Mapping of the fibrinogen-binding site on the staphylocoagulase C-terminal repeat region

Fibrin (Fbn) deposits are a hallmark of staphylocoagulase (SC)-positive endocarditis. Binding of the N terminus of Staphylococcus aureus SC to host prothrombin triggers formation of an active SC·prothrombin∗ complex that cleaves host fibrinogen to Fbn. In addition, the C-terminal domain of the prototypical SC contains one pseudorepeat (PR) and seven repeats (R1 → R7) that bind fibrinogen/Fbn fragment D (frag D) by a mechanism that is unclear. Here, we define affinities and stoichiometries of frag D binding to C-terminal SC constructs, using fluorescence equilibrium binding, NMR titration, alanine scanning, and native PAGE. We found that constructs containing the PR and single repeats bound frag D with KD ∼50 to 130 nM and a 1:1 stoichiometry, indicating a conserved binding site bridging the PR and each repeat. NMR titration of PR–R7 with frag D revealed that residues 22 to 49, bridging PR and R7, constituted the minimal peptide (MP) for binding, corroborated by alanine scanning, and binding of labeled MP to frag D. MP alignment with the PR–R and inter-repeat junctions identified critical conserved residues. Full-length PR–(R1 → R7) bound frag D with KD ∼20 nM and a stoichiometry of 1:5, whereas constructs containing the PR and various three repeats competed with PR–(R1 → R7) for frag D binding, with a 1:3 stoichiometry. These findings are consistent with binding at PR–R and R–R junctions with modest inter-repeat sequence variability. CD of PR–R7 and PR–(R1 → R7) suggested a disordered flexible structure, allowing binding of multiple fibrin(ogen) molecules. Taken together, these results provide insights into pathogen localization on host fibrin networks.

secreted virulence factors, originally grouped into 12 serotypes based on their genetic diversity (3), and with recent serotype additions now a total of 15 (4,5). SC contributes to abscess formation, typical of S. aureus infection (6)(7)(8). Interaction of host fibrinogen (Fbg) with the S. aureus virulence factors, clumping factor A and B, FbnA, efb, and SC, was reported previously (9)(10)(11)(12)(13); however, several mechanisms underlying these specific Fbg interactions are not yet defined at the molecular level. Coagulase-positive S. aureus adheres to exposed subendothelium on heart valves damaged by turbulent blood flow, where deposition of platelets and fibrin (Fbn) creates adhesion foci for circulating bacteria (1). Fbg also facilitates interactions between S. aureus and platelets (14). Our previous fluorescence equilibrium binding and native PAGE studies revealed for the first time that Fbg and Fbn bind free SC as well as its complex with host prothrombin (ProT); and that the fragment D (frag D) domain of Fbg/Fbn harbors major interaction sites for binding to the C-terminal repeat region of SC (15,16). The current work identifies the binding motifs in the C-terminal SC region, essential for interaction with frag D, and defines the stoichiometry of frag D binding to free and ProT-complexed SC.
SC is a bifunctional protein that uses its N terminus for conformational activation of ProT, the precursor of the central clotting protease, thrombin (T). Classical serine proteinase activation requires proteolytic cleavage of the Arg 15 -Ile 16 peptide bond in the zymogen (chymotrypsinogen numbering), with release of a new N terminus that inserts in the Ile 16 -binding pocket to trigger formation of the active site (17,18). In contrast, our structure-function studies showed that formation of the active site in ProT by SC is a nonproteolytic mechanism, in which SC inserts its N terminus into the zymogen activation pocket, thereby forming a critical salt bridge required for conformational expression of the active site in ProT (19,20). In the absence of SC, Fbg binds to T at exosite I and is cleaved as a substrate. In the SC⋅ProT* (* denotes a fully expressed active site) and SC⋅T complexes, (pro)exosite I is occupied by the tight-binding N-terminal D2 domain, and a new binding site is expressed on these complexes for Fbg as a substrate (21). The SC(1-325) fragment, containing the D1 and D2 domains, is sufficient for expression of this novel Fbg-binding site on the SC⋅ProT* and SC⋅T complexes and triggering of proteolytic Fbg cleavage to form Fbn clots.
In addition to the N-terminal D1 and D2 ProT-binding domains, the prototypical full-length SC(1-660) from the S. aureus strain Newman D2 Tager 104 also contains a central region with unknown function and a C-terminal domain consisting of one 32-residue pseudorepeat (PR) and seven 27-residue tandem repeats, R1 → R7 ( Fig. 1) with highly conserved inter-repeat sequences. Our studies use the Tager 104 sequence with GenBank accession number AY225090.2 as identified by us (19). SC from other strains contains the PR and four to nine repeats (Fig. S1) (22,23). Fbg binding to SC was recognized previously (14,24,25), and we narrowed down this interaction to Fbg frag D binding to the C-terminal domain (15,16,26,27). Binding of Fbg and frag D to the SC C-terminal domain has been studied by turbidity assays, ELISA, isothermal titration calorimetry, and solid-phase Fbg binding of an SC C-terminal domain construct (25,28,29). The present work identifies a conserved minimal peptide (MP), required for binding of PR-R constructs to frag D, by equilibrium binding, NMR titration, and alanine (Ala) scanning. The MP sequence bridges the PR and R junctions, exhibits alignment with inter-repeat junctions, and contains a set of conserved residues that are critical for frag D binding. Knowledge of this minimal binding site is useful for future development of mechanism-based antibody drugs targeting the SC C-terminal domain.
We present an in-depth analysis of the potential of SC to bind multiple frag D units at its C-terminal repeat domain, independent of substrate Fbg binding to the catalytic SC⋅ProT* complex. We show that the full-length C-terminal repeat construct, PR-(R1 → R7) binds frag D with a stoichiometry of 5 frag D per construct; and constructs containing a leading PR and three R units in consecutive or arbitrary order bind frag D with a stoichiometry of 3. Shortening of the fulllength repeat construct or changing the order of repeats did not alter the ability of repeats to bind frag D, reaffirming the requirement of aligned conserved residues at the repeat junctions for formation of multiple binding sites. Consistent with these observations, our data of in vivo localization of SC⋅ProT* complexes to areas rich in Fbg show involvement of a template mechanism that utilizes the SC C-terminal domain (15). The high affinity of multiple binding sites for Fbg/Fbn D domains on a single SC molecule contributes to a mechanism of selective anchoring of proteolytically active SC⋅ProT* complexes on host Fbn-bacteria vegetations subjected to arterial shear stress.

Sequence alignment of the SC C-terminal repeats
The C-terminal repeat alignment of SC(1-660) S. aureus Newman D2 Tager 104 was performed using an online box shade alignment program, http://embnet.vital-it.ch/software/ BOX_form.html. Figure 2 shows the sequence alignment of PR, 32 amino acid residues, with R1 to R7, each 27 residues in length. PR and the repeats show a 30 to 40% conserved sequence, and there is >80% pairwise identity among the repeats. Repeats R2, R4, and R5 are identical. We identified the MP sequence as described later, and in Figures 3-11, the MP alignment is shown with the bridging sequences between repeats and conserved residues (red) at positions G 37 , Y 41 , A 43 , R 44 , P 45 , K 49 , and P 50 (PR-R7 numbering). The residues G 28 and I 31 are present only in the PR and are replaced with similar size hydrophobic residues A and V in the repeats. PR-Rx constructs used in this study were created with a 10-residue SC sequence preceding the PR to preserve the integrity of a putative binding site; five residues of the SC sequence were added at the Figure 1. Domain organization of full-length SC and multirepeat constructs. Full-length SC secreted by the Staphylococcus aureus Newman D2 Tager 104 contains three major regions: the N-terminal D1 and D2 domains (I), a central region (II), and the C-terminal repeat region (III), consisting of one pseudorepeat (PR) and seven repeats (R1-R7). The five residues at the C-terminal end are conserved (IV). SC, staphylocoagulase.
