Structural and functional analysis of an anchorlessfibronectin-binding protein FBPS from Gram-positivebacterium Streptococcus suisAbednego Moki Musyokia,b,1, Zhongyu Shia,b,1, Chunling Xuana,b,1, Guangwen Lua,c,d, Jianxun Qia, Feng Gaoe,Beiwen Zhengf, Qiangmin Zhanga, Yan Lia, Joel Haywooda,b, Cuihua Liua, Jinghua Yana, Yi Shia,b,g,and George F. Gaoa,b,g,h,2

aCAS Key Laboratory of Pathogenic Microbiology and Immunology, Institute of Microbiology, Chinese Academy of Sciences, Beijing 100101, China;bUniversity of the Chinese Academy of Sciences, Beijing 100049, China; cWest China Hospital Emergency Department (WCHED), State Key Laboratory ofBiotherapy, West China Hospital, Sichuan University, Sichuan 610041, China; dCollaborative Innovation Center of Biotherapy, Chengdu, Sichuan 610041,China; eInstitute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing 100101, China; fState Key Laboratory for Diagnosis andTreatment of Infectious Diseases, the First Affiliated Hospital, School of Medicine, Zhejiang University, Hangzhou 310003, China; gResearch Network ofImmunity and Health, Beijing Institutes of Life Science, Chinese Academy of Sciences, Beijing 100101, China; and hNational Institute for Viral DiseaseControl and Prevention, Chinese Center for Disease Control and Prevention, Beijing 102206, China

Edited by Scott J. Hultgren, Washington University School of Medicine, St. Louis, MO, and approved October 14, 2016 (received for review May 31, 2016)

The anchorless fibronectin-binding proteins (FnBPs) are a groupof important virulence factors for which the structures are notavailable and the functions are not well defined. In this study weperformed comprehensive studies on a prototypic member of thisgroup: the fibronectin-/fibrinogen-binding protein from Strepto-coccus suis (FBPS). The structures of the N- and C-terminal halves(FBPS-N and FBPS-C), which together cover the full-length proteinin sequence, were solved at a resolution of 2.1 and 2.6 Å, respec-tively, and each was found to be composed of two domains withunique folds. Furthermore, we have elucidated the organization ofthese domains by small-angle X-ray scattering. We further showedthat the fibronectin-binding site is located in FBPS-C and that FBPSpromotes the adherence of S. suis to host cells by attaching thebacteria via FBPS-N. Finally, we demonstrated that FBPS functionsboth as an adhesin, promoting S. suis attachment to host cells, andas a bacterial factor, activating signaling pathways via β1 integrinreceptors to induce chemokine production.

fibronectin-binding protein of Streptococcus suis | structure | novel fold |fibronectin-binding property | function

The Streptococcus suis serotype 2 (S. suis 2) is an importantzoonotic pathogen causing swine infections (1). Occasionally

S. suis can also cause human infections that result in meningitis,septicemia, arthritis, and other mild diseases or, in some extremecases, severe postinfection sequelae or death, consequently gen-erating worldwide public concern (2–4). In 2005, a large out-break (215 cases) of human S. suis infections was reported inSichuan Province, China (5–7). A previous outbreak of humanS. suis infections occurred in Jiangsu Province, China in 1998 (8, 9).Most cases of the disease, in both swine and humans, are caused byS. suis 2, and therefore almost all studies on virulence factors andthe pathogenesis of infection focus on this serotype (10–12).S. suis can adhere to and invade eukaryotic cells; this adherence

is likely a prerequisite for the bacterium invasion and establish-ment of disease in the host (1, 3). As with other Gram-positivebacteria, S. suis can express specific cell-surface componentscalled “adhesins” to mediate adherence to host cells (13, 14).Most of these adhesins function by binding to various com-ponents of the host extracellular matrix (ECM) (14, 15). One ofthe most common adhesin–ECM interactions involves the re-cruitment of fibronectin, a ubiquitous extracellular protein (14)which is abundant in the circulation system and at various ex-tracellular sites (16). Intriguingly, fibronectin can bind to bothhost cells and bacteria and therefore is considered an essentialmolecule for mediating the adherence of Gram-positive bacteriato host organisms. Moreover, via its interaction with integrins,

fibronectin also plays a role in triggering the signal transductionevents that facilitate bacterial invasion into eukaryotic cells (17).Bacterial pathogens are able to use host fibronectin for path-

ogenesis by expressing multiple fibronectin-binding proteins(FnBPs). These FnBPs are microbial surface components rec-ognizing adhesive matrix molecules (MSCRAMM) (13) and canbe categorized into two groups based on their surface-anchoringmechanisms (18). One group of FnBPs is covalently anchored tothe bacterial surface. Members of this group, such as strepto-coccal fibronectin-binding protein 1 (SfbI) of Streptococcuspyogenes (19) and FnBPA of Staphylococcus aureus (20), containan N-terminal signal peptide for secretion, a C-terminal hydro-phobic region, and a charged tail within which a hydrophobic do-main contains the LPXTGmotif for covalent anchorage to cell-wallpeptidoglycan. In these proteins the fibronectin-binding activity islocated in the C-terminal half of the molecules. Here sequence

Significance

Gram-positive bacteria have evolved to use host fibronectin viamolecules called “fibronectin-binding proteins” (FnBPs) to ex-ecute their host-interaction functions. The anchorless FnBPs,for which neither structural information nor a well-definedfunction is available, were recently proposed to be importantvirulence factors. Our work illustrates the organization of fi-bronectin/fibrinogen-binding protein from Streptococcus suis(FBPS), a representative member of the anchorless FnBP groupfrom S. suis, by small-angle X-ray scattering and describes twoterminal-half structures at high resolution. The C-terminal halfof FBPS interacts with fibronectin and the N-terminal half at-taches to the bacterial surface. Functionally, FBPS contributesto the bacterial pathogenesis both as an adhesin and asa chemokine stimulator.

