Calcium Binding to Leptospira Outer Membrane Antigen LipL32 Is Not Necessary for Its Interaction with Plasma Fibronectin, Collagen Type IV, and Plasminogen*

LipL32 is the most abundant outer membrane protein from pathogenic Leptospira and has been shown to bind extracellular matrix (ECM) proteins as well as Ca2+. Recent crystal structures have been obtained for the protein in the apo- and Ca2+-bound forms. In this work, we produced three LipL32 mutants (D163–168A, Q67A, and S247A) and evaluated their ability to interact with Ca2+ and with ECM glycoproteins and human plasminogen. The D163–168A mutant modifies aspartate residues involved in Ca2+ binding, whereas the other two modify residues in a cavity on the other side of the protein structure. Loss of calcium binding in the D163-D168A mutant was confirmed using intrinsic tryptophan fluorescence, circular dichroism, and thermal denaturation whereas the Q67A and S247A mutants presented the same Ca2+ affinity as the wild-type protein. We then evaluated if Ca2+ binding to LipL32 would be crucial for its interaction with collagen type IV and plasma proteins fibronectin and plasminogen. Surprisingly, the wild-type protein and all three mutants, including the D163–168A variant, bound to these ECM proteins with very similar affinities, both in the presence and absence of Ca2+ ions. In conclusion, calcium binding to LipL32 may be important to stabilize the protein, but is not necessary to mediate interaction with host extracellular matrix proteins.

Leptospirosis is a zoonosis caused by spirochetes belonging to the genus Leptospira. The disease is more prevalent in tropical countries, due to climatic and environmental conditions (1). The main symptoms of the disease include fever, jaundice, and tubule-interstitial nephritis (2). LipL32 is a lipoprotein highly expressed on the surface of pathogenic Leptospira (3). Genome sequencing analyses have shown that saprophytic strains of Leptospira do not harbor genes orthologous to LipL32 (4), suggesting an important role of this protein in leptospiral pathogenesis. Although a knock-out of the lipL32 gene in pathogenic Leptospira did not prevent colonization of the renal tubules in a rat model of chronic infection (5), it is important to keep in mind that animal models may not reproduce all relevant conditions of infection and colonization in the host. Therefore, the precise role of LipL32 in Leptospira biology remains unclear. The use of LipL32 as a vaccine antigen in animal models has been shown to confer partial protection (6 -8). Studies performed with recombinant LipL32 revealed that this highly immunogenic protein can induce a robust inflammatory response in renal proximal tubule cells (9,10). Furthermore, LipL32 may contribute to host colonization, because it has been demonstrated that it interacts with extracellular matrix (ECM) 3 components, such as collagen type IV, plasma fibronectin, and laminin (11,12) and binds human plasminogen (13).
Our group has recently demonstrated that LipL32 specifically binds Ca 2ϩ ions and that this binding increases LipL32 conformational stability (14). The crystal structure of apo-LipL32 (14,15) and the calcium-bound structure (16) led to the characterization of the significant structural changes associated with metal binding that includes strand-coil transitions and a 10-Å translocation of an aspartate-rich loop to form part of the Ca 2ϩ -binding site. These observations raised the question regarding the role of Ca 2ϩ in the binding of LipL32 to its target proteins. Binding assays involving different portions of LipL32 indicated that its C-terminal region (residues 185-272) is responsible for mediating interaction with ECM proteins. Although Hoke et al. (11) and Hauk et al. (12) showed that C-terminal fragments from LipL32, which lack the Ca 2ϩ -binding site, mediate its interactions with ECM proteins, a recent study by Tung et al. (16) reported that LipL32 affinity for fibronectin is enhanced 4-fold by Ca 2ϩ . Therefore, the role of Ca 2ϩ in the interactions of LipL32 with specific host proteins remains unclear. To directly address this question, we produced a set of recombinant LipL32 mutants including one variant (D163-168A) in which five aspartate residues within the acidic loop directly involved in Ca 2ϩ binding were changed to alanine. Although the Q67A and S247A mutants presented the same Ca 2ϩ affinity as the wild-type protein, the D163-168A protein failed to bind Ca 2ϩ in a variety of independent assays. ECM binding assays were performed with the three LipL32 mutants. Wild-type LipL32 and all three mutants tested all bound to ECM macromolecules with affinities that did not change significantly in the presence or absence of Ca 2ϩ . These results indicate that LipL32 binds to target proteins in a manner that does not depend on metal binding to the Ca 2ϩ -binding site.