C-terminal end of constructs ending in R7 to simplify expression, and a Cys residue was introduced for labeling purposes, as detailed in the "Experimental procedures" section.
Size-exclusion chromatography and light scattering SC(1-325) containing only the N-terminal domains D1 and D2, and full-length SC(1-660) that includes the C-terminal repeats, were chromatographed separately and in mixtures with frag D. Figure 3A shows the elution profile of SC(1-325) (black), frag D (red), and their mixture (blue). Light scattering indicated that SC(1-325) (Mr = 38,000 Da) eluted with an apparent molecular mass of 32,500 Da. Frag D (relative molecular mass [M r ], 88,000 Da) showed a hydrodynamic property-related anomaly, eluting in the same apparent molecular mass range, in part because of interaction of frag D with the column matrix. The SC(1-325) and frag D mixture eluted at a position similar to SC(1-325) and frag D alone, suggesting that SC(1-325) does not form a high M r complex with frag D. Figure 3B shows the elution profile of SC(1-660) (black), frag D (red), and their mixture (blue). Light scattering indicated that SC(1-660) (M r , 74,390 Da) eluted with apparent molecular mass of 81,200 Da. The mixture of SC(1-660) and frag D showed a new peak eluting at 165,000 Da, suggesting that the C-terminal region of SC(1-660) binds frag D.
Native PAGE of frag D binding to SC(1-660), [5F]PR-R1, the SC(1-660)⋅[5F]ProT complex, and labeled C-terminal repeat constructs Formation of multimeric SC(1-660)⋅frag D complexes was confirmed by native PAGE (Fig. 3C). Frag D (lane 10, M r , 88 KDa) runs faster than SC(1-660) (lane 1, M r , 74.4 KDa) because of combined differences in isoelectric point, hydrodynamic properties, secondary structure, and native charge density, unlike in SDS-PAGE. Minor frag D proteolysis bands were due to Fbg digestion by trypsin during frag D preparation. Upon binding of frag D to SC(1-660), the SC(1-660)⋅frag D complex band shifted, starting at twofold molar excess frag D, and the shift and band intensity became more pronounced at increasing frag D (lanes 2-9). Unbound frag D was observed starting at sevenfold molar excess (lanes 7-9), suggesting that the C-terminal region of SC(1-660) binds multiple molecules of frag D, possibly five to six.
Tight-binding complexes of SC(1-660) with ProT, covalently labeled at the active site with ([5F]ProT) (30), recruited  Mapping the fibrinogen-binding site on the SC C terminus multiple frag D molecules at the C-terminal SC domain. In incubations of 2.5 μM [5F]ProT with 3.75 μM SC, [5F]ProT binding was saturated because of the tight K D of this interaction. Subsequent incubation of the complex with molar excess frag D yielded higher order complexes (Fig. S3, A and B). All the [5F]ProT was bound in the SC(1-660)⋅[5F]ProT binary complex, with 1:1 M stoichiometry (lane 3). Binding to multiple frag D molecules occurred, with unbound frag D observed at eightfold molar excess (lane 9).
Frag D bound to [5F]PR-(R1 → R7) containing all seven repeats in order (Fig. S4, A and B). Higher order complex formation was indicated by an upward shift of fluorescence and Coomassie-stained bands with increasing frag D. Unbound frag D appeared at an approximately eightfold molar excess (lane 7), suggesting a binding stoichiometry of 7.
Frag D bound to [5F]PR-R1R6R7 and [5F]PR-R1R2R3, suggesting the presence of binding sites bridging continuous and noncontinuous repeat sequences, and a binding role of conserved residues contained in the bridging areas, independent of the order of the repeats (Figs. S5 and 11). Unbound frag D was observed above 3.5-fold molar excess, suggesting binding of 3 frag D molecules per construct, consistent with binding sites formed at the PR-Rx and Rx-Ry junctions.
Equilibrium binding of [5F]-labeled C-terminal repeat peptides to frag D No increase in fluorescence intensity was observed when [5F]PR and [5F]R1 were titrated with frag D, suggesting that these peptides do not bind frag D (Fig. 4A), consistent with our native PAGE findings. In contrast, [5F]PR-R1, [5F]PR-R2, [5F]PR-R3, [5F]PR-R6, and [5F]PR-R7 bound to frag D, with 1:1 stoichiometry and dissociation constants (K D ) ranging from 49 to 130 nM (Fig. 4, B-F). The results for [5F]PR-R1 were in excellent agreement with our previously reported K D of 36 ± 8 nM for frag D binding (15). The K D values, ΔF max /F o , Titrations and data fitting were performed as described in the "Experimental procedures" section. The parameters are given in Table 1. frag D, fibrinogen fragment D; PR, pseudorepeat; R, repeat.
Mapping the fibrinogen-binding site on the SC C terminus and stoichiometries obtained by quadratic fit are listed in Table 1. Errors are given by 2× SD (95% confidence interval).
The results indicate the presence of one single binding site for frag D that bridges PR and the individual Rx repeats. Because R4 and R5 are identical to R2, their PR-Rx constructs were not tested for frag D binding. Frag D is a trimer of three polypeptide chains, α, β, and γ. The β and γ chains contain holes at their C-terminal end. The peptide Gly-Pro-Arg-Pro (GPRP) is known to bind and block the β-holes and γ-holes of Fbg and frag D (31)(32)(33).
[5F]PR-R1 was titrated with frag D at saturating GPRP. In the presence of 5 mM GPRP, [5F]PR-R1 still bound frag D with K D of 18 ± 10 nM, and ΔF max /F o 0.08 ± 0.01 (Fig. 5), indicating that [5F]PR-R1, and most likely also the other PR-Rx constructs do not bind the β-chain and γ-chain holes of frag D, but utilize a different binding site on frag D, and that the SC C-terminal domain does not interfere with Fbn polymerization.
NMR assignment of PR-R7, and titration of PR-R7 with frag D The 1 H 15 N heteronuclear single quantum coherence (HSQC) spectrum with the assignments of all PR-R7 residues, including the additional N-terminal and C-terminal residues, was recently published (27). Chemical shift data are retrievable from the Biological Magnetic Resonance Data Bank databank (http://www.bmrb.wisc.edu) under accession number 27036.  The assignment for each correlation peak in the 1 H-15 N HSQC spectrum provides residue-specific information on an atomic level for each amide in the peptide. This allows observation of residue-specific changes of PR-R7 exposed to frag D and results in a clearly defined interaction map. Full assignment was achieved with amide-based experiments, supplemented with novel 13 C-direct detect methods, allowing for full N, NH, and C 0 resonance assignments, with the exception of the N-terminal and C-terminal one (27). The 13 C direct detect experiments allowed to unambiguously assign residues that were problematic in amide-based experiments, such as prolines; Glu 2 , Thr 3 , Tyr 24 , Tyr 34 , Arg 44 , and Val 72 because of heavy peak overlap; low-intensity peaks for His 8 , Asn 16 , Asn 54 , His 61 , Ala 62 , and Ser 70 ; and Tyr 9 as it was between the unassigned His 8 and Pro 10 , and therefore, no connections could be assigned (Fig. S6). Peak overlap and lowintensity peaks remained an issue for the analysis of the HSQC-based titration experiments, but unambiguously knowing the exact position of these peaks from the 13 C direct detect experiments was crucial during the refinement of the titration analysis. Given the very limited chemical shift dispersion of 8.2 ± 0.27 ppm in the proton dimension, which led to the peak overlap, it was anticipated that the PR-R7 peptide had limited to no secondary structure. With the backbone assignment in hand, secondary structure prediction using the program Talos+ (34) confirmed that there was no consistent secondary structure to be anticipated in PR-R7 alone (Fig. S7).
