Host Cell Interactome of Tyrosine-Phosphorylated Bacterial Proteins

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Cell Host & Microbe Resource

Matthias Selbach,1,2,* Florian Ernst Paul,2,6 Sabine Brandt,3,6 Patrick Guye,4,6 Oliver Daumke,2 Steffen Backert,5 Christoph Dehio,4 and Matthias Mann1,* 1Max-Planck-Institute of Biochemistry, Department Proteomics and Signal Transduction, Am Klopferspitz 18, D-82152 Martinsried, Germany 2Max Delbru¨ ck Center for Molecular Medicine, Robert-Ro¨ ssle-Str. 10, D-13092 Berlin, Germany 3Otto-von-Guericke-Universita¨ t Magdeburg, Institute for Medical Microbiology, Leipziger Strabe 44, D-39120 Magdeburg, Germany 4Biozentrum of the University of Basel, Focal Area Infection Biology, Klingelbergstrasse 70, CH-4056 Basel, Switzerland 5University College Dublin, School of Biomolecular and Biomedical Science, Ardmore House, Dublin-4, Ireland 6These authors contributed equally to this work *Correspondence: [email protected] (M.S.), [email protected] (M.M.) DOI 10.1016/j.chom.2009.03.004

SUMMARY et al., 1999). Once phosphorylated, these proteins can recruit cellular binding partners with SH2 domains to perturb host cell Selective interactions between tyrosine-phosphory- function (Backert and Selbach, 2005). However, only few interac- lated proteins and their cognate, SH2-domain con- tion partners have been identified so far. taining ligands play key roles in mammalian signal Systematic mapping of protein-protein interactions can transduction. Several bacterial pathogens use secre- provide valuable insights into biological systems (Kocher and tion systems to inject tyrosine kinase substrates into Superti-Furga, 2007; Stelzl and Wanker, 2006). However, current host cells. Upon phosphorylation, these effector large-scale screening methods fail to provide information about interactions regulated by posttranslational modifications. Quan- proteins recruit cellular binding partners to manipu- titative proteomics has recently emerged as a powerful late host cell functions. So far, only a few interaction approach to study protein-protein interactions (Gingras et al., partners have been identified. Here we report the 2007; Ong and Mann, 2005). Here, we employed quantitative results of a proteomic screen to systematically iden- proteomics to systematically identify host cell proteins interact- tify binding partners of all known tyrosine-phosphor- ing with bacterial effector proteins in a phosphorylation-depen- ylated bacterial effectors by high-resolution mass dant manner. spectrometry. We identified 39 host interactions, all mediated by SH2 domains, including four of the RESULTS five already known interaction partners. Individual phosphorylation sites recruited a surprisingly high SH2 domains recognize phosphotyrosine residues in the context number of cellular interaction partners suggesting of a short peptide motif (Schlessinger and Lemmon, 2003). Thus, that individual phosphorylation sites can interfere we reasoned that synthetic peptides corresponding to the pY with multiple cellular signaling pathways. Collec- sites in the bacterial effector proteins should be able to pull out specific interaction partners from lysates of host cells (Schulze tively, our results indicate that tyrosine-phosphoryla- et al., 2005; Schulze and Mann, 2004). To distinguish between tion sites of bacterial effector proteins have evolved specific interaction partners and nonspecific contaminants, as versatile interaction modules that can recruit a a pervasive problem in protein interaction mapping, we em- rich repertoire of cellular SH2 domains. ployed a quantitative proteomic screen (Figure 1). Cells are metabolically labeled by cultivating them in growth medium INTRODUCTION that contains heavy stable isotopes of lysine and arginine (SILAC, stable isotope labeling with amino acids in cell culture) Protein-protein interactions play a central role for molecular path- (Mann, 2006; Ong et al., 2002). Lysate from the heavy-labeled ogenesis of infectious diseases (Bhavsar et al., 2007; Pizarro- cells is combined with the tyrosine-phosphorylated peptide, Cerda and Cossart, 2006). Several bacterial pathogens use type while lysate from the light-labeled cells is incubated with the non- III and type IV secretion systems to inject effector proteins into phosphorylated peptide of the same amino acid sequence. the cytoplasm of host cells (Cascales and Christie, 2003; Galan Specific interaction partners of the tyrosine-phosphorylated and Wolf-Watz, 2006). After translocation, these effectors target peptide are more abundant in the heavy form, while all nonspe- various components of eukaryotic signal transduction pathways, cific contaminants should have a 1:1 ratio. which subvert host cell functions for the benefit of the pathogen. We evaluated the screen with the translocated intimin receptor Intriguingly, four bacterial pathogens, enteropathogenic Escheri- (Tir) of EPEC. Tir is phosphorylated on tyrosine 474 and recruits the chia coli (EPEC), Helicobacter pylori, Chlamydia trachomatis, and SH2 domain of the adaptor protein Nck, which induces actin poly- Bartonella henselae, inject effector proteins, which are phosphor- merization at the site of intimate adhesion (Gruenheid et al., 2001; ylated on tyrosine residues by kinases of the host cell (Clifton Phillips et al., 2004; Swimm et al., 2004). We synthesized a peptide et al., 2004; Kenny et al., 1997; Schulein et al., 2005; Segal corresponding to this phosphorylation site as a 15-mer with

