A Stat3-interacting protein (StIP1) regulates

Robert G. Collum*, Siska Brutsaert†, Gladys Lee*, and Christian Schindler*†‡

Departments of †Microbiology and *Medicine, Columbia University, New York, NY 10032

Edited by James E. Darnell, Jr., The Rockfeller University, New York, NY, and approved June 28, 2000 (received for review May 12, 1999) Genetic and biochemical studies have led to the identification of Materials and Methods the Stat3-Interacting Protein StIP1. The preferential association of Cell Culture and Transfections. 293T, M1, HepG2, NIH 3T3, and StIP1 with inactive (i.e., unphosphorylated) Stat3 suggests that it SF9 cells were acquired from the American Type Culture may contribute to the regulation of Stat3 activation. Consistent Collection and grown as reported (22–24). Before stimulation with this possibility, StIP1 also exhibits an affinity for members of with IL-6 (15 ng͞ml; PeproTech, Rocky Hill, NJ) or IFN-␥ (66 the Janus kinase family. Overexpression of the Stat3-binding units͞ml for 15 min; Pestka Research Labs, New Brunswick, NJ), domain of StIP1 blocks Stat3 activation, nuclear translocation, and cells were starved in 0.1% serum. SF9 insect cells were infected Stat3-dependent induction of a reporter gene. These studies indi- with a baculovirus encoding Stat3 and Jak1 as described (22). cate that StIP1 regulates the -dependent activation of Stat3, HepG2 and 293T cells were transfected by the calcium phos- potentially by serving as a scaffold protein that promotes the phate method after a 5-min pretreatment with 25 mM Chloro- interaction between Janus kinases and their Stat3 substrate. The quine (Sigma) as described (24). The cells were stimulated with ability of StIP1 to associate with several additional members of the IL-6 24 h later. signal transducer and activator of transcription family suggests that StIP1 may serve a broader role in cytokine-signaling events. Yeast Two-Hybrid Assay. Two-hybrid screens were performed as described in Y153 cells with a ‘‘bait’’ plasmid (pAS-Stat3) he ability of IFN-␣ to rapidly induce new genes led to the encoding amino acids 103–330 of murine Stat3 (25, 26). A mouse Tidentification of the signal transducer and activator of tran- myelomonocytic cDNA library (cloned into pGADNOT) served scription (STAT) family of transcription factors. Subsequent as the ‘‘prey’’ (a generous gift from S. Goff, Columbia Univer- studies determined that IL-6, as well as all other , sity, New York; ref. 27). Hisϩ͞␤-galactosidaseϩ prey plasmids transduce vital signals through members of the STAT family (1, underwent secondary screens, including retransformation into 2). On binding ligand, -associated kinases from the Y153 cells with pAS1 (empty vector) or a Gal4-Cyclophillin Janus kinase (JAK) family become activated. STATs are in turn fusion (a generous gift from J. Luban, Columbia University, New recruited to the receptor, where they become activated in a York) to eliminate false positives. To confirm interactions, prey JAK-dependent tyrosine phosphorylation event. Activated plasmids were retransformed into CTY10–5d cells along with a STATs are released from the receptor and dimerize. These bait plasmid, where the DNA-binding domain had been changed dimers then translocate to the nucleus and bind to members of to that of LexA (pSH2–1; ref. 28). the interferon-gamma activation site (GAS) family of enhancers. One member of the STAT family, Stat3, has been implicated DNA and RNA. Full-length StIP1 cDNAs were isolated from a in a wide range of biological processes including nephrogenesis, murine library (25). StIP-1 glutathione S-transferase (GST) gliogenesis, hepatogenesis, proliferation, inflammation, fusion proteins were generated by cloning the appropriate and oncogenesis (3–9). Many of these Stat3-dependent re- restriction fragments into pGEX-2T and confirmed by sequenc- sponses are triggered by members of the IL-6 family of cytokines, ing. GST-StIP1A contains the original sequence identified in the which transduce their signals through a common gp130 receptor yeast two-hybrid screen (StIP1 amino acids 511–831; Fig. 1 and chain. Consistent with these pleiotropic effects, Stat3 knockout see Fig. 3D). Additionally, GST-StIP1 fusion proteins included: mice exhibit an embryonic lethal phenotype (10). More recent GST-StIP1B, amino acids 511–724; GST-StIP1C, amino acids studies have begun to identify additional molecules that regulate 511–683; GST-StIP1D, amino acids 554–683; GST-StIP1E, Stat3 signaling. Some of these molecules potentiate the tran- amino acids 554–634; GST-StIP1G, amino acids 254–593; and scriptional activity of Stat3 (11, 12) whereas others antagonize GST-StIP1H, amino acids 254–395. GST-StIP1G and GST- Stat3-dependent signals (13, 14). StIP1H were derived