With the HSQC spectrum assigned, binding of frag D to PR-R7 was investigated in sodium phosphate buffer (pH 7.0). The initial spectrum contained 65 μM of 15 N-labeled PR-R7 peptide alone and exhibited all expected nitrogen-proton resonances. Upon successive addition of aliquots from a 240 μM stock solution of nonlabeled frag D, most resonances became weaker as more frag D was added (Fig. 6). Individual titration points were chosen at 0.5, 1.0, 2.0, 3.21, and 4.95 molequivalents (me) of frag D to PR-R7 peptide. Decreasing intensity of most resonances in the HSQC spectrum was attributed to intermediate exchange, and formation of a complex, affecting the correlation time of the molecule substantially broadening those resonances. The titration was analyzed by measuring the peak intensity for each residue of  the first three titration spectra, recording them as a percentage of the reference spectrum (Fig. 7). The peak intensity change follows the same trend for each individual residue after each addition of frag D. Only the first three titration points at 0.5, 1.0, and 2.0 me of frag D contained meaningful information for the majority of residues and were analyzed. At higher frag D concentrations, most resonances were exchange broadened or affected by the slower tumbling of the larger frag D⋅PR-R7 complex to the point where individual peaks were no longer contributing in a relevant matter to the analysis. Only about a dozen high-intensity peaks remained visible in the spectra with 3.21 and 4.95 me of frag D. Manual analysis of overlapping peaks assured correct intensity representation. HSQC spectra of residue pairs Glu 2 -Tyr 9 , Ser 7 -Thr 53 , Lys 20 -Asp 26 , Tyr 34 -Phe 39 , and Glu 33 -His 61 were overlapping, with mutual influencing of peak intensities, which complicated independent analysis. The Glu 2 -Tyr 9 and Glu 33 -His 61 pairs merited further analysis. Glu peaks, typically weak to begin with, are often exchange broadened in these spectra. The normally stronger Tyr 9 was weakened by frag D binding, making it look similar to the Glu 2 peak. Similarly, Glu 33 was in addition broadened by its position in the binding interface, whereas the overlapping His 61 was exchange broadened from the beginning, and this effect is accentuated in this environment, causing the intensities of these two peaks to decrease at similar rates. Figure 7 shows that the intensity of Glu 2 and His 61 is clearly smaller than anticipated compared with their neighbors, whereas Tyr 9 and Glu 33 fit in well with their neighbors, indicating that the intensity fit indeed was correct for those residues, but for different reasons. The intensities of other overlapping peaks matched well with those of their neighbors and had no impact on the overall interpretation. The correlation peaks in the HSQC spectrum at 2.0 me of frag D for Lys 17 , Thr 29 , and Asn 35 were no longer detectable, but the first two titration points showed clear signal intensities that were analyzed. More interestingly, Figure 7 reveals a trend where the stretch of residues 20 to 49 shows consistently more diminished signal intensity in all three titration points, compared with the C-terminal residues 51 to 75 and N-terminal residues 2 to 18 that retain a stronger intensity on average. The N-terminal residues three to seven are affected the least, and their signals remain visible even at the highest frag D addition. The most plausible explanation is that frag D has its main binding interaction with PR-R7 within the PR-R7 residues 20 to 49, which suggests a binding site overlapping PR and R7. Significantly less signal reduction was observed for the peaks in the C-terminal residues 51 to 75 leading to the conclusion that this portion of the peptide might contribute less to the interaction with frag D, although weak binding of that sequence with frag D can still be anticipated. The stretch of residues from 22 to 50 was determined to constitute the MP that is required for binding to frag D. Because the NMR studies required 20 mM sodium phosphate, 150 mM NaCl, pH 7.0 buffer, equilibrium binding of frag D to [5F]PR-R7 was also performed in phosphate buffers with corresponding ionic strength at pH 7.0 and 7.4 to determine whether the phosphate buffer and pH affect binding. Titration of [5F]PR-R7 with stoichiometries fixed at 1, yielded K D values of 3 ± 2 μM and 27 ± 75 nM at pH 7.0 and 7.4, respectively. The data at pH 7.4 corresponded well with the results for frag D binding in Hepes buffer. Titrations in phosphate buffer were somewhat noisier than in Hepes buffer, presumably because of probe-buffer interactions. Equilibrium binding of labeled PR-R7 at pH 7.0 was weaker, possibly because of the combination of differences in protonation states of the two His imidazole groups (35) in this construct, and the fluorescein probe with a pK a of 6.4. However, our NMR titration of the labeled peptide, collected at micromolar reactant concentrations, clearly indicated a distinct region of interaction under these experimental conditions.

Equilibrium binding of frag D to the 5F-labeled MP, [5F]MP
The 29-residue MP, identified by NMR studies, was synthesized, with an additional C-terminal Cys for attachment of a 5F probe. Titration of [5F]MP (23 nM) with frag D demonstrated binding, with K D of 5 ± 4 μM, ΔF max /F o 0.17 ± 0.02, and a fixed stoichiometry of 1 (Fig. 8). This indicates that the MP sequence is sufficient to bind frag D.
Equilibrium binding of Ala scanning mutants of PR-R7 to frag D Ala scanning mutagenesis and equilibrium binding further determined the specific PR-R7 residues involved in frag D binding. Sequential mutation of four to six amino acid residue stretches was done in the MP region and beyond, until the end of repeat R7. 10 Ala mutant constructs of [5F]PR-R7 were titrated with frag D. Ala mutants of VKYRDA, GTGIR, EYNDG in PR, and of ARPT, YKKP, and HADG in R7 abolished binding (Figs. 9 and 10A), indicating essential contributions of these sequences to frag D binding. VKYRDA, GTGIR, and EYNDG form a contiguous stretch located at the C-terminal region of the PR. ARPT and YKKP are located at the N-terminal region of R7, whereas HADG is located toward the C terminus of R7. Mutations at TFGYE, SETNA, YNVTT, and TATYG still allowed binding of frag D, with respective K D values of 500 ± 400, 120 ± 40, 350 ± 150, and 260 ± 100 nM (Figs. 9 and 10, B-E). Ala substitution at SETNA resulted in a binding affinity similar to that of [5F]PR-R7, suggesting limited involvement of these residues in binding. Maximum fluorescence intensity increases of these four constructs were comparable to that of native [5F]PR-R7. The sequence alignment of the corresponding MP regions in other interrepeat regions showed strictly conserved G 37 , Y 41 , A 43 , R 44 , P 45 , K 49 , and P 50 (Fig. 11). G 28 and I 31 are present only in the PR, and the corresponding residues A and V in other repeats are hydrophobic and similar in size. Y 41 is highly conserved; however, Ala substitution still allows for weak binding of frag D.
