type I IFN production in cells. These functions of cFLIP_L may be useful considering that there is differential expression of IFNβ and IFNα by different cell types. For example, although IFNβ is produced largely by fibroblasts (73), the major expressers of IFNα are pDCs (6). Indeed, cFLIP is expressed in these cells, suggesting that cFLIP has evolved to control type I IFN production across various cell types (74). However, it appears that cFLIP_L is not a pan-IRF inhibitor; cFLIP_L did not inhibit IRF5-controlled il12p40-based luciferase activity in our hands.

There is one previous report that shows that cFLIP_S inhibits type IFNα and IFNβ production (29). Buskiewicz et al. (29) proposed that cFLIP modulates the MAVS complex to inhibit IFNβ production, but the mechanism for inhibition of IFNα expression was not elucidated. We show here that both cFLIP_S and cFLIP_L inhibit IR7 activation and IFNα production. It is possible that this IKKα-binding property of cFLIP is responsible for the inhibition of IFNα production that was observed by Buskiewicz et al. (29).

There remains some controversy with respect to the roles of IRAK1 and IKKα as IRF7 kinases. Of course, each protein is critical for IFNα production (17, 19). However, it remains unknown whether IRAK1 phosphorylates IKKα, which then goes on to phosphorylate and activate IRF7, or whether IRAK1 and IKKα each phosphorylate IRF7 at different residues to activate IRF7 (52). In our hands, cFLIP_L significantly reduces IRF7 phosphorylation while still allowing TRAF6–IRF7 interactions. Because TRAF6–IRF7 interactions occur downstream of IRAK1 kinase activity (17, 53), IRAK1 signaling events are probably not compromised in the presence of cFLIP_L. Thus, we currently suspect that cFLIP_L targets IKKα but not IRAK1.

We show here that cFLIP_L co-immunoprecipitates with IKKα, resulting in a block in IRF7 activation. Neumann et al. (54) report that the p43 form of cFLIP_L binds to IKKα and that this interaction activates the NF-κB pathway. It is unlikely that cFLIP-induced NF-κB activation indirectly contributed to IRF7 inhibition because NF-κB activation stimulates IRF7 expression during TNF stimulation (75). Nevertheless, it is quite interesting that cFLIP_L and p43 appear to have diametrically opposed functions: cFLIP_L inhibits IRF3 and IRF7, whereas p43 activates NF-κB (27, 54, 76). Thus, cFLIP may down-regulate type I IFN responses while still allowing expression of other cytokine or chemokine genes controlled by NF-κB. How this may balance an appropriate immune response remains a mystery.

Several groups target silencing of the cFLIP gene (cflar) to activate apoptosis in tumor cells that overexpress cFLIP (59, 60, 77). However, our data raise the possibility that overexpression of cFLIP may prove useful as a treatment for some types of autoimmune diseases to down-regulate IFNα production (7, 78). Thus, cFLIP_L may be one protein that could be manipulated in more than one way to the benefit of human health.

Experimental procedures

Cell lines

The human embryonic kidney 293T, human cervical HeLa, and monocytic THP-1 human cell lines were obtained from the American Type Culture Collection. The CAL-1 plasmacytoid dendritic human cell line was kindly provided by Dr. Klinman (NCI, National Institutes of Health) and Dr. Maeda (Nagasaki University) (65). 293T and HeLa cells were cultured in minimum Eagle's medium supplemented with 10% FBS (Thermo Fisher Scientific) and 1% penicillin–streptomycin (Thermo Fisher Scientific). THP-1 and CAL-1 cells were cultured in RPMI medium supplemented with 10% FBS (Thermo Fisher Scientific) and 1% penicillin–streptomycin (Thermo Fisher Scientific).