from the cDNA missing WD40 repeat Stat3 consists of several conserved domains that are important number 7. A dominant-negative StIP1 mutant, pCGN0.8 (amino for STAT function. This includes well-characterized DNA bind- acids 554–789) was amino terminally hemagglutinin-tagged ing, linker, Src homology 2, and transcriptional activation do- (pCGN, a generous gift from S. Goff, Columbia University, New mains (15, 16). In addition, there are two well-conserved amino- York; ref. 27). RcCMV-Stat3 and pMT2T-Jak1K896R (‘‘kinase terminal domains whose functions are less well characterized. dead’’) expression constructs were generous gifts from J. Darnell The more amino proximal domain has been shown to promote cooperativity in DNA binding (17). The larger and more amino distal domain has been shown to form a coiled coil that projects This paper was submitted directly (Track II) to the PNAS office. laterally from the core STAT structure (15, 16). This generates Abbreviations: StIP1, Stat3-interacting protein; JAK, Janus kinase; STAT, signal transducer a large exposed surface area. Consistent with this, several recent and activator of transcription; GAS, interferon-gamma activation site; GST, glutathione S-transferase; IRF, IFN regulatory factor. studies have identified proteins that interact with the coiled coil ͞ Data deposition: The sequence reported in this paper has been deposited in the GenBank of Stat3 and or other STATs (12, 18–20). database (accession no. AF291064). In an effort to characterize the conserved coiled-coil domain ‡To whom reprint requests should be addressed. E-mail: [email protected]. in Stat3, a yeast two-hybrid screen was performed. This led to The publication costs of this article were defrayed in part by page charge payment. This identification of StIP1. This protein encodes 12 WD40 repeats, article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. which are known to mediate protein–protein interactions (21). §1734 solely to indicate this fact. Biochemical studies suggest that StIP1 may regulate the activa- Article published online before print: Proc. Natl. Acad. Sci. USA, 10.1073͞pnas.170192197. tion of Stat3 as well as other STATs. Article and publication date are at www.pnas.org͞cgi͞doi͞10.1073͞pnas.170192197

10120–10125 ͉ PNAS ͉ August 29, 2000 ͉ vol. 97 ͉ no. 18 Downloaded by guest on October 1, 2021 Fig. 2. Expression pattern of StIP1 transcripts. Expression of StIP1 was evaluated by a reverse transcription–PCR assay (Roche Molecular Biochemi- Fig. 1. The StIP1 sequence. Comparison of conceptually translated murine cals). RNA was prepared from each of the indicated tissues and amplified with (mu) and human (hu) StIP1 cDNA sequences. WD40 motifs are shaded and primers from repeat 6 and 8 (Fig. 3A). PCR products corresponding to the labeled 1–12. Dots represent absence and dashes represent identity of corre- full-length (FL) or ⌬7 transcripts are indicated. Controls samples (Cntl) were sponding amino acids. Arrows indicate borders of the spliced exon in ⌬7 StIP1. amplified from cDNA templates. Murine StIP1 has a predicted molecular mass of 93 kDa.

Stat3 (1:250 dilution of C-20; Santa Cruz Biotechnology) and (The Rockefeller University, New York) (25) and J. Krolewski hemagglutinin epitope tag (1:250 dilution of 12CA5; Roche (University of California, Irvine) (29). RNA was prepared by Molecular Biochemicals) primary antibodies and then Cy3- guanidium thiocyanate lysis (30). For reverse transcription– conjugated (1:500; Jackson ImmunoResearch) and Alexa- PCR, RNA was reverse transcribed (AMV-RT; Roche Molec- conjugated (1:500; Molecular Probes) secondary antibodies. ular Biochemicals) with a StIP1-specific oligonucleotide primer NIH 3T3 cells were stained with the StIP1 antiserum (1:100) and corresponding to amino acids 384–391 and then amplified with visualized with the Cy3-conjugated secondary antibody. Because primers flanking WD40 repeat number 7 (i.e., amino acids the StIP1 antibody does not recognize human StIP1, it was first 296–305 and 378–383; see Fig. 3D). preabsorbed on HeLa cells. Cells were observed at ϫ400 under a Nikon Eclipse TE300 microscope after excitation at 495 nm Cell Extracts and Protein Assays. GST pulldown assays were per- (Alexa) or 550 nm (Cy3). formed as described (22). Immobilized GST fusion proteins (i.e., bound to glutathione agarose; Sigma) then were incubated for Results 2–16 h with whole-cell extracts from HepG2 and M1 cells (23, 31) A yeast two-hybrid screen, where the coiled-coil domain of Stat3 ͞ or recombinant Stat3 from insect cells infected with Stat3 and or (amino acids 103–330) served as ‘‘bait’’, identified a novel 1.3-kb Jak1 baculoviruses (22). Bound proteins were eluted either with murine cDNA, referred to as StIP1. Full-length (i.e., Ϸ3 kb) ͞ 25 mM glutathione or by boiling in SDS Laemmli buffer, cDNAs were isolated and found to encode an 831-aa protein ͞ fractionated on 7–8% SDS PAGE, and then immunoblotted whose ORF corresponds to the reading frame identified in the with Stat3 (C-20; Santa Cruz Biotechnology), phospho-specific yeast two-hybrid ‘‘hit’’ (Fig. 1). One notable feature of the StIP1 Stat3 (New England Biolabs), or Jak1 (Upstate Biotechnology, sequence is the presence of 12 WD40 repeat motifs. These motifs IMMUNOLOGY Lake Placid, NY) antibodies. Immunoblots were visualized by form a defined propeller structure that is known to mediate enhanced chemiluminescence (Amersham Pharmacia). Coim- protein–protein interactions in a number of important cellular munoprecipitations were preformed as described (22) with Stat3 processes (21). A second cDNA, with an internal 132-bp deletion (C20; Santa Cruz Biotechnology), Jak1 (Upstate Biotechnolo- that leads to a loss of WD40 repeat number 7, also was recovered. gy), Jak2 (Upstate Biotechnology), Tyk2 (a generous gift of S. Northern blotting studies identified an Ϸ3-kb transcript in all Pellegrini, Institut Pasteur, Paris) antibodies, or with a StIP1- tissues (data not shown). A subsequent reverse transcription– specific antibody. The StIP1 antibody was generated by immu- PCR study with primers flanking the deleted region demon- nizing rabbits with purified GST-StIP1C fusion protein (amino strated that both full-length and the shorter (i.e., ⌬7) transcript acids 511–683; see Fig. 3D; ref. 30; Covance, Inc., Princeton). It are expressed in all examined tissues (Fig. 2). The full-length was used at 1:3000 in Western blots and 1:100 in immunopre- transcript, however, predominates significantly. cipitations. Electrophoretic mobility shift assays were performed A search of the National Center for Biotechnology Informa- with an IFN regulatory factor 1 (IRF-1) GAS probe (gatcGATT- tion sequence database identified homologous human and nem- TCCCCGAAAT; Oligos Etc., Guilford, CT) as described (32, atode expressed sequence tags, but no additional functional 33). Reporter assays used a GAS-driven luciferase reporter and domains. The high degree of similarity between the murine StIP1 were standardized by evaluating ‘‘null promoter’’-driven renilla and an ORF compiled from overlapping human expressed luciferase activity (Dual Luciferase; Promega) in a Turner sequence tags indicated that they are likely to encode the human Luminometer. StIP1 homologue (Fig. 1). As is the case with other WD40- containing proteins, the similarity was most striking between Immunofluorescence. HepG2 and NIH 3T3 cells were stained as corresponding WD40 repeats. Nonrepeat regions were more reported (24). Briefly, cells were fixed in 3.7% formaldehyde for poorly conserved (21). 20 min at 22°C, permeabilized in 0.2% Triton X-100 (2–4 min), washed in 1ϫ Tris-buffered saline ϩ0.05% Tween 20 (137 mM StIP1 GST Fusion Proteins Bind STATs. To further characterize the NaCl͞25 mM Tris⅐HCl, pH 7.4), and blocked in 3% BSA or 10% interaction between StIP1 and Stat3, a series of partially over- milk in Tris-buffered saline (30 min). Samples were stained with lapping GST StIP1 fusion proteins were generated (Fig. 3A).

Collum et al. PNAS ͉ August 29, 2000 ͉ vol. 97 ͉ no. 18 ͉ 10121 Downloaded by guest on October 1, 2021 most notable difference was a modest increase in relative affinity for GST-StIP1G. To extend this line of investigation, extracts were prepared from a myeloid cell line that expresses abundant quantities of Stat5, a more distant member of the family (30). Consistent with the previous pulldowns, significant amounts of Stat5 were recovered. However, of the doublet of Stat5b bands normally seen after IL-3 stimulation (Fig. 3E, lane 2), GST- StIP1A appeared to preferentially bind the faster migrating and nontyrosine phosphorylated (30) Stat5 isoform (Fig. 3E, lane 4). The interaction with unphosphorylated Stat5 (i.e., amino acids 71–315) was confirmed in a yeast two-hybrid assay (data not shown). Because a preferential interaction between StIP1 and unphosphorylated STATs also was observed with the M1 ex- tracts (Fig. 3 C and D; data not shown), another set of GST pulldowns was performed with more abundant preparations of Stat3. A panel of GST-StIP1 fusion proteins were either incubated with unphosphorylated or tyrosine-phosphorylated preparations of recombinant Stat3. Again, GST-StIP1B and GST-StIP1E-bound Stat3 avidly, in both unphosphorylated and tyrosine-phosphorylated preparations (Fig. 3F). It is important to note that less than 50% of Stat3 was tyrosine phosphorylated in the activated preparation (data not shown). When these filters were reprobed with an antibody that is specific for tyrosine- phosphorylated