Clotting of Fbg by the SC(1-325)⋅ProT QQQ * and SC(1-660)⋅ ProT QQQ * complexes Cleavage of Fbg to Fbn causes turbidity increase because of formation and polymerization of Fbn with a lower solubility. We compared the rates of Fbn clotting by T, and the active SC(1-325)⋅ProT QQQ * and SC(1-660)⋅ProT QQQ * complexes. ProT(R155Q,R271Q,R284Q), or ProT QQQ , is a human ProT variant with its prothrombinase and T cleavage sites mutated to prevent proteolytic processing (36). It can be activated proteolytically to a meizothrombin form as well as conformationally by triggering formation of an active site by occupation of its Ile 16 pocket by the N terminus of SC. T, as a control reaction, initiated immediate cleavage of Fbg, whereas addition of ProT QQQ in mixtures with SC(1-325) and SC(1-660) showed a lag phase during which the SC(1-325)⋅ ProT QQQ * and SC(1-660)⋅ProT QQQ * complexes are formed (Fig. 12). In the T-initiated or SC(1-325)⋅ProT QQQ *-initiated cleavage of Fbg at 0.5 mg/ml, there was rapid and comparable increase in turbidity reaching a maximal level, indicating clot stabilization. Clot formation by SC(1-660)⋅ProT QQQ * was slower, and the initial rates of turbidity increase, and absorbance readings at clot stabilization were dependent on the Fbg concentration. SC(1-325) conformationally activates ProT but lacks the C-terminal domain for frag D binding. This additional interaction of SC(1-660) may sequester Fbg molecules at the C-terminal domain, thereby decreasing the effective in vitro Fbg concentration available for cleavage by SC(1-660)⋅ ProT QQQ * or posing conformational restrains on cleavage of Fbg bound in the substrate mode, either with or without simultaneous tethering to the SC repeat domain.

Fluorescence anisotropy titrations of repeat constructs binding to frag D
The full-length repeat construct, PR-(R1 → R7), contains the highly conserved PR and seven repeats, of which R2, R4, and R5 are identical. Constructs for direct or competitive binding, compared with full-length SC(1-660), are shown in Figure 1. Using fluorescence anisotropy, we titrated three independently prepared batches of [5F]PR-(R1 → R7) with frag D. Their [5F]PR-(R1 → R7)⋅frag D complexes were used in competitive titrations with three independently prepared PR-R1R2R3 batches to determine K D values for PR-R1R2R3 and to validate the assay and reproducibility. We performed competitive titrations of the [5F]PR-(R1 → R7)⋅frag D complex with PR-R1R6R7, PR-R3R4R7, and PR-R3R6R7  Mapping the fibrinogen-binding site on the SC C terminus  Table 2). Competitive binding of frag D to PR-R1R2R3, PR-R1R6R7, PR-R3R4R7, and PR-R3R6R7 gave stoichiometries of 3 as expected for a construct containing three inter-repeat sequences with conserved residues essential for frag D binding and binding parameters comparable to those of [5F]PR-(R1 → R7). Swapping of the repeat order did not affect the position of critical and conserved residues in sequences bridging the repeats and did not weaken the binding of frag D significantly.
CD spectroscopy of SC repeat peptides and frag D CD spectroscopy of full-length SC(1-660) (blue), SC(1-325) (brown), PR-R7 (black), PR-(R1 → R7) (red), and frag D (green) was performed to determine the presence of secondary structure content (Fig. S9). Frag D showed major helical content with typical minima at 208 and 222 nm. The negative band at 224 nm in the SC(1-325) spectrum was suggestive of helix (37), consistent with its known crystal structure (19). The profiles of PR-R7, PR-(R1 → R7), and SC(1-660) indicated an extended or irregular structure. The 222/208 nm ratio of >1.1 for frag D suggested the presence of coiled-coil helices, consistent with its published structure (32). However, the PR-(R1 → R7) spectrum was more ambiguous, with a positive band at 212 nm suggesting random coil (38,39). Although the 222/208 nm ratio was >1, secondary structure predictions by GOR IV (40) and PSIPRED (41) gave 70% random coil and 30% extended strand content for PR-(R1-R7) and PR-R7, with no helical content. The prediction based on the NMR data confirms the calculations and indicates no presence of secondary structure (Fig. S7). The respective helix, extended

Discussion
Coagulase-positive S. aureus employs various Fbg-binding virulence factors, including SC, to facilitate rapid colonization and spreading inside the human host (43)(44)(45)(46)(47)(48). We showed previously that the complex of SC and SC(1-325) with conformationally activated host ProT forms a new binding site for substrate Fbg, distinct from ProT proexosite I, and proteolytically converts it to Fbn clots that serve as focal points for bacterial adhesion (21). In addition, localization of SC on host fibrin(ogen) is facilitated by the SC C-terminal repeat sequence (15,28,29,49). SC serotypes have four to nine highly conserved repeats, preceded by a PR containing a TFGYE sequence not present in the repeats (3,19,22,50) (Figs. 2 and  S1). These repeats, in part the result of selective pressure, increase the Fbg-binding efficiency of SC. Our studies used SC of the prototypical Newman D2 Tager 104 strain that has seven repeats, with 100% identical R2, R4, and R5, suggesting duplication in the genome (19).
We used the frag D domain of Fbg/Fbn to demonstrate binding to free and ProT-complexed full-length SC(1-660); Cterminal constructs containing PR and one or more repeats; the MP identified by NMR; and to demonstrate the absence of binding to PR, the single repeats, or specific Ala-scanned PR-R7 constructs. The PR-R constructs with one repeat exhibited similar affinities for frag D, suggesting a fairly conserved binding site, even with minor sequence variability among the repeats. We chose PR-R7, with the highest expression levels, for independent and mutually validating binding approaches because NMR typically requires fairly high experimental concentrations.
We identified the PR sequence VKYRDAGTGIREYNDG, strictly conserved among all serotypes, and ARPTYKKP and its respective similar sequences in the N-terminal portion of the repeats, as essential determinants in frag D binding. The PR sequence differs from the previously identified Fbgbinding Coa-RI sequence (28) and represents a partial novel binding site (Fig. 11). Ala scanning including nine conserved residues in the MP abolished or severely weakened frag D binding to PR-R7, supporting frag D-binding sites at the junctions of PR-R and R-R. HADG at the C-terminal end of PR-R7 aligns with YNDG in the MP, containing the conserved G 37 . Disruption of frag D binding by Ala substitution of either sequence (Figs. 9 and 10A) indicates that this G residue is a part of multiple frag D-/Fbg-binding sites in the C-terminal repeat region. PR and R1 by themselves did not bind frag D with measurable affinity, whereas all the PR-R constructs, containing the conserved G/A 28 , I/V 31 , and G 37 residues aligned in the R-R bridging sequences, bound frag D with a 1:1 stoichiometry and K D values in the 50 to 130 nM range. These findings, together with our Ala scanning results, are consistent with a simultaneous requirement for both binding subsites. Tighter binding of PR-Rx compared with MP suggests multiple binding interactions for frag D, and our findings suggest as many as five to six binding sites, in the complete repeat domain of SC (15). The capacity to bind multiple Fbg/Fbn molecules simultaneously may facilitate bacterial localization on blood clots, evasion of the host immune system (51), and may impair the efficacy of antibiotic therapy.
