Serine/threonine acetylation of TGFβ-activated kinase (TAK1) by Yersinia pestis YopJ inhibits innate immune signaling

YopJ Blocks the IMD Innate Immune Signaling Pathway.

To develop Drosophila as a model system for the study of YopJ-mediated innate immune inhibition, S2* stable cell lines inducibly expressing either wild-type YopJ (YopJ^WT) or a catalytically inactive mutant (C172A, YopJ^CA) under the control of the metallothionein promoter were generated. In these cells, the addition of copper sulfate induced YopJ expression, whereas in the absence of copper, no protein was detected (Fig. 1A, Lower). These cells were stimulated with Spätzle-C106 (22) or DAP-type PGN to activate the Toll or IMD pathways, respectively. Toll signaling was monitored by Northern blotting for Drosomycin message, whereas IMD signaling was monitored by probing for Diptericin (and, to a lesser degree, Drosomycin) (Fig. 1A). YopJ had no effect on the Spätzle-induced expression of Drosomycin, whereas PGN-induced Diptericin and Drosomycin expression was dramatically inhibited. The failure to induce Diptericin was clearly linked to the presence of YopJ^WT, because YopJ^CA was not inhibitory and without copper pretreatment Diptericin induction was robust. Together, these results demonstrate that YopJ is a potent inhibitor of IMD, but not Toll, signaling in Drosophila cells.

For studies in whole animals, we generated YopJ transgenic flies. Using the dual GMR/UAS promoter system (pGUS; ref. 23) (Fig. S1A), we generated flies constitutively expressing wild-type or mutant YopJ in the eye, and at any other location/time with appropriate Gal4 “drivers.” To monitor the effect of YopJ on the humoral systemic immune response, the fat body-specific yolk-Gal4 driver was used. After infection with live Escherichia coli, control animals (female yolk-gal4 driver flies or YopJ^WT flies containing no Gal4 driver) showed robust IMD pathway activation, as assayed by Northern blotting for Diptericin. However, the yolk-gal4;UAS-YopJ^WT females exhibited markedly reduced Diptericin expression (Fig. S1B), consistent with our cell culture data. Together these observations show that YopJ blocks IMD signaling in both tissue culture and whole animals.

Interestingly, moderate expression of YopJ^WT in the developing eye imaginal disk (from the GMR element in the pGUS transgene) led to a rough and reduced eye phenotype, which was not observed in YopJ^CA flies (Fig. S1C). Multiple transgenic insertion lines were generated for both wild-type and mutant YopJ. Although the penetrance of the rough eye phenotype varied among the YopJ^WT lines (as shown in Fig. 1C, columns 1 and 2), no rough/reduced eye phenotype was present in any lines expressing YopJ^CA. The rough and reduced eyes found in YopJ^WT flies were further enhanced by driving higher levels of YopJ^WT via an additional GMR-GAL4 driver (Fig. S1C, Lower). Unlike the Toll pathway, the IMD pathway is not involved in development and most IMD pathway mutants are viable to adulthood, with normal eyes. Therefore, it seems likely that the developmental defects seen in YopJ^WT flies are a consequence of inhibition of other MAP kinase pathways important for eye development. Further work will be required to identify these kinases.

Having established that YopJ^WT inhibits the IMD signaling pathway, we next sought to determine which component(s) of this signaling pathway are targeted. We have shown that the cleavage and subsequent ubiquitination of the imd protein are crucial events during signaling (16). Because YopJ was proposed to act as a ubiquitin-protein protease (7, 8), we examined whether either of these IMD modifications were altered in stable cell lines expressing YopJ. After stimulation with PGN, comparable levels of IMD cleavage and ubiquitination were detected in both the YopJ^WT and YopJ^CA expressing cells, in the presence or absence of copper, indicating that YopJ does not function as an IMD-specific ubiquitin-protein protease (Fig. 1B).

