Crystal Structure of a Guanine Nucleotide Exchange Factor Encoded by the Scrub Typhus Pathogen Orientia Tsutsugamushi

Crystal structure of a guanine nucleotide exchange factor encoded by the scrub typhus pathogen Orientia tsutsugamushi

Christopher Lima,1, Jason M. Berka,1, Alyssa Blaisea, Josie Birchera, Anthony J. Koleskea, Mark Hochstrassera,2, and Yong Xionga,2

aDepartment of Molecular Biophysics & Biochemistry, Yale University, New Haven, CT 06520

Edited by Peter J. Novick, University of California San Diego, La Jolla, CA, and approved October 7, 2020 (received for review September 3, 2020) Rho family GTPases regulate an array of cellular processes and are IpgB2 GEF protein activates RhoA and causes characteristic often modulated by pathogens to promote infection. Here, we membrane ruffles that are critical for Shigella invasion of the host identify a cryptic guanine nucleotide exchange factor (GEF) do- cell (7). Bacterial effector GEFs belong to either the WxxxE main in the OtDUB protein encoded by the pathogenic bacterium family (named for a conserved motif important for folding and Orientia tsutsugamushi. A proteomics-based OtDUB interaction structural integrity) or the SopE family (SopE, SopE2, and BopE). screen identified numerous potential host interactors, including These bacterial effectors share no sequence or structural homol- the Rho GTPases Rac1 and Cdc42. We discovered a domain in ogy to eukaryotic Rho GEFs, which predominantly belong to the OtDUB with Rac1/Cdc42 GEF activity (OtDUBGEF), with higher ac- Dbl homology (DH) family of GEFs that adopt a six-helix bundle tivity toward Rac1 in vitro. While this GEF bears no obvious se- with an elongated, kinked “chaise lounge” fold (8, 9). Rather, quence similarity to known GEFs, crystal structures of OtDUBGEF bacterial effector GEFs adopt a characteristic compact V-shaped alone (3.0 Å) and complexed with Rac1 (1.7 Å) reveal striking con- fold, yet activate the Rho GTPases via the same contact regions in vergent evolution, with a unique topology, on a V-shaped bacte- the GTPases that are crucial for nucleotide exchange by DH- rial GEF fold shared with other bacterial GEF domains. Structure- family GEFs (10). While substantial effort has been exerted in guided mutational analyses identified residues critical for activity detailing the molecular determinants of bacterial GEF activities and a mechanism for nucleotide displacement. Ectopic expression BIOCHEMISTRY of OtDUB activates Rac1 preferentially in cells, and expression of and specificities, no bacterial effector GEFs have been identified outside of the WxxxE or SopE-like families. the OtDUBGEF alone alters cell morphology. Cumulatively, this work reveals a bacterial GEF within the multifunctional OtDUB Recently, we identified and characterized a putative effector that co-opts host Rac1 signaling to induce changes in cytoskeletal protein, OtDUB, from the obligate intracellular bacterium that structure. causes scrub typhus, Orientia tsutsugamushi. Despite extensive characterization of the OtDUB deubiquitylase (DUB) domain Orientia tsutsugamushi | X-ray crystallography | guanine nucleotide (residues 1–259), the function of the extensive C-terminal region, exchange factor | scrub typhus | Rac1 encompassing more than 1,000 amino acids, remained elusive (11). Here, we report that OtDUB encodes a GEF domain, he Ras homologous (Rho) family of GTPases is part of the TRas superfamily of small G proteins. Rho family GTPases Significance are molecular switches that control intracellular actin dynamics and regulate a diverse array of cellular processes from cytoki- Scrub typhus is a neglected tropical disease caused by the nesis to cell migration and wound healing (1–3). These small bacterium Orientia tsutsugamushi. Although O. tsutsugamushi ∼21-kDa proteins are highly conserved in all eukaryotes, with is an emerging public health threat, its pathogenic mechanisms three founding family members that have been extensively remain markedly understudied. Bacterial pathogens subvert studied: Rac1, Cdc42, and RhoA. Each Rho family GTPase exerts host actin dynamics by encoding guanine nucleotide exchange specific effects on the actin cytoskeleton, and constitutive activa- factors (GEFs) as effector proteins, which activate cellular Rho tion of each protein leads to characteristic cellular phenotypes. GTPases. Here, we identify a GEF domain within an O. tsutsu- The signaling activity of a GTPase is controlled by its bound gamushi protein that activates the host GTPase Rac1. While the nucleotide. When GDP is bound, the GTPase is in the “inactive” overall shape of the GEF is similar to that of other bacterial state, and loading of a GTP promotes the “active” conformation effectors, the primary sequence, topology, and catalytic mecha- of the G protein. Interaction with downstream effector proteins nism are completely distinct, suggesting convergent evolution. and subsequent actin reorganization only occurs when the GTPase Our studies reveal a cryptic GEF domain encoded by O. tsutsu- is in the GTP-bound active state. The intrinsic nucleotide ex- gamushi and provide the groundwork to probe the role of cy- change (GDP to GTP) and hydrolysis (GTP to GDP) rates of Rho toskeletal modulation in this neglected pathogen. family GTPases alone are slow. Nucleotide exchange occurs on Author contributions: C.L., J.M.B., A.B., J.B., A.J.K., M.H., and Y.X. designed research; C.L., the order of 1.5 per hour (4, 5), and the intrinsic hydrolysis rate is J.M.B., A.B., and J.B. performed research; C.L., J.M.B., A.J.K., M.H., and Y.X. analyzed data; ∼0.15 per min (6). Rapid regulation of GTPases, therefore, is and C.L. and J.M.B. wrote the paper. controlled by two classes of proteins that either switch them “on” The authors declare no competing interest. or “off”: guanine nucleotide exchange factors (GEFs) promote the This article is a PNAS Direct Submission. dissociation of GDP and allow loading with GTP, and GTPase Published under the PNAS license. activating proteins (GAPs) accelerate the intrinsic GTP hydrolysis 1C.L. and J.M.B. contributed equally to this work. by the G protein. 2To whom correspondence may be addressed. Email: [email protected] or Bacterial pathogens such as certain species of Salmonella, [email protected]. Shigella and enteropathic Escherichia coli, encode and secrete This article contains supporting information online at https://www.pnas.org/lookup/suppl/ effector proteins that modulate small GTPases to benefit the doi:10.1073/pnas.2018163117/-/DCSupplemental. bacterium during infection. For instance, the Shigella flexineri