Stat3, it was clear that none of the ‘‘pulled down’’ Stat3 was tyrosine phosphorylated, despite adequate quantities of this species being present in the input material (Fig. 3F, lane 6). These observations demonstrate that StIP1 has a strong preferential affinity for the unphosphorylated (i.e., inac- tive) STATs. Fig. 3. In vitro association of StIP1 with Stat1, Stat3, and Stat5. (A) Schematic diagram of seven GST-StIP1 fusion proteins (GST-StIP1A–GST-StIP1H). (B) GST- StIP1 Coimmunoprecipitates with Stat3. To determine whether StIP1 fusion proteins (lanes 3–6) or GST alone (lane 2) were incubated with HepG2 extracts. Bound proteins were detected with a Stat3-specific antibody StIP1 and Stat3 associate within the cell, an antiserum specific (C-20; Santa Cruz Biotechnology). Lane 1 represents 1͞20 of the input material for murine StIP1 was prepared. Consistent with the predicted for each pulldown. The mobility of Stat3 is indicated. (C) Extracts prepared molecular weight (Fig. 1), this antibody recognized a doublet of from unstimulated M1 cells were incubated with indicated GST-StIP1 fusion bands of Ϸ92 kDa (Fig. 4A). It is not known whether these two proteins (lanes 3–6) or GST alone (lane 2) and detected with a Stat3-specific isoforms arise by pre- or posttranslational processing. To con- antibody (C-20; Santa Cruz Biotechnology). Lane 1 represents 1͞3 of the input firm that StIP1 preferentially interacts with inactive Stat3, material for each pulldown. (D) The filter from C was reprobed with a extracts were prepared from M1 cells both before and after Stat1-specific antibody (M-22; Santa Cruz Biotechnology). (E) Extracts from stimulation with IL-6. As anticipated, the StIP1 antibody was unstimulated (lanes 1, 3, and 5) or IL-3-stimulated (lanes 2, 4, and 6) 32D cells able to coimmunoprecipitate Stat3 (Fig. 4A Top). Notably, none were incubated with GST-StIP1A (lanes 3 and 4) or GST alone (lane 5), and then detected with a Stat5b-specific antibody (C-17; Santa Cruz Biotechnology). of the coprecipitated Stat3 was tyrosine phosphorylated, indi- Input extracts were loaded directly onto the gel in lanes 1 and 2. Lane 2 shows cating that StIP1 has a much higher avidity for unphosphorylated the appearance of a more slowly migrating band corresponding to the IL-3- Stat3 (Fig. 4A Middle). Likewise, in a reciprocal immunopre- stimulated tyrosine-phosphorylated isoform of Stat5b (confirmed by an- cipitation, Stat3 antibodies coprecipitated StIP1, but consider- tiphosphotyrosine blotting; data not shown). (F) Recombinant Stat3 either ably more robustly in the unstimulated samples (Fig. 4A Top). inactive͞nontyrosine phosphorylated (rStat3; lanes 1–5) or active͞tyrosine These studies verify that StIP1 exhibits a strong binding prefer- phosphorylated (rStat3*P; lanes 6–10) was prepared from insect cells infected ence for the inactive (unphosphorylated) isoform of Stat3. either with a Stat3 or Stat3 plus Jak1 baculovirus. These recombinant proteins were incubated with the indicated GST-StIP1 fusion proteins (lanes 3–5 and StIP1 Associates with JAKs. Because WD40 repeats are known to 8–10) or GST alone (lanes 2 and 7) and then detected by Western blotting with a Stat3-specific antibody (C-20; Santa Cruz Biotechnology). The filter then was mediate the formation of multiprotein complexes (21), the reprobed with a phosphotyrosine-specific Stat3 antibody (Lower; New En- ability of StIP1 to interact with another major component of the gland Biolabs). JAK-STAT signaling pathway was evaluated. Antibodies di- rected against all three JAKs implicated in IL-6 signaling (i.e., Jak1, Jak2, and Tyk2; ref. 34) were able to coimmunoprecipitate When these GST-fusion proteins were incubated with extracts StIP1 from both unstimulated and IL-6-stimulated M1 cell B prepared from HepG2 cells, abundant quantities of Stat3 were extracts (Fig. 4 ). In reciprocal experiments, StIP1 antibodies were able to coimmunoprecipitate JAK kinases (data not recovered with both the StIP1 sequences encoded by the original shown). To further characterize the interactions between StIP1 1.3-kb StIP1 ‘‘hit’’ (i.e., GST-StIP1A), as well as shorter con- and JAKs, an additional set of GST pulldowns were performed structs (Fig. 3B). In contrast, GST alone was ineffective at and then evaluated with our most informative JAK antibody, binding Stat3. Studies with additional GST-StIP1 fusion proteins ␣-Jak1. The pattern of Jak1 binding to these GST-StIP1 fusion (i.e., GST-StIP1E to GST-StIP1H) determined that WD40 proteins was quite different from that observed for Stat3, with repeat number 10 was most important in mediating the inter- GST-StIP1H demonstrating the highest affinity (Fig. 4C). No action with Stat3, in both human HepG2 cells (data not shown) differences were observed between unstimulated and stimulated and murine M1 cells (Fig. 3C). extracts. Analogous results were obtained with recombinant Next, the ability of StIP1 to interact with other STATs was Jak1 (data not shown). Hence, StIP1 interacts with JAKs in a evaluated. GST-StIP1 fusion proteins were able to recover pattern that is distinct from that observed with Stat3. abundant quantities of Stat1, a close relative of Stat3 (25). The The ability of StIP1 to associate with both inactive Stat3 and