However, downstream signaling events in both the JNK and NF-κB arms of the IMD pathway were clearly inhibited by YopJ. For example, the Drosophila JNK pathway, as assayed by immunoblotting for PGN-induced phospho-JNK, was blocked by YopJ^WT (Fig. 1C). In the NF-κB/Relish branch, activation of the Drosophila IKK complex was also inhibited by YopJ^WT, but not by YopJ^CA (Fig. 1D). Together with the inhibition of AMP gene induction, these data indicate that YopJ acts by inhibiting the IMD pathway downstream of IMD cleavage and ubiquitination but upstream or at the level of JNK phosphorylation and IKK activation. Although Orth and colleagues have argued that both NF-κB and JNK signaling are inhibited by YopJ-mediated acetylation of multiple MAP2 kinases (10), an alternate possibility is that YopJ also inhibits a common upstream protein that is required for both NF-κB and JNK activation, such as dTAK1.

YopJ Inhibits dTAK1 Activation.

To examine whether YopJ may be directly inhibiting the kinase dTAK1, we next coexpressed both dTAK1 and YopJ (WT or CA) in double-stable S2* cells. In these cells, dTAK1 and IKK kinase activities were monitored with IP-kinase assays by using recombinant MKK6^K82A (a mammalian TAK1 target) or recombinant Relish as substrates, respectively. In S2* cell lines, overexpression of dTAK1 was sufficient to activate both dTAK1 and the downstream IKK complex (Fig. 2A, rows 1 and 4). When both dTAK1 and YopJ^WT were coexpressed, the activity of both kinases was inhibited, whereas coexpression of YopJ^CA resulted in no discernible reduction in signal. Interestingly, dTAK1, expressed alone or in combination with YopJ^CA, migrated as a tight doublet at a molecular mass just below 105 kDa, whereas in concert with YopJ^WT, dTAK1 migrated noticeably faster, closer to 85 kDa (Fig. 2A, row 2). When optimized for better resolution, immunoblot analysis showed that overexpressed dTAK1 alone, or in the presence of YopJ^CA, migrates as a doublet at ∼95 and 105 kDa. In the presence of YopJ^WT, dTAK1 runs faster at a molecular mass of ∼85 kDa (Fig. 2A, row 3). The predicted molecular mass for Drosophila TAK1 is 76 kDa. Together, these data strongly indicate that YopJ directly interferes with the activity of dTAK1 and that this interference is likely the result of posttranslational modification(s).

To determine whether the observed Drosophila TAK1 bands represent phosphorylated forms, we treated lysates with λ-protein phosphatase. Phosphatase treatment resolved the doublets observed in the dTAK1 alone and dTAK1 with YopJ^CA samples from ∼95/105 kDa to ∼85/93 kDa. In the presence of YopJ^WT, no change in the migration of dTAK1 was detected with phosphatase treatment (Fig. 2B). These data show that active dTAK1, (when expressed alone or with the inactive YopJ^CA) is phosphorylated; however in the presence of YopJ^WT, dTAK1 is inactive and unphosphorylated.

To analyze the effect of YopJ on endogenous dTAK1, IP-kinase assays were undertaken with an anti-dTAK1 antibody (16). Cells expressing YopJ^WT showed little PGN-induced TAK1 kinase activity, whereas cells similarly expressing YopJ^CA showed activity comparable to parental S2* cells (Fig. 2C). Although our anti-dTAK1 antibody is effective for immunopreciptiation, as monitored by endogenous dTAK1 IP-kinase assays, it is not very useful for immunoblotting (16). Therefore, we used an RNAi approach to validate this assay. RNAi treatment targeting dTAK1 was able to inhibit endogenous dTAK1 kinase activity in this IP-kinase assay and accumulation of phospho-JNK after immune induction (Fig. S2). Together, the results with both endogenous or overexpressed dTAK1 strongly suggest that YopJ interferes with this MAP3 kinase.

YopJ Acetylates dTAK1.

An alignment of Drosophila and mammalian TAK1 shows that three phospho-acceptor sites found in the activation loop of mammalian TAK1 (T184, T187, and S192) and most of the surrounding residues are highly conserved (Fig. 3A). In mammals, alanine substitution of any of these serine or threonine residues is sufficient to ablate kinase activity and block downstream signaling (24–28). To identify the phosphorylated residues of Drosophila TAK1 and map the molecular changes induced by YopJ, FLAG-TAK1 was isolated from lysates prepared from cells expressing dTAK1 alone or in combination with YopJ^WT or YopJ^CA. These samples were then subjected to microcapillary reverse-phase HPLC nano-electrospray tandem mass spectrometry (MS/MS).