N-terminal cleavage of the Fbg α-chains and β-chains to form Fbn monomer releases the fibrinopeptides AαBβ, with generation of new Gly-Pro-Arg N termini that, respectively, bind the γ-holes and β-holes to form the Fbn network. The tetrapeptide GPRP binds to Fbg and frag D holes with an affinity of 5 mM (31, 32) and is routinely used to prevent Fbn polymerization in solution studies of Fbn monomer. Blocking the frag D holes with GPRP had no weakening effect on its binding to PR-R1 (Fig. 5), excluding them as potential interaction sites. This implies that interaction sites of frag D are located elsewhere in the frag D molecule, possibly in the coiled-coiled region. The modestly tighter affinity for frag D binding (18 nM) compared with that in the absence of GPRP (36-49 nM) may reflect a combination of subtle conformational changes and batch-to-batch PR-R1 and frag D variability. 5.2 ± 0.3 12.0 ± 5.0 0.14 ± 0.01 0.10 ± 0.01 4.8 ± 0. 6 7.4 ± 6.0 0.13 ± 0.01 0.09 ± 0.01 5.1 ± 0. 5 32.0 ± 9.0 0.15 ± 0.01 0.10 ± 0.01 Three independent [5F]PR-(R1 → R7) preparations were titrated with frag D, and the dissociation constant (K D ), stoichiometric factor (n), initial anisotropy (r o ), and maximum anisotropy (r max ) were obtained by fitting to the quadratic binding equation. Competitive binding of PR-R1R2R3 (three independent preparations), PR-R1R6R7, PR-R3R4R7, and PR-R3R6R7 to frag D was measured by titration of two fixed [5F]PR-(R1 → R7)/frag D mixtures with the unlabeled peptides. The two curves of each dataset were fit simultaneously with the two direct [5F]PR-(R1 → R7) binding curves, using the cubic equation, to obtain K D and n for the competitor peptides. Experimental error represents ±2 SD. Binding studies and data analysis were performed as described under the "Experimental procedures" section.
The NMR data further confirmed our hypothesis of a functional binding unit comprised of residues from both PR and the repeats. Decreased NMR peak intensity upon titration with frag D, and residue-specific response showed that the MP, a continuous stretch of 29 amino acids in PR-R7, binds frag D. Residue-specific analysis of the binding interaction between PR-R7 and frag D required initial backbone assignment of this peptide. Very limited dispersion of the NMR signals in the proton dimension raised the suspicion that PR-R7 had limited secondary structure in its apo (unbound) state, which subsequently was shown by the chemical shift prediction using Talos+. This limited dispersion led to substantial signal overlap, and full backbone assignment was ultimately achievable with the aid of novel 13 C-direct detect measurements geared toward such peptides. A very careful and manual analysis of the titration experiments ensured true representation of the peak intensities for each residue. Based on these results, it is clear that an important interaction of frag D consists of residues 20 to 49 on PR-R7. The C-terminal end of the interaction site is defined more clearly than the N-terminal end, although Ala scanning identified both to be required for binding. Both ends are flanked by prolines, and at this point, their role at either side of the MP is unclear. Residues 51 to 75 were significantly less impacted by the presence of frag D, but a secondary interaction with those residues cannot be ruled out as indicated by the Ala scan of that portion of PR-R7. The HADG sequence and aligned YNDG/HANG/HGNG in other repeats may actually participate in subsequent R/R binding of frag D/Fbg.
A previous study used SC from S. aureus strain 8325-4 (22), with a PR and five repeats, for investigation of C-terminal Fbg interactions (28). ELISA assays and microcalorimetry, respectively, demonstrated Fbg and frag D binding to recombinant fragments of the C-terminal SC repeats. Sequence similarities in the SC PR and R1-R2 junction with another S. aureus virulence factor, the extracellular Fbg-binding protein (Efb), suggested a common Fbg binding motif, absent in the virulence factor von Willebrand factor-binding protein that uses another mechanism to recruit Fbg (29). Efb uses Fbg binding to facilitate phagocytosis escape (13), and the pathogen co-opts host Fbg binding by both SC and Efb to evade the host immune system. In ELISA competition assays, short peptides inhibited binding of a C-terminal SC fragment to Fbg-coated microtiter wells. This fragment included the PR and repeats R1 to R5. The competitive peptides were Coa-R0, which has a 32-residue sequence bridging PR and R1; the 27-residue peptides Coa-RI, Coa-RI2, Coa-RI3, and Coa-RI4, all bridging R1 and R2; and Coa-RV1 with a partial R5 sequence. The sequence of repeat R5 in SC 8325-4 largely corresponds to that of repeat R7 in SC Tager 104. Binding of Coa-RV1 to immobilized Fbg and frag D by isothermal titration calorimetry cannot readily be explained by potentially different exposure of frag D epitopes in conformationally heterogeneous immobilized Fbg (52) and in solution. However, three other peptides, Coa-RV2, Coa-RV3, and Coa-RV4 largely corresponding to R5 were not competitive, consistent with our hypothesis that most single repeats may not have a high affinity for frag D. Binding of Coa-R1 and Coa-RI3 to frag D, measured by isothermal titration calorimetry, yielded affinities between 77 and 140 nM, in agreement with our fluorescence equilibrium binding data for the affinities of labeled PR-R constructs (Table 1). Ala scanning of ETNAYNVTTHANGQVSY-GARPTYKKPS bridging R1 and R2 in Coa-RI (28) indicated critical (bold, underlined) and important (bold) residues. Our NMR analysis and Ala scanning encompassed a different interdomain sequence, comprised of the C-terminal VKYR-DAGTGIREYNDGTFGYE residues of the PR, and the N-terminal part ARPTYKKPSETNAYNVTTHADGTATYG of R7. Figure 11 shows the alignment of critical conserved residues in the PR and the repeat junctions (critical Coa-RI and MP residues underlined in different colors). HADG is conserved in the last repeat of the serotypes (50), aligns with YNDG in the conserved PR and also with HANG in Coa-RI. Ala scanning of the PR sequence EYNDG 37 completely abolished fragD binding to PR-R7 (Fig. 9), suggesting that alignment of G 37 with the conserved G residues in the inter-repeat sequences plays a critical role. PR residues G/A 28 , I/V 31 , G 37 , and Y 41 are aligned with the C-terminal sequences in the individual repeats. Ko et al. (28) identified essential N, Y, T, H, and G residues in SC 8325-4 repeat R1, flanking the aligned PR residues, and these may represent a complementary and conserved binding motif in the repeat junctions. Unexpectedly, Ala scanning of HADG outside the MP region in PR-R7 abolished frag D binding, even though a complete MP sequence was present. In the absence of a crystal structure of the frag D-PR-R7 complex, PSIPRED analysis of the HADG-(Ala) 4 mutant does not indicate significant change in propensity for strand or coil structure in this region. However, conformational changes in this coil sequence and the substitution of D by a nonpolar residue may contribute to a change in overall intramolecular or intermolecular contacts of PR-R7, resulting in the loss of frag D binding (K D 112 nM). These potential structural and charge constraints might be absent in the weaker binding (K D 5 μM) of the isolated MP to frag D. Further structural analysis of the construct is required to identify the cause of the lack of binding.
Although competitive ELISA binding to immobilized fulllength Fbg may detect additional secondary binding sites for the SC C-terminal repeat region, it is clear that frag D binding represents a major site of interaction on Fbg. Multiple frag D binding to SC may indicate the potential of multiple full-length Fbg binding and a role in establishing conformational interactions that may favor Fbg binding in the substrate mode to the complex of SC and host ProT.