10122 ͉ www.pnas.org Collum et al. Downloaded by guest on October 1, 2021 Fig. 4. Association of StIP1 with Stat3 and JAKs. (A) Extracts were prepared from M1 cells either before or after stimulation with IL-6 (200 units͞ml). They Fig. 5. Stat3, Jak1, and StIP1 do not form a stable trimolecular complex. (A) were immunoprecipitated either with a StIP1 or Stat3 (C-20, Santa Cruz 293T cells were cotransfected with 5 ␮g each of RcCMV-Stat3 (25), pMT2T- K896R Biotechnology) antibody and subsequently immunoblotted with either a Jak1 (29), and pCNG0.8 as indicated. Whole-cell extracts (10 ␮l), prepared Stat3 (Š; C-20, Santa Cruz Biotechnology), StIP1 (), or antiphosphotyrosine after 24 h, were diluted to 500 ␮l and incubated with 5 ␮lof␣-Stat3 (C-20; (4G10; Upstate Biotechnology) antibodies as indicated. (B) M1 extracts were Santa Cruz Biotechnology). Precipitates were immunoblotted with either immunoprecipitated with Jak1 (lanes 1 and 2; Upstate Biotechnology), Jak2 ␣-StIP1 (Top), ␣-Jak1 (Middle; Upstate Biotechnology), or ␣-StIP1 (Bottom). (lanes 3 and 4; Upstate Biotechnology), or Tyk2 (lanes 5 and 6) antibodies and Mobilities of StIP1, Jak1, and Stat3 are indicated in the left margin and Mr in sequentially immunoblotted with StIP1 or the appropriate JAK antibodies as the right margin. (B) Samples, prepared as in A, were immunoprecipitated indicated. (C) Extracts from unstimulated (lanes 1–6) and IL-6-stimulated with 5 ␮lof␣-StIP1 and immunoblotted with ␣-Jak1 (Upper)or␣-StIP1 (lanes 7–12) M1 cells were incubated with GST-StIP1 fusion proteins (lanes 3–6 (Lower). Mobilities of Jak1 and StIP1 are indicated in the left margin and Mr and 9–12) or GST alone (lanes 2 and 8) and bound proteins detected with a in the right margin. (C) Model of the putative Stat3-StIP1-Jak1 trimolecular Jak1-specific antibody (˜; Signal Transduction Laboratories, Lexington, KY). complex.

JAKs suggested that StIP1 might serve as a scaffold protein, stimulated a robust induction of our IRF-1 GAS-driven reporter potentiating the functional interaction between kinase and sub- in HepG2 cells (Fig. 6A). However, cotransfection with a StIP1 strate. A number of signaling pathways recently have been shown construct expressing the WD40 repeats that bind STATs (i.e., to use scaffold proteins (35–37). To determine whether a stable repeats 10–12; pCGN0.8), led to a substantial (and dose- complex between Stat3, StIP1, and Jak1 can be identified (see dependent; data not shown) reduction of IL-6-dependent lucif- model in Fig. 5C), 293T cells were transfected with cDNAs for erase expression (Fig. 6A). We speculated that pCGN0.8 func-

Stat3 (25), StIP1, and kinase-dead Jak1 (29). Kinase-dead Jak1 tioned as an effective dominant-negative mutant, binding to and IMMUNOLOGY was selected to increase the chance of ‘‘trapping’’ Stat3 in a sequestering unphosphorylated Stat3. Cotransfection of full- putative trimolecular complex. As anticipated, abundant quan- length StIP1 lead to a more modest reduction in IL-6-dependent tities of StIP1 were recovered when Stat3 was immunoprecipi- luciferase expression. The failure of full-length StIP1 to poten- tated from cells transfected with those corresponding cDNAs tiate signaling is consistent with recent observations on other (Fig. 5A, lanes 1, 2, and 4). Hence, StIP1 and Stat3 stably overexpressed scaffold proteins, presumably because scaffold associated in these extracts (see model in Fig. 5C). However, proteins exhibit a relatively high affinity for individual compo- overexpression of Stat3 led to only a modest increase in recov- nents of their respective signaling cascade (12, 38, 39). ered StIP1, most likely because the Stat3 antibody was rate Next, the idea that overexpressed StIP1 domains block sig- limiting. (Stat3 also was readily recovered in the appropriate naling by preventing Stat3 activation was evaluated. Expression StIP-1 immunoprecipitations; data not shown.) Likewise, when constructs for Stat3 (25) and pCGN0.8 (our most potent StIP1 was immunoprecipitated, abundant quantities of Jak1 dominant-negative StIP1 mutant) were introduced into 293T were recovered in cells transfected with those corresponding cells. These cells were selected because they express modest cDNAs (Fig. 5B lane 4). Hence, StIP1 and