Within the activation loop of Drosophila TAK1, a single phosphorylation, at S176 (which aligns to S192 of the mammalian protein), was identified by MS/MS (Fig. 3B). A number of other phosphorylation sites were found outside the activation loop (see Table S1 for the full list of phosphoresidues). Interestingly, no phosphorylation was detected in the activation loop of dTAK1 when coexpressed with YopJ^WT, and overall phosphorylation was greatly reduced. When coexpressed with YopJ^CA, dTAK1 was phosphorylated on a number of residues, including all of the sites identified when TAK1 was expressed alone and some additional sites outside the activation loop (Table S1). The relevance of these additional phosphorylation sites is unclear; however, it is important to note that when expressed with YopJ^CA, dTAK1 remains fully functional. Given that YopJ is thought to act as a serine/threonine acetyl-transferase (9, 10), mass spectrometry was also used to detect acetylation. In the presence of YopJ^WT, dTAK1 showed clear acetylation of both T168 and T171 and possible acetylation of S167 and S176, all within the activation loop and at residues highly conserved with mammalian TAK1 (Fig. 3B). However, dTAK1 alone or in the presence YopJ^CA displayed no detectible acetylation (see Table S2 for a full list of acetyl residues). Together, these data suggest that acetylation of dTAK1 threonines T168 and T171 (and possibly S167 and S176) by YopJ inactivates the kinase, which leaves it in an unphosphorylated state.

Previous studies with mammalian TAK1 indicate that the residues equivalent to Drosophila TAK1 T168, T171, and S176 (in mammalian TAK1: T184, T187, and S192) play important roles in the kinase activity of hTAK1 (24–28). To elucidate the role that these residues play in Drosophila TAK1 activity, we generated various substitution mutations at these sites. As shown previously, overexpression of wild-type dTAK1 is sufficient to activate dTAK1 kinase activity, as monitored by cold-IP-kinase assay (e.g., immunoblotting with anti-pMKK6; Fig. 3C). Interestingly, substitution of T168 to alanine did not affect kinase activity, and this mutant is still subject to YopJ-mediated inhibition, likely because T171 is still a target for acetylation and, as such, is sufficient for inhibition of the kinase. Conversely, substituting either T171 or S176 to alanine renders dTAK1 inactive (Fig. 3C), indicating that these residues are necessary for proper dTAK1 function. These data are consistent with studies of mammalian TAK1 that indicate the corresponding residues, T187 and S192, are essential for activation of the kinase, whereas T184 plays a minor role (24–28). In an attempt to bypass YopJ-mediated acetylation of these residues, we also generated a second set of Drosophila TAK1 mutants in which these serine and threonines were changed to a phosphomemetic acidic residue. Unfortunately, substitution of T168, T171, or S176 to glutamic acid rendered TAK1 inactive (Fig. 3D). All together, this analysis of dTAK1 substitution mutants indicates that residues T171 and S176 are essential for kinase activity, and that substitution or acetylation of these residues impairs the function of dTAK1.

YopJ Inhibits and Acetylates Mammalian TAK1.

With the discovery that YopJ inactivates Drosophila TAK1, we sought to determine whether YopJ had a similar effect on mammalian TAK1. To that end, transient expression of wild-type mammalian TAK1 in human 293T cells resulted in kinase activation and autophosphorylation, as monitored by cold-kinase assay using MKK6 substrate or immunoblotting for phospho-hTAK1, respectively (Fig. 4A). Interestingly, when wild-type YopJ was coexpressed with hTAK1, the kinase failed to activate, similar to the effects observed with YopJ and Drosophila TAK1. As controls, an inactive hTAK1 mutant (TAK^K63W) also failed to autophosphorylate or activate, whereas the coexpression of YopJ^CA had no effect.