www.pnas.org/cgi/doi/10.1073/pnas.2018163117 PNAS Latest Articles | 1of11 Downloaded by guest on September 26, 2021 OtDUBGEF. Using biochemical, structural, and cellular methods, bound uniquely to GST-OtDUB275–1369 and not to GST- we demonstrate that OtDUBGEF predominantly activates Rac1 OtDUB675–1369 or the GST protein control (Fig. 1C and SI Ap- in vitro and in cell culture. While the primary sequence of pendix,TableS1). Notably, several small GTPases, such as Rac1 OtDUBGEF is unrelated to WxxxE or SopE GEFs, the OtDUBGEF and Cdc42, and proteins of related functions were significantly crystal structure reveals a similar V-shaped fold despite an entirely enriched in the GST-OtDUB275–1369–bound sample compared to different topological and helical arrangement, suggesting convergent the GST-OtDUB675–1369 and GST controls (Fig. 1D). We focused evolution. We further determined the OtDUBGEF:Rac1 complex on the small GTPases as there are numerous examples of path- crystal structure and demonstrate that OtDUBGEF interacts with ogens hijacking GTPase signaling (12, 13). Rac1 at key common loci in the GTPase. The complex structure also suggested a distinct mechanism for GDP displacement, unique Identification of GTPase-Binding and GEF Activity in OtDUB. We among all GEFs characterized to date. Our work reveals that O. carried out coimmunoprecipitation (co-IP) assays using lysates of tsutsugamushi has evolved a GEF domain that expands the molec- HeLa cells ectopically expressing several OtDUB fragments to ular repertoire of bacterial effectors and suggests a critical function verify the putative interactions identified by mass spectrometry. for OtDUB in regulating Rac1 to benefit the pathogen during Flag-tagged OtDUB fragments were immunoprecipitated fol- infection. lowing incubation of whole cell lysates, and bound proteins were resolved by sodium dodecyl sulphate-polyacrylamide gel elec- Results trophoresis (SDS-PAGE) and immunoblotted for Rac1, Cdc42, The OtDUB C-Terminal Segment Is Toxic in Yeast and Interacts with and RhoA. These co-IP experiments confirmed that full-length GTPases. The OtDUB N-terminal region contains an active DUB (FL) OtDUB and OtDUB275–1369 bound Rac1 and Cdc42 but and a high-affinity ubiquitin binding domain (UBD) within the not RhoA (Fig. 2A). Further truncations (OtDUB1–675 and first 259 residues (11). However, the remainder of the 1,369- OtDUB675–1369) abolished the interaction. To demonstrate that residue protein is devoid of any computationally predicted do- this association does not require other cellular components, we mains. To examine how the OtDUB might affect eukaryotic cells, performed direct binding assays between purified E. coli- we first generated a series of truncations (Fig. 1A) and expressed expressed recombinant OtDUB fragments and recombinant the proteins in the yeast Saccharomyces cerevisiae (Fig. 1B). Re- GST-tagged Rac1 or Cdc42. In agreement with the co-IP re- markably, expression of the full-length protein caused a complete sults with HeLa cell lysates, OtDUB275–1369-Flag bound to both block to growth even if the DUB was inactivated by mutation of Rac1 and Cdc42 in anti-Flag immunoprecipitation assays, the catalytic cysteine, C135A (Fig. 1B and SI Appendix,Fig.S1A). whereas OtDUB675–1369-Flag did not. The reciprocal GST Expression of a fragment excluding the N-terminal DUB and pulldowns produced analogous results (Fig. 2B). UBD domains (OtDUB275–1369) also exhibited severe toxicity in GAPs and GEFs are the most common binding partners for yeast, but toxicity was no longer observed with the shorter GTPases. Several observations suggested that the OtDUB frag- OtDUB675–1369. The OtDUB1–675 truncation caused a lesser but ment might harbor GEF activity. First, certain intracellular still substantial growth deficit that was partially DUB dependent. bacterial pathogens encode GEFs to subvert GTPase signaling These data suggest that a region within residues 275–1369 is the and alter actin networks during infection (14). Additionally, the principal source of toxicity when expressed in yeast, but in the preceding pulldowns required EDTA, a condition that promotes absence of this toxic domain, an additional DUB-dependent nucleotide unloading and enhances GEF:GTPase interactions (15). growth defect is uncovered. To test whether OtDUB has GEF activity, we measured the To determine if specific human host proteins bind the OtDUB rate of dissociation of a fluorescent GDP analog (BODIPY-GDP) “toxic region,” we isolated proteins from HeLa cell lysates that from Rac1 and Cdc42 (16). BODIPY-GDP only fluoresces specifically bound glutathione-S-transferease (GST)–tagged OtDUB strongly when bound to protein. In the absence of OtDUB, Rac1 fragments and identified them using liquid chromatography and and Cdc42 released very little BODIPY-GDP (Fig. 2C). In con- tandem mass spectrometry (LC-MS/MS) (SI Appendix,Fig.S1B and trast, OtDUB275–1369 (50 nM) greatly accelerated nucleotide dis- C and SI Materials and Methods). We identified several proteins that sociation from Rac1. The same fragment also promoted GDP

Fig. 1. OtDUB275–1369 is toxic in yeast and binds multiple proteins in HeLa cell lysates. (A) Cartoon diagram of various OtDUB fragments used for ectopic expression in yeast, mammalian cell cultures, and bacteria. (B) Growth of W303 yeast expressing various OtDUB fragments from the galactose-inducible promoter in p416GAL1. Yeast cultures were serially diluted in 10-fold steps and spotted on SD lacking uracil containing either galactose or glucose asthe

carbon source and grown for 3 d at 30 °C. (C) Venn diagram of total proteins identified between GST, OtDUB275–1369, and OtDUB675–1369.(D) Candidate interactors of indicated OtDUB fragments from LC-MS/MS. Total peptide counts are shown.

2of11 | www.pnas.org/cgi/doi/10.1073/pnas.2018163117 Lim et al. Downloaded by guest on September 26, 2021 A B 275-1369FLAG 675-1369FLAG

Rac1,2,3

OtDUB Cdc42 GST-Cdc42 GST-Rac1

RhoA PD: FLAG GST IP: Flag

OtDUB FLAG GST-Cdc42 GST-Rac1 PD: GST

GST Rac1,2,3

C OtDUB275-1369 Cdc42 1.0 Rac1 / 0 nM GEF RhoA 0.8 Cdc42 / 0 nM GEF 0.6 Inputs (7%) 0.4 FLAG 0.2 Rac1 / 50 nM GEF Relative GDP dissociation GDP Relative 0.0 Cdc42 / 250 nM GEF 0 500 1000 1500 Time (s)

Fig. 2. OtDUB binds Rac1 and Cdc42 and catalyzes nucleotide exchange in vitro. (A) Inputs and anti-Flag immunoprecipitates of lysates from HeLa cells ectopically expressing the indicated Flag-tagged OtDUB fragments. Proteins were resolved by SDS-PAGE and immunoblotted for Rho GTPases (Rac1,2,3; Cdc42; RhoA). (B) FLAG (Upper) and reciprocal GST (Lower) pulldown experiments between purified recombinant FLAG-tagged OtDUB fragments and GST- tagged Rac1 or Cdc42. Proteins were resolved by SDS-PAGE and stained with Coomassie Blue. (C) Time course of the dissociation of BODIPY-GDP from Rac1 or BIOCHEMISTRY

Cdc42 in the presence of OtDUB275–1369 (0, 50, or 250 nM) as measured by loss of BODIPY-GDP fluorescence. Excitation and emission wavelength were 488 nm and 535 nm, respectively.

release from Cdc42, albeit less efficiently, requiring five times binds to and promotes nucleotide exchange on both Rac1 and more OtDUB (250 nM) to achieve similar levels of dissociation. In Cdc42. line with our initial binding studies, OtDUB275–1369 did not catalyze GDP dissociation from RhoA, and the shorter OtDUB675–1369 Determination of a Minimal, GTPase-Binding OtDUBGEF Domain. fragment displayed no exchange activity with Rac1 (SI Appendix, Bacterial GEFs in the WxxxE or SopE families exclusively target Fig. S1C). Together, these data indicate that OtDUB275–1369 Rho GTPases yet are divergent in both sequence and structure

A B C OtDUBGEF vs. Rac1 GST fragments OtDUBGEF 275-1369 1 2 3 4 5 1.0 275-913 Rac1 0 nM 366-913 482-913 mix 0.8 526-913 548-913 580-1369 0.6 1nM 548-808 548-759 released nucleotide 0.4 3 nM A260/280 ~ 1.48 10 nM Relative GDP dissociation 30 nM 0.2 0 500 1000 1500 Time (s)

Michaelis-Menten plot 1.0×10-2

1 2 3 4 5 . -3 5 -1 -1 7.5×10 kcat/KM = 2.6x10 M s ) -1 -3 (s 5.0×10 fragments obs GST-OtDUB k BOUND 2.5×10-3

OtDUBGEF 0.0 Rac1 Rac1 0102030 Rac1 (nM)

Fig. 3. A 200-residue OtDUB subdomain sufficient for binding Rac1. (A) OtDUB truncations used for mapping studies (Upper) and GST pulldown experiment using GST-OtDUB fragments as bait and Rac1 as prey (Lower). Gel was stained with Coomassie Blue. N-terminal truncation to residue 580 abolished binding,

whereas the 548–759 fragment (OtDUBGEF domain) retained full binding capacity. (B) SEC of OtDUBGEF:Rac1 mixtures demonstrates stable complex formation (Upper). Column fractions were evaluated by SDS-PAGE and protein staining (Lower). (C) Time course of the dissociation of BODIPY-GDP from Rac1 in the

presence of increasing amounts of OtDUBGEF (residues 548–759). Raw fluorescence curves were fit to a single exponential decay (Upper), and initial rates were 5 −1 −1 plotted against Rac1 concentration (Lower). Linear fit of the initial velocities yielded a kcat/KM of 2.6 ± 0.3 × 10 M ·s .