Cell Host & Microbe 5, 397–403, April 23, 2009 ª2009 Elsevier Inc. 397 Cell Host & Microbe Interactome of Tyrosine-Phosphorylated Effectors

Figure 1. A Proteomic Screening for Inter- action Partners of Tyrosine-Phosphorylated Bacterial Effectors (A) Several bacterial pathogens inject proteins into mammalian cells where they are phosphorylated on tyrosine residues by a host cell kinase and recruit cellular interaction partners. (B) To identify the cellular binding partners, cells are labeled by growing them in the presence of light or heavy amino acids (SILAC). Peptides corresponding to the bacterial tyrosine phosphorylation sites are synthesized in a phosphorylated and nonphosphorylated form. These peptides are used to pull down interaction partners from whole-cell lysates of the labeled cells. Samples are combined and analyzed by quantitative proteomics (LC-MS/MS). The schematic mass spec- trum shows how interaction partners are detected by their higher abundance in the heavy form. (C) A peptide derived from the known interaction partner Nck is about ten times more abundant in the heavy form than in the light form, indicating that Nck binds specifically to Tir phosphorylated on Y474. (D) In a crossover experiment (swapped SILAC labels) the abundance ratio of the same peptide is inverted. (E) Most other proteins have a 1:1 ratio, indicating that they are unspecific binders to the peptides or the magnetic beads used in the pull-down. (F) A plot of the abundance ratios shows that Nck is a significant outlier in both the Y474 pull-down and the crossover experiment. Every dot desig- nates one protein.

demonstrate that peptide pull-downs combined with quantitative proteomics can confidently identify cellular interaction partners of tyrosine-phosphorylated effector proteins. Next, we applied the screen to all known tyrosine-phosphorylated bacterial effectors from different pathogens. All effectors have more than one known (Tir, CagA) or predicted (Tarp, BepD-F) phos- tyrosine in the central position and immobilized it on magnetic phorylation site yielding a total number of 12 unique sites. The beads. Phosphorylated and nonphosphorylated peptides amino acid sequences of the phosphorylation sites were derived were incubated with whole-cell lysates from heavy and light from well-characterized model strains of the four pathogens SILAC-labeled HeLa cells to pull out interaction partners (Figures (Table S1). In order to cover a range of host cells, we performed 1C–1F) and high-resolution MS analysis identified more than experiments with two cell lines, HeLa (human cervix epithelium) 100 proteins, with very high confidence indicated by the and AGS (human gastric epithelium). In total, we performed 50 absence of protein hits in a reversed sequence database (see interaction screens yielding more than 1,000,000 tandem mass Experimental Procedures). The vast majority of these proteins spectra. Altogether, we identified 39 interaction partners for had SILAC-abundance ratios of 1:1, indicating that they bind the 12 pY sites (Figure 2 and Table S1). From the five interactions to the peptides or the beads in a nonspecific manner. In reported in the literature (Gruenheid et al., 2001; Higashi et al., contrast, the known interaction partner Nck was 13 times 2002; Mimuro et al., 2002; Tsutsumi et al., 2003) we detected more abundant in the heavy form than in the light form, demon- four. We did not identify Crk as a binding partner of CagA, prob- strating that it binds specifically to the phosphorylated tyrosine ably because Crk interacts only weakly with CagA from the residue 474 of Tir. As a further control we also performed H. pylori strain 26695 used in this study (Suzuki et al., 2005)or a crossover experiment with swapped SILAC labels, which re- interacts indirectly via the Abl kinase (Tammer et al., 2007). Collec- sulted in reciprocal abundance ratios for Nck. These results tively, the results suggest that our screen covers a substantial

398 Cell Host & Microbe 5, 397–403, April 23, 2009 ª2009 Elsevier Inc. Cell Host & Microbe Interactome of Tyrosine-Phosphorylated Effectors

