Allosteric Regulation of Fibronectin/α5β1 Interaction by Fibronectin-Binding MSCRAMMs

Adherence of microbes to host tissues is a hallmark of infectious disease and is often mediated by a class of adhesins termed MSCRAMMs (Microbial Surface Components Recognizing Adhesive Matrix Molecules). Numerous pathogens express MSCRAMMs that specifically bind the heterodimeric human glycoprotein fibronectin (Fn). In addition to roles in adhesion, Fn-binding MSCRAMMs exploit physiological Fn functions. For example, several pathogens can invade host cells by a mechanism whereby MSCRAMM-bound Fn bridges interaction with α5β1 integrin. Here, we investigate two Fn-binding MSCRAMMs, FnBPA (Staphylococcus aureus) and BBK32 (Borrelia burgdorferi) to probe structure-activity relationships of MSCRAMM-induced Fn/α5β1integrin activation. Circular dichroism, fluorescence resonance energy transfer, and dynamic light scattering techniques uncover a conformational rearrangement of Fn involving domains distant from the MSCRAMM binding site. Surface plasmon resonance experiments demonstrate a significant enhancement of Fn/α5β1 integrin affinity in the presence of FnBPA or BBK32. Detailed kinetic analysis of these interactions reveal that this change in affinity can be attributed solely to an increase in the initial Fn/α5β1 on-rate and that this rate-enhancement is dependent on high-affinity Fn-binding by MSCRAMMs. These data implicate MSCRAMM-induced perturbation of specific intramolecular contacts within the Fn heterodimer resulting in activation by exposing previously cryptic α5β1 interaction motifs. By correlating structural changes in Fn to a direct measurement of increased Fn/α5β1 affinity, this work significantly advances our understanding of the structural basis for the modulation of integrin function by Fn-binding MSCRAMMs.


Introduction
Fibronectin (Fn) is a multidomain glycoprotein found in blood plasma, other bodily fluids and the extra-cellular matrix (ECM) and serves as a natural ligand for several integrins including α 5 β 1 , α v β 3 , and α 4 β 1 [1]. The functional roles of Fn are diverse and include ECM assembly, angiogenesis, wound-repair, and oncogenesis [2]. Fn is secreted as a C-terminally disulfidelinked heterodimer composed of three structurally defined repeating units termed Fn domains or modules (Fig 1A). Major forms of Fn are comprised of 12 type I modules (FnI) and two type II modules (FnII), while splice variation results in 15 to 17 type III modules (FnIII). Fn is found in two predominant forms; cellular Fn, which is tissue localized and assembled into a fibrillar matrix, and the hepatocyte expressed, soluble plasma Fn that is secreted and maintained in blood at~0.3 mg ml -1 [3]. Despite having independent functional roles from cellular Fn, it is significant that plasma Fn accounts for a large fraction of Fn found in tissue ECM [3,4].
To date, over one-hundred bacterial Fn-binding proteins (FnBPs) have been reported, and a majority of these belong to a protein family termed Microbial Surface Components Recognizing Adhesive Matrix Molecules (MSCRAMMs) [2]. Many Fn-binding MSCRAMMs contribute to virulence as in vivo studies demonstrate a critical role for these proteins in development of infectious diseases such as endocarditis, mastitis, and wound-infection [8][9][10]. A substantial body of evidence suggests that FnBPs can also manipulate physiological functions of Fn and thus contribute to pathogenesis in ways beyond mediating bacterial adhesion [6,[11][12][13][14][15][16]. Indeed, several studies over the past decade involving the F1 and SfbI proteins from Streptococcus pyogenes [11,12,17], and more recently, BBK32 from Borrelia burgdorferi [6], have led to the development of a model whereby certain endogenous Fn activities are activated allosterically by Fn-binding MSCRAMMs. Specific intramolecular interactions exist in native plasma Fn which hold its solution conformation in a relatively compact state [18]. These interactions are mediated, in part, by FnI modules from the Fn-NTD fragment and FnIII modules originating from the distant Fn-CBD fragment (Fig 1A) [19][20][21][22]. Intrinsically disordered sequences from SfbI, F1, and BBK32 engage the Fn-NTD fragment via a tandem β-zipper model of binding [6,[23][24][25] and, in doing so, compete for these intramolecular Fn contacts [6,12,24]. This competition results in a conformational expansion of Fn which has been directly measured using dynamic light scattering (DLS) upon binding of SfbI [12]. Interestingly, the SfbI-induced conformational change in Fn mirrors the transition of compact Fn to an elongated structure in solutions of increasing ionic strength [18].