Author contributions: G.L., B.Z., Q.Z., C.L., J.Y., Y.S., and G.F.G. designed research; A.M.M.,Z.S., and C.X. performed research; B.Z., Q.Z., Y.L., and C.L. contributed new reagents/analytic tools; A.M.M., Z.S., C.X., G.L., J.Q., F.G., J.H., and Y.S. analyzed data; and A.M.M.,Z.S., C.X., G.L., J.H., Y.S., and G.F.G. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.

Data deposition: Crystallography, atomic coordinates, and structure factors reported inthis paper have been deposited in the Protein Data Bank [ID codes 5H3X (FBPS-N) and5H3W (FBPS-C)].1A.M.M., Z.S., and C.X. contributed equally to this work.2To whom correspondence should be addressed. Email: gaof@im.ac.cn.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1608406113/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1608406113 PNAS | November 29, 2016 | vol. 113 | no. 48 | 13869–13874

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

19, 2

020

repeats, characterized as the high-affinity fibronectin-bindingrepeats (FnBRs) (21), were found to assemble into tandemβ-zippers which align with sequential type I modules in fibro-nectin (22, 23). In recent years, another group of FnBPs, rep-resented by PavA of Streptococcus pneumoniae (24) and Fbp54of S. pyogenes (25), has been identified in many Gram-positivebacteria. Distinct from the LPXTG-mediated attachment, mem-bers of this FnBP group lack a canonical signal peptide and anLPXTG-like motif and instead use an unknown mechanism forsurface localization (18). These anchorless adhesins have impor-tant biological functions and form a new class of virulence factors(26–29). However, unlike the LPXTG-anchored FnBPs, therecurrently is a paucity of structural and functional data regardingthe anchorless FnBPs, which share only a low sequence homology.Here we have characterized the structural and functional fea-

tures of the anchorless FnBP of S. suis 2 (FBPS). The structure wassolved in its N- and C-terminal halves, FBPS-N and FBPS-C, re-spectively, which unexpectedly revealed two protein folds. Fur-thermore, we have elucidated the organization of these domains bysmall-angle X-ray scattering (SAXS) and show that the FBPSN-terminal half attaches to the bacterial surface, whereas theC-terminal half mediates adhesion to the host cells. Moreover, wehave characterized the role of FBPS as both an adhesin and avirulence-contributing factor by comparing the attachment of WTand Δfbps mutant bacteria to HEp-2 cells and by analyzing FBPS-mediated activation of downstream signaling pathways. These re-sults demonstrate that FBPS can both promote the adherence of S.suis to HEp-2 cells and activate host kinases to induce IL-6 andIL-8 production via integrin receptors. We further comparedthe structure of FBPS-N with that of another family member, theFnBP of S. aureus, whose structure had been deposited in theProtein Data Bank (PBD) but had not yet been analyzed, revealingtheir common structural characteristics and distinct features.

ResultsCrystal Structure of the N-Terminal Half of the FBPS. The FBPS ofS. suis 2 consists of 552 amino acids and exhibits high sequencesimilarity to PavA of S. pneumoniae (74% identity) and Fbp54 ofS. pyogenes (70% identity) (Fig. S1). To explore the structuralfeatures of FBPS, we first used the full-length protein for crys-tallization trials. The solved structure unexpectedly containedonly FBPS-N. The final model, which is refined to 2.1-Å reso-lution (Table S1), contains 267 amino acids extending from S2 toK268. The overall structure involves two distinct domains (do-mains I and II), one extra small helix (αn7), and two extremelylong loops connecting the domains and the helix. Domain I is anine-stranded β-barrel core wrapped by several surface helicesand loops. Topologically, the domain residues first assemble intoan α-helix (αn1) and a four-stranded front-sheet with four con-secutive antiparallel β-strands (βn1–βn4) and then proceed via aloop-connecting bridge over the face of the βn1–βn4 strands toform another helix (αn2) and a rear-sheet composed of theremaining five core-barrel strands (βn5–βn9) arranged in anantiparallel manner. Sterically, the two helices locate betweenand clamp the two barrel-sheets on the opposite sides. Domain Iis linked to domain II by a long βn9/αn3 interloop that flowsunderneath the βn5–βn9 strands (Fig. 1 A and B).In contrast to domain I, which is predominantly β-stranded,

domain II of FBPS-N is a purely α-helical structure. In this do-main, four helices (αn3–αn6) cluster together, forming an in-dependent bundle-like fold. Following this domain, the αn6/7interloop flows back toward the βn4–αn2 bridge loop of domain Iforming two short β-strands (βn10–βn11) arranged in an anti-parallel manner defining a small β-hairpin structure that formstyrosine-mediated hydrogen bonds (K237–Y264 and Y239–Y265) with the final α-helical structure of the FBPS-N (αn7).Moreover, the αn7 helix is further stabilized by hydrogen bond-ing to the βn4–αn2 bridge of domain I. (Fig. 1 A and B).