Lim et al. PNAS Latest Articles | 3of11 Downloaded by guest on September 26, 2021 from all eukaryotic Rho GEFs. Intriguingly, no region of the previously resolved bacterial effector GEFs is their shared “back- OtDUB protein sequence aligns to WxxxE or SopE sequences nor and-forth” topology of helices (14). In SopE and Map, for ex- to any eukaryotic GEF sequence. Hence, we employed a trunca- ample, α1/4/5 form the respective “left” lobes, and α2/3/6 form tion mapping strategy to further define the boundaries of a po- the “right” lobes of both proteins (Fig. 4 B, Center). By contrast, tential GEF domain in OtDUB. Guided by secondary structure OtDUBGEF splits sequentially into an N-terminal lobe (α1–3) predictions, we generated nine OtDUB constructs and tested and a C-terminal lobe (α4–9), with only a single “crossing” of the them for Rac1 binding in an in vitro GST pulldown assay polypeptide chain between lobes (Fig. 4B). The divergent to- (Fig. 3A). Consistent with our co-IP results, the ∼1,000-residue pology of OtDUBGEF explains why primary sequence alignment (OtDUB275–1369) fragment strongly associated with Rac1. Neither to the WxxxE or SopE GEFs was impossible prior to solving the C-terminal truncation (275–913) nor N-terminal truncations up to structure, despite having related functions. Notably, the “cata- residue 548 reduced binding to Rac1; however, N-terminal trun- lytic loops” present in the WxxxE and SopE families (Fig. 4B, cation to residue 580 abolished binding. Further truncation from bolded lines between α3 and α4) are required for nucleotide the C terminus revealed a minimal, ∼200-residue fragment (resi- exchange (10), but the crossover loop between α3 and α4of dues 548–759, hereafter referred to as OtDUBGEF)thatretained OtDUBGEF is not on the GTPase-interacting face of the enzyme full binding to Rac1. (see below) owing to its unrelated topology. The apo structure We employed size exclusion chromatography (SEC) to con- therefore suggests that OtDUBGEF has evolved a different firm that OtDUBGEF behaved well in solution and formed a structural motif to dissociate nucleotide (discussed in detail be- stable complex with Rac1. Indeed, free OtDUBGEF eluted as a low). The striking differences in topology and sequence for a monodisperse peak, and its mixture with Rac1 resulted in coe- common fold strongly indicates convergent evolution between lution of both proteins at the expected volume for a 1:1 complex OtDUBGEF and the SopE /WxxxE families of GEFs. (Fig. 3B). Furthermore, another new peak appeared in the mixture fractionation at an elution volume corresponding to that Crystal Structure of the OtDUBGEF:Rac1 Complex. To gain insight into of a small molecule; the high A260/280 ratio of this peak (∼1.5) the mechanism of Rac1 activation by OtDUBGEF, we determined suggested it represents released nucleotide that had copurified a crystal structure of the OtDUBGEF:Rac1 complex to 1.7-Å res- with Rac1. Similar SEC profiles were obtained for Cdc42 (SI olution (Fig. 5A and Table 1). OtDUBGEF does not undergo major Appendix, Figs. S2 A and B), demonstrating stable complex for- conformational changes upon binding to Rac1, with overall rmsds mation. We used isothermal titration calorimetry (ITC) to quan- ranging from 0.1 Å to 1.9 Å when compared to the six independent tify the binding affinity of OtDUBGEF toward Rac1 or Cdc42 (for copies of the molecule found in the asymmetric unit of the apo details, see SI Appendix, SI Materials and Methods). The apparent crystal (SI Appendix, Fig. S5B), suggesting a rigid structure evolved dissociation constant (Kd, app) between OtDUBGEF and Rac1 is ∼5 to be primed for interaction. Like other Rho GEFs, OtDUBGEF μM, similar to those of other GEF:GTPase pairs (SI Appendix, forms contacts with three key surfaces of the Rac1 GTPase: the Fig. S2C)(17–19). The measured Kd, app for Cdc42 is ∼16 μΜ.Of switch I and II regions, residues near the nucleotide-binding cleft note, WT GTPases with copurified nucleotides were used for ITC (the catalytic pocket), and the β2–3hairpin“interswitch” region, 2 analysis, meaning the measured Kd, app for Rac1 and Cdc42 is a burying a total surface area of more than 1,800 Å per molecule result of multiple processes (GEF:GTPase binding, nucleotide (Fig. 5A). Each interacting region is described in detail below. release and rebinding, and GTP hydrolysis); other factors such as The switch I interface (salmon in Fig. 5 B, Left) is composed of 564 607 subcellular localization or effector protein recruitment may also a hydrophobic pocket formed by OtDUBGEF-Phe /Val / dictate GTPase specificity. Nevertheless, the threefold difference Val612 that accommodates Rac1-Val36. This hydrophobic cleft is 568 in apparent affinities may partially explain the weaker activation of flanked on one side by OtDUBGEF-Arg interacting with the main 37 Cdc42 by OtDUBGEF. chain carbonyl of Rac1-Phe , and on the other by OtDUBGEF- 561 35 32 Finally, to ensure that the OtDUBGEF fragment fully accounts Glu forming a polar network with Rac1-Thr and Tyr .Atthe for catalytic activity, we tested various concentrations of this switch II interface (yellow in Fig. 5 B, Right), OtDUBGEF forms fragment for GEF activity against Rac1. In agreement with re- extensive electrostatic interactions, coordinating Rac1-Arg66 via 590 593 sults from the pulldown assay, OtDUBGEF catalyzed GDP dis- OtDUBGEF-Asn and Glu .Furthermore,OtDUBGEF resi- sociation from Rac1 at all concentrations tested (Fig. 3C). dues Ser594,Ser601, and Glu603 form charged-polar interactions Extraction of initial dissociation rates from this titration resulted with Rac1 residues Tyr64,Asp63, and the backbone of Glu62,re- 5 −1 −1 in a catalytic efficiency of kcat/KM = 2.6 × 10 M ·s , compa- spectively. Finally, near the nucleotide-binding pocket, a network 5 −1 −1 572 rable in magnitude to SopE acting on Rac1 (5.0 × 10 M ·s ) of ordered water molecules is coordinated by OtDUBGEF-Glu , (20). In contrast, OtDUBGEF was a relatively poor GEF against which interact with multiple Rac1 residues (Fig. 5 B, Center). Cdc42, with a catalytic efficiency ∼15-fold worse than that for To confirm the importance of these residues, we mutated Rac1 (SI Appendix, Fig. S2 D and E). Together, our in vitro OtDUBGEF residues at each interface and tested these mutants findings demonstrate that the ∼200-residue OtDUBGEF fragment for both binding and catalytic activities against Rac1. Compared is sufficient to promote nucleotide exchange in two different Rho to WT OtDUBGEF, both switch I (E561A) and switch II (N590A/ GTPases. The remainder of our studies focused solely on Rac1 E593A) mutants reduced apparent binding in the GST pulldown because of the clear preference of OtDUBGEF for this GTPase. assay, whereas little defect in catalysis was observed (Fig. 4C). By contrast, mutation of the central catalytic residue (E572A) ab- Crystal Structure of OtDUBGEF. All prokaryotic GEF effectors with rogated binding entirely and reduced GDP dissociation activity available structures diverge from the most common eukaryotic to background levels, highlighting the importance of this central Rho GEF family, the Dbl-homology (DH) GEFs, which typically residue for interaction with Rac1. This mutational analysis adopt an extended helical bundle, or “chaise lounge” fold (9). confirms that OtDUB harbors a bona fide GEF domain with a We crystallized OtDUBGEF (residues 548–759 of OtDUB) in its unique topology. apo form and determined the structure to 3.0-Å resolution The OtDUBGEF:Rac1 complex further revealed the structural (Table 1). The structure reveals that OtDUBGEF is composed basis for its selectivity among Rho-family GTPases. Previously, a exclusively of alpha helices with intervening linkers (Fig. 4A and “lock-and-key” mechanism for substrate selection has been SI Appendix, Fig. S5A) and adopts the familiar V-shaped fold proposed for discrimination between Rac1, Cdc42, and RhoA seen in WxxxE and SopE GEFs. Despite their structural simi- (21, 22). In this model, complementary pairing of amino acids larity at the level of the overall protein fold, the topology of generates favorable interactions for target GTPases in the β2–3 OtDUBGEF is very distinct. One conserved feature of all hairpin “selectivity patch,” and steric hindrance precludes interaction