Figure 2. Summary of the Identified Interaction Partners Previously known interactions are represented in black and white; interactions identified by this study are colored. Possible downstream signaling events elicited by the interactions are indicated (see text for details). fraction of the interactome of tyrosine-phosphorylated bacterial (Figure 3C). While both wild-type BepE and BepE-Y37F are tyro- effectors. As most of the interaction partners were not previously sine-phosphorylated in HEK293T cells, only wild-type BepE known, the data greatly extend the information about tyrosine- coimmunoprecipitated with Csk, demonstrating specific interac- phosphorylated effector proteins and should provide a valuable tion with this site. c-Src overexpression enhanced BepE tyro- resource for the scientific community. Several of the interactions sine-phosphorylation, suggesting that BepE is a c-Src substrate immediately suggest functional implications for host pathogen (Figure 3D). Importantly, c-Src overexpression also increased interaction (Figure 2). binding of BepE to Csk and Shp-2. Hence, our in vitro screen de- To test whether the newly identified proteins are relevant tected interactions that directly depend on the phosphorylation during infection we selected a few interaction partners for exper- state of the effector in vivo. imental validation. We infected gastric epithelial cells (AGS) with An intriguing result of our study was that several of the bacte- EPEC strains expressing tagged Tir. PI3K specifically coprecipi- rial phosphorylation sites recruit a surprisingly high number of tated with wt Tir, demonstrating that PI3K and Tir interact in vivo cellular interaction partners. This observation is not an artifact (Figure 3A). Mutagenesis of the major tyrosine-phosphorylation of our peptide-based screen because the same method revealed site Y474 did not disrupt this interaction. In contrast, when we greater specificity in the case of endogenous tyrosine-phosphor- mutated the recently identified second phosphorylation site ylation sites (Schulze et al., 2005). To compare how bacterial and Y454 (Campellone and Leong, 2005) Tir was unable to recruit endogenous tyrosine-phosphorylation sites interact with SH2 PI3K. These results are in excellent agreement with our proteo- domains, we measured their binding affinities. We selected mic screen, which detected specific binding of PI3K to the Tir SH2 domains of PI3K and Csk as these proteins interacted Y454 phosphorylation site. Interaction partners of CagA were with many bacterial phosphorylation sites. We expressed and validated by AGS cell infection. PI3K, Shp-1 and Grb-7 specifi- purified both SH2 domains, combined them with phosphopepti- cally coimmunoprecipitated with CagA, demonstrating that all des, and measured binding energy by isothermal titration calo- three proteins are indeed cellular interaction partners rimetry (Figure 4). The SH2 domain of PI3K bound with high

(Figure 3B). Association of BepE phosphorylated on tyrosine afﬁnity (Kd = 11.4 ± 2.9 nM) to a phosphopeptide derived from 37 with Csk was conﬁrmed by expressing FLAG-tagged BepE insulin receptor substrate 1 (IRS-1), a known interaction partner

Cell Host & Microbe 5, 397–403, April 23, 2009 ª2009 Elsevier Inc. 399 Cell Host & Microbe Interactome of Tyrosine-Phosphorylated Effectors

Figure 3. Validation of the Interactions (A) Infection of AGS cells with EPEC expressing HA-tagged Tir shows that PI3K interacts with Tir in a Y454-dependent manner. (B) Infection of AGS cells with wild-type (WT) H. pylori or an isogenic cagA knockout mutant conﬁrms interaction of CagA with PI3K, Shp-1, and Grb-7. (C) Transfection of FLAG-tagged wild-type (WT) BepE or BepE Y37F in HEK293T cells shows that the interaction of Csk with BepE depends on Y37. Empty vector was used as control. (D) Cotransfection of FLAG-BepE and c-Src demonstrates that BepE binds Csk and Shp-2 in a phosphorylation-dependent manner. Antibody heavy chains (IgG hc) are indicated (C and D).

DISCUSSION

The cellular targets identified here have direct implications for the molecular pathogenesis of infections. For example, as EPEC Tir is a substrate of redundant kinases including the Src family (Bom- marius et al., 2007; Phillips et al., 2004), the interaction of Src with phosphorylated Y474 could stimulate further Tir phosphorylation through a positive feed- back mechanism. Tir is also phosphorylated on Y454, although to a lesser extent. Importantly, Y454 phosphorylation leads to recruitment of N-WASP independently of Nck (Brady et al., 2007). Our finding that two central modulators of the actin cytoskeleton, Cbl and PI3K, interact with Y454 provides a possible mechanism how this could be achieved. H. pylori injects CagA into gastric epithelial cells, where it is phosphorylated by Src and Abl tyrosine kinases (Poppe et al., 2007; Selbach et al., 2002; Stein et al., 2002; Tammer et al., 2007). Phosphory- of this SH2 domain (Songyang et al., 1993). In contrast, the IRS-1 lated CagA has been reported to interact with the SH2 domains phosphopeptide bound with 1000-fold reduced affinity to the of Shp-2, Csk, Grb2, and Crk (Higashi et al., 2002; Mimuro et al.,