A monoclonal antibody (mAbIII-10), which recognizes a conformationally-sensitive epitope within 10 FnIII [26] has also been used to monitor structural changes induced in Fn by both streptococcal and borrelial FnBPs [6,12,17]. These studies reveal a conformational rearrangement in Fn that occurs in domains located far outside of the FnBP/Fn binding site. The conformational expansion of Fn induced by binding of SfbI, F1, or BBK32 has been linked to three primary effects; (i) the 10 th FnIII module of Fn which harbors the "RGD" integrin recognition motif exposes a previously cryptic mAbIII-10 epitope, (ii) the motogenic "IGD" motifs of the 7 th and 9 th FnI modules of the Fn-GBD fragment become exposed, and (iii) in the case of SfbI, binding results in the blocking of Fn assembly into fibrils. In addition to binding to the Fn-NTD fragment, SfbI, F1 and BBK32 also harbor an Fn-GBD binding site. However, Fn-NTD interaction alone is sufficient to induce allosteric changes in Fn [12]. Consistent with this is the recent discovery of the Streptococcus equi FnBP termed SFS, which only binds Fn via Fn-GBD interaction, yet fails to cause conformational expansion of Fn [27].
While S. aureus is typically considered an extracellular pathogen, intracellular S. aureus is associated with several chronic and reoccurring infections [47][48][49][50] and can establish infection even in the presence of the vancomycin [51]. Thus, S. aureus internalized by non-professional phagocytes represents a bacterial reservoir protected against antibiotics and innate host defense Fn is a large C-terminally disulfide linked heterodimer that is modular in nature. Alternative splicing results in Fn monomers which are similar but not identical [5]. For clarity, only the larger monomer is shown here. Fn consists of three domain types each with distinct folds: FnI, FnII and FnIII (Uniprot ID: P02751). Splice variation results in a variable region (V) in a single Fn monomer. Several functionally relevant protease-stable Fn fragments are known, including the Nterminal domain (Fn-NTD), the gelatin-binding domain (Fn-GBD), and the cell-binding domain (Fn-CBD). The Fn-CBD fragment harbors the canonical α 5 β 1 "RGD" interaction motif within the 10 th FnIII module and "synergy site" in the 9 th FnIII module. Two free cysteines (SH) exist in each Fn monomer which were used to site-specifically label the FRET probes used in this study and are located in the 7 th and 15 th FnIII modules. Arrows indicate specific intramolecular interactions formed between domains of the Fn-NTD fragment and the distant Fn-CBD fragment. (B) Domain organizations of mature FnBPA from S. aureus. The extracellular region of FnBPA consists of an A domain that interacts with fibrinogen, followed by a sequential arrangement of 11 Fn binding repeats, the proline-rich (PR) and cell wall (W) spanning regions, and a C-terminal sortase A recognition motif (LPETG) (Uniprot ID: P14738). Sequence alignment of Fn-binding repeats used in this study are shown. Highlighted in cyan are the residues within each repeat that bind to or are presumed to engage Fn-NTD. (C) Domain organizations of mature BBK32 from B. burgdorferi (Uniprot ID: O50835). The residues involved in binding to Fn-NTD are shown in cyan. Unlike FnBPA BBK32 also encodes a binding site for Fn-GBD indicated by the underlined sequence [6]. The recently identified C1 binding domain is also indicated [7]. All domains are drawn to the scale using IBS (Illustrator for Biological Sequences, http://ibs. biocuckoo.org).