The PDB contains a structure, the FnBP-N of S. aureus (PDBID code 3DOA) resolved by Structural Genomics Consortiumthat is a homolog of our FBPS-N but whose analysis had notbeen described. We compared the two structures and found thatFBPS-N has a fold similar to that in FnBP-N and that theiroverall structures are similar (Fig. 1 C and D). Detailed super-imposition analysis revealed that domain I is almost exactly thesame in FBPS-N and FBPS-N, but domain II is moderately dif-ferent in the two structures. The αn4/αn5 loop, αn5/αn6 loop, andthe β-hairpin are shorter in the FBPS-N structure than in theFnBP-N structure (Fig. 1 C and D).

Structure of the C-Terminal Half of the FBPS. The crystal structure ofFBPS-C starting with residue A269 was resolved at a resolutionof 2.6 Å (Table S1). Clear electron densities can be traced un-ambiguously from the fourth residue (D272) to the very endresidue (I552), with the exception of amino acids T320 to P354,which were devoid of any traceable densities. The FBPS-Cstructure also can be divided into two domains (domains I andII). Domain I is a purely helical structure that consists of threetopologically consecutive α-helices. The first two helices, αc1 andαc2, are extremely long, with an estimated length of 74 and 64 Å,respectively, and are arranged in a cross shape with an interangleof ∼20° to form a long coiled coil. The domain II-distal ends ofthese domain I helices are linked via a long and flexible loop (αc1/αc2 loop) that is formed by the 33 untraceable residues (T320–P354). The final relatively small third helix (αc3) of FBPS-C do-main I is located on one longitudinal side of the coiled structure,proximal to domain II, and assembles into a stable three-helixbundle with helices αc1 and αc2 (Fig. 1 E and F).Domain II of FBPS-C consists of amino acids E432 to I552,

forming an overall globular fold. The domain is composed of sixantiparallel β-strands (βc1–βc6) and three α-helices with αc5located in the center of a ring formed by the βc1–βc6 strands andαc6 and αc4 forming stabilizing hydrogen bond interactions withdomain I αc3 (Fig. 1E). Topologically, this domain exhibits acharacteristic structure with a tandem arrangement of three ββαmotifs (Fig. 1F).

Architecture of Full-Length FBPS. Having solved the individualFBPS-N and FBPS-C crystal structures, we were able to examinethe architecture of the full-length protein in solution further bySAXS. The FBPS-N and FBPS-C structures fit well into theSAXS map, and from the side view we can see that FBPS-N andFBPS-C are organized in a cis orientation, like a human arm, andthat the axis lines of these two N- and C-terminal structures meetat an angle of about 90° (Fig. 2). From the top view, we can seethat the FBPS-N domain II and FBPS-C domain II are on oneside, and the FBPS-N domain I and FBPS-C domain I are on theother side (Fig. 2).

FBPS Promotes S. suis Adherence by Attaching Bacteria via FBPS-Nand Adhering to Host Cells with FBPS-C. A typical feature of FnBPsis their ability to promote bacterial adherence to host cells bybridging the bacteria and the ECM. Thus we also tested whetherFBPS could function as such an adhesin in S. suis pathogenesis.The fbps gene was knocked out from the virulent S. suis strain05ZYH33 (7) to yield a mutant strain, Δfbps, and successfulconstruction of the deletion mutant was confirmed by a PCR-based assay (Fig. S2). Overall, deletion of the fbps gene did notaffect the in vitro growth rate of the bacterial cells in Todd–Hewitt broth supplemented with 2% yeast extract (THY me-dium) (Fig. S3). The elimination of FBPS did not completelyabrogate the cell adherence of S. suis; however, a statisticallysignificant decrease in surface attachment to HEp-2 cells wasobserved for Δfbps compared with the WT strain. Upon com-plementation, the mutant strain regained its adhesion capacity toa level similar to that of WT S. suis (Fig. 3A).

13870 | www.pnas.org/cgi/doi/10.1073/pnas.1608406113 Musyoki et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

19, 2

020

We further probed the binding mechanism of S. suis by in-vestigating the binding of the full-length protein and the twoterminal halves to bacterial and host cells. As expected, the full-length protein was able to adhere to both the S. suis and theHEp-2 cells. Interestingly, the bacterial surface-attachment ac-tivity of FBPS lies in the N-terminal half of the protein and isspecific to the S. suis cells. Neither S. pneumonia nor Group BStreptococcus (GBS) could pull down FBPS or FBPS-N (Fig. 3B).On the other hand, FBPS uses its C-terminal half for host-cellattachment. FBPS-C, rather than FBPS-N, was pulled down bythe HEp-2 cells (Fig. 3C).