Production of Recombinant LipL32 and Mutants-Plasmids and
Escherichia coli strains used in this work are described in Table 1. Wild-type LipL32 with or without an N-terminal His tag fusion and a recombinant C-terminal fragment corresponding to residues 185-272 plus an N-terminal His tag (LipL32 185-272_His tag ) were produced as described (12,14). Mutations were introduced into the lipL32 gene by a PCR-driven overlap extension protocol (17) using as template the previously described plasmid lipL32ϩpAE (14). Mutagenic primers and flanking primers ( Table 2) were used to generate intermediate overlapping PCR products that were combined to produce a full-length product using flanking primers. Three full-length fragments (corresponding to amino acids 21-273 of LipL32) were produced that code for the mutant proteins LipL32 Q67A , LipL32 S247A , and LipL32 D163-168A (in this last mutant residues 163-168 were changed from DDDGDD to AAAGAA). These fragments were cloned into pGEM-T Easy vector (Promega) and subcloned into the pAE expression vector (18) between XhoI and HindIII sites to produce lipL32 Q67A ϩpAE His tag , lipL32 S247A ϩpAE His tag , and lipL32 D163-168A ϩpAE His tag . These three constructs allow expression of LipL32 mutants with a N-terminal polyhistidine tag (LipL32 Q67A_His tag , LipL32 S247A_His tag , and LipL32 D163-168A_His tag ) ( Table 1). After digestion of these three constructs with NdeI and re-ligation, three new constructs were produced (lipL32 Q67A ϩpAE, lipL32 S247A ϩpAE, and lipL32 D163-168A ϩpAE), which allowed the expression of LipL32 mutants without the polyhistidine tag (LipL32 Q67A , LipL32 S247A , and LipL32 D163-168A ) ( Table 1). In these three constructs the initiation codon methionine is followed immediately by the codon for Gly-21 and therefore corresponds to the mature LipL32 protein minus its N-terminal lipid-anchored cysteine residue. The recombinant proteins lacking a His tag were used in the spectroscopy-based calcium-binding and denaturation assays, whereas the His-tagged LipL32 mutants were used in ECM binding assays. The LipL32 mutant plasmid constructs coding for LipL32 Q67A , LipL32 S247A , LipL32 D163-168A , LipL32 Q67A_His tag , LipL32 S247A_His tag , and LipL32 D163-168A_His tag were transformed into E. coli BL21 (SI) (19). The transformed E. coli strains were grown in 1 liter of 2YTON ampicillin until the culture optical density at 600 nm reached 0.6 at which time protein expression was induced by incubation with 300 mM NaCl for 3 h at 30°C. Cells were collected by centrifugation at 8400 ϫ g for 10 min and resuspended in 100 ml of 20 mM triethanolamine (pH 7.8) for proteins without His tags or 150 mM Tris-HCl (pH 8.0) for proteins with His tags and lysed in a French pressure cell (Thermo Spectronic). The soluble and insoluble fractions were isolated by centrifugation at 8400 ϫ g for 10 min. Purification of LipL32 mutants without a His tag fusion proceeded as previously described for wild-type LipL32 (20) followed by dialysis against 10 mM Tris-HCl (pH 8.0), 50 mM KCl. For the purification of LipL32 mutants with N-terminal His tag fusions, the soluble fraction of the cell lysate was applied onto a 5-ml Ni 2ϩ -charged chelating Sepharose bed (1 cm diameter column, GE Healthcare) equilibrated with 150 mM Tris-HCl (pH 8.0). Bound proteins were eluted using a 0 -500 mM imidazole gradient over 12 column volumes. Fractions were analyzed by 15% SDS-PAGE and those lacking contaminating E. coli proteins were pooled and dialyzed against 150 mM Tris-HCl (pH 8.0).