SH2 domain of Csk (Kd = 12.5 ± 2.2 mM). Conversely, a phospho- 2002; Suzuki et al., 2005; Tsutsumi et al., 2003). We identified peptide derived from Csk binding protein (Cbp), which is known PI3K, Shp-1, Ras-GAP1, and Grb7 as additional cellular interac- to bind Csk (Kawabuchi et al., 2000), specifically interacted with tion partners providing insights into how CagA could induce this SH2 domain (Kd = 1.15 ± 0.05 mM) but showed a greatly rearrangements of the actin cytoskeleton and promote inflam- reduced affinity to the SH2 domain of PI3K (Kd = 22 ± 0.7 mM). mation. CagA was recently shown to block apoptosis via the These results highlight the specificity of cellular phosphotyrosine CagA / Grb2 / Ras / Raf / Mek / Erk pathway, which signal transduction cascades. In contrast, the bacterial phos- dampens gut epithelial self-renewal (Mimuro et al., 2007), and phopeptide CagA-Y918 interacted with the SH2 domains of PI3K is well-known for its antiapoptotic effects (Franke et al., both PI3K (Kd = 350 ± 70 nM) and Csk (1.12 ± 0.07 mM). Csk 1997). BepD, BepE, and BepF effector proteins of Bartonella bound phosphorylated CagA-Y918 and its endogenous binding henselae are thought to be important in invasion, inhibition of partner Cbp with very similar affinity. While PI3K preferred the apoptosis, cytokine release, and cytotoxic effects. Consistent endogenous interaction partner IRS-1, phosphorylated CagA- with these diverse roles, we identified seven different cellular Y918 still interacted with PI3K remarkably well. Thus, a single interaction partners of the Beps, indicating that these proteins bacterial phosphopeptide can interact with different SH2 can interfere with cellular signaling in multiple and redundant domains with considerable affinity. ways. Finally, the Tarp of Chlamydia trachomatis appears to

400 Cell Host & Microbe 5, 397–403, April 23, 2009 ª2009 Elsevier Inc. Cell Host & Microbe Interactome of Tyrosine-Phosphorylated Effectors

Figure 4. Interaction of SH2 Domains with Endoge- nous and Bacterial Phosphopeptides (A and B) Binding of the SH2 domain of Csk (A) and the C-terminal SH2 domain of PI3K (B) to phosphopeptides was analyzed by ITC. Both SH2 domains selectively interacted with their cognate endogenous interaction partners (Cbp- pY317 and IRS-1-pY989). Cross-reactivity with the respective other endogenous phosphopeptide was negligible. In contrast, the CagA-derived phosphopeptide (CagA-pY918) bound to both SH2 domains with nanomolar to low micro- molar afﬁnity. Nonphosphorylated CagA (CagA-Y918) served as a negative control.

of SH2 domains. While a few sites are rather specific for a limited set of SH2 domains (e.g., BepF-Y149 and BepF-Y213), most have a rather broad specificity. Multiple sequence alignments of all effector proteins reveal remarkable similarity of the phosphorylation sites with a consensus motif EPxYAxV (Figure S1A). Except BepF and Tir-Y474, all of the bacterial phosphorylation sites were more similar to other bacterial phosphorylation sites than to any of the optimal binding motifs for the SH2 domains. This is surprising, as we expected the phosphorylation sites of bacterial effectors to be evolutionarily optimized to be most similar to the optimal recognition motifs of the SH2 domains that they interact with. It is tempting to speculate that bacterial phosphorylation sites evolved as promiscuous ‘‘master keys’’ that can recruit a wide range of cellular proteins. If so, such sequences would only be evolved by exogenous organisms and selected against by host cell proteins, as they would interfere with specificity of cellular signaling pathways. Indeed, mammalian proteomes are significantly depleted of this sequence motif (Figure S2). Collectively, our results indicate that tyrosine- phosphorylation sites of bacterial effector proteins have evolved as versatile interaction modules that can recruit a surprisingly rich repertoire of cellular play a role in cell invasion (Clifton et al., 2004; Lane et al., 2008). SH2 domains. While our assay is very sensitive, we cannot The identified interaction partners indicate possible mechanisms exclude that we missed some interaction partners for technical of host cell entry: Cbl is a ubiquitin ligase with a central role in reasons. For example, some relevant in vivo interaction partners endocytosis of activated growth-factor receptors. Binding of may not be expressed in the two cell lines we used. Hence, the Cbl to tyrosine-phosphorylated growth factor receptors medi- number of interaction partners of individual bacterial phosphor- ates internalization of the receptors into endocytic vesicles. ylation sites may be even larger. This promiscuity may allow the C. trachomatis might hijack this machinery to force its own pathogens to manipulate several signaling cascades in parallel uptake into host cells. The interaction partner PI3K implies a and thus to gain greater control of the host cell. Targeting of role for PIP3 during cell invasion, which agrees well with recent multiple pathways in parallel appears to be a common strategy results (Lane et al., 2008). of bacterial pathogens (Bhavsar et al., 2007). As new tyrosine- The interaction between SH2 domains and the cognate phosphorylated bacterial effector proteins are discovered, they pY-containing linear peptide sequences in their cellular binding can rapidly be assessed for interactions using the screen partners is generally considered to proceed in a modular, lock- described here. For example, the method is ideally suited to and-key fashion (Seet et al., 2006). Specificity is important, as identify interaction partners of AnkA from Anaplasma phagocyto- binding to other SH2 domains would activate the inappropriate philum and LspA from Haemophilus ducreyi (Deng et al., 2008; signaling pathway in response to stimulus. We found that indi- Lin et al., 2007). Similar methods can be used to address interac- vidual bacterial pY sites can bind to a surprisingly diverse set tions of full-length bacterial proteins with host cells (Selbach and