FBPS Binds to Fibronectin Predominantly via its C-Terminal Half andContributes Directly to S. suis Adherence. We further characterizedthe fibronectin-binding properties of FBPS using a surfaceplasma resonance (SPR) binding assay. The N-terminal 30-kDaportion of human fibronectin (FnN30) was used in the experi-ment, because this protein fragment has been shown to be in-volved in the engagement of bacterial ligands (17). As expected,a potent binding to immobilized FnN30 was observed for the full-length FBPS (Fig. 3D). However, although FBPS-C displayedbinding kinetics similar to that observed for the full-length protein,with a dissociation constant (Kd) of 370 nM (Fig. 3F), a negligible

interaction was detected for FBPS-N (Fig. 3E). These results in-dicate that FBPS-C, rather than FBPS-N, represents the pre-dominant domain for fibronectin-binding activity and recognizesthe N-terminal fragment of human fibronectin for engagement.Our results also are consistent with a previous study showing thatC-terminal truncations of S. pneumoniae PavA cause the proteinto lose its capacity for fibronectin interaction (22).We then sought to investigate whether binding to fibronectin

by FBPS contributes directly to S. suis adherence. Therefore wecompared the ability of the WT, Δfbps, and complementedbacterial strains to bind to the immobilized human fibronectin.Similar to the cell-based adhesion assay, S. suis binding to fi-bronectin was impaired dramatically after knockout of the fbpsgene; the observed binding level of Δfbps was approximately halfthat observed for the WT or the complemented strains (Fig. 3G).

FBPS Contributes to IL-6/-8 Production via β1 Integrin Receptors. Anumber of surface or secreted components of Staphylococci spp.and Streptococci spp. strongly induce proinflammatory responses(30–32). We therefore investigated whether FBPS contributesto the production of proinflammatory cytokines by testing theamounts of IL-6 and IL-8 secreted by HEp-2 cells after S. suisinfection. A strong production of both cytokines was detected

Fig. 1. Crystal structure of FBPS-N and FBPS-C.(A) Overall fold of FBPS-N. Domain I, domain II, andthe C-terminal α-helix are colored green, cyan, andmagenta, respectively. The domain linkers are inorange for the loop connecting domains I and II andblue for the loop linking domain II with the terminalhelix. The secondary elements are labeled accordingto their occurrence within the structure. (B) Topo-logical view of the FBPS-N structure. The α-helicesand β-strands are colored as in A. (C) Comparison ofour FBPS-N of S. suiswith FnBP-N of S. aureus (PDB IDcode 3DOA). Domain I of FBPS-N is highly similar todomain I of FnBP-N; however, in domain II, the αn4/αn5 loop, the αn5/αn6 loop, and the β-hairpin areshorter in FBPS-N than in FnBP-N. (D) Topologicalview of the FnBP-N structure. (E ) Overall fold ofFBPS-C. Two molecules of essentially the samestructure are observed in the asymmetric unit; onlyone of them is multicolored and labeled. Thestructure can be divided into two domains, whichare colored magenta and yellow, respectively. A33-residue loop in domain I, which is devoid oftraceable electron densities, is indicated by adashed curve. The secondary structure elementsare successively marked based on their occurrencewithin the structure. (F ) Topological view of thesame structure.

Musyoki et al. PNAS | November 29, 2016 | vol. 113 | no. 48 | 13871

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

19, 2

020

when the cells were infected with WT S. suis compared withcontrol cells. The Δfbps bacterium also induced a rapid expres-sion of IL-6 and -8 in HEp-2 cells but at a decreased level.Complementation with the intact fbps gene faithfully restoredthe HEp-2 secretion of IL-6 and -8 (Fig. 4 A and B). Moreover,by blocking the β1 integrin receptors with a specific antibody, wewere able to show that these receptors are crucial for the in-duction of this cytokine release, because a decreased secretion ofIL-6 and -8 by HEp-2 cells, which abolished the difference incytokine levels between WT and complemented strains with theknockout, was observed in the presence of this specific antibodybut not in the presence of an isotype control (Fig. 4 A and B).We further explored the host signaling pathway underlying

FBPS-induced cytokine production. The MAPKs such as ERKand p38 are shown to be phosphorylated upon bacterial infectionand play an important role in IL-8 production induced by YadA,an FnBP identified in Yersinia pseudotuberculosis (33). Wetherefore examined whether ERK and p38 are activated byphosphorylation upon S. suis infection and compared the acti-vation pattern of the two kinases in WT and Δfbps strains. Rapidand strong activation of both kinases was observed when HEp-2cells were incubated with WT S. suis. The phosphorylation was

maximally increased 10–20 min after infection and remained de-tectable for at least 45 min after infection. The Δfbps infection alsoactivated ERK and p38, but the levels were significantly lowerthan those observed in the WT strain over the same time period.In addition, deletion of the fbps gene caused a dramatic delay inkinase activation, with maximal phosphorylation occurring 45 minpostinfection (Fig. 4 C and D). The two kinases also were acti-vated effectively by purified FBPS and FBPS-C proteins but not byBSA or FBPS-N (Fig. 4 E and F), demonstrating the importantrole of FBPS-C in activating host proinflammatory responses.