Jak1 stably associated levels of endogenous Stat3 and can be transfected efficiently. in these extracts (see model in Fig. 5C). To determine whether The cells then were stimulated with IL-6 and evaluated for Stat3 a stable trimolecular complex exists, Stat3 immunoprecipitates activation by a sensitive electrophoretic mobility shift assay (Fig. were probed for associated Jak1. None was identified, suggesting 6B). IL-6 potently induced Stat3 homodimer DNA binding. This that these molecules do not form a stable trimolecular complex activity was inhibited in a dose-dependent manner by the in 293T cells (Fig. 5A Middle). coexpression of pCGN0.8. Stat1 homodimer binding, which arose predominantly from the Ϸ65% nontransfected cells, Dominant-Negative Mutants of StIP1 Block Stat3 Activation. To served as a control in these studies. These results are consistent determine whether the interactions between StIP1 and STATs with the reporter assays and indicated that the expression of (or JAKs) are important for signal transduction, the effect of pCGN0.8 blocked ligand-dependent Stat3 activation. overexpressing the STAT-binding domain of StIP1 on IL-6 The requirement of tyrosine phosphorylation for STAT nu- signaling was evaluated in a reporter assay. As anticipated, IL-6 clear translocation provided an additional assay to evaluate

Collum et al. PNAS ͉ August 29, 2000 ͉ vol. 97 ͉ no. 18 ͉ 10123 Downloaded by guest on October 1, 2021 Because the protein encoded by pCGN0.8 was hemagglutinin- tagged, it was possible to evaluate its expression immunohisto- chemically as well. Consistent with the cellular distribution of endogenous (Fig. 6 G and H) or overexpressed StIP1 (data not shown), pCGN0.8 was expressed predominately in the of both unstimulated and stimulated cells (Fig. 6 C and E). Importantly, IL-6-stimulated nuclear translocation of endoge- nous Stat3 was blocked in cells expressing pCGN0.8 (Fig. 6F). These studies provide further evidence that expression of pCGN0.8 blocks IL-6-stimulated activation of Stat3. Discussion The ability of protein kinases to change the activation state of substrates has led to their exploitation in signaling cascades. Both the number and complexity of kinases has increased markedly during the evolution of more complex organisms. In many cases, a single kinase has evolved to regulate more than one signaling pathway. A classic example of this is STE11, a yeast mitogen-activated homologue, which plays a critical role in both the mating response and osmo- regulation (35, 36). Notably, these two pathways function com- pletely independently. This is achieved through the use of two distinct scaffolding proteins, STE5 and Pbs2, respectively. In each case, the scaffold proteins nucleate the formation of a specific multiprotein ‘‘transducisome’’ that consists of STE11 and the appropriate downstream targets. Not only does this facilitate signaling, but it also prevents the activation of inap- propriate pathways. More recently, analogous scaffold proteins have been identified for mammalian mitogen-activated protein kinases (JIP-1 and MP1; refs. 39 and 40), as well as for several other important mammalian kinases including c-Src, Raf, I␬B kinase, and (38, 41–43). However, to date, no scaffold proteins have been reported for the STAT signaling cascade. To identify proteins involved in the regulation of STAT signaling, a yeast two-hybrid screen was set up for Stat3. The conserved coiled-coil domain, whose function is poorly under- stood, was selected as ‘‘bait.’’ Its large exposed surface area suggested that it was likely to mediate protein interactions. Fig. 6. Overexpression of StIP1 mutants blocks cytokine signaling. (A) HepG2 Moreover, recent studies have suggested that this domain un- cells were cotransfected with an IRF-1 GAS-driven reporter (32) and either empty vector, full-length StIP1, or pCGN0.8. Extracts were assayed for lucif- dergoes conformational changes during activation, potentially erase activity either before (open bars) or after (filled bars) treatment with regulating the interaction with other proteins (24, 44). Consis- IL-6. Data represent results from one experiment done in triplicate, but similar tent with these observations, several groups recently have iden- results were obtained in three independent studies. Firefly luciferase activity tified proteins that interact with STAT coiled-coil domains. The was normalized to renilla luciferase activity expressed by a control vector, best characterized is IRF-9 (p48), which binds to the coiled-coil pRLnull (Promega). (B) 293T cells were cotransfected with 1 ␮g of a Stat3 domains of Stat1 and Stat2 (18, 19). Additional interacting expression vector (25) and increasing amounts (i.e., 0.5, 1.0, 2.0, and 5.0 ␮g; proteins include