4of11 | www.pnas.org/cgi/doi/10.1073/pnas.2018163117 Lim et al. Downloaded by guest on September 26, 2021 Table 1. Data collection and refinement statistics

OtDUBGEF apo OtDUBGEF:Rac1

Data collection

Space group P63 P1 Cell dimensions a, b, c, Å 110.53, 110.53, 251.24 50.06, 54.16, 94.00 α, β, γ, ° 90, 90, 120 83.37, 76.34, 62.52 Resolution, Å 3.0 (3.06–3.00)* 1.7 (1.73–1.70)

Rsym or Rmerge 0.069 (1.610) 0.033 (0.644) I/σI 19.7 (0.8) 21.0 (1.0) Completeness, % 99.6 (96.7) 81.7 (79.5)

CC1/2 0.997 (0.372) 0.996 (0.295) Redundancy 8.4 (7.3) 1.6 (1.6)

Refinement Resolution, Å 47.9–3.0 48.1–1.7 No. of reflections 37,727 (3,645) 91,794 (8,930)

Rwork/Rfree 0.19/0.24 (0.22/0.26) 0.18/0.22 (0.20/0.24) No. of atoms Protein 10,288 6,266 Ligand/ion 1 1 Water 1 313 B-factors, Å2 Protein 88.9 28.0 Ligand/ion 91.7 67.9 Water 77.1 38.0 rms deviations

Bond lengths, Å 0.016 0.018 BIOCHEMISTRY Bond angles, ° 2.0 2.0 Ramachandran statistics, % Favored 99.0 98.9 Allowed 1.0 1.1 Outlier 0.0 0.0

One crystal for each dataset was used for data collection and structure determination. *Statistics for the highest-resolution shell are shown in parentheses.

with nontarget family members. For instance, where Rac1 and and SopE GEFs extrude flexible “catalytic loops” between α3and Cdc42 encode small amino acids, Ala3 and Thr3, respectively, α4 that rest in a cleft formed between switch I and switch II of the RhoA places a bulky, charged Arg5 residue at the same position GTPase (SI Appendix,Fig.S4B, Center). These catalytic loops (SI Appendix, Fig. S3A). In agreement with the dual specificity share some sequence similarity, requiring an Ala followed by a observed in the in vitro GEF assay, OtDUBGEF residues are polar Asn/Gln residue to facilitate nucleotide exchange. Strikingly, compatible with both selectivity patches of Rac1 and Cdc42. In the unique topology of OtDUBGEF necessarily precludes the ex- contrast, the bulky, charged amino acids of RhoA (Arg5 and Glu54) istence of a homologous catalytic loop, suggesting that OtDUB- are predicted to clash with several rigid regions of OtDUBGEF (SI GEF utilizes a distinct catalytic mechanism. Appendix,Fig.S3B). “ ” We next analyzed whether OtDUBGEF makes similar contacts OtDUBGEF Utilizes a Carbonyl Catalytic Lever to Dissociate Nucleotide. with Rac1 when compared other bacterial GEFs and their cog- In all DH GEF:Rho GTPase structures, highly conserved and nate Rho-family GTPases: SopE and Cdc42 (Protein Data Bank characteristic conformational changes in switch II are responsible [PDB] ID code: 1GZS), Map and Cdc42 (PDB ID code: 3GCG), for dissociation of GDP. For instance, the canonical GEF-bound and IpgB2 and RhoA (PDB ID code: 3LW8, complex A). structure of the TIAM1:Rac1 complex (PDB ID code: 1FOE, Overall, the footprints of the bacterial effector GEFs and slate) reveals that the sidechains of Glu62 and Ala59 of Rac1 both OtDUBGEF on their GTPases are very similar (SI Appendix, Fig. rotate nearly 180° toward the nucleotide-binding cleft relative to S4). However, comparison of the interactions made by WxxxE their positions before GEF binding, i.e., in the GDP-bound state and SopE GEFs (hereafter referred to collectively as bacGEFs) (PDB ID code: 5N6O, gray) (Fig. 6A). This allows Rac1-Glu62 to 16 β and by OtDUBGEF revealed that the latter has distinct contacts interact with Rac1-Lys , competing for -phosphate binding, and with Rac1. Interactions with switch I are the most similar: Where Rac1-Ala59 displaces Mg2+ in a hydrophobic pocket (Fig. 6A). bacGEFs use an invariant acidic Asp residue to make backbone Identical conformations are observed in all other bacterial GEF 36 38 contacts with Val and Thr (Rac1/Cdc42 numbering), OtDUBGEF complex structures such as SopE:Cdc42 and Map:Cdc42, and in- positions Glu561 to interact with Rac1-Tyr32 and Thr35 (SI Ap- teractions with their catalytic loops facilitate this conformational pendix,Fig.S4B, Left). Switch II interactions are slightly more change (23, 24). Thus, all known Rho GEFs use a similar catalytic diverse in bacGEFs, with SopE and Map using a Gln residue to mechanism that drives the GTPase switch II region through a make polar contacts with Asp65, and IpgB2 positioning an Asp conserved set of main chain and side chain motions toward the 66 62 sidechain to interact with Arg of RhoA. Here, OtDUBGEF co- bound GDP. Accordingly, mutation of Rac1-Glu abrogates ordinates Rac1-Arg66 with Asn590/Glu593 and interacts with Tyr64 dissociation activity for all Rho GEFs tested to date, as well as for instead of Asp65 (SI Appendix,Fig.S4B, Right). Finally, WxxxE most Ras and Ran GEFs (25).

Lim et al. PNAS Latest Articles | 5of11 Downloaded by guest on September 26, 2021 Fig. 4. Crystal structure of the apo OtDUBGEF domain reveals a unique topology. (A) Overall structural comparison of apo OtDUBGEF and other Rho GEFs: Map (PDB ID code: 3GCG), SopE (PDB ID code: 1GZS), TIAM1 (PDB ID code: 1FOE). All structures shown with helices as cylinders and transparent surface. (B) To-

pology diagrams of the same GEFs color-ramped from N terminus (blue) to C terminus (red) to highlight the unique topology of OtDUBGEF. Helices are shown as rectangles, beta strands are shown as arrows, and loop regions are shown as lines.