DiscussionFnBPs have been implicated in the pathogenesis of many im-portant bacterial pathogens (17). Known FnBPs can be catego-rized into two groups: those with an LPXTG motif for surfacelocalization and those lacking similar anchoring motifs (18). Inthis study, we illustrate the structural features of the latter an-chorless group using a representative member from S. suis,FBPS. The structure of FBPS was solved in its N- and C-terminalhalves, and its full-length architecture in solution was de-termined by SAXS. FBPS-N and FBPS-C are independent instructure. Furthermore, our functional assays revealed thatFBPS-N attaches to the bacterial surface, whereas FBPS-C bindsto host cells and interacts with fibronectin. Therefore it is areasonable inference that the anchorless FnBPs are composed oftwo relatively independent halves with both structural andfunctional variances. It is interesting that FBPS-N ends with anα-helix (αn7) and FBPS-C begins with another α-helix (αc1).Therefore it is possible that helices αn7 and αc1 are actually onehelix that is digested at a potential trypsin-recognition site(K268/A269) during crystallization. In our experience, however,such an arrangement is energetically unfavorable, and our SAXSdata suggest that a small, flexible interdomain loop connects theN- and C- terminal domains. In addition, the flexible charac-teristics of a loop could confer steric motility to FBPS-C (relative

Fig. 3. FBPS contributes to S. suis adherence. (A) Adhesion of WT S. suis (black bar), the Δfbps mutant (white bar), and complemented S. suis (gray bar) toHEp-2 cells. Data are presented as means ± SD of three independent experiments. **P < 0.005. (B and C) Pull-down assays characterizing the binding ca-pacities of different FBPS proteins to the bacterial or the host cells. (B) Proteins pulled down by the indicated bacterial cells. (C) Proteins pulled down by HEp-2cells. (D–F) BIAcore analysis of the binding between FnN30 and different FBPS proteins. Gradient concentrations of the indicated S. suis proteins were flowedover immobilized FnN30. Kinetic profiles are shown. (D) FBPS binding to FnN30. (E) FBPS-N binding to FnN30. (F) FBPS-C binding to FnN30. (G) Adhesion of WTS. suis (black bar), Δfbps mutant (white bar), and complemented S. suis (gray bar) cells to immobilized fibronectin. Data are presented as means ± SD of threeindependent experiments. ***P < 0.001.

Fig. 2. Architecture of full-length FBPS. The FBPS-N and FBPS-C structureshave been fitted into the SAXS envelopes of FBPS.

13872 | www.pnas.org/cgi/doi/10.1073/pnas.1608406113 Musyoki et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

19, 2

020

to FBPS-N), thereby allowing the protein to fulfill its role as abridging adhesin more effectively. Despite a long-term trial forcrystallization of full-length FBPS, we repeatedly got the cleavedN/C crystals separately, and the site-mutagenesis construct yieldedeither the same crystals or none, indicating the separate foldingof the N/C domains.Our SPR analysis clearly illustrates that FBPS is able to bind

to fibronectin and that this binding capacity can be largely at-tributed to the C-terminal half of the protein. However, theobserved binding affinity is weaker than that for the other groupof FnBPs, which are normally at one- or two-digit nanomolarlevels (21). We therefore believe that FBPS represents a rela-tively low-affinity fibronectin ligand (26). Because FBPS also hasbeen reported to bind to fibrinogen, we investigated by SPRwhether FBPS could interact with other ECM components usinghuman collagen I–V as representative examples and found thatFBPS binding to the ECM protein fibronectin/fibrinogen ap-pears to be selective, because low levels of binding were observedwith collagens I–V and the nonspecific protein control BSA (Fig.S4). Functionally, FnBPs have been characterized as importantadhesion molecules that can facilitate bacterial attachment tohost cells (13). FBPS fulfills the role as a bacterial adhesin in ouradhesive analysis. However, deletion of the fbps gene neitherabolished S. suis adherence to HEp-2 cells nor abrogated thebacterial attachment to immobilized fibronectin. Therefore, it isprobable that other adhesins in S. suis aid in bacterial adherence(30). Nevertheless, the anchorless FBPS may be present in largeamounts as a secreted protein, freely contacting fibronectin andthereby enabling the activation of multiple integrins (31). Dif-ferent signaling pathways, as reported in other bacteria (33, 34),have been shown to be modulated in this way to facilitate bac-terial pathogenesis. Moreover, we have shown that FBPS canactivate downstream signaling pathways by phosphorylatingERK and p38 and induce the production of IL-6 and -8 via β1integrin receptors. Interestingly, a pathogenetic study on GBSrevealed that enhanced chemokine expression leads to increasedneutrophil infiltration and promotes bacterial dissemination inCNS (32). Given the previous observation that FBPS contributesto S. suis pathogenesis in pigs (26), it is tempting to speculatethat FBPS represents an effective virulence-contributing factor,contributing to CNS diseases caused by S. suis, such as menin-gitis, and could represent an important target for structure-baseddrug development.

Materials and MethodsCloning, Expression, and Purification of Native and Selenomethionine-LabeledProteins. The fbps gene encoding FBPS was PCR-amplified from the genomicDNA of S. suis 05ZYH33 (7) with the primers FBPS-F and FBPS-R (for se-quences, see Table S2). The coding fragment for FBPS-N (residues 1–268) wasamplified with primers FBPS-F and FBPS-N-R, and that for FBPS-C (residues269–552) was amplified with primers FBPS-C-F and FBPS-R. The productswere cloned into pET-21a and transformed into Escherichia coli strain BL21(DE3) for expression. To facilitate purification, these plasmids contained anextra His-tag–coding sequence encoded by the vector.