Semiquantitative native PAGE analysis and quantitative fluorescence equilibrium binding titrations suggested stoichiometries for frag D binding largely consistent with the number of inter-repeat sequences containing conserved motifs, that is, 3 for the PR-R1R2R3, PR-R1R6R7, PR-R3R4R7, and PR-R3R6R7 constructs, and 5 to 7 for PR-(R1 → R7). This small five to seven discrepancy for PR-(R1 → R7) may reflect different conformational binding site constraints under both experimental conditions. Titrations of all PR-R constructs and the MP titration were fit very well by the 1:1 binding model, whereas higher order interactions exhibited a typical stoichiometric titration shape and could only be fit satisfactorily by an unconstrained stoichiometry parameter. Equations for the direct and competitive binding analysis are given in Figure S11. The K D values for frag D binding to the repeat constructs represent affinities based on the assumption of independent and noncooperative binding. K D values ranged from 7 to 32 nM for constructs containing the naturally occurring PR-R1 sequence and were only marginally higher (42 nM) for those containing a PR-R3 sequence. Similarly, modest differences were observed for frag D binding to labeled PR-R1 and PR-R3 constructs containing only one repeat (50 and 100 nM, respectively, Table 1). Multiple PR-R1R2R3 and [5F]PR-(R1-R7) preparations exhibited acceptable variability in affinities, with good consistency in the stoichiometry and amplitude parameters. This variability is to be expected in nonlinear regression analysis with no constraint imposed on any of the parameters and using multiple independent protein preparations ([5F]PR-(R1 → R7), PR-R1R2R3, and frag D) in direct and competitive titrations. Minor differences in sample preparation and limitations of assay sensitivity and of fitting interactions with high stoichiometry factors may all contribute to this variability. Frag D affinities for constructs with multiple repeats were higher than those for PR-R constructs containing only one inter-repeat sequence, suggesting a cumulative effect. The naturally occurring PR-R1 sequence appears to correlate with modestly tighter binding, and the slight sequence variability of the individual repeats might affect the affinity somewhat. However, binding of frag D to our constructs irrespective of the order of repeats is consistent with the recurring pattern of conserved residues in the inter-repeat regions (Fig. S1).
The frag D subunit is considerably smaller than Fbg/Fbn, and in vivo maximizing Fbg/Fbn occupancy of the SC interrepeat sequences with minimal sterical hindrance may require a highly organized molecular arrangement such as found in Fbn networks. A hypothetical scheme is shown in Figure S10. These networks serve as anchoring points and concealing structures for S. aureus to evade the host immune system (28,53).
The coil or disordered structure of PR-(R1 → R7) may allow sufficient conformational freedom for multiple frag D binding. The much larger fibrin(ogen) molecules might associate with these junctions in staggered arrays and favor transition of the disordered SC repeat domain into more organized complex structures such as seen in Fbn protofibril association. This transition may facilitate formation of Fbn/bacteria vegetations in vivo. In solution, Fbg binding may be substoichiometric with regard to the number of inter-repeat sequences and allow for secondary interactions in addition to the frag D domains. Further structural studies are required to characterize these higher order complexes.
Binding multiple ligand molecules is a characteristic property of intrinsically disordered proteins (54) and intrinsically disordered protein regions (55). Our CD analysis showed an extended or random coil secondary structure confirmation (Fig. S9) of PR-R7 and PR-(R1 → R7), and secondary structure prediction of full-length SC identified major intrinsically disordered protein regions in the C-terminal repeat region (40)(41)(42). This facilitates binding of multiple ligands, consistent with our observations. Our binding data did not suggest positive or negative cooperativity and were fit very well by simple quadratic-or cubic-binding equations (Fig. S11).
Capturing circulating host Fbg by the C-terminal repeat domain may provide a mechanism of recruiting Fbg for presentation as a substrate of the SC(1-660)⋅ProT* complex in vivo; however, Fbg binding to the complex as a substrate for activated ProT* may occur independently of binding to the C-terminal domain of SC. In turbidity assays of Fbg cleavage (Fig. 12), Fbg cleavage by T (control) and the SC(1-325)⋅ProT* complex was rapid, compared with Fbn formation by the SC(1-660)⋅ProT* complex at a similar Fbg substrate concentration (Fig. 12B). Binding to the C-terminal SC domain may sequester the available Fbg for cleavage by the SC(1-660)⋅ ProT* complex, resulting in decreased rates compared with those observed for the SC(1-325)⋅ProT* complex, lacking the C-terminal domain. Rates and amplitudes of Fbn formation by the SC(1-660)⋅ProT* complex were dependent on the substrate Fbg concentration, consistent with this hypothesis.
In conclusion, we demonstrated a clear correlation between the specific and highly conserved inter-repeat sequences in the C-terminal domain of SC and the stoichiometry of frag D binding to these repeat sequences. We also demonstrated for the first time that the conserved PR sequence contains important fibrin(ogen)-binding motifs. The disordered nature of the C-terminal SC domains may favor identification of conserved linear epitopes for developing therapeutic antibodies to combat S. aureus-associated infections. These antibodies may be effective across the complete spectrum of serotypes.

Preparation of frag D from human Fbg
Frag D was purified from human Fbg (56), with some modifications. Fbg (400 mg) was dissolved in 150 mM NaCl, 50 mM imidazole buffer, pH 7.2, to a final concentration of 5 to 10 mg/ml and dialyzed against the same buffer. CaCl 2 was added to a final concentration of 5 mM. The Fbg solution was treated with 5 mM iodoacetamide for 15 min at room temperature to inhibit crosslinking by trace FXIIIa. Fbg was digested by trypsin (0.01 mg trypsin/mg Fbg) for 4 h at room temperature. The reaction was terminated by adding soybean trypsin inhibitor to a concentration of 0.03 mg/mg Fbg and incubating for 1 h on ice. The digested solution was precipitated by slowly adding ice-cold ammonium sulfate solution to a final concentration of 1.18 M and stirring for 10 min on ice. The solution was centrifuged at 10,600g for 1 h at 4 C, and the pellet was discarded. The supernatant was further precipitated by slowly adding ice-cold ammonium sulfate solution to a final concentration of 1.61 M and stirring for 10 min. The solution was centrifuged for 10,600g for 1 h. The pellet was resuspended in 20 mM Tris buffer, pH 8.0, containing 10 μM D-Phe-Phe-Arg-chloromethyl ketone (FFR-CK), 10 μM D-Phe-Pro-Arg-chloromethyl ketone (FPR-CK), and 100 μM 4-benzenesulfonyl fluoride hydrochloride, and dialyzed against the same buffer at 4 C. The dialyzed solution was loaded onto a pre-equilibrated 1 ml Resource Q column, washed with 10 column volumes of 20 mM Tris buffer, pH 8.0, and eluted with a 20 mM Tris, 0 to 250 mM NaCl gradient, pH 8.0. Fractions containing frag D were pooled and concentrated. The solution was dialyzed against 20 mM sodium phosphate, 150 mM NaCl, pH 7.0 (for NMR); or 50 mM Hepes and 125 mM NaCl buffer, pH 7.4 (all other experiments) and stored at −80 C until use. The frag D concentration was calculated using M r 88,000 and absorbance at 280 nm extinction coefficient of 2.08 ml mg −1 cm −1 (57).