Cell Host & Microbe 5, 397–403, April 23, 2009 ª2009 Elsevier Inc. 401 Cell Host & Microbe Interactome of Tyrosine-Phosphorylated Effectors

Mann, 2006). As shown here, these host-pathogen interactomes by Not and BamHI, and ligated into the pFLAGCMV2 plasmid (Schulein et al., will suggest a large number of testable hypotheses for bacterial 2005). To generate the Y37F mutant we performed site-directed mutagenesis pathogenesis and intervention. by megaprime PCR. FLAG-BepE constructs were transfected into HEK293T cells for 36 hr, harvested, immunoprecipitated with anti-FLAG antibodies, and analyzed by western blotting. EXPERIMENTAL PROCEDURES ITC Measurements Pull-Down Experiments Expression constructs encoding GST fusion proteins with the SH2 domains of HeLa cells (CCL-2) and AGS cells (CRL-1739) were cultivated in lysine and Csk or the C-terminal SH2 domain of PI3K were kindly provided by Luis Serrano arginine-free DMEM and RPMI, respectively, supplemented with 10% dialyzed (Centre de Regulacio Genomica; Barcelona, Spain). Proteins were expressed fetal bovine serum (GIBCO). ‘‘Heavy’’ and ‘‘light’’ RPMI for SILAC was in E. coli, SH2 domains were cleaved off, purified and dialyzed against ITC 13 15 13 15 prepared by adding 84 mg/l C6 N4 L-arginine and 40 mg/l C6 N2 L-lysine buffer (100 mM HEPES pH 7.5, 100 mM NaCl, 2 mM DTT). Isothermal titration (Sigma Isotec) or the corresponding nonlabeled amino acids, respectively. calorimetry (ITC) was performed on a Microcal VP-ITC instrument (MicroCal, Desthiobiotinylated peptides were synthesized as described (Schulze et al., Northampton, MA). We used three purified phosphopeptides: IRS-1 (amino 2005). Cells were lysed in 1% (v/v) Nonident P-40, 150 mM sodium chloride, acids 985–994, Acetyl-SRGDpYMTMQ-NH2), Cbp (amino acids 313–321, 50 mM Tris-HCl (pH 7.5), protease inhibitors (Complete tablets, Roche), and Acetyl-ISAMpYSSVN-NH2) and CagA-Y918 (Acetyl-EEPIpYTQVA-NH2). 1 mM sodium orthovanadate. Cleared lysates were incubated with immobi- Control experiments were performed with unphosphorylated CagA-Y918. All lized peptides for 6 hr at 4 C using 1–2 mg of total protein per pull-down. In peptides were purified by HPLC and dissolved in ITC buffer. At a concentration order to facilitate peptide pair recognition during quantitation with MSQuant, of 400 mM peptides were titrated from a syringe into the sample cell containing we added 5% of the light cell lysate to the heavy lysate and vice versa in a30mM solution of the respective SH2 domain. Experiments were performed at some experiments. After washing in lysis buffer the corresponding beads (tyro- an equilibrium temperature of 20C with 50 injections of a volume of 5 ml each. sine-phosphorylated peptide and control peptide) were combined, washed Peaks of the obtained data were integrated and further analyzed by ORIGIN three more times, and eluted with 200 ml of 20 mM biotin for 30 min at room software supplied by Microcal. temperature. Eluates were precipitated with ethanol (final concentration 70% EtOH in 100 mM sodium acetate [pH 5.0]). SUPPLEMENTAL DATA