Plasmids and transfections

Plasmid pCI was obtained from Promega. Plasmids encoding a FLAG-tagged human cFLIP_L (pcFLIP_L) or cFLIP_S (pcFLIP_S) were published previously (27). Plasmid pCLD encodes a FLAG-tagged caspase-like domain of cFLIP_L (residues 178–480) and was a gift from Dr. Condorelli (University of Naples, Naples, Italy). Plasmid pIRF3CA, which expresses a constitutively active IRF3, pMAVS, which expresses the MAVS protein, and psnp11, which expresses the porcine respiratory virus nsp11 protein, were kind gifts from Dr. Yoo (University of Illinois). Plasmid pMyD88 was obtained from Dr. Richard Tapping (Department of Microbiology, University of Illinois). A plasmid encoding a GFP-tagged IRF3 (pIRF3) was a kind gift from Dr. Michelle Arnold (Louisiana State University Health Sciences Center, Shreveport, LA). Plasmids IRF7 (pIRF7) and pIFNα were provided by Dr. Fanxiu Zhu (Florida State University, Tallahassee, FL). Plasmid pIRF7CA, which expresses a constitutively active IRF7, and pIRF7DN, which expresses a dominant-negative IRF7, were kind gifts from Dr. Luciana Castiello (Instituto Pasteur, Rome, Italy). A plasmid encoding a myc-tagged TRAF6 protein was used in this work. Plasmid pIKKα encodes a FLAG-tagged IKKα protein and was a kind gift from Dr. Ulrich Siebenlist (National Institutes of Health, Bethesda, MD). Plasmid pAIP encodes a myc-tagged AIP protein and was a kind gift from Dr. Harhaj (Johns Hopkins University, Baltimore, MD). Plasmid pRL-TK was purchased from Promega. Plasmid pifna6-luc was kindly provided by Dr. Sun (Shanghai Institutes for Biological Sciences, Shanghai Shi, China). Plasmids pil12p40-luc, pVpx (encoding a FLAG-myc-hemagglutinin–tagged Vpx protein) and pIRF5 (encoding a GFP-tagged IRF5) were kindly provided Dr. Ratner (Washington University, St. Louis, MO). Plasmid DNA was transfected into cells using TransIT-2020 transfection reagent (Mirus Bio) following the protocol of the manufacturer.

Luciferase assays

Subconfluent 293T cellular monolayers were transfected with 50 ng of pRL-TK, 450 ng of pil12p40-luc, and either 500 ng of pIRF3CA, 500 ng of pIRF7, or 250 ng of pIRF5 and 250 ng of pTRAF6 to quantify IRF5 transcriptional activation. In this case, cells were additionally co-transfected with 1000 ng of pCI, pcFLIP_L, or pVpx. To detect IRF7-specific induction of gene expression, 293T cells were transfected with 50 ng of pRL-TK, 450 ng of pinfa6-luc, and either 500 ng of pIRF3CA, 500 ng of pIRF7, or 250 ng of pIRF5 and 250 ng of pTRAF6. In this case, cells were additionally co-transfected with 1000 ng of pCI, pcFLIP_L, or pAIP. To mimic myddosome-mediated, IRF7-driven gene expression, 293T cells were transfected with 50 ng of pRL-TK, 450 ng of pinfa6-luc, and either 1000 ng of pCI or 500 ng of pIRF7 and 500 ng of pMyD88, or 250 ng of pIRF7, 250 ng of pMyD88, 250 ng of pIKKα, and 250 ng of pTRAF6. In these cases, cells were additionally co-transfected with 1000 ng of pCI, pcFLIP_L, or pIRF7DN. Additionally, 293T cells were co-transfected with 50 ng of pRL-TK, 450 ng of pinfa6-luc, either 500 ng of pCI or pIRF7CA, and 1000 ng of pCI, pcFLIP_L, or pAIP. To detect IRF7 activation in HeLa cells, subconfluent cellular monolayers were transfected with 50 ng of pRL-TK, 450 ng of pinfa6-luc, 250 ng of pIFNα, and 1000 ng of pCI, pcFLIP_L, or pAIP. 24 h post-transfection, HeLa cells were incubated in medium lacking or containing 3 μm CpG-A (ODN-2216, Invivogen) for 3 h. These same conditions were used to examine the effect of cFLIP_S and CLD on IRF7 activation. In this case, HeLa cellular monolayers were transfected with 50 ng of pRL-TK, 450 ng of pinfa6-luc, 250 ng of pIFNα, and 1000 ng of pCI, pcFLIP_L, pcFLIP_S, or pCLD.