p300, members of the protein inhibitor of lanes 3–6) of a pCGN0.8 expression construct with a His-epitope tag, as activated STAT family, and Nmi (12, 20, 45). Our yeast two- indicated. DNA-binding activity was evaluated with an IRF-1 GAS probe either hybrid screen identified a novel Stat3-interacting protein, StIP1. before or after stimulation with IL-6 (200 Units͞ml) plus soluble IL-6 receptor (0.5 ␮g͞ml; a generous gift from F. Horn, RWTH, Aachen, Germany; ref. 34). Subsequent studies have determined that StIP1 has a similar The mobility of STAT DNA-binding complexes is indicated in the right margin. affinity for other members of the STAT family (e.g., Stat1, Stat3, (C–F) HepG2 cells were transfected with 5.0 ␮g of pCGN0.8 and then evaluated and Stat5; Fig. 3), suggesting StIP1 may play a more general role by dual immunostaining with anti-hemagglutinin (C and E; 1° antibody ϭ in STAT signaling. However, we were unable to detect a stable 12CA5, Roche Molecular Biochemicals; and 2° antibody ϭ goat anti-mouse- interaction with Stat2 (data not shown). This is reminiscent of Alexa, Molecular Probes), and anti-Stat3 (D and F; 1° antibody ϭ C20, Santa studies demonstrating that Nmi associates with the coiled-coil Cruz Biotechnology; and 2° antibody ϭ donkey anti-rabbit-Cy3, Jackson Im- domain of all STATs, except Stat2 (12). These observations munoResearch), either before (C and D) or after an 8-min stimulation with IL-6 ͞ suggest that the Stat2 coiled-coil domain may exhibit some (E and F;15ng ml, PeproTech, Rocky Hill, NJ). (G and H) The localization of unique structural properties. Consistent with this possibility, endogenous StIP1 in NIH 3T3 cells was evaluated by immunostaining with a murine-specific StIP1 antibody either before or after stimulation with ligand. recent studies suggest that the STAT coiled-coil domain may regulate DNA-binding activity (C. Leung and C.S., unpublished observation). Perhaps the inability of Stat2 to bind DNA (2) is related to a structurally divergent coiled-coil domain. pCGN0.8 under more physiological conditions. HepG2 cells StIP1 is a novel protein that consists of 12 WD40 repeats. were transfected with pCGN0.8 and then stimulated with IL-6 These repeats have been shown to mediate the formation of (Fig. 6 C–F). As reported, when unstimulated cells were stained multiprotein complexes that play important roles in the regula- for Stat3, the protein was predominately cytoplasmic, with tion of signal transduction, transcription, and targeted proteol- modest levels in the nucleus (Fig 6D; ref. 31). Stimulation with ysis (21). To determine whether StIP1 might interact with other IL-6 promoted a dramatic increase in nuclear Stat3 (Fig. 6F). components of the JAK-STAT signaling cascade, three JAKs

10124 ͉ www.pnas.org Collum et al. Downloaded by guest on October 1, 2021 that are important in IL-6 signaling were evaluated. Not only did unable to isolate a Stat3–StIP1–Jak1 complex, there are these studies confirm an interaction with these JAKs, but several lines of evidence that support this model. First, the domain-mapping studies also indicate that they exhibit a distinct interaction between Stat3 and StIP1 depends on the phos- pattern of binding to StIP1 (i.e., vs. Stat3). These observations, phorylation state of Stat3. Second, StIP1 exhibits a distinct along with the strong preference of StIP1 for unphosphorylated pattern of affinity for the JAK kinases. Third, StIP1 is a Stat3, suggest that StIP1 plays a role early in the ligand- cytosolic protein. Fourth, overexpression of StIP1 domains dependent activation of Stat3. Consistent with this, overexpres- block IL-6-stimulated Stat3 activation, preventing dimeriza- sion of StIP1, or StIP1 domains that bind inactive Stat3, leads to tion͞DNA binding, nuclear translocation, and the induction of a block in IL-6-dependent Stat3 activation. Moreover, the af- a reporter gene. finity of StIP for other STATs suggests that it may serve a similar role for other STATs. We thank Jeremy Luban and Steve Greenberg for reagents and advice One appealing model is that StIP1 functions as a scaffold and Bill Kim and Wei Liu for superb technical assistance. These studies protein, potentiating the interaction between Stat3 and JAKs. were supported by National Institutes of Health Grant HL55413, the Once phosphorylated, Stat3 loses its affinity for StIP1 and Leukemia Society of America, the Lucille P. Markey Foundation, and the dissociates from the receptor complex. Although we have been Pfizer Postdoctoral Fellowship Program.