Mass Spectrometry Supplemental Data include one table, two figures, and Supplemental Refer- Protein pellets were solubilized in 6 M urea/2 M thiourea in 10 mM HEPES (pH ences and can be found online at http://www.cell.com/cellhostandmicrobe/ 8.0), reduced with 1 mg DTT, alkylated with 5 mg iodoacetamide, and digested supplemental/S1931-3128(08)00095-X. with LysC (Wako) and trypsin (Promega) after dilution in 50 mM NH4HCO3. Peptide mixtures were analyzed by C18 reversed phase nanoflow HPLC on an ACKNOWLEDGMENTS Agilent 1100 LC system coupled to an LTQ-FT or LTQ-Orbitrap (Thermo Fisher). The mass spectrometers were operated in the data-dependent mode with We thank John Leong (University of Massachusetts) for providing us with a high-resolution full scan in the FT part of the instruments and 5 (Orbitrap) or HA-tagged Tir constructs and Luis Serrano (Centre de Regulacio Genomica, 3 (LTQ-FTICR) parallel MS/MS scans in the LTQ. In the LTQ-FTICR we Barcelona, Spain) for the SH2-domain GST fusion proteins. We would also recorded additional high mass-accuracy precursor SIM (Single Ion Monitoring) like to thank B. Blagoev, S. Hanke, B. Macek, J. Olsen, W. Schulze, and the scans in the FT. Automatic gain control target values were as follows: 5E6 (FT other members of the department Department of Proteomics and Signal Trans- full scan), 8E4 (FT SIM scan), 1E6 (Orbitrap full scan), and 5E3 (LTQ MS/MS duction in Martinsried and the Center of Experimental Bioinformatics in scan). Odense/Denmark for fruitful discussions and assistance. The work was sup- ported by the Interaction Proteome project of the European Union (M.M.), the Data Analysis Swiss National Science Foundation (3100A0-109925/1) and the Howard Raw MS data files were converted into peak lists and searched against the Hughes Medical Institute (55005501) (C.D.), and the Deutsche Forschungsge- human IPI database (http://www.ebi.ac.uk) on an in house Mascot search meinschaft (Ba1671/3-1 and Ba1671/8-1) (S.B.). engine (http://www.matrixscience.com). Carbamidomethylation was used as 13 15 fixed, oxidation of methionine, N-acetylation of the protein, C6 N4 arginine, Received: February 27, 2008 13 15 and C6 N2 lysine as variable modifications. We required full tryptic speci- Revised: December 15, 2008 ficity and allowed up to two missed cleavages with mass tolerance 5 ppm Accepted: March 4, 2009 for the parent ions and 0.5 Da for MS/MS peaks. Only proteins identified Published: April 22, 2009 with at least two peptides (Mascot score > 20) were considered. Quantitation and validation of protein hits was performed with MSQuant (http://msquant. REFERENCES sourceforge.net). Protein ratios were normalized by division through the median of all protein ratios. Interaction partners of a tyrosine-phosphorylation Backert, S., and Selbach, M. (2005). Tyrosine-phosphorylated bacterial site were identified by their significantly increased abundance ratio (3s above effector proteins: the enemies within. Trends Microbiol. 13, 476–484. the median corresponding to a confidence interval of more than 99.7%). Backert, S., Moese, S., Selbach, M., Brinkmann, V., and Meyer, T.F. (2001). Proteins were only validated as specific interaction partners when they were Phosphorylation of tyrosine 972 of the Helicobacter pylori CagA protein is also identified in crossover experiment (swapped SILAC labels) with a recip- essential for induction of a scattering phenotype in gastric epithelial cells. rocal ratio. Mol. Microbiol. 42, 631–644. Bhavsar, A.P., Guttman, J.A., and Finlay, B.B. (2007). Manipulation of host-cell Infection, Transfection, and Immunoprecipitation pathways by bacterial pathogens. Nature 449, 827–834. We used the EPEC tir knockout strain E2348-196 recomplemented with HA-tagged wild-type tir, TirY474F, or TirY454F constructs kindly provided by Bommarius, B., Maxwell, D., Swimm, A., Leung, S., Corbett, A., Bornmann, W., John Leong (Campellone and Leong, 2005). H. pylori infections were performed and Kalman, D. (2007). Enteropathogenic Escherichia coli Tir is an SH2/3 with strain P1DcagA recomplemented with cagA from the strain 26695 (Backert ligand that recruits and activates tyrosine kinases required for pedestal forma- et al., 2001). After infection of AGS cells for 6 hr cells were harvested in lysis tion. Mol. Microbiol. 63, 1748–1768. buffer, immunoprecipitated with anti-HA or anti-CagA antibodies, and analyzed Brady, M.J., Campellone, K.G., Ghildiyal, M., and Leong, J.M. (2007). Entero- by western blotting. BepE was amplified (primers ATAAGAATGCGGCCGC haemorrhagic and enteropathogenic Escherichia coli Tir proteins trigger GATGAAAAGAAATCAACCACCCC and CGGGATCCTTAGATGGCGAAAGC a common Nck-independent actin assembly pathway. Cell. Microbiol. 9, TATTGCC) from chromosomal DNA of B. henselae strain RSE247, cleaved 2242–2253.