The native proteins were expressed by the addition of 1 mM isopropylβ-d-1-thiogalactopyranoside (IPTG) to exponential phase cultures. After a 4-hincubation at 37 °C, the cells were harvested by centrifugation, disrupted bysonication, and clarified by centrifugation. FBPS, FBPS-N, and FBPS -C thenwere purified further by Ni-NTA affinity chromatography and gel filtration.

For the production of selenomethionine (SeMet)-labeled proteins, bac-teria were grown in M9 minimal medium until midlog phase. Then L-lysine,L-phenylalanine, L-threonine, L-isoleucine, L-leucine, L-valine, and L-SeMetwere added to the culture 15 min before the addition of 1 mM IPTG. Fol-lowing induction, the SeMet-labeled proteins were purified in the same wayas native proteins.

Crystallization and Data Collection. All crystallization experiments were car-ried out using the hanging-drop vapor-diffusion method at 18 °C. Typically,a 5- to 15-mg/mL protein solution was mixed in a 1:1 ratio with reservoir so-lution. The FBPS-N crystals were obtained with a solution containing 0.1 MNa citrate (pH 5.6), 20% (wt/vol) PEG 4,000, and 0.2 M NaCl. The FBPS-Cprotein was crystallized in a solution containing 0.1 M Hepes (pH 7.5), 10%(wt/vol) PEG 4,000, and 500 mM NaCl. Crystals of the SeMet-labeled FBPS-Nand FBPS-C were grown under identical conditions. Before data collection,crystals were transferred for no more than 5 s into a cryoprotectant bufferconsisting of 25% (vol/vol) glycerol and 75% (vol/vol) reservoir solution.Crystals then were mounted in cryoloops and were flash-cooled in liquidnitrogen. The dataset for FBPS-N was collected at the Beijing SynchrotronRadiation Facility Beamline 1W2B, and that for FBPS-C was collected at theShanghai Synchrotron Radiation Facility Beamline 17U. The collected in-tensities were indexed, integrated, and scaled using HKL2000 (32).

Structure Determination and Refinement. The structures of both FBPS-N andFBPS-C were determined by the single-wavelength anomalous diffraction(SAD) method. The expected Se atoms were located by SHELXD (33), andinitial phases were calculated using PHASER (35). The real space constraintswere further applied to the electron density map in DM (36). The initialmodels were built with Autobuild in PHENIX package (34). The nativedatasets then were used for structure-solving by the molecular replacementmethod using PHASER from the CCP4 suite (37). Extensive model building andrestrained refinement were performed with COOT (38) and REFMAC5 (39).

Fig. 4. FBPS-dependent cytokine secretion andMAPK activation. (A and B) Secretion of IL-6 (A) andIL-8 (B) by HEp-2 cells upon infection with WT (blackbar), Δfbps mutant (white bar), and complemented(gray bar) S. suis strains. Concentrations of IL-6 andIL-8 in HEp-2 supernatants in the absence of anyantibodies or in the presence of anti–β1-integrinantibody or an isotype control antibody were mea-sured using ELISA. **P < 0.005, ***P < 0.001; NS, notsignificant. (C and D) Detection of phosphorylatedand nonphosphorylated forms of ERK (C) and p38(D) in HEp-2 cells infected with WT or Δfbps S. suis. (Eand F) Detection of phosphorylated and non-phosphorylated forms of ERK (E) and p38 (F) in HEp-2 cells incubated with individual FBPS proteins orBSA (an irrelevant protein control). The proteinswere visualized at the indicated time points usingspecific antibodies. A representative Western blotresult is shown.

Musyoki et al. PNAS | November 29, 2016 | vol. 113 | no. 48 | 13873

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

19, 2

020

Further rounds of refinement were performed using the phenix.refine (36).The stereochemical quality of the final models was assessed with the pro-gram PROCHECK (40).

Data collection and processing statistics are summarized in Table S1. ThePDB ID codes are 5H3X for FBPS-N and 5H3W for FBPS-C.

SAXS Data Acquisition and Analysis. Three concentrations of FBPS (2.5 mg/mL,5 mg/mL, and 10mg/mL) were prepared along with the buffer as backgroundcontrol. SAXS experiments were performed at beamline 19U2 of NationalCenter for Protein Science Shanghai at the Shanghai Synchrotron RadiationFacility. The wavelength, λ, of X-ray radiation was set as 1.033 Å. ScatteredX-ray intensities were measured using a Pilatus 1M detector (DECTRIS Ltd). Thesample-to-detector distance was set so that the detecting range of momentumtransfer q [=4π sin θ/λ, where 2θ is the scattering angle] of SAXS experimentswas 0.01–0.5 Å−1. To reduce the radiation damage, a flow cell made of a cy-lindrical quartz capillary with a diameter of 1.5 mm and a wall of 10 μm wasused, and the exposure time was set to 1 s. The X-ray beam with a size of0.40 × 0.15 (horizontal × vertical) mm2 was adjusted to pass through thecenters of the capillaries for every measurement. To obtain good signal-to-noise ratios, 10 images were taken for each sample and buffer. The 2D

scattering images were converted to 1D SAXS curves by azimuthal averagingafter solid angle correction and then normalizing with the intensity of thetransmitted X-ray beam, using the software package BioXTAS RAW (41). TheATSAS package (42) and GNOM (43) were used for the subsequent data pro-cessing and rigid-body modeling. The ab initio models were calculated using theapplication DAMMIN (44). Consensus models and the normalized spatial dis-crepancy values were calculated by averaging 20 ab initio models using theapplication DAMAVER (45). Final statistics for data collection and scattering-derived parameters are presented in Table S3.