systems. Although it is evident that FnBPA, Fn, and α 5 β 1 integrin work in concert to promote an intracellular "life-style" for S. aureus [14][15][16]40], detailed knowledge of the underlying molecular events that lead to invasion have remained elusive. A mechanistic understanding of how FnBPA manipulates Fn function to enable entry of S. aureus into non-phagocytic cell types, will benefit renewed efforts [51] to target this form of the pathogen. With this in mind, the goals of this study were to determine if the allosteric model of Fn activation, which has been proposed for streptococcal and borrelial FnBPs, applies to staphylococcal FnBPA. Furthermore, we sought to better understand the nature of structural changes induced by Fn-binding MSCRAMMs and correlate these changes to quantitative and direct measurements of Fn/ α 5 β 1 integrin interaction. Therefore, in parallel with our studies of FnBPA, we have also evaluated the effect of borrelial BBK32 (Fig 1C) on the solution structure of Fn and its effect on Fn/α 5 β 1 interaction. Herein, we apply a cross-disciplinary approach to dissect the structural basis for MSCRAMM-induced Fn activation using circular dichroism (CD), DLS, and fluorescence resonance energy transfer (FRET) to examine the conformational status of native Fn or MSCRAMM-bound Fn in solution. Surface plasmon resonance (SPR) is used to quantitatively measure the effects of FnBPA and BBK32 on direct Fn/α 5 β 1 interaction. Together these experiments provide a detailed structure-activity relationship of MSCRAMM/host interaction and strongly suggest an allosteric mode of Fn/α 5 β 1 affinity-enhancement for Fn-binding MSCRAMMs.

Materials and Methods Materials
Human Fn was purified from freshly drawn citrated plasma (Gulf Coast Regional Blood Center, Houston, TX) using gelatin affinity chromatography combined with arginine affinity chromatography as described previously [52]. Fn was stored at 4°C in Tris-buffered saline (TBS: 50 mM Tris-HCl, pH 7.4, 150 mM NaCl), and used within one month. Fn dimer concentration was calculated from OD 280 nm with EC 280 (1%) = 12.8 and molecular weight of 500 kDa. The ectodomain of integrin α 5 β 1 -Fc fusion protein, which contains the N-terminal 613 residues of α 5subunit and fragment 121-455 of β 1 -subunit, was generated and purified as previously reported [53]. Fn-NTD and Fn-CBD were purified as previously described [54]. Expression and purification of recombinant GST-FnBPAs, FnBPA-10 peptide and BBK32  was performed as previously described [46,55]. Rabbit anti-Fn pAb (polyclonal antibody) was purchased from ICN Pharmaceuticals, Inc. (Costa Mesa, California). BSA (bovine serum albumin), heparin sodium (from porcine intestinal mucosa, 17,000-19,000 Da) was from Sigma.

Dynamic light scattering
DLS measurements were performed using DynaPro Titan Ambient laser unit (Wyatt Technology Corporation) with wavelength at 828.5 nm and 10% to 12% laser power. Samples were dissolved in TBS and filtered through 0.22 μm filter to eliminate any particles. Fn concentration was maintained at 0.8 mg ml -1 (1.6 μM as dimer) in TBS (50 mM Tris-HCl pH 7.4, 150 mM NaCl) for the conditions tested. After equilibration at room temperature for 15 to 30 min, the samples were measured three times, with each measurement lasting 100 seconds (10 acquisitions, 10 seconds each). Data analysis was performed using Dynamics V6 software.

Circular Dichroism
CD measurements in the near UV region (350-250 nm) were carried out at ambient temperature on a Jasco J-720 spectropolarimeter (Easton, MD) with a 1 cm cell. Ten scans were collected and averaged at a scan speed of 200 nm min -1 , with a time constant of 1 s and band width of 1 nm. A mean residue molecular weight of 110 was used in the calculation of mean residue ellipticity. The spectra of the Fn solution were background-corrected with the CD signal obtained from the buffer or molecules mixed with Fn.