To confirm that YopJ acetylates mammalian TAK1, hTAK1 was expressed in the presence or absence of YopJ and then analyzed by tandem mass spectrometry. Similar to dTAK1, hTAK1 showed no acetylation when expressed alone; however, in the presence of YopJ, T184 and T187 within the activation loop of hTAK1 were acetylated (Fig. 4B). Because phosphorylation of T187 is critical for activation of hTAK1 (24–28), acetylation of this residue likely blocks phosphorylation and activation of the kinase. A number of other residues in the activation loop and beyond were also acetylated in the presence of YopJ, although definitive localization of some of the sites could not be unambiguously identified (Table S3).

Furthermore, we tested the YopJ-dependent acetylation of a number of mammalian MAP2 kinases downstream of TAK1 in mammalian systems. Previously, YopJ has been shown to acetylate MKK6 on ser207 and thr211, the activating phosphoacceptor sites in the activation loop (10). Consistent with published findings, YopJ blocked MKK6 phosphorylation on ser207 (Fig. S3A, Top). In addition to acetylating ser207 and thr211, YopJ also acetylates a neighboring lysine (lys210) (10). Although this lysine modification is dispensable for the inhibitory effect of YopJ, it provides a unique opportunity to detect YopJ-mediated acetylation by immunoblotting with an acetyl-lysine antibody (Fig. S3A, Middle). MKK4 and MKK7, the MAP2Ks that specifically activate JNK, also contain at least one lysine in their activation motif. Exploiting these nearby lysines revealed that YopJ acetylates both MKK4 and MKK7 (Fig. S3B). Consistent with this acetylation, YopJ^WT, blocked JNK phosphorylation by MKK7 overexpression (Fig. S3C). Although YopJ has been shown to interact with MKK4 or MKK7 (29, 30), these data provide direct evidence for YopJ-mediated acetylation and inhibition of these JNK-specific MAP2Ks. Our results argue that that YopJ is capable of modifying both a MAP3 kinase (i.e., TAK1) and MAP2 kinases (i.e., MKK4 and MKK7) in a single pathway (i.e., JNK) (Fig. 5). We therefore propose that this two-pronged attack underlies the potent inhibitory effects YopJ exerts on multiple signaling pathways.

To date, previous work has failed to identify a kinase not affected by the acetyltransferase activity of YopJ. Our prior study on YopJ demonstrated that RIG-I–mediated IRF3 activation is impervious to YopJ-mediated inhibition (7). In this pathway, TBK1 is an essential IRF3 kinase (31–33). Thus, the ability of YopJ to interfere with hTAK1 and TBK1 kinase activities was compared directly. Consistent with the inability of YopJ to disrupt RIG-I signaling, it failed to inhibit TBK1 kinase activity while clearly blocking hTAK1 (Fig. 4C, row 1). Furthermore, phosphorylation of the TBK1 activation loop was not blocked by YopJ, whereas hTAK1 phosphorylation was inhibited (Fig. 4C, row 4). These results suggest that, despite the broad inhibitory profile of YopJ, this effector targets a subset of host kinases.

Cell Culture.

S2* cells were cultured in Schneiders media (Gibco) supplimented with 10% (vol/vol) FBS (Valley Biomedical), 1% (vol/vol) glutamine (Gibco), and 0.2% Pen/Strep (Gibco). Cells were treated with 1 μM 20-hydroxyecdysone (Sigma) for 24–30 h before stimulation with 100 ng/mL peptidoglycan (Invivogen) or 5 mM Spätzle (22).

RNA Analysis.

Total RNA was isolated with the TRIzol reagent (Invitrogen) as described (35) and expression of Diptericin and Rp49 was analyzed by Northern blot analysis followed by autoradiography.

Protein and Immunoprecipitation Assays.

Proteins were precipitated and analyzed as described (7, 16). See SI Materials and Methods for more details.

Kinase Assays.

Kinases were assayed as described (16, 35, 36), with modifications for cold kinase assays (see SI Materials and Methods for more details). GST-IRF3 aa380-427 was used as a substrate for TBK1 as described (32).

Peptide Identification by Tandem Mass Spectrometry.

See SI Materials and Methods for more details.