ACKNOWLEDGMENTS. We thank Hao Song, Ming Li, Jun Liu, Zheng Fan,Yuanyuan Chen, and Shuijun Zhang for their excellent assistance. This workwas supported by Strategic Priority Research Program of the Chinese Acad-emy of Sciences Grant XDB08020100 and External Cooperation Program ofthe Chinese Academy of Sciences Grant GJHZ1307. Y.S. is supported by theExcellent Young Scientist Program from the National Natural Science Foun-dation of China (NSFC) Grant 81622031, the Excellent Young Scientist Pro-gram of the Chinese Academy of Sciences and Youth Innovation PromotionAssociation of the Chinese Academy of Sciences Grant 2015078. G.F.G. is aleading principal investigator of the NSFC Innovative Research Groupawarded Grant 81321063.

1. Staats JJ, Feder I, Okwumabua O, Chengappa MM (1997) Streptococcus suis: Past andpresent. Vet Res Commun 21(6):381–407.

2. Ma Y, Feng Y, Liu D, Gao GF (2009) Avian influenza virus, Streptococcus suis serotype2, severe acute respiratory syndrome-coronavirus and beyond: Molecular epidemiology,ecology and the situation in China. Philos Trans R Soc Lond B Biol Sci 364(1530):2725–2737.

3. Lun ZR, Wang QP, Chen XG, Li AX, Zhu XQ (2007) Streptococcus suis: An emergingzoonotic pathogen. Lancet Infect Dis 7(3):201–209.

4. Gottschalk M, Segura M (2000) The pathogenesis of the meningitis caused by Strep-tococcus suis: The unresolved questions. Vet Microbiol 76(3):259–272.

5. Yu H, et al.; Streptococcus suis study groups (2006) Human Streptococcus suis out-break, Sichuan, China. Emerg Infect Dis 12(6):914–920.

6. Tang J, et al. (2006) Streptococcal toxic shock syndrome caused by Streptococcus suisserotype 2. PLoS Med 3(5):e151.

7. Chen C, et al. (2007) A glimpse of streptococcal toxic shock syndrome from compar-ative genomics of S. suis 2 Chinese isolates. PLoS One 2(3):e315.

8. Normile D (2005) Infectious diseases. WHO probes deadliness of China’s pig-bornedisease. Science 309(5739):1308–1309.

9. Feng Y, Zhang H, Ma Y, Gao GF (2010) Uncovering newly emerging variants ofStreptococcus suis, an important zoonotic agent. Trends Microbiol 18(3):124–131.

10. Berthelot-Hérault F, Morvan H, Kéribin A-M, Gottschalk M, Kobisch M (2000) Pro-duction of muraminidase-released protein (MRP), extracellular factor (EF) and suilysinby field isolates of Streptococcus suis capsular types 2, 1/2, 9, 7 and 3 isolated fromswine in France. Vet Res 31(5):473–479.

11. Zhang Q, et al. (2009) Crystal structures of Streptococcus suis mannonate dehydratase(ManD) and its complex with substrate: Genetic and biochemical evidence for a cat-alytic mechanism. J Bacteriol 191(18):5832–5837.

12. Lu Q, Lu H, Qi J, Lu G, Gao GF (2012) An octamer of enolase from Streptococcus suis.Protein Cell 3(10):769–780.

13. Patti JM, Allen BL, McGavin MJ, Höök M (1994) MSCRAMM-mediated adherence ofmicroorganisms to host tissues. Annu Rev Microbiol 48:585–617.

14. Chhatwal GSPK (2006) Extracellular matrix interaction with Gram-positive bacteria.Gram-Positive Pathogens, eds Fischetti VA, Novick RP, Ferretti JJ, Portnoy DA, Rood JI(ASM Press, Washington, DC), pp 89–99.

15. Nitsche-Schmitz DP, Rohde M, Chhatwal GS (2007) Invasion mechanisms of Gram-positive pathogenic cocci. Thromb Haemost 98(3):488–496.

16. Pankov R, Yamada KM (2002) Fibronectin at a glance. J Cell Sci 115(Pt 20):3861–3863.17. Schwarz-Linek U, Höök M, Potts JR (2004) The molecular basis of fibronectin-medi-

ated bacterial adherence to host cells. Mol Microbiol 52(3):631–641.18. Chhatwal GS (2002) Anchorless adhesins and invasins of Gram-positive bacteria: A

new class of virulence factors. Trends Microbiol 10(5):205–208.19. Towers RJ, et al. (2003) Evolution of sfbI encoding streptococcal fibronectin-binding

protein I: Horizontal genetic transfer and gene mosaic structure. J Clin Microbiol41(12):5398–5406.

20. Peacock SJ, Day NP, Thomas MG, Berendt AR, Foster TJ (2000) Clinical isolates ofStaphylococcus aureus exhibit diversity in fnb genes and adhesion to human fibro-nectin. J Infect 41(1):23–31.

21. Schwarz-Linek U, Höök M, Potts JR (2006) Fibronectin-binding proteins of gram-positive cocci. Microbes Infect 8(8):2291–2298.