1. Darnell, J. E. (1997) Science 277, 1630–1635. T., Horn, F., Kashleva, H., Ihle, J., et al. (1994) Immunity 1, 457–468. 2. Schindler, C. & Strehlow, I. (2000) Cytokines and STAT Signaling (Academic, 24. Strehlow, I. & Schindler, C. (1998) J. Biol. Chem. 273, 28049–28056. San Diego). 25. Zhong, Z., Wen, Z. & Darnell, J. E. (1994) Science 264, 95–98. 3. Barasch, J., Yang, J., Ware, C. B., Taga, T., Yoshida, K., Erdjument-Bromage, 26. Durfee, T., Becherer, K., Chen, P., Yeh, S., Yang, Y., Kilburn, A. E., Lee, W. H., Tempst, P., Parravicini, E., Malach, S., Aranoff, T., et al. Cell 99, 377–386. & Elledge, S. J. (1993) Genes Dev. 7, 555–569. 4. Boccaccio, C., Ando, M., Tamagnone, L., Bardelli, A., Michieli, P., Battistini, 27. Dunaief, J. L., Strober, B. E., Guha, S., Khavari, P. A., Alin, K., Luban, J., C. & Comoglio, P. M. (1998) Nature (London) 391, 285–288. Begemann, M., Crabtree, G. R. & Goff, S. P. (1994) Cell 79, 119–130. 5. Bonni, A., Sun, Y., Nadal-Vicens, M., Bhatt, A., Frank, D. A., Rozovosky, I., 28. Luban, J., Bossolt, K. L., Franke, E. K., Kalpana, G. & Goff, S. P. (1993) Cell Stahl, N., Yancopoulos, G. D. & Greenberg, M. E. (1997) Science 278, 477–483. 73, 1067–1078. 6. Bromberg, J. F., Wrzeszczynska, M. H., Devgan, G., Zhao, Y., Pestell, R. G., 29. Krishnan, K., Pine, R. & Krolewski, J. J. (1997) Eur. J. Biochem. 247, 298–305. Albanese, C. & Darnell, J. E. (1999) Cell 98, 295–303. 30. Azam, M., Erdjument-Bromage, H., Kreider, B. L., Xia, M., Quelle, F., Basu, 7. Garcia, R. & Jove, R. (1998) J. Biomed. Sci. 5, 79–85. R., Saris, C., Tempst, P., Ihle, J. N. & Schindler, C. (1995) EMBO J. 14, 8. Sano, S., Itami, S., Takeda, K., Tarutani, M., Yamaguchi, Y., Miura, H., 1402–1411. Yoshikawa, K., Akira, S. & Takeda, J. (2000) EMBO J. 18, 4657–4668. 31. Zhang, X., Blenis, J., Li, H.-C., Schindler, C. & Chen-Kiang, S. (1995) Science 9. Takeda, K., Clausen, B. E., Kaisho, T., Tsujimura, T., Terada, N., Fo¨rster, I. 267, 1990–1994. 10, & Akira, S. (1999) Immunity 39–49. 32. Pine, R., Canova, A. & Schindler, C. (1994) EMBO J. 13, 158–167. 10. Yoshida, K., Taga, T., Saito, M., Suematsu, S., Kumanogoh, A., Tanaka, T., 33. Park, C. & Schindler, C. (1998) Methods Companion Methods Enzymol. 15, Fujiwara, H., Hirata, M., Yamagami, T., Nakahata, T., et al. (1996) Proc. Natl. 175–188. Acad. Sci. USA 93, 407–411. 34. Lu¨tticken, C., Wegenka, U. M., Yuan, J., Buschmann, J., Schindler, C., 11. Wen, Z., Zhong, Z. & Darnell, J. E. (1995) Cell 82, 241–250. Ziemiecki, A., Harpur, A. G., Wilks, A. F., Yasukawa, K., Taga, T., et al. (1994) 12. Zhu, M.-H., John, S., Berg, M. & Leonard, W. J. (1999) Cell 96, 121–130. Science 263, 89–92. 13. Schmitz, J., Weissenbach, M., Haan, S., Heinrich, P. C. & Schaper, F. (2000) 35. Faux, M. C. & Scott, J. D. (1996) Cell 85, 9–12. J. Biol. Chem. 275, 12848–12856. Trends Biochem. Sci. 23, 14. Yu, C. L., Jin, Y. J. & Burakoff, S. J. (2000) J. Biol. Chem. 275, 599–604. 36. Whitmarsh, A. J. & Davies, R. J. (1998) 481–485. 90, 15. Becker, S., Groner, B. & Mu¨ller, C. W. (1998) Nature (London) 394, 145–151. 37. Zou, H., Henzel, W. J., Liu, X., Lutschg, A. & Wang, X. (1997) Cell 405–413. 16. Chen, X., Winkemeier, U., Zhao, Y., Jeruzalmi, D., Darnell, J. E. & Kuriyan, 38. Cohen, L., Henzel, W. J. & Baeuerle, P. A. (1998) Nature (London) 395, J. (1998) Cell 93, 827–839. 292–296. 17. Vinkemeier, U., Cohen, S. L., Moarefi, I., Chait, B. T., Kuriyan, J. & Darnell, 39. Yasuda, J., Whitmarsh, A. J., Cavanagh, J., Sharma, M. & Davies, R. J. (1999) J. E. (1996) EMBO J. 15, 5616–5626. Mol. Cell. Biol. 19, 7245–7254. 18. Martinez-Moczygemba, M., Glutch, M. J., French, D. L. & Reich, N. C. (1997) 40. Schaeffer, H. J., Catling, A. D., Eblen, S. T., Collier, L. S., Krauss, A. & Weber, J. Biol. Chem. 272, 20070–20076. M. J. (1998) Science 281, 1668–1674. 19. Horvath, C. M., Stark, G. R., Kerr, I. M. & Darnell, J. E. (1996) Mol. Cell. Biol. 41. Luttrell, L. M., Ferguson, S. S. G., Daaka, Y., Miller, W. E., Maudsley, S., IMMUNOLOGY 16, 6957–6964. Rocca, G. J. D., Lin, F.-T., Kawakatsu, H., Owada, K., Lutrell, D. K., et al. 20. Chung, C. D., Liao, J., Liu, B., Rao, X., Jay, P., Berta, P. & Shuai, K. (1997) (1999) Science 283, 655–660. Science 278, 1803–1805. 42. Therrien, M., Wong, A. M. & Rubin, G. M. (1998) Cell 95, 343–353. 21. Smith, T. F., Gaitatzes, C., Saxena, K. & Neer, E. J. (1999) Trends Biochem. Sci. 43. Schillace, R. V. & Scott, J. D. (1999) J. Clin. Invest. 103, 761–765. 24, 181–185. 44. Sekimoto, T., Imamoto, N., Nakajima, K., Tachibana, T., Hirano, T. & Yoneda, 22. Gupta, S., Yan, H., Wong, L. H., Ralph, S., Krolewski, J. & Schindler, C. (1996) Y. (1997) EMBO J. 16, 7076–7077. EMBO J. 15, 1075–1084. 45. Bhattacharya, S., Eckner, R., Grossman, S., Oldread, E., Arany, Z., D’Andrea, 23. Rothman, P., Kreider, B., Azam, M., Levy, D., Wegenka, U., Eilers, A., Decker, A. & Livingston, D. (1996) Nature (London) 383, 344–347.

Collum et al. PNAS ͉ August 29, 2000 ͉ vol. 97 ͉ no. 18 ͉ 10125 Downloaded by guest on October 1, 2021