62 Strikingly, the conformation of Rac1-Glu in our OtDUBGEF:Rac1 of the electrostatic surface potential of the catalytic lever of complex positions the sidechain away from the nucleotide pocket OtDUBGEF near the nucleotide-binding cleft of Rac1 revealed and instead resembles the GDP-bound conformation (Fig. 6A). significant negative charge formed not by side chains, but by three Our structure was obtained after nucleotide dissociation (Fig. 3C) main chain carbonyls pointed toward the nucleotide-binding and clearly shows no nucleotide present (SI Appendix,Fig.S5B), pocket (Fig. 6C). We propose that these polar carbonyl moieties suggesting that we captured the complex in a postexchange con- are inserted in close proximity to the bound nucleotide and re- formation despite the observed position of Rac1-Glu62.Thisdis- pulse the negatively charged phosphates of GDP, facilitating nu- crepancy led us to hypothesize that Rac1-Glu62 would be dispensable cleotide exchange. The conformations of the catalytic lever in the for GDP dissociation by OtDUBGEF.Indeed,OtDUBGEF-catalyzed OtDUBGEF apo and Rac1-complexed crystal structures are re- nucleotide dissociation was indistinguishable between WT and markably similar (SI Appendix,Fig.S5D,pairwiseCα rmsdlever = E62A Rac1 (Fig. 6A). There are exceptions to the dependence 0.04–0.25 Å when comparing the two independent copies of the on Glu62 (or its equivalent), as in the case of the GEF domain levers in the complex crystal with the six independent copies in the of Rabex-5 and its GTPase Rab21, where Rabex-5 supplies an apo crystal), suggesting the catalytic lever is rigid and primed for acidic residue in trans to catalyze exchange on Rab21 (26). direct interaction with Rac1. The attribution of catalysis activity to However, to our knowledge OtDUBGEF is the only known ex- the catalytic lever highlights a unique mechanism utilized by ample of a Rho-specific GEF whose catalytic mechanism does not OtDUBGEF to promote nucleotide dissociation. rely on Glu62. How does OtDUBGEF promote nucleotide exchange in the OtDUBGEF Specifically Activates Rac1 in Cells. To complement these absence of a canonical catalytic loop? In addition to the unusual structural and in vitro activity results, we examined the state of 62 conformation of Glu in our complex structure, we also observed Rac1 and Cdc42 in HeLa cells ectopically expressing OtDUBFL a short loop in OtDUBGEF (601SQEKGAVS608, hereafter referred or the GEF impaired mutant OtDUBE572A. We utilized a resin- to as the “catalytic lever”) that protrudes out of α2 near the ca- bound GST fusion of the p21-binding domain (PBD) of Pak1 to nonical catalytic loop position (Fig. 6B). The deep projection of specifically enrich the active forms of Cdc42 and Rac1 (27). ± the OtDUBGEF lever into the nucleotide-binding cleft led us to OtDUB-expressing cells exhibited a 2.7 0.3-fold increase in hypothesize that this loop may be critical for nucleotide exchange. activated Rac1 compared to control cells (Fig. 7A). Importantly, Deletion of the catalytic lever in OtDUBGEF completely abro- the E572A mutant, which abolished OtDUBGEF nucleotide ex- gated nucleotide exchange activity compared to wild type in vitro change activity toward Rac1, no longer raised active Rac1 levels (Fig. 6B). The compromised catalytic activity cannot be attributed in cells above baseline levels. Consistent with the lower in vitro to misfolding of the protein or to loss of binding Rac1 since activity of OtDUBGEF on Cdc42 (Fig. 2C and SI Appendix, Fig. the OtDUBGEF Δlever protein was monodisperse and continued S2E), there was no significant increase in the active Cdc42 when to form a complex with Rac1 (SI Appendix,Fig.S5C). Inspection OtDUB was ectopically expressed (1.1 ± 0.2-fold increase).

6of11 | www.pnas.org/cgi/doi/10.1073/pnas.2018163117 Lim et al. Downloaded by guest on September 26, 2021 BIOCHEMISTRY

Fig. 5. Biochemical function and structure of the OtDUBGEF:Rac1 complex. (A) Three orthogonal views of the complex with OtDUBGEF in purple and Rac1 in cyan. Switch I and II loops are in salmon and yellow, respectively, and the β2–3 hairpin selectivity interface is green. (B) Close-up views of each key interface

between OtDUBGEF and Rac1 showing selected residues, with hydrogen bonds and electrostatic interactions shown as dashes and water molecules as red spheres. (C) GST pulldown assays (Left) of GST-OtDUBGEF (WT or charge-neutralizing mutations at each interface) incubated with Rac1 and analyzed by SDS/ PAGE and Coomassie Blue staining. Corresponding BODIPY-GDP release assays are shown at Right.

These data both validate that the Rho GTPase GEF activity is Expression of OtDUBGEF leads to statistically significant cell contained within the OtDUBGEF domain and show a clear pref- morphology changes. Rac1 regulation of the actin cytoskeleton erence for Rac1 over Cdc42 in cells. promotes cell edge protrusion, and Rac1 activity is required for To examine possible downstream effects of OtDUBGEF-acti- cell spreading on adhesive surfaces in many cell types (28–30). vated Rac1 on the actin cytoskeleton, we turned to a model cell Cells expressing the WT OtDUBGEF-GFP were slightly smaller, line, mouse 3T3 fibroblasts. Ectopic expression of full-length less elongated, and had fewer filopodial extensions compared to OtDUB-GFP (WT or E572A) caused massive cell death and control cells expressing GFP alone (Fig. 7B). Quantification of GFP signal could not be detected. In contrast, expression of a cell shape metrics revealed that overall cell area, cell perimeter, and GFP fusion of the GEF domain alone was tolerated by fibro- major axis length decreased upon expression of the OtDUBGEF- blasts and expression was readily detected (Fig. 7B). Expression GFP (Fig. 7C). Critically, cells expressing the catalytically inactive of OtDUBGEF did not globally modulate filamentous actin OtDUBGEF-E572A-GFP mutant were not different from GFP structures, as revealed by phalloidin staining and confocal mi- control. Together, these data suggest that OtDUBGEF activity is croscopy (SI Appendix, Fig. S6A). responsible for the observed morphological phenotypes and that its

Lim et al. PNAS Latest Articles | 7of11 Downloaded by guest on September 26, 2021 Fig. 6. OtDUBGEF utilizes a unique carbonyl “catalytic lever” to catalyze nucleotide exchange. (A) Close-up views of switch II in GDP-bound Rac1 (gray, PDB ID code: 5N6O), three overlaid GEF-bound structures (slate is TIAM1:Rac1, PDB ID code: 1FOE; red is SopE:Cdc42, PDB ID code: 1GZS; and yellow is Map:Cdc42, 62 PDB ID code: 3GCG), and the OtDUBGEF-bound Rac1 structure (cyan). Unlike the three overlaid GEF-bound conformations, Glu in the OtDUBGEF-bound structure is most similar to that found in GDP-bound Rac1. GEFs have been removed for clarity, and switch I residues are shown as sticks. BODIPY-GDP ex- 62 change assay reveals that Glu is not required for efficient exchange catalyzed by OtDUBGEF.(B, Upper) A loop of OtDUBGEF that interrupts α2 acts as a “catalytic lever” to promote release of GDP. OtDUBGEF is shown as cartoon (purple), and Rac1 is shown as transparent surface (cyan). GDP (not present in our structure) is modeled based on PDB ID code 5N6O for reference. The steric clash and electrostatic repulsion are indicated by the red lines. (B, Lower) Nu- cleotide exchange assay reveals a strict requirement in the lever segment for exchange of BODIPY-GDP. (C) Calculated electrostatic surface potential map

(unit kBT/e) shows high negative charge in the OtDUBGEF lever formed by a tricarbonyl motif at the vicinity of the diphosphate group of GDP.

GEF activity potentially regulates the host cytoskeletal machinery to objective for bacterial intracellular pathogens, namely, the ma- produce distinct morphological defects. nipulation of host GTPases to promote the survival and proliferation of the pathogen. Sequence comparison of the OtDUBGEF Discussion domain against the nonredundant National Center for Biotech- The results presented here uncover an unexpected GEF activity nology Information protein database revealed clear conservation in the multidomain OtDUB protein from the pathogen O. tsut- only within the Orientia genus. Interestingly, a duplicated segment sugamushi and provide insight into our understanding of the (residues 283–499) within the OtDUB protein itself shares 48% regulation of Rho GTPases. The cryptic GEF domain harbors identity with the OtDUBGEF domain (residues 548–759; SI Ap- GEF activity toward both Rac1 and Cdc42, with a clear prefer- pendix,Fig.S7A). This duplicated region does not have Rho ence for Rac1 both in vitro and in living cells. High-resolution GTPase GEF activity nor does it bind Rho GTPases (Fig. 2A) crystal structures of the OtDUBGEF domain alone and in com- despite similarity at critical Rac1-interacting residues (SI Appen- plex with Rac1 reveal a GEF topology that nevertheless adopts a dix, Fig. S7A); however, we cannot rule out that this region does conserved V-shaped GEF fold. Importantly, OtDUBGEF utilizes not bind other small GTPases identified in our proteomics screen a mechanism to activate Rho GTPases involving direct, steric (Fig. 1E and SI Appendix,TableS1). Furthermore, the ortholog exclusion of GDP and distinct molecular determinants. from Orientia chuto (OcDUB, WP_052694629.1) contains three The unique topology of OtDUBGEF compared to all other regions with homology to OtDUBGEF: two regions (OcDUB426–641, bacterial effector GEFs represents a unique solution to a common 47% identity and OcDUB925–1137, 49% identity) that are mostly