Intrinsic Tryptophan Fluorescence and Ca 2ϩ Binding Studies-For calcium-binding studies, proteins were decalcified as previously described (14). Briefly, proteins were incubated for 60 min at 25°C in 3 mM EDTA followed by buffer exchange with chelating Sepharose-treated Tris-HCl buffer. All buffer solutions used for calcium-binding studies were passed through chelating Sepharose resin and stored in plastic containers. Fluorescence spectroscopy studies were carried out using a 1 cm path-length quartz cell and an Aviv ATF105 Fluorimeter or a Hitachi F4500 Fluorimeter with 1-nm excitation bandwidth and 10-nm emission bandwidth. Intrinsic tryptophan fluorescence was measured with an excitation wavelength set at 285 nm and emission spectra were recorded between 310 and 400 nm using 2 M LipL32 (or mutants) in 10 mM Tris-HCl (pH 8.0), 50 mM KCl in the presence or absence of 1 mM CaCl 2 (14). Calcium titration experiments were performed using 2 M LipL32 wild-type or mutant LipL32 in 10 mM Tris-HCl (pH 7.0), 50 mM KCl, 0.5 mM EGTA. CaCl 2 was added to achieve the desired free Ca 2ϩ concentration, expressed as pCa ϭ Ϫlog[Ca 2ϩ free ], based on the Ca 2ϩ -EGTA buffer system (21). The calcium-induced reduction in intrinsic tryptophan fluorescence between 338 and 342 nm (285 nm excitation) was used to accompany calcium binding.
Ca 2ϩ -induced Changes in Protein Stability-Temperaturedependent unfolding of LipL32 and its mutants was monitored using circular dichroism (CD) and extrinsic ANS (1amino-2-naphtol-4-sulfonic acid) fluorescence. Far-UV CD measurements were carried out on a J-810 circular dichro- ism spectropolarimeter (Jasco) coupled to a Peltier Jasco PFD-425S system for temperature control. Spectra were recorded by measuring [] at 216 nm at 2°C temperature steps between 40 and 70°C. LipL32 proteins were decalcified as described previously (14). Assay conditions were: 2 M protein, 10 mM Tris-HCl (pH 8.0), 50 mM KCl, in the absence or presence of 1 mM CaCl 2 . Extrinsic ANS fluorescence measurements were carried out in a 1-cm path-length quartz cell using an Aviv ATF105 Fluorimeter with 0.25-nm excitation bandwidth, 10-nm emission bandwidth, and excitation and emission wavelengths of 380 and 470 nm, respectively. Temperature was varied from 40 and 70°C at a rate of 0.5°C/ min. Assay conditions were: 2 M protein, 10 mM Tris-HCl (pH 8.0), 50 mM KCl, 8 M ANS, and 3 mM EDTA in the absence or presence of 5 mM CaCl 2 . The pre-transition and post-transition baselines were used to calculate the fractional change in signal at each temperature, assuming that the CD and fluorescence signals of the folded and unfolded states are linear functions of temperature in the transition region (22).

Binding of Wild-type LipL32, LipL32 C-terminal Fragment, and LipL32 Mutants to ECM Components and Human
Plasminogen-Human plasma fibronectin and its 30-kDa proteolytic fragment (F30, heparin-binding domain), human plasminogen, and collagen type IV from the basement membrane of Engelbreth-Holm-Swarm mouse sarcoma were purchased from Sigma. LipL32 binding to these macromolecules was performed according to a previously published protocol with some modifications (12). Briefly, enzyme-linked immunosorbent assay (ELISA) plate wells (Nunc-Immuno Plate MaxiSorp Surface) were coated with 1 g of plasma fibronectin, F30, collagen type IV, or human plasminogen in 100 l of PBS and incubated overnight at 4°C. The wells were washed twice in 150 mM Tris-HCl (pH 8.0), 1 mM Mg 2ϩ , 0.05% Tween 20, and then blocked with 200 l of 1% bovine serum albumin, 150 mM Tris (pH 8.0), 1 mM Mg 2ϩ for 2 h at 37°C. Dose-dependent attachment of LipL32 and its mutants to ECM molecules and human plasminogen in the absence or presence of calcium ions was then assessed by adding to the wells 0 to 4 M of recombinant LipL32 proteins in 150 mM Tris-HCl (pH 8.0), 0.5 mM EGTA, 1 mM Mg 2ϩ (buffer without Ca 2ϩ ) or in 150 mM Tris-HCl (pH 8.0), 0.5 mM EGTA, 1 mM Mg 2ϩ , 1 mM Ca 2ϩ (incubation buffer with Ca 2ϩ ). Proteins were allowed to attach to the different substrates for 90 min at 37°C. Reactions incubated in the presence of Ca 2ϩ were washed six times in the incubation buffer with Ca 2ϩ containing 0.05% Tween 20, and those in which Ca 2ϩ was omitted were washed five times with buffer without Ca 2ϩ containing 0.05% Tween 20 and one more time in buffer with Ca 2ϩ and 0.05% Tween 20. From this step on, all washes and incubations were performed in incubation buffer containing Ca 2ϩ (this was found to be important to avoid Ca 2ϩ -dependent variations in subsequent antibody binding steps as explained in supplemental Figs. S1 and S2 and under "Results" and "Discussion"). Bound proteins were detected by adding 100 l of a 1:10,000 dilution of mouse anti-LipL32 serum. Incubation proceeded for 1 h and after three washes 100 l of a 1:5,000 dilution of horseradish peroxidase-conjugated goat anti-mouse immunoglobulin G (IgG) were added per well and the plate was incubated for 1 h. All incubations took place at 37°C. The wells were washed three times and o-phenylenediamine (0.04%) in citrate phosphate buffer (pH 5.0) plus 0.01% H 2 O 2 (w/v) was added. The reaction was allowed to proceed for 15 min and then interrupted by the addition of 50 l of 8 M H 2 SO 4 . The absorbance at 492 nm was determined in a microplate reader (Labsystems Uniscience Multiskan EX). Three independent experiments were performed, each one in duplicate. Student's two-tailed t test was employed for statistical analyses.