Preparation of SC C-terminal repeat variants
Combination constructs of PR with single and multiple repeats were amplified by PCR from the SC(1-660) gene of S. aureus Newman D2 Tager 104 using degenerate and specific primers and cloned into a modified pET 30b(+) (Novagen) expression vector containing an N-terminal His 6 tag followed by a tobacco etch virus (TEV) cleavage site (15,58) and the SC residues SETTEASHYP preceding PR, to preserve proper folding and binding properties of PR. The constructs ending in R7 included the C-terminal SC sequence SRVTK of our reference construct AY225090.2. PCR was performed in 50 μl reaction mixtures containing 10 ng of SC(1-660) template DNA, 1 ng of primer, 0.8 mM dNTPs, and 1 unit of highfidelity polymerase (Stratagene). An additional C-terminal Cys residue for 5-IAF labeling was introduced by site-directed mutagenesis (QuikChange) in constructs for fluorescence binding, and all sequences were confirmed by DNA sequencing. The constructs were transformed into Escherichia coli Rosetta 2 (DE3) pLysS. Positive colonies were identified by DNA sequencing. LB (150 ml containing 0.1 mg/ml of kanamycin) was inoculated with single colonies and cultured overnight. LB media containing 0.1 mg/ml of kanamycin were inoculated (1:40) with the overnight cultures and grown for 2 to 3 h with shaking at 250 RPM and 37 C until absorbance of 0.6 at 600 nm was reached. After induction with D-lactose (10 mg/ml) and an additional 4 h growth, cells were centrifuged at 5000g for 30 min and lysed in 50 mM Hepes, 125 mM NaCl, 1 mg/ml PEG, 1 mM EDTA buffer, pH 7.4, containing 100 μM phenyl methyl sulfonyl fluoride (PMSF), and 10 μM FFR-CK and FPR-CK, by three freeze-thaw cycles in liquid nitrogen. The lysate was centrifuged at 39,200g for 45 min, and the constructs were released from the inclusion bodies in the pellet by suspension in the same buffer containing 3 M NaSCN. The solution was centrifuged and dialyzed against 50 mM Hepes, 325 mM NaCl, 50 mM imidazole buffer, pH 7.4 before chromatography on Ni 2+ -iminodiacetic acid-Sepharose (5 ml), and elution with a gradient of 0 to 500 mm imidazole (19). The N-terminal His 6 tag was removed by incubation with recombinant TEV protease (15,59,60). The C-terminal Cys thiol groups of unlabeled peptides for competitive binding were blocked with 5-iodoacetamide. Peptides (2-4 mg) were initially reduced by adding 2 mM DTT for 30 min on ice and further treated with fourfold molar excess iodoacetamide for 2 h in the dark at 20 C. Excess blocking agent was removed by dialysis against 50 mM Hepes, 125 mM NaCl, and pH 7.4. Because the experimentally determined molar extinction coefficients ε 280 of the constructs differed by 20% from the theoretical values, molar absorptivities ε 205 of the peptides at 205 nm were determined directly from their amino acid sequence (https://spin.niddk.nih.gov/ clore/) and by using measured absorbance at 205 and 280 m, good estimates for ε 280 = ε 205 (absorbance at 280 nm/absorbance at 205 nm) were obtained (61). The peptide concentrations were determined from the absorbance at 280 nm using molar extinction coefficients ε Purified peptides destined for labeling were reduced with DTT at 1 mM final concentration and dialyzed against 5 mM 2-(N-morpholino)ethanesulfonic acid, 150 mM NaCl buffer, pH 6.0, before labeling with 5-IAF. Cysthiol incorporation was measured as described (15). The pH of peptide solutions was raised to 7 by addition of 0.1 v 1 M Hepes, pH 7.0, in the presence of a twofold molar excess of 5-IAF. After addition of 0.9 volumes of 100 mM Hepes, 100 mM NaCl buffer, pH 7.0, and incubation for 1 to 2 h at 25 C, the reaction was quenched by adding DTT to a final concentration of 1 mM. The mixture was centrifuged for 15 min at 20,820g. Excess probe was removed by dialysis against 5 mM 2-(N-morpholino)ethanesulfonic acid, 150 mM NaCl buffer, and pH 6.0. The labeled peptides were further purified by C-18 (5 μm, 4.6 × 150 mm; Beckman Coulter, Inc) reverse-phase HPLC. Peptides were eluted using a linear gradient of 0.1% TFA and 100% acetonitrile with 0.1% TFA. The absorbance was measured at 205 nm (peptide bond) and 440 nm (5F probe absorbance). Peak fractions corresponding to 205 and 440 nm were pooled and lyophilized. The lyophilized powder was resuspended in water, and the labeling ratio was measured. Incorporation of fluorescein was determined by absorbance at 280 and 498 nm with an absorption coefficient of 84,000 m −1 cm −1 for fluorescein and an absorbance at 280 and 498 nm ratio of 0.19 in 6 m guanidine, 100 mm Tris-Cl, 1 mm EDTA buffer, pH 8.5 (62), and determined to be 0.75 to 1.1 mol fluorescein/mol of peptide.
The full-length SC gene SC(1-660) containing a His 6 tag and TEV cleavage site at the N terminus was amplified using specific primers from genomic DNA of S. aureus Newman D2 Tager 104 and cloned into the same pET30b(+) vector. His 6 -SC(1-660) was expressed in E. coli Rosetta 2 (DE3) pLysS and purified on Ni 2+ -iminodiacetic acid-Sepharose, followed by removal of the His 6 tag by TEV protease cleavage. The concentration of SC(1-660) was measured using an experimentally determined extinction coefficient of 0.989 M −1 cm −1 and a molecular mass of 74,390 Da. Our previously prepared SC(1-325) construct was also used in the current study (19,30).

Expression and purification of 15 N PR-R7 peptides
Single colonies of Rosetta 2 (DE3) pLysS containing the PR-R7 construct were inoculated into 100 ml sterile LB media containing 0.1 mg/ml kanamycin and grown overnight at 37 C with shaking at 250 RPM. Overnight cultures (60 ml) were centrifuged for 15 min, and the pellets were resuspended in 50 ml sterile milliQ water. Sterile flasks containing 900 ml of milliQ water (two per construct) were each inoculated with 25 ml of resuspended culture. BioExpress cell growth media (Cambridge Isotope Laboratories, Inc) was added (2 × 100 ml of 10× U-15N, 98% for the 15 N peptide), and cultures were grown at 37 C with shaking at 250 RPM for 2 to 3 h in the presence of 0.1 mg/ml of kanamycin until an absorbance reading of 0.6 at 600 nm was reached. Peptide expression was induced by adding 10 mg/ml D-lactose, and the cultures were grown for four more hours at 37 C with shaking at 250 RPM. Cultures were centrifuged, and the peptides were purified by HPLC as described previously.

Expression of PR-R7 Ala constructs
Wildtype PR-R7 with an N-terminal His 6 tag followed by a TEV cleavage site, and an engineered C-terminal Cys residue, cloned into the pET30b(+) vector was used as a starting construct. Residues 22 to 49, constituting the MP region, and the subsequent R7 residues 50 to 69 were converted groupwise (four to five residues, consecutively) to Ala residues through Quik-Change site-directed mutagenesis and confirmed by DNA sequencing. The constructs were transformed into E. coli Rosetta 2 (DE3), expressed, purified by His 6 tag and TEV cleavage, labeled with 5F, and purified by HPLC as described previously. The peptide concentration was determined from the absorbance at 280 nm using molar extinction coefficients of 10,430 M −1 cm −1 for PR-R7 constructs with sequences VKYRDA, EYNDG, TFGYE, YKKP, YNVTT, and TATYG changed to Ala residues; and 11,920 M −1 cm −1 for PR-R7 constructs with sequences GTGIR, ARPT, SETNA, and HADG changed to Alas.