8of11 | www.pnas.org/cgi/doi/10.1073/pnas.2018163117 Lim et al. Downloaded by guest on September 26, 2021 GFP-OtDUB GFP-OtDUB A B GEF GEF GFP (ctrl) WT E572A

Rac1

Cdc42 GST-Pak1BD (PonS)

OtDUB PD: GST 200 μm INPUT (2%) C

p = 0.0006

Fig. 7. OtDUBGEF specifically activates Rac1 and modulates cell morphology. (A) Lysates from HeLa cells carrying empty vector or plasmids expressing WT or E572A OtDUB were subjected to GST-Pak1PBD pulldowns to enrich active Rac1 and Cdc42. Representative experiment (Upper) from four independent experiments that were quantified (Lower) relative to input levels and the empty vector control. Bars represent mean and SD P value determined from a two- tailed unpaired Student t test. Outlier from Rac1 wild type (5.4-fold increase in activity) excluded one trial. Outlier was identified using Grubbs algorithm with BIOCHEMISTRY alpha = 0.05. (B) Representative epifluorescence images of fibroblasts carrying the vector expressing only GFP or plasmids expressing WT or E572A OtDUBGEF. (C) Quantification of cell area, perimeter, and major axis length.

conserved at Rac1-interacting sites and one segment (OcDUB665–883, fashion as the SopE and SopE2 GEFs from Salmonella, which 37% identity) that is more weakly conserved (SI Appendix,Fig.S7B). are secreted into the host cell cytoplasm to activate Rac1/Cdc42, Further experiments will be required to determine whether OcDUB inducing membrane ruffling and facilitating uptake of the bac- harbors a GEF activity similar to OtDUB. terium (33). Rather, it is tempting to speculate that the To our knowledge, the nucleotide exchange mechanism employed OtDUBGEF activates Rac1 after the bacterium escapes the vac- 62 by OtDUBGEF is unique among known GEFs. The role of Glu uolar compartment and accesses the cytoplasmic space, roughly has been studied extensively by Gasper et al. (25); whereas DH/ 2 h postinfection (34). One potential indicator of Rac1 activation Dbl Rho GEFs (e.g., Dbs, TIAM1, and p190) and Cdc25 domain- during postendosomal escape is the activation of the MAPK containing Ras GEFs (e.g., Sos and Rap) are absolutely depen- signaling pathway (35), a known downstream target of Rac1 dent on Glu62 for nucleotide exchange, Sec7-like Arf GEFs (e.g., (36–38). Additionally, IQGAP1, a downstream effector of Rac1/ Rabex-5 and Sec7) circumvent this requirement by supplying an Cdc42, was found enriched in the OtDUB275–1369 mass spec- acidic residue in trans (26, 31). Notably, OtDUBGEF exchange trometry analysis (SI Appendix, Table S1). IQGAP1 is a scaf- activity is independent of Glu62, and the structure does not im- folding protein with numerous binding partners and roles (39) plicate any nearby acidic residue from the GEF for interaction and is also required for efficient infection by Salmonella (40). with Lys16, in contrast to Sec7-like GEFs. Instead, the catalytic Cumulatively, the biochemical, bioinformatic, and structural lever of OtDUBGEF inserts directly into the nucleotide pocket, data presented here demonstrate that a recently evolved GEF do- revealing an unusual steric mechanism to promote GDP unload- main with a unique sequence, topology, and nucleotide-exchange ing. From the level of protein topology to the mechanistic details mechanism exists in the O. tsutsugamushi scrub typhus pathogen. of nucleotide exchange, OtDUBGEF is a fascinating example of The characterization of nearly half a dozen distinct and evolution- convergent evolution and represents a unique solution for inter- arily unrelated GEF families in eukaryotes indicates that numerous action with a highly conserved regulator of the host cytoskeleton, protein folds can support guanine nucleotide exchange activity. The Rac1. unique topology of OtDUBGEF further suggests that the sequence While it remains to be demonstrated that OtDUB is in fact a space of GEFs may be incompletely characterized. Finally, our work bona fide effector protein, the robust Rac1 activation in cultured suggests that OtDUB GEF activity may be crucial for O. tsutsuga- cells further bolsters this possibility. The combination of the DUB, mushi infection and opens the door for future investigations in this UBD, and GEF activities—all of which specifically interact with neglected pathogen. eukaryotic cellular signaling molecules that are absent from prokaryotic systems—strongly suggests that OtDUB is exposed to the Materials and Methods host cellular milieu. We note that ectopic expression of OtDUB Plasmid Construction and Protein Purifications. All plasmid constructs and with inactivated DUB, UBD, and GEF domains (OtDUB-C135A, proteins generated for in vitro assays and crystallization were expressed in BL21(DE3) E. coli cells and purified by affinity chromatography. Details can V203D, E572A) still causes complete growth inhibition in yeast, be found in SI Appendix, SI Materials and Methods. suggesting additional regions of unknown function in OtDUB that can also interfere with host physiology (SI Appendix,Fig.S8). Yeast Spot Assays. Yeast were grown under standard conditions at 30 °C using Due to a lack of early Rac1 activation during O. tsutsugamushi yeast rich (yeast extract-peptone-dextrose or YPD) or minimal (SD) media infection (32), it is unlikely that the OtDUBGEF acts in the same (41). Isolated transformants carried p416GAL1-based plasmids (42) in the

Lim et al. PNAS Latest Articles | 9of11 Downloaded by guest on September 26, 2021 W303 (MHY2416) (43) S. cerevisiae background was grown overnight in SD dithiothreitol (DTT), supplemented with 2 mM ethylenediaminetetraacetic lacking uracil (SD-URA) medium supplemented with casamino acids. Cultures acid (EDTA) to promote unloading of copurified nucleotide. Loading reac-

were diluted in sterile water to 0.2 OD600 from which 10-fold dilutions were tions were incubated for 1 h at 25 °C and then quenched by addition of made. Diluted cultures were spotted onto SD-URA plates supplemented with MgCl2 to a final concentration of 25 mM. Exchange reactions were initiated either 2% glucose or 2% galactose. Plates were incubated at the indicated by addition of a solution containing the OtDUB, 2 mM MgCl2, and 1 mM temperature for 3 d. Steady-state levels of each OtDUB polypeptide was verified GTP. Reactions were mixed rapidly, loaded into a multiwell plate, and real- by back diluting 2.5 OD units of SD-URA + 2% raffinose cultures into SD-URA + time fluorescence was collected on a plate reader for 30 min at 25 °C, using 2% galactose, growing for 4 h at 30 °C and processing 2.5 OD units by alkaline excitation at 488 nm and emission at 535 nm (16). lysis for subsequent Western immunoblot analysis using anti-Flag antibodies.