RESULTS
Production of LipL32 Mutants-The resolution of the apo-LipL32 structure (14,15) led to the prediction and demonstration of its ability to bind Ca 2ϩ (14). Because the apo-structure did not unambiguously reveal the Ca 2ϩ -binding site, we produced three LipL32 mutants designed to disrupt two originally proposed candidate binding sites on opposite sides of the protein: (i) LipL32 Q67A and LipL32 S247A have mutations in a cavity that coincides with the Ca 2ϩ -binding site from the ColG collagenase from Clostridium histolyticum and (ii) LipL32 D163-168A has five aspartates in a conserved acidic loop all changed to alanines. The residues mutated in these three mutants are shown in Fig. 1. The resolution of the Ca 2ϩ -LipL32 structure by Tung et al. (16) unambiguously showed that the Ca 2ϩ site is made up of the acidic loop residues Asp-164 and Asp-165 as well as residues Asp-132, Thr-133, and Tyr-178 (Fig. 1B).
LipL32 D163-168A Does Not Bind Ca 2ϩ -The intrinsic tryptophan fluorescence emission spectrum of wild-type LipL32 undergoes a significant blue shift and a reduction in fluorescence intensity upon addition of Ca 2ϩ ( Fig. 2A) (14). The same Ca 2ϩ -induced shifts were observed in the emission spectra of LipL32 Q67A and LipL32 S247A (Fig. 2, B and D, respectively). However, the fluorescence of LipL32 D163-168A was unaffected by the addition of 1 mM CaCl 2 (Fig. 2C). The fluorescence maximum ( max ) for LipL32 D163-168A was observed to be 330 nm. This value is the same as that observed for the apo form of wild-type LipL32 (Fig. 2A). We then used a Ca 2ϩ -EGTA buffer system (21) to monitor the Ca 2ϩ -induced fluorescence change

LipL32 Binds to ECM in the Absence of Calcium
We have previously shown that the binding of Ca 2ϩ to LipL32 results in a significant increase in the conformational stability of the protein as determined by thermal denaturation experiments monitored by changes in extrinsic ANS fluorescence and CD (14). We therefore tested whether the addition of Ca 2ϩ could induce similar increments in the stability of the three LipL32 mutants. In the presence of Ca 2ϩ , the T m of the wild-type protein increases significantly, by ϳ7 degrees, in thermal denaturations monitored by both extrinsic ANS fluorescence (Fig. 4A) or CD (Fig. 5A). The denaturation profiles were relatively insensitive to pH or ionic strength: no significant difference was observed in experiments performed at pH 8.0 or 7.4 and exchanging 50 mM KCl to 140 mM NaCl resulted in only slight reductions (0 -1°C) in the T m (data not shown). This Ca 2ϩ -induced stabilization was also observed for LipL32 Q67A (Figs. 4B and 5B) and LipL32 S247A (Figs. 4D and 5D). However, the LipL32 D163-168A mutant denaturation profiles were unaltered by the addition of Ca 2ϩ (Figs. 4C and 5C) and its T m was very close to that observed for the wild-type protein in the absence of Ca 2ϩ (Figs. 4A and  5A). The results