22. Schwarz-Linek U, et al. (2003) Pathogenic bacteria attach to human fibronectinthrough a tandem beta-zipper. Nature 423(6936):177–181.

23. Bingham RJ, et al. (2008) Crystal structures of fibronectin-binding sites from Staphy-lococcus aureus FnBPA in complex with fibronectin domains. Proc Natl Acad Sci USA105(34):12254–12258.

24. Holmes AR, et al. (2001) The pavA gene of Streptococcus pneumoniae encodes afibronectin-binding protein that is essential for virulence. Mol Microbiol 41(6):1395–1408.

25. Courtney HS, Li Y, Dale JB, Hasty DL (1994) Cloning, sequencing, and expression of afibronectin/fibrinogen-binding protein from group A streptococci. Infect Immun62(9):3937–3946.

26. Torelli R, et al. (2012) The PavA-like fibronectin-binding protein of Enterococcusfaecalis, EfbA, is important for virulence in a mouse model of ascending urinary tractinfection. J Infect Dis 206(6):952–960.

27. Pracht D, et al. (2005) PavA of Streptococcus pneumoniae modulates adherence, in-vasion, and meningeal inflammation. Infect Immun 73(5):2680–2689.

28. de Greeff A, et al. (2002) Contribution of fibronectin-binding protein to pathogenesisof Streptococcus suis serotype 2. Infect Immun 70(3):1319–1325.

29. Courtney HS, Dale JB, Hasty DI (1996) Differential effects of the streptococcal fibro-nectin-binding protein, FBP54, on adhesion of group A streptococci to human buccalcells and HEp-2 tissue culture cells. Infect Immun 64(7):2415–2419.

30. Li W, Wan Y, Tao Z, Chen H, Zhou R (2013) A novel fibronectin-binding protein ofStreptococcus suis serotype 2 contributes to epithelial cell invasion and in vivo dis-semination. Vet Microbiol 162(1):186–194.

31. Leiss M, Beckmann K, Girós A, Costell M, Fässler R (2008) The role of integrin bindingsites in fibronectin matrix assembly in vivo. Curr Opin Cell Biol 20(5):502–507.

32. Otwinowski Z, Minor W (1997) Processing of X-ray diffraction data collected in os-cillation mode. Methods Enzymol 276:307–326.

33. Usón I, Sheldrick GM (1999) Advances in direct methods for protein crystallography.Curr Opin Struct Biol 9(5):643–648.

34. Adams PD, et al. (2010) PHENIX: A comprehensive Python-based system for macro-molecular structure solution. Acta Crystallogr D Biol Crystallogr 66(Pt 2):213–221.

35. Read RJ (2001) Pushing the boundaries of molecular replacement with maximumlikelihood. Acta Crystallogr D Biol Crystallogr 57(Pt 10):1373–1382.

36. Cowtan KD, Zhang KY (1999) Density modification for macromolecular phase im-provement. Prog Biophys Mol Biol 72(3):245–270.

37. Collaborative Computational Project, Number 4 (1994) The CCP4 suite: Programs forprotein crystallography. Acta Crystallogr D Biol Crystallogr 50(Pt 5):760–763.

38. Emsley P, Cowtan K (2004) Coot: Model-building tools for molecular graphics. ActaCrystallogr D Biol Crystallogr 60(Pt 12 Pt 1):2126–2132.

39. Murshudov GN, Vagin AA, Dodson EJ (1997) Refinement of macromolecular struc-tures by the maximum-likelihood method. Acta Crystallogr D Biol Crystallogr 53(Pt 3):240–255.

40. Laskowski RA, MacArthur MW, Moss DS, Thornton JM (1993) PROCHECK: A programto check the stereochemical quality of protein structures. J Appl Cryst 26:283–291.

41. Nielsen SS, Møller M, Gillilan RE (2012) High-throughput biological small-angle X-rayscattering with a robotically loaded capillary cell. J Appl Cryst 45(Pt 2):213–223.

42. Petoukhov MV, et al. (2012) New developments in the ATSAS program package forsmall-angle scattering data analysis. J Appl Cryst 45(Pt 2):342–350.

43. Semenyuk AV, Svergun DI (1991) Gnom - a program package for small-angle scat-tering data-processing. J Appl Cryst 24:537–540.

44. Svergun DI (1999) Restoring low resolution structure of biological macromoleculesfrom solution scattering using simulated annealing. Biophys J 76(6):2879–2886.

45. Semenyuk AV, Svergun DI (2003) Uniqueness of ab-initio shape determination insmall-angle scattering. J Appl Cryst 36:860–864.

46. Zhang N, et al. (2011) Binding of herpes simplex virus glycoprotein D to nectin-1exploits host cell adhesion. Nat Commun 2:577.

47. Li M, et al. (2008) SalK/SalR, a two-component signal transduction system, is essentialfor full virulence of highly invasive Streptococcus suis serotype 2. PLoS One 3(5):e2080.

48. Vanier G, Segura M, Friedl P, Lacouture S, Gottschalk M (2004) Invasion of porcinebrain microvascular endothelial cells by Streptococcus suis serotype 2. Infect Immun72(3):1441–1449.

13874 | www.pnas.org/cgi/doi/10.1073/pnas.1608406113 Musyoki et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

19, 2

020