All cells were harvested 24–27 h post-transfection and lysed. Luciferase activities were detected using the Dual-Luciferase reporter assay system (Promega) and quantified using the Clarity luminescence microplate reader (BioTek Instruments). Analysis of firefly and sea pansy luciferase activities was performed as described previously (27). Values were normalized to those of untreated cells transfected with empty vectors. Values are shown as mean ± S.D. Student's t test was used to determine the statistical significance of inhibition of luciferase activity. A portion of each lysate was also analyzed for protein expression by immunoblotting. Luciferase assays are representative of three technical replicates, and all luciferase assays were performed at least three times.

Co-immunoprecipitations

To examine potential IRF7–cFLIP_L interactions, subconfluent 293T cells were co-transfected with 500 ng of pIRF7 or 500 ng of pIRF3 and 1000 ng of pcFLIP_L or pAIP. For HeLa cells, subconfluent monolayers were co-transfected with 1000 ng of pcFLIP_L or pAIP. In experiments that examined IRF7–IKKα interactions, subconfluent 293T cells were co-transfected with 500 ng of pIRF7 and either 500 ng of pIKKα or pTRAF6 and 1000 ng of pCI or pcFLIP_L. For co-immunoprecipitations of IKKα, 293T cells were transfected with 500 ng of pIKKα and 1000 ng of pCI, pcFLIP_L, pIRF7, pcFLIP_S, or pCLD. For CAL-1 cells, 10⁸ control or cFLIP_L-expressing transduced cells were treated with 10 μm CpG-A for 3 h. In all cases, cells were lysed in whole-cell lysis buffer (Abcam) 24 h post-transfection or after CpG-A treatment. Clarified supernatants were collected. A portion of each lysate was set aside for the purpose of detecting protein expression. The remaining sample was used for co-immunoprecipitations. Lysates were incubated with rabbit anti-IRF7 (Cell Signaling Technology), anti-IRF3 (Cell Signaling Technology), anti-IKKα (Cell Signaling Technology), or rabbit nonspecific IgG (Cell Signaling Technology) for 16 h at 4 °C. Protein G–Sepharose beads (Invitrogen) in a 50% slurry were added to each sample and incubated with rotation for 6 h. Beads were collected and washed three times. Pelleted beads were suspended in Laemmli buffer containing 5% 2-mercaptoethanol and boiled for 5 min. Samples were analyzed for the presence of proteins by using immunoblotting.

Immunoblotting

For all immunoblotting assays, the protein concentration of each lysate was determined by the 660-nm protein assay (Pierce). For phosphorylation assays, HeLa cells were seeded in 10-cm² dishes, and samples were lysed in 100 μl of lysis buffer to concentrate protein levels. An equal amount of protein from each lysate was electrophoretically separated by SDS-PAGE. Proteins were transferred to polyvinylidene difluoride membranes (Millipore). Antibody–antigen reactions were detected by using chemiluminescence reagents (Amersham Biosciences and Thermo Scientific) and autoradiography. Primary antibodies included the following: monoclonal rabbit anti-IRF3 (Cell Signaling Technology), monoclonal mouse anti-FLAG (Sigma-Aldrich), monoclonal rabbit anti-FLAG (Sigma-Aldrich), monoclonal mouse β-actin (Calbiochem), monoclonal mouse anti-myc (Cell Signaling Technology), monoclonal rabbit anti-myc (Cell Signaling Technology), monoclonal rabbit anti-FLIP (Cell Signaling Technology), monoclonal mouse anti-FLIP (7F10, Enzo), mouse anti-GFP (Sigma-Aldrich), rabbit anti-IKKα (Cell Signaling Technology), mouse anti-IKKα (Santa Cruz Biotechnology), rabbit anti-IRF7 (Cell Signaling Technology), and rabbit anti-phospho-IRF7 (Cell Signaling Technology).