SEC. Each OtDUBGEF construct (75 μM) was mixed with the appropriate OtDUB:Rho GTPases co-IPs. HeLa cell pellets were collected 24 h after trans- GTPase (150 μM) or diluted alone in a 500-μL reaction and incubated for fection and resuspended in lysis buffer (50 mM Tris·HCl pH 7.5, 150 mM NaCl, 30 min at 4 °C. Potential aggregates were removed by pelleting at 21,000 × g 0.2% Triton X-100, 2 mM phenylmethanesulfonyl fluoride (PMSF), cOmplete for 5 min prior to loading. Protein mixtures were injected into a Superdex 75 protease inhibitor tablet [Roche]) and incubated on ice for 30 min with in- GL column (GE Healthcare) preequilibrated in 50 mM Tris pH 8.0, 100 mM termittent vortexing. The insoluble fraction was then pelleted by centri- NaCl, and 0.1 mM tris(2-carboxyethyl)phosphine (TCEP). Peak fractions were fuging at 21,000 × g for 15 min. Protein concentrations were determined by identified by monitoring absorbance at both 280 and 260 nm, resolved by the Bradford assay (Bio-Rad) and normalized to 1 mg/mL with lysis buffer. IPs SDS/PAGE, and visualized with Coomassie Blue staining. were carried out by rotating 850 μL of lysate with 15 μL of M2 anti-Flag resin

(Sigma) at 4 °C for 4 h. The resin was washed three times with 0.5 mL of lysis Crystallization and Structure Determination. Crystals of both apo OtDUBGEF μ buffer, followed by protein elution at 37 °C for 15 min with 50 L of triple- and OtDUBGEF:Rac1 complex were grown at 25 °C using the microbatch Flag peptide elution buffer (0.4 mg/mL of peptide in lysis buffer). The eluent under-oil method (45). Various seeding methods were used to generate was isolated with a Quik-Spin Column (Bio-Rad), mixed with Laemmli sample large, well-diffracting crystals, and crystals were cryo-protected by addition buffer, boiled for 3 min, and 25% was resolved by SDS/PAGE followed by of crystallization buffer supplemented with 25% (vol/vol) glycerol before immunoblotting. Detailed procedures for immunoblotting can be found in being flash frozen in liquid N2. Details on data collection and structure de- SI Appendix, SI Materials and Methods. termination can be found in SI Appendix, SI Materials and Methods.

Active Rac1/Cdc42 Enrichment. At 32 h posttransfection, the HeLa cell medium Confocal Microscopy. After spreading on coated coverslips for 24 h, mouse 3T3 was replaced with serum-free DMEM to reduce the basal levels of active (WT32) fibroblast cells were fixed in 2% paraformaldehyde in cytoskeleton Rac1/Cdc42 (44). At 48 h posttransfection, cells were harvested and resus- buffer (10 mM 2-(N-morpholino)ethanesulfonic acid [MES] pH 6.8, 138 mM pended in Pak1PBD buffer (50 mM Tris·HCl pH 7.5, 150 mM NaCl, 0.2% KCl, 3 mM MgCl2, 2 mM EGTA, 320 mM sucrose), rinsed in Tris-buffered Triton X-100, 5 mM MgCl2, 2 mM PMSF, cOmplete EDTA-free protease in- saline (TBS), permeabilized, blocked, and incubated with primary goat hibitor tablet [Roche]). Lysates were clarified, quantified, and normalized as anti-GFP antibody (Rockland 600-101-215) at 4 °C overnight. The next μ above. Purified GST-Pak1PBD (20 g) was added to 1 mL of 1.0 mg/mL lysates morning, cells were incubated with secondary Alexa Fluor 488 donkey anti- μ along with 15 L of glutathione resin (ThermoFisher) and rotated for 90 min goat antibody (Abcam ab150129) and phalloidin Atto 647N (Sigma-Aldrich at 4 °C. Resin was washed three times with 0.5 mL of Pak1PBD buffer and 65906). After drying, coverslips were sealed using clear nail polish and im- resuspended in Laemmli sample buffer. Samples were boiled for 3 min, and aged by epifluorescence on a microscope (model TE2000-S, Nikon) using a 50% of eluted proteins were resolved by SDS/PAGE followed by immuno- 20× objective, or with a 40× objective on a spinning disk confocal microscope blotting. Detailed procedures for immunoblotting can be found in SI Ap- (UltraVIEW VoX spinning disk confocal [Perkin-Elmer], Nikon Ti-E-Eclipse). pendix, SI Materials and Methods. CellProfiler (46) was used to detect and edit cell edges to compute all morphological metrics. For each condition, 36–58 cells were measured and an In Vitro Affinity Pulldown Assays. For pulldown assays between GST-Rac1/ ordinary one-way ANOVA with multiple comparisons was used to determine Cdc42 and OtDUB-Flag polypeptides, protein combinations were diluted to significant differences between groups. Further details can be found in SI 1 μM in binding buffer (50 mM Tris·HCl, pH 7.5, 150 mM NaCl, 0.01% Triton Appendix, SI Materials and Methods. X-100) to a final volume of 500 μL. After incubating at 37 °C for 30 min, the μ samples were spiked with 30- L glutathione resin or M2-Flag resin and ro- Data Availability. Coordinates and structure factors have been deposited tated for an additional 30 min at room temperature. Resin pellets were in the PDB (ID codes 6X1H [OtDUBGEF apo] and 6X1G [OtDUBGEF:Rac1 washed three times with binding buffer and eluted with Laemmli sample complex]). buffer or with 3xFlag peptide. For pulldown assays used to determine the minimal Rac1-interacting domain and validation point mutations in OtDUB, ACKNOWLEDGMENTS. We thank Titus Boggon (Yale University) for the Rac1 μ μ GST-tagged OtDUB fragments (20 M) were mixed with Rac1 (60 M) in and Cdc42 template DNA, Jonathan Chernoff (Fox Chase Cancer Center) for binding buffer (50 mM Tris·HCl pH 7.5, 150 mM NaCl, 0.1 mM TCEP) to a final the pGEXTK-Pak1 plasmid (Addgene #12217), and the W.M. Keck Founda- volume of 250 μL. Reactions were incubated with 200 μL of glutathione resin tion Biotechnology Resource Laboratory (Yale University) for mass spec- at 4 °C for 1 h. The beads were washed with 15 CV binding buffer and eluted trometry data collection and analysis. This work was supported by NIH with binding buffer supplemented with 10 mM reduced L-glutathione. Elu- Grants R01 AI116313 (to Y.X.); R01 GM046904 and R35 GM136325 (to M.H.); ates were boiled and resolved by SDS/PAGE and stained with Coomassie and R01 MH115939, R01 NS105640, and R21 NS112121 (to A.J.K.). Additional Blue. training support was provided by an NSF Fellowship DGE1122492 (to C.L.), NIH R01 MH115939-03S1 (to A.B.), and NIH F31 NS113511 (to J.B.). We are grateful to our host Igor Kourinov at the NE-CAT 24-ID-C beamline for BODIPY-GDP Nucleotide Exchange Assays. GTPases at a final concentration of diffraction data collection at the Advanced Photon Source of Argonne National 10 μM were loaded with substoichiometric (1/4 [GTPase]) concentrations of Laboratory. This work was supported by NE-CAT beamlines (P30 GM124165) at GDP-BODIPY-FL (Thermo Scientific) in 50 mM Tris pH 7.5, 50 mM NaCl, 1 mM the APS (DE-AC02-06CH11357).

1. K. Burridge, K. Wennerberg, Rho and Rac take center stage. Cell 116, 167–179 8. K. L. Rossman, C. J. Der, J. Sondek, GEF means go: Turning on RHO GTPases with (2004). guanine nucleotide-exchange factors. Nat. Rev. Mol. Cell Biol. 6,167–180 2. A. B. Jaffe, A. Hall, Rho GTPases: Biochemistry and biology. Annu. Rev. Cell Dev. Biol. (2005). 21, 247–269 (2005). 9. D. K. Worthylake, K. L. Rossman, J. Sondek, Crystal structure of Rac1 in complex with 3. I. R. Vetter, A. Wittinghofer, The guanine nucleotide-binding switch in three di- the guanine nucleotide exchange region of Tiam1. 408, 682–688 (2000). mensions. 294, 1299–1304 (2001). 10. R. Bulgin et al., Bacterial guanine nucleotide exchange factors SopE-like and WxxxE 4. M. J. Davis et al., RAC1P29S is a spontaneously activating cancer-associated GTPase. effectors. Infect. Immun. 78, 1417–1425 (2010). Proc. Natl. Acad. Sci. U.S.A. 110, 912–917 (2013). 11. J. M. Berk et al., A deubiquitylase with an unusually high-affinity ubiquitin-binding 5. B. Zhang, Y. Zhang, Z. Wang, Y. Zheng, The role of Mg2+ cofactor in the guanine domain from the scrub typhus pathogen Orientia tsutsugamushi. Nat. Commun. 11, nucleotide exchange and GTP hydrolysis reactions of Rho family GTP-binding pro- 2343 (2020). teins. J. Biol. Chem. 275, 25299–25307 (2000). 12. M. R. Popoff, Bacterial factors exploit eukaryotic Rho GTPase signaling cascades to 6. L. C. Haeusler, L. Blumenstein, P. Stege, R. Dvorsky, M. R. Ahmadian, Comparative promote invasion and proliferation within their host. Small GTPases 5, e983863 functional analysis of the Rac GTPases. FEBS Lett. 555, 556–560 (2003). (2014). 7. N. M. Alto et al., Identification of a bacterial type III effector family with G protein 13. M.-P. Stein, M. P. Müller, A. Wandinger-Ness, Bacterial pathogens commandeer Rab mimicry functions. Cell 124, 133–145 (2006). GTPases to establish intracellular niches. Traffic 13, 1565–1588 (2012).