402 Cell Host & Microbe 5, 397–403, April 23, 2009 ª2009 Elsevier Inc. Cell Host & Microbe Interactome of Tyrosine-Phosphorylated Effectors

Campellone, K.G., and Leong, J.M. (2005). Nck-independent actin assembly is SILAC, as a simple and accurate approach to expression proteomics. Mol. mediated by two phosphorylated tyrosines within enteropathogenic Escheri- Cell. Proteomics 1, 376–386. chia coli Tir. Mol. Microbiol. 56, 416–432. Phillips, N., Hayward, R.D., and Koronakis, V. (2004). Phosphorylation of the Cascales, E., and Christie, P.J. (2003). The versatile bacterial type IV secretion enteropathogenic E. coli receptor by the Src-family kinase c-Fyn triggers actin systems. Nat. Rev. Microbiol. 1, 137–149. pedestal formation. Nat. Cell Biol. 6, 618–625. Clifton,D.R.,Fields, K.A., Grieshaber, S.S., Dooley, C.A., Fischer, E.R., Mead, D.J., Pizarro-Cerda, J., and Cossart, P. (2006). Bacterial adhesion and entry into Carabeo, R.A., and Hackstadt, T. (2004). A chlamydial type III translocated protein host cells. Cell 124, 715–727. is tyrosine-phosphorylated at the site of entry and associated with recruitment of Poppe, M., Feller, S.M., Romer, G., and Wessler, S. (2007). Phosphorylation actin. Proc. Natl. Acad. Sci. USA 101, 10166–10171. of Helicobacter pylori CagA by c-Abl leads to cell motility. Oncogene 26, Deng, K., Mock, J.R., Greenberg, S., van Oers, N.S., and Hansen, E.J. (2008). 3462–3472. Haemophilus ducreyi LspA proteins are tyrosine phosphorylated by macro- Schlessinger, J., and Lemmon, M.A. (2003). SH2 and PTB domains in tyrosine phage-encoded protein tyrosine kinases. Infect. Immun. 76, 4692–4702. kinase signaling. Sci. STKE 2003, RE12. Franke, T.F., Kaplan, D.R., and Cantley, L.C. (1997). PI3K: downstream AKTion Schulein, R., Guye, P., Rhomberg, T.A., Schmid, M.C., Schroder, G., blocks apoptosis. Cell 88, 435–437. Vergunst, A.C., Carena, I., and Dehio, C. (2005). A bipartite signal mediates Galan, J.E., and Wolf-Watz, H. (2006). Protein delivery into eukaryotic cells by the transfer of type IV secretion substrates of Bartonella henselae into human type III secretion machines. Nature 444, 567–573. cells. Proc. Natl. Acad. Sci. USA 102, 856–861. Gingras, A.C., Gstaiger, M., Raught, B., and Aebersold, R. (2007). Analysis Schulze, W.X., and Mann, M. (2004). A novel proteomic screen for peptide- of protein complexes using mass spectrometry. Nat. Rev. Mol. Cell Biol. 8, protein interactions. J. Biol. Chem. 279, 10756–10764. 645–654. Schulze, W.X., Deng, L., and Mann, M. (2005). Phosphotyrosine interactome of Gruenheid, S., DeVinney, R., Bladt, F., Goosney, D., Gelkop, S., Gish, G.D., the ErbB-receptor kinase family. Mol. Syst. Biol. 1, 2005.0008. Pawson, T., and Finlay, B.B. (2001). Enteropathogenic E. coli Tir binds Nck to initiate actin pedestal formation in host cells. Nat. Cell Biol. 3, 856–859. Seet, B.T., Dikic, I., Zhou, M.M., and Pawson, T. (2006). Reading protein modi- ﬁcations with interaction domains. Nat. Rev. Mol. Cell Biol. 7, 473–483. Higashi, H., Tsutsumi, R., Muto, S., Sugiyama, T., Azuma, T., Asaka, M., and Hatakeyama, M. (2002). SHP-2 tyrosine phosphatase as an intracellular target Segal, E.D., Cha, J., Lo, J., Falkow, S., and Tompkins, L.S. (1999). Altered of Helicobacter pylori Cag. Protein Sci. 295, 683–686. states: involvement of phosphorylated CagA in the induction of host cellular growth changes by Helicobacter pylori. Proc. Natl. Acad. Sci. USA 96, Kawabuchi, M., Satomi, Y., Takao, T., Shimonishi, Y., Nada, S., Nagai, K., 14559–14564. Tarakhovsky, A., and Okada, M. (2000). Transmembrane phosphoprotein Cbp regulates the activities of Src-family