Transduction of cells with lentiviruses

Lentiviruses containing either cFLIP_L (lenti-FLIP) or no transgene (lenti-con) were produced by co-transfecting 293T cells with the packaging plasmids pCMV-dR8.2 (Addgene, 4.5 μg) and pCMV–VSV-G (Addgene, 1.8 μg), and either an empty vector (pTRIP-IRES-GFP-control, 6 μg) or a plasmid containing the cFLIP_L gene (pTRIP-cFLIP_L-IRES-GFP, 6 μg) (61). 48 h post-transfection, lentiviruses were isolated from cellular supernatants. Lentiviruses were concentrated with Lent-X Concentrator (Clontech). The THP-1 or CAL-1 cell line was inoculated with lentiviruses by using spinfection. Briefly, 1 × 10⁶ cells, 50 μl of concentrated virus, and 10 μg of Polybrene in 1 ml of virus medium (RPMI with 1% FBS) were centrifuged at 800 × g for 45 min at 37 °C. After spinfection, the medium was aspirated, and cells were resuspended in 1 ml of fresh medium (RPMI with 10% FBS) with 50 μl of concentrated virus and incubated at 37 °C. 24–72 h post-infection, GFP expression was used as a visual marker of transduction. Cellular populations with >80% GFP expression were passaged for use as stably transduced cell lines (THP-1 cells) or used immediately for experimentation (CAL-1 cells). Transduced THP-1 cells were passaged no more than four times, checking for GFP expression after each passage.

Quantitative RT-PCR

THP-1 cells were incubated in medium without or containing 10 ng/ml PMA for 16 h to differentiate cells into macrophage-like cells (62). Differentiated THP-1 or CAL-1 cells were stably transduced with a control lentivirus (lenti-con) or a lentivirus expressing cFLIP_L (lenti-FLIP). Transduced cells were stimulated with 10 μm CpG-A for 5 h to stimulate the IRF7 signal transduction pathway (64). For il12p40 expression, transduced cells were stimulated with 10 μm CpG-B (ODN-2006, Invivogen) for 5 h. Total RNA was extracted from cells using the RNAeasy extraction kit (Qiagen). cDNA was generated using Moloney murine leukemia virus (M-MuLV) reverse transcriptase and poly(dT) oligonucleotides (New England Biolabs). Quantitative PCR was performed using a Mastercycler Realplex EP (Eppendorf) and SoFast EvaGreen Super Mix (Bio-Rad) according to the instructions of the manufacturer. The following primers were used to PCR-amplify cDNA: β-actin forward (5′-AGTTGCGTTACACCCTTTCT-3′), β-actinreverse (5′-ACCTTCACCGTTCCAGTTT-3′), ifna4 forward (5′-GATACTCCTGGCACAAATGG-3′), ifna4 reverse (5′-TCATGGAGGACAGAGATGG-3′), ifna6 forward (5′-CAGTTCCAGAAGGCTGAAG-3′), ifna6 reverse (5′-GAGTCCTTTGTGCTGAAGAG-3′), il12p40 forward (5′-AGAGCAGTGAGGTCTTAGG-3′), and il12p40 reverse (5′-CTTTGTGACAGGTGTACTGG-3′). Changes in gene expression levels were calculated by the 2ΔΔCt method (79). For normalization, respective β-actin mRNA quantities for each cDNA sample were measured, and then each value was normalized to that of unstimulated control cells, whose value was set to one. For all samples, data are presented as the mean ± S.D. from three independent experiments. Student's t test was used to determine statistically significant differences in mRNA expression levels compared with unstimulated cells.

Author contributions

L. T. G.-T. and J. L. S., conceptualization; L. T. G.-T., resources; L. T. G.-T., formal analysis; L. T. G.-T. and J. L. S., supervision; L. T. G.-T. and J. L. S., funding acquisition; L. T. G.-T., validation; L. T. G.-T., investigation; L. T. G.-T., visualization; L. T. G.-T., writing-original draft; J. L. S., methodology; J. L. S., project administration; J. L. S., writing-review and editing.