10 of 11 | www.pnas.org/cgi/doi/10.1073/pnas.2018163117 Lim et al. Downloaded by guest on September 26, 2021 14. R. C. Orchard, N. M. Alto, Mimicking GEFs: A common theme for bacterial pathogens. 31. J. Goldberg, Structural basis for activation of ARF GTPase: Mechanisms of guanine Cell. Microbiol. 14,10–18 (2012). nucleotide exchange and GTP-myristoyl switching. Cell 95, 237–248 (1998). 15. M. P. Müller, R. S. Goody, Molecular control of Rab activity by GEFs, GAPs and GDI. 32. B.-A. Cho, N.-H. Cho, S.-Y. Seong, M.-S. Choi, I.-S. Kim, Intracellular invasion by Ori- Small GTPases 9,5–21 (2018). entia tsutsugamushi is mediated by integrin signaling and actin cytoskeleton re- 16. L. Hemsath, M. R. Ahmadian, Fluorescence approaches for monitoring interactions of arrangements. Infect. Immun. 78, 1915–1923 (2010). – Rho GTPases with nucleotides, regulators, and effectors. Methods 37, 173 182 (2005). 33. W. D. Hardt, L. M. Chen, K. E. Schuebel, X. R. Bustelo, J. E. Galán, S. typhimurium 17. S. Schoebel, L. K. Oesterlin, W. Blankenfeldt, R. S. Goody, A. Itzen, RabGDI displace- encodes an activator of Rho GTPases that induces membrane ruffling and nuclear ment by DrrA from Legionella is a consequence of its guanine nucleotide exchange responses in host cells. Cell 93, 815–826 (1998). activity. Mol. Cell 36, 1060–1072 (2009). 34. H. Chu et al., Exploitation of the endocytic pathway by Orientia tsutsugamushi in 18. K. R. Abdul Azeez, S. Knapp, J. M. P. Fernandes, E. Klussmann, J. M. Elkins, The crystal nonprofessional phagocytes. Infect. Immun. 74, 4246–4253 (2006). structure of the RhoA-AKAP-Lbc DH-PH domain complex. Biochem. J. 464, 231–239 35. J.-E. Koo et al., Activation of mitogen-activated protein kinases is involved in the (2014). 19. C. Klebe, H. Prinz, A. Wittinghofer, R. S. Goody, The kinetic mechanism of Ran– induction of interferon beta gene in macrophages infected with Orientia tsutsuga- – Nucleotide exchange catalyzed by RCC1. Biochemistry 34, 12543–12552 (1995). mushi. Microbiol. Immunol. 53, 123 129 (2009). 20. A. Friebel et al., SopE and SopE2 from Salmonella typhimurium activate different sets 36. K. Jiang et al., Pivotal role of phosphoinositide-3 kinase in regulation of cytotoxicity in of RhoGTPases of the host cell. J. Biol. Chem. 276, 34035–34040 (2001). natural killer cells. Nat. Immunol. 1, 419–425 (2000). 21. A. E. Karnoub et al., Molecular basis for Rac1 recognition by guanine nucleotide 37. J. A. Frost et al., Cross-cascade activation of ERKs and ternary complex factors by Rho exchange factors. Nat. Struct. Biol. 8, 1037–1041 (2001). family proteins. EMBO J. 16, 6426–6438 (1997). 22. J. T. Snyder et al., Structural basis for the selective activation of Rho GTPases by Dbl 38. Y. Tang, J. Yu, J. Field, Signals from the Ras, Rac, and Rho GTPases converge on the exchange factors. Nat. Struct. Biol. 9, 468–475 (2002). Pak protein kinase in Rat-1 fibroblasts. Mol. Cell. Biol. 19, 1881–1891 (1999). 23. G. Buchwald et al., Structural basis for the reversible activation of a Rho protein by 39. A. M. Abel et al., IQGAP1: Insights into the function of a molecular puppeteer. Mol. the bacterial toxin SopE. EMBO J. 21, 3286–3295 (2002). Immunol. 65, 336–349 (2015). 24. Z. Huang et al., Structural insights into host GTPase isoform selection by a family of 40. M. D. Brown, L. Bry, Z. Li, D. B. Sacks, IQGAP1 regulates Salmonella invasion through bacterial GEF mimics. Nat. Struct. Mol. Biol. 16, 853–860 (2009). interactions with actin, Rac1, and Cdc42. J. Biol. Chem. 282, 30265–30272 (2007). 25. R. Gasper, C. Thomas, M. R. Ahmadian, A. Wittinghofer, The role of the conserved 41. C. Guthrie, G. R. Fink, “Guide to yeast genetics and molecular biology” in Methods in switch II glutamate in guanine nucleotide exchange factor-mediated nucleotide ex- Enzymology (Elsevier, San Diego, CA, 1991), pp. 12–14. – change of GTP-binding proteins. J. Mol. Biol. 379,51 63 (2008). 42. D. Mumberg, R. Müller, M. Funk, Regulatable promoters of Saccharomyces cerevisiae: 26. A. Delprato, D. G. Lambright, Structural basis for Rab GTPase activation by VPS9 Comparison of transcriptional activity and their use for heterologous expression. domain exchange factors. Nat. Struct. Mol. Biol. 14, 406– 412 (2007). Nucleic Acids Res. 22, 5767–5768 (1994). 27. V. Benard, B. P. Bohl, G. M. Bokoch, Characterization of rac and cdc42 activation in 43. B. J. Thomas, R. Rothstein, Elevated recombination rates in transcriptionally active chemoattractant-stimulated human neutrophils using a novel assay for active GTPa- DNA. Cell 56, 619–630 (1989). ses. J. Biol. Chem. 274, 13198–13204 (1999). 44. A. Chatterjee, L. Wang, D. L. Armstrong, S. Rossie, Activated Rac1 GTPase translocates 28. L. Vidali, F. Chen, G. Cicchetti, Y. Ohta, D. J. Kwiatkowski, Rac1-null mouse embryonic fibroblasts are motile and respond to platelet-derived growth factor. Mol. Biol. Cell protein phosphatase 5 to the cell membrane and stimulates phosphatase activity – 17, 2377–2390 (2006). in vitro. J. Biol. Chem. 285, 3872 3882 (2010).

29. C. M. Wells, M. Walmsley, S. Ooi, V. Tybulewicz, A. J. Ridley, Rac1-deficient macro- 45. N. E. Chayen, P. S. Stewart, D. M. Blow, Microbatch crystallization under oil - a new BIOCHEMISTRY phages exhibit defects in cell spreading and membrane ruffling but not migration. technique allowing many small-volume crystallization trials. J. Cryst. Growth 122, J. Cell Sci. 117, 1259–1268 (2004). 176–180 (1992). 30. J. F. Aranda et al., MYADM regulates Rac1 targeting to ordered membranes required 46. A. E. Carpenter et al., CellProfiler: Image analysis software for identifying and for cell spreading and migration. Mol. Biol. Cell 22, 1252–1262 (2011). quantifying cell phenotypes. Genome Biol. 7, R100–R111 (2006).

Lim et al. PNAS Latest Articles | 11 of 11 Downloaded by guest on September 26, 2021