tyrosine kinases. Nature 404, 999–1003. Selbach, M., and Mann, M. (2006). Protein interaction screening by quantitative immunoprecipitation combined with knockdown (QUICK). Nat. Methods Kenny, B., DeVinney, R., Stein, M., Reinscheid, D.J., Frey, E.A., and Finlay, B.B. 3, 981–983. (1997). Enteropathogenic E. coli (EPEC) transfers its receptor for intimate adherence into mammalian cells. Cell 91, 511–520. Selbach, M., Moese, S., Hauck, C.R., Meyer, T.F., and Backert, S. (2002). Src is the kinase of the Helicobacter pylori CagA protein in vitro and in vivo. J. Biol. Kocher, T., and Superti-Furga, G. (2007). Mass spectrometry-based functional Chem. 277, 6775–6778. proteomics: from molecular machines to protein networks. Nat. Methods 4, 807–815. Songyang, Z., Shoelson, S.E., Chaudhuri, M., Gish, G., Pawson, T., Haser, W.G., Lane, B.J., Mutchler, C., Al Khodor, S., Grieshaber, S.S., and Carabeo, R.A. King, F., Roberts, T., Ratnofsky, S., Lechleider, R.J., et al. (1993). SH2 domains (2008). Chlamydial entry involves TARP binding of guanine nucleotide recognize speciﬁc phosphopeptide sequences. Cell 72, 767–778. exchange factors. PLoS Pathog. 4, e1000014. Stein, M., Bagnoli, F., Halenbeck, R., Rappuoli, R., Fantl, W.J., and Covacci, A. Lin, M., den Dulk-Ras, A., Hooykaas, P.J., and Rikihisa, Y. (2007). Anaplasma (2002). c-Src/Lyn kinases activate Helicobacter pylori CagA through tyrosine phagocytophilum AnkA secreted by type IV secretion system is tyrosine phos- phosphorylation of the EPIYA motifs. Mol. Microbiol. 43, 971–980. phorylated by Abl-1 to facilitate infection. Cell. Microbiol. 9, 2644–2657. Stelzl, U., and Wanker, E.E. (2006). The value of high quality protein-protein Mann, M. (2006). Functional and quantitative proteomics using SILAC. Nat. interaction networks for systems biology. Curr. Opin. Chem. Biol. 10, 551–558. Rev. Mol. Cell Biol. 7, 952–958. Suzuki, M., Mimuro, H., Suzuki, T., Park, M., Yamamoto, T., and Sasakawa, C. Mimuro, H., Suzuki, T., Tanaka, J., Asahi, M., Haas, R., and Sasakawa, C. (2005). Interaction of CagA with Crk plays an important role in Helicobacter (2002). Grb2 is a key mediator of helicobacter pylori CagA protein activities. pylori-induced loss of gastric epithelial cell adhesion. J. Exp. Med. 202, Mol. Cell 10, 745–755. 1235–1247. Mimuro, H., Suzuki, T., Nagai, S., Rieder, G., Suzuki, M., Nagai, T., Fujita, Y., Swimm, A., Bommarius, B., Li, Y., Cheng,D., Reeves, P., Sherman,M., Veach,D., Nagamatsu, K., Ishijima, N., Koyasu, S., et al. (2007). Helicobacter pylori Bornmann, W., and Kalman, D. (2004). Enteropathogenic Escherichia coli use dampens gut epithelial self-renewal by inhibiting apoptosis, a bacterial redundanttyrosine kinases to form actin pedestals. Mol. Biol. Cell 15, 3520–3529. strategy to enhance colonization of the stomach. Cell Host Microbe 2, Tammer, I., Brandt, S., Hartig, R., Konig, W., and Backert, S. (2007). Activation 250–263. of Abl by Helicobacter pylori: a novel kinase for CagA and crucial mediator of Ong, S.E., and Mann, M. (2005). Mass spectrometry-based proteomics turns host cell scattering. Gastroenterology 132, 1309–1319. quantitative. Nat. Chem. Biol. 1, 252–262. Tsutsumi, R., Higashi, H., Higuchi, M., Okada, M., and Hatakeyama, M. (2003). Ong, S.E., Blagoev, B., Kratchmarova, I., Kristensen, D.B., Steen, H., Pandey, A., Attenuation of Helicobacter pylori CagA x SHP-2 signaling by interaction and Mann, M. (2002). Stable isotope labeling by amino acids in cell culture, between CagA and C-terminal Src kinase. J. Biol. Chem. 278, 3664–3670.

Cell Host & Microbe 5, 397–403, April 23, 2009 ª2009 Elsevier Inc. 403