Size-exclusion chromatography and light scattering
Size-exclusion chromatography demonstrated that fulllength SC(1-660), but not SC(1-325), interacts with frag D. The size exclusion profile was recorded using a Superdex-200 column (GE, 10 × 300 mm) pre-equilibrated with 50 mM Hepes, 125 mM NaCl, 5 mM CaCl 2 , and pH 7.4 buffer. In a first set of experiments, 100 μl each of the samples of SC(1-325) (52.67 μM or 2.0 mg/ml); frag D (24.7 μM or 2.1 mg/ml); and 100 μl of a 1:1 mixture of SC(1-325) and frag D in 50 mM Hepes and 125 mM NaCl buffer, pH 7.4, were passed on the column separately. Subsequently, 100 μl each of the samples of SC(1-660) (24.2 μM or 1.8 mg/ml); frag D (24.7 μM/2.1 mg/ml); and 100 μl of a 1:1 mixture of SC(1-660) and frag D in 50 mM Hepes and 125 mM NaCl, pH 7.4, were passed separately on the pre-equilibrated column and eluted using the same buffer as described previously. Elution profiles (absorbance at 280 nm in milliabsorbance versus elution volume) were recorded continuously for both sets. Light scattering data were acquired with a PD2010 multidetection light scattering system to obtain the apparent molecular mass (M r ) of the eluted complexes and proteins.
[5F]PR-(R1 → R7) NMR spectroscopy, data collection, and chemical shift assignments Titration NMR experiments were performed at 25 C on a Bruker AV-III 600.13 MHz spectrometer equipped with quadruple resonance cryogenically cooled CPQCI probe. Standard 1 H-15 N HSQC experiments were used containing a flip-back pulse and Watergate sequence to suppress the water signal (63)(64)(65). Titration experiments were done using 200 μl of 65 μM protein solution plus 5 μl D 2 O in a 3 mm NMR tube. Various amounts of a 240 μM frag D stock in the same buffer solution were added. All experiments were referenced against the 4,4-dimethyl-4-silapentane-1-sulfonic acid standard, processed with Topspin 3.5 (Bruker BioSpin), and analyzed with NMRView (One Moon Scientific, Inc). Backbone resonance assignments were utilized as described (27). and stoichiometric factor (n) for the peptide constructs. The error estimates (2SD) represent the 95% confidence interval. Fitting of the Ala constructs was done with the stoichiometric factor (n) fixed to 1, appropriate for titrations with probe concentration below the K D of the interaction.
To determine whether the repeats bind to the frag D β-holes and γ-holes, [5F]PR-R1 was titrated with frag D in the presence of saturating 4.9 mM GPRP under experimental conditions described previously. The K D for binding of GPRP to frag D is 20 μM (31). The K D and maximum fluorescence intensity for frag D binding to [5F]PR-R1 were obtained by fitting the data with n fixed to 1 (Fig. 5).
To determine the effect of pH and NMR buffer conditions, equilibrium binding at pH 7.0 and 7.4 of [5F]PR-R7 (17 and 848 nM, respectively) to frag D was measured in 20 mM sodium phosphate, 150 mM NaCl, and 1 mg/ml PEG-8000 buffers containing 10 μM FPR-CK inhibitor, using λ excitation of 490 nm (4 nm band pass at pH 7.0; 2 nm at pH 7.4) and λ emission of 514 nm (8 nm band pass at pH 7.0; 4 nm at pH 7.4). The K D and maximum fluorescence intensity values were obtained as described previously, with n fixed to 1.
Fluorescence anisotropy titrations were performed in the same buffer. Anisotropy measurements were corrected for total intensity changes using a modified Lakowicz equation (67).
[5F]PR-(R1 → R7) was titrated with frag D at excitation wavelength λ excitation of 495 nm and emission wavelength λ emission of 514 nm with band passes ranging from 2 to 6 nm depending on the [5F]PR-(R1 → R7) concentration. Three different [5F]PR-(R1 → R7) preparations were titrated with three different frag D preparations to show reproducibility (Figs. 13 and S8; Table 2). Measurements of r obs as a function of frag D concentration were analyzed by the quadratic binding equation (66) to obtain the dissociation constant K D and stoichiometric factor n of ligand binding, initial (r o ) and maximal (r max ) anisotropies, and the change in anisotropy (Δr = r max − r o ).
In competitive titrations, mixtures of a fixed concentration of [5F]PR-(R1-R7) with two different concentrations of frag D were each titrated with unlabeled PR-R1R2R3 (three independent batches), PR-R1R6R7, PR-R3R4R7, and PR-R3R6R7 to obtain two data curves for each competitor (Figs. 13 and S8; Table 2). Concentrations of [5F]PR-(R1-R7) and frag D in the competitive titrations are given in the figure legends. Respective sets of direct and competitive titrations were analyzed simultaneously by nonlinear least-squares fitting of the cubic equation in SCIENTIST (MicroMath) (66,67) (Fig. S11) to obtain the K D and stoichiometric factor n, initial (r o ) and maximal (r max ) anisotropies, and the maximal change in anisotropy (Δr = r max − r o ) for competitor binding to frag D. Error estimates (±2SD) represent the 95% confidence interval.
Clotting of Fbg by the SC(1-325)⋅ProT QQQ * and SC(1-660)⋅ ProT QQQ * complexes Fbg cleavage by T and the SC(1-325)⋅ProT QQQ * and SC(1-660)⋅ProT QQQ * complexes was measured as described previously (60). Assays were performed in 50 mM Hepes, 110 mM NaCl, 5 mM CaCl 2 , 1 mg/ml PEG-8000, and pH 7.4 reaction buffer. Increases in turbidity were measured at 450 nm, 25 C in a microtiter plate reader, for mixtures of Fbg  37 μM), and frag D (3.93 μM) were measured using a Jasco J-810 CD spectrometer from Beckman Coulter, Inc. The spectra were measured in 20 mM sodium phosphate and 150 mM NaCl, pH 7.0 buffer, at peptide/protein concentrations of 0.1 mg/ml. Solutions were inspected for the absence of turbidity. Spectra were measured from 260 to 200 nm with high sensitivity, bandwidth set to 1 nm, response time of 2 s, and scanning speed of 50 nm/min at 25 C. Each scan represents the average of three individual scans. Data between 190 and 200 nm were noisy, prohibiting accurate analysis in this region. Analysis was performed as described (68) using the formula, [θ] = (100(signal))/cnl, where [θ] is the mean residue ellipticity in deg cm 2 decimol −1 , signal is raw output in millidegrees, c is the peptide or protein concentration in millimolar, n is the number of amino acids, and l is the cell path length in centimeter. Ellipticity data were plotted without using a smoothing algorithm (Fig. S9). GOR IV, PSIPRED, and PONDR-FIT secondary structure predictions were performed online as described (40)(41)(42).
All other data are contained within the article and Supporting Information.
Supporting information-This article contains supporting information.