Fluorescence resonance energy transfer
Fluorescence labeling of human plasma Fn was performed at room temperature as described previously [56]. The fluorescence probes used in this study were purchased from Molecular Probes. Briefly, newly purified human plasma Fn at 1.2 mg ml -1 in TBS was partly denatured in 4 M GdnHCl (Sigma) for 15 min to expose the free cysteines in the Fn dimer. The denatured Fn was mixed with thiol-reactive Alexa Fluo 1 546 C5-maleimide (AF546,~15-fold molar excess over Fn monomer concentration) and allowed to incubate for 2 hr with gentle rocking. The acceptor fluorophore AF546-conjugated Fn (~0.9 mg ml -1 ) was labeled with a second probe, the amine-reactive Alexa Fluo 1 488 carboxylic acid-succinimidyl ester-mixed-isomers (AF488,~20-fold molar excess). After each labeling, the unbound dye was removed using a PD-10 desalting column (GE Healthcare). The labeling ratios were~8 AF488 donors and~4 AF546 acceptors in each Fn dimer.
FRET within donor and acceptor-labeled Fn was measured using spectrofluorimeter LS 50B (Perkin-Elmer) at room temperature with excitation wavelength at 493 nm. The sensitivity of FRET response to the Fn unfolding was evaluated under mild denaturant (up to 2 M guanidinium chloride (GdnHCl)). Measurement of MSCRAMM-induced FRET within Fn was performed by titrating 0.5, 2 and 5 μl of 50 μM of BBK32 or FnBPA-10 in to 500 μl of labeled Fn. Subsequently, the emission spectra collected for each titration was recorded and normalized to the donor emission peak so that changes in energy transfer were reflected only by changes in the acceptor peak. Control experiments were conducted using an equimolar solution of donorlabeled Fn with acceptor-labeled Fn in the presence and absence of 2M GdnHCl to confirm the absence of energy transfer between adjacent Fn molecules in solution.

SPR-based Biacore analysis of Fn and α 5 β 1 interaction
Experiments were performed at 25°C on a Biacore 3000 (GE Healthcare Bio-Sciences AB, Uppsala, Sweden) using HBS-T (10 mM HEPES, pH 7.3, 150 mM NaCl, 0.005% Tween-20) containing 1 mM MnCl 2 . To prepare sensor surface for kinetic experiments,~1400 RU of α 5 β 1 was immobilized to a CM5 chip by amine coupling. Briefly, 18 μL of protein (10 μg ml -1 in 10 mM sodium acetate, pH 4.8) was injected onto an activated (4 min activation) surface at a flow rate of 5 μl min -1 . BSA was coupled to the adjacent flow cell and served as a reference surface (~5000 RU). A flow rate of 30 μL min -1 was used for all binding experiments. The sensor chip surface was regenerated to remove bound Fn by injection of 1 M NaCl for 1 min. All measurements were baseline corrected by subtracting the response from the reference surface. For kinetic analysis, signals from buffer blanks were subtracted.
Kinetic constants were obtained from curve fitting to the predefined bivalent analyte model using BIAevaluation software (Version 4.1). Heterodimeric Fn has two identical binding sites for integrin α 5 β 1 . The reaction between soluble Fn analyte (A) and immobilized integrin ligand (L) can be described by this equation: Binding to the first ligand molecule is described by a single set of rate constants (k a1 , k d1 ), so that the two sites on the analyte are equivalent in the first binding step. Binding of the second ligand molecule is described by a second set of rate constants (k a2 , k d2 ), which is reported in an unconventional units of RU -1 s -1 . The first set of rate constants were used for calculating the dissociation constant K D (K D = k d1 /k a1 ).

ELISA-type binding assays
Experiments were conducted at room temperature and each sample was assayed in triplicate. Immunoassay microplate Immulon 2HB (Thermo Scientific, Waltham, MA) was coated with 2 μg of integrins dissolved in TBS for 1 hour. The plate was then blocked for 1 h with 1% (w/v) BSA in a buffer "A" (TBS containing 1 mM MnCl 2 ). Citrated plasma (stored at -20°C for less than 6 months) was fast-thawed at 37°C and spun to remove aggregates. The supernatant was diluted 10-fold in buffer "A" and then mixed with varying concentrations of BBK32 to a final concentration of 5% plasma. The sample mixtures were added onto the plate and incubated for 1 h. The plate was washed three times with buffer "B" (TBS with 1 mM, MnCl 2 , and 0.05% Tween-20) before addition of rabbit anti-Fn antibody. Following washing with buffer "B", APconjugated goat-anti-rabbit IgG (BioRad, Hercules, CA) was added to the plate and incubated for 1 h. All antibodies were diluted 2000-fold in buffer "B" containing 1% BSA. Alkaline phosphatase substrate PNPP (Pierce, Rockford, IL) was added after washing, and the absorbance was measured at 405 nm using a Thermo Max plate reader (Molecular Devices, Sunnyvale, CA).

Results
The tertiary structure of plasma Fn is altered by FnBPA and BBK32 Although high resolution crystal structures have been solved for a number of smaller Fn fragments, a complete structure of full-length Fn remains to be determined. However, a wide range of biophysical studies using DLS [18,[57][58][59][60][61], CD [59,62,63], steady-state fluorescence spectroscopy [59,63,64], small-angle X-ray scattering (SAXS) [65] and small-angle neutron scattering (SANS) [60,65,66] have yielded a wealth of information about the conformational states and structure of native Fn in solution. Conformational changes induced by FnBPs underlie the allosteric model of Fn activation proposed for streptococcal and borrelial FnBPs [6,12,17], and a previous study using DLS indicates an expansion of Fn structure upon SfbI binding which is similar to what is observed when Fn is incubated in a high salt medium [12]. To understand if analogous conformational changes are induced by staphylococcal FnBPA we used DLS to measure the hydrodynamic radii (R h ) of Fn in the presence or absence of the FnBPA Fn-binding repeat FnBPA-10 (Table 1). Fn adopted an extended conformation in the presence of high salt  [12,18]. Addition of staphylococcal FnBPA-10 (R h = 12.73 ± 0.69) or the borrelial FnBP, BBK32 (R h = 13.00 ± 0.38) resulted in a significantly increased R h , suggesting that Fn indeed adopts an extended structure upon binding to these MSCRAMMs. Native Fn contains large numbers of aromatic residues in its type I and III domains which exist in common hydrophobic environments and give rise to a signature near-UV CD spectrum characterized by two strong negative peaks centered at 291 nm and 299 nm, and a weaker negative peak at 282 nm (Fig 2) [67,68]. Binding of Fn by the glycosaminoglycan heparin has been shown to alter Fn tertiary structure which can be monitored by shifts in this signature spectrum (Fig 2) [69]. When Fn was incubated in the presence of FnBPA-10 or BBK32, both complexes produce significantly different spectra relative to native Fn (Fig 2). Interestingly we noted the similarity in the near-UV spectra for heparin-bound Fn when compared to FnBPA-10/Fn (Fig 2), whereas, the BBK32 spectrum is reminiscent of the partial unfolding of Fn in 4M urea [63]. Importantly, the MSCRAMM fragments used in these experiments are intrinsically disordered polypeptides that do not contain tryptophan residues and therefore lack near-UV optical activity [70] (Fig 2). However, since the Fn-NTD is engaged by BBK32 and FnBPA-10, such interaction could potentially cause near-UV CD optical activity changes within NTD domain. As shown in S1 Fig, BBK32 and FnBPA-10 caused intrinsic Trp fluorescence quenching in Fn (S1A Fig) and Fn-NTD (S1B Fig), suggesting that Trp environments in the NTD and GBD (in the case for BBK32) are affected by the binding and may attribute partially to the near-UV CD changes in Fn.
Data obtained using an anti-Fn antibody (mAb10-III) that recognizes a conformational sensitive epitope indicates that structural changes distant from the streptococcal and borrelial FnBP binding site are induced in Fn [6,12,17] on MSCRAMM binding. To learn if FnBPA causes similar allosteric changes in Fn upon binding, and to confirm apparent conformational changes in solution Fn indicated by CD (Fig 2), we next employed a FRET-based approach. To this end, Fn was site-specifically labelled by conjugating an acceptor fluorophore to free cysteine residues using a technique previously described [56]. Native Fn contains only two free cysteines (4 per dimer molecule) and these are found in the C-terminally positioned 7 FnIII and 15 FnIII modules [71] (denoted in Fig 1A). When Fn is denatured by introducing increasing amounts of GdnHCl (Fig 3A and 3B), or incubated in the presence of increasing ionic strength (Fig 3C), the distance between the amine coupled donor fluorophores and the cysteine coupled acceptor fluorophores increased and the acceptor peak was diminished. There is not FRET  observed between adjacent or denatured Fn molecules (Fig 3D). To determine if the MSCRAMM-induced structural changes measured by CD and DLS involve Fn domains outside of the MSCRAMM binding site, spectra were then collected in the presence of various concentrations of BBK32 ( Fig 3E) and FnBPA-10 ( Fig 3F) GST fusion proteins. These experiments showed a clear dose-dependent loss of acceptor peak signal for MSCRAMM-bound Fn. Importantly, an identical effect is observed when a 36 amino acid peptide lacking the GST fusion and corresponding to the high-affinity FnBPA repeat FnBPA-5 is used (Fig 3G). GST alone or a GST fusion to the low-affinity FnBPA repeat FnBPA-3 exhibits no effect in this assay system (Fig 3H). Taken together, the DLS, CD, and fluorescence based assays indicated that BBK32 and FnBPA profoundly modify the native solution structure of Fn by promoting an extended Fn conformation that results in large scale rearrangement of Fn domains at sites distant from the MSCRAMM binding site.

Fn recognition of the α 5 β 1 integrin is enhanced by FnBPA and BBK32
Fn serves as a primary ECM substrate for several adhesion molecules including the major Fn cell surface receptor, α 5 β 1 integrin [1]. We hypothesized that conformational changes induced upon binding by FnBPA or BBK32, as shown in Figs 2 and 3 and Table 1, may result in modified Fn/integrin interactions. Previous studies on streptococcal and borrelial Fn-binding MSCRAMMs have used the conformational Fn antibody (mAb10-III) as a surrogate for potential effects on Fn/α 5 β 1 integrin interaction. To test this more directly, we developed an SPRbased system to study the effects of FnBPA and BBK32 on the recognition of α 5 β 1 integrin by Fn. Briefly, a recombinant α 5 β 1 -Fc fusion protein representing the minimal functional unit of the integrin [53] was immobilized on the surface of a Biacore CM5 chip. Fn interacts with the α 5 β 1 biosensor and binding is enhanced by Mn 2+ as previously reported (S1A Fig) [53]. Further validation of this experimental approach was performed by using Fn fragments including the cell-binding domain (Fn-CBD) possessing the canonical α 5 β 1 -binding "RGD" ( 10 FnIII) and PSHRN synergy sites ( 9 FnIII) [72,73]. As expected, the Fn-CBD fragment exhibited dose- dependent interaction with α 5 β 1 , while the Fn-NTD fragment lacking the integrin interaction domains failed to bind the integrin (S1B and S1C Fig).
Using this approach we measured a moderate affinity of~1 μM for Fn/α 5 β 1 (Fig 4A-4D) that is in good agreement with affinities reported for fibroblast cell interaction with plasma Fn [74]. To understand if MSCRAMM binding to Fn affects the Fn/α 5 β 1 interaction, we next injected mixtures of BBK32/Fn or FnBPA-10/Fn (fixed MSCRAMM + varied Fn concentrations) over the immobilized α 5 β 1 integrin. A remarkable enhancement of the binding response was observed for MSCRAMM-bound Fn relative to native Fn (Fig 4B-4D). To obtain a quantitative measure of the increase in affinity, kinetic parameters were obtained by fitting SPR response curves to a bivalent analyte binding model (Fig 4B-4D). This model was selected on the basis of closeness of fit (χ 2 1, Table 1) and the expectation that dimeric Fn contributes one identical α 5 β 1 binding site per subunit. Analysis for each interaction revealed that the approximately five-fold increase in apparent affinity (K D ) (K D Fn/α5β1 = 1000 nM vs. K D BBK32/Fn/α5β1 = 170 nM & K D FnBPA-10/Fn/α5β1 = 260 nM) was due to an increase in the initial association rate ([k a1 = (1.5, 6.1, & 7.5) x10 4 M -1 s -1 ], respectively). Interestingly, all other rate constants (k d1 , k a2 and k d2 ) ( Table 2) were similar across experiments, indicating that the Fn/α 5 β 1 binding site present in native Fn is the same for MSCRAMM-bound Fn.
To address the specificity of enhanced Fn/α 5 β 1 activity and to further dissect the kinetic observations presented above, we next injected a mixture of varied MSCRAMM and fixed Fn concentrations (Fig 5A-5C). Again we observed an increase in response that behaved in a dose-dependent and saturable manner, similar to the results obtained in Fig 4B-4D. Likewise, only the initial association rate k a1 was changed, while the stability of the complex (k d1 ), the second rate step (k a2 and k d2 ), and maximum response when all the binding sites on active α 5 β 1 are saturated (R max ) were unaffected by the presence of MSCRAMMs (Fig 5B-5D). The initial  Table 2. There was no detectable binding when FnBPA-10 (1 μM) or BBK32 (0.5 μM) were injected over the same α 5 β 1 surface (data not shown). High-affinity FnBPA/Fn-NTD interaction is required for enhancement of Fn/α 5 β 1 activity Of the eleven Fn-binding repeats found in FnBPA, six have been characterized as high-affinity binders (FnBPA-1, -4, -5, -9, -10, -11). The higher affinity of these repeats has been attributed to their ability to interact with four sequential FnI modules of the Fn-NTD fragment [46]. For example, while FnBPA-10 contains correctly spaced motifs to interact with 2-5 FnI in a β-tandem zipper model of binding, the low-affinity repeat FnBPA-3 appears to lack the 5 FnI interaction motif. Intriguingly, these particular Fn modules have been implicated in mediating longrange intra-molecular contacts within native Fn dimers (Fig 1A) [19,21,22]. To understand if high-affinity interaction is required for an MSCRAMM-induced increase in Fn/α 5 β 1 recognition, we next tested a panel of high and low-affinity FnBPA repeats (Fig 6). All high-affinity Association (k a1 , k a2 ) and dissociation rate (k d1 , k d2 ) constants obtained from fitting SPR response curves (Fig 4B-4D)  repeats studied (FnBPA-1, -5, -9, -10, and -11) increased the Fn/α 5 β 1 binding response, while the low-affinity repeat FnBPA-3 exhibited control-level signal.

MSCRAMMs enhance Fn/α 5 β 1 interaction in human plasma
The SPR-based activity assay utilizing purified native and recombinant reagents allowed for a quantitative kinetic assessment of the MSCRAMM-induced enhancement of Fn/α 5 β 1 binding. We next asked if this effect could be recapitulated using human plasma as a source of native Fn. Indeed, an MSCRAMM-dependent increase in plasma Fn/α 5 β 1 binding was observed in the presence of FnBPA-10 or and to a greater extent for BBK32 when 5% human plasma (~15 μg ml -1 or equivalently~30 nM Fn) was used in an ELISA-type binding assay (Fig 7).

Discussion
Modeling the allosteric activation of Fn/α 5 β 1 by Fn-binding MSCRAMMs By manipulating the conformation of key Fn functional domains, Fn-binding MSCRAMMs enable pathogens to hijack normal host physiology. In the case of S. aureus, enhanced Fn/α 5 β 1 interaction leads to recruitment of focal contact-associated proteins and subsequent integrin clustering at the bacterial attachment site initiates intracellular signaling through the focal adhesion kinase (FAK) and Src kinases [33,43,75]. This appears to confer an advantage to S. aureus by facilitating immune evasion and serves as a bacterial reservoir in chronic infections Relationship of high affinity Fn-NTD fragment interaction and α 5 β 1 integrin activation. SPR sensorgrams were generated by injecting 50 nM Fn in the presence of GST-fused FnBPA repeats (0.5 μM for high-affinity repeats -1, -5, -9, -10, -11 and 15 μM for low affinity repeat -3) over immobilized α 5 β 1 . The K D reported for direct binding of Fn-NTD by high affinity FnBPA repeats are provided for reference and were obtained from previously published isothermal titration calorimetry (ITC) experiments [46]. The weak affinity between FnBPA-3 and Fn-NTD was measured by SPR (this work, data not shown). None of the repeats bound directly to α 5 β 1 (data not shown).
doi:10.1371/journal.pone.0159118.g006  [33,51]. In light of the observations above, we propose a model for the allosteric enhancement of Fn/α 5 β 1 binding by staphylococcal FnBPA (Fig 8), similar to that put forward for streptococcal and borrelial FnBPs [6,12,13,17]. We speculate that dimeric native Fn exists as an array of conformers in solution. At equilibrium, predominate forms of Fn are compact and sterically occlude α 5 β 1 interaction domains ( 9 FnIII-10 FnIII) by long-range intramolecular interactions, thereby impeding Fn/α 5 β 1 recognition. The equilibrium of Fn conformers shifts upon highaffinity binding of FnBPA and native intramolecular contacts are disrupted exposing previously cryptic α 5 β 1 binding sites and promoting Fn/α 5 β 1 binding. Domains involved in integrin binding by Fn are located at a site distant from the MSCRAMM binding site. It appears then that a critical feature of MSCRAMM-induced activation of Fn/α 5 β 1 is allostery and allosteric modes of host protein modulation have been previously reported for S. aureus [76]. Our findings also suggest that the conformation of integrin ligands, like Fn, may fine-tune affinities and therefore contribute to integrin "activation" traditionally associated with conformational change of the integrin ectodomains [77]. The data presented here provides a foundation to study the potential involvement of Fnbinding MSCRAMMs in several aspects of Fn physiology. In addition to Fn/integrin interaction, cryptic functional sites in Fn play a fundamental role in mechanoresponsive and motogenic pathways [19,78]. Interestingly, polymorphisms have been recently reported in two highaffinity FnBPA repeats (FnBPA-5 and -9) and S. aureus strains harboring these mutations are highly correlated with infection of cardiovascular devices [79,80]. How these FnBPA mutations impact bound Fn tertiary structure is currently unknown, however, we predict that sequence changes to residues in FnBPA repeats could result in different global Fn structure and thus function. Finally, as the clinical relevance of intracellular forms of S. aureus are becoming  [6,12,17], we constructed the above model for allosteric affinity enhancement of the Fn/α 5 β 1 interaction by staphylococcal FnBPA. Fn is represented in its native state (left) or bound by FnBPA (right). The native Fn model including intermolecular contacts are adapted from a previously proposed model based on multiple independent studies. In a native state, major solution conformations of Fn occlude the canonical α 5 β 1 interaction motifs located in 9-10 FnIII domains. Upon binding to Fn, FnBPA disrupts specific intermolecular contacts in the Nterminal domain of Fn (Fn-NTD) resulting in global structural rearrangements at sites distant from the FnBPA binding site. Previously cryptic integrin interaction sites are exposed, promoting interaction of Fn with α 5 β 1 . This initiates 'outside-in' signaling events leading to cytoskeletal rearrangement and eventual internalization of Fn-coated S. aureus. doi:10.1371/journal.pone.0159118.g008 clearer [51], elucidation of the molecular mechanisms of Fn-binding MSCRAMMs, like FnBPA, stands to provide a major contribution towards improving the treatment of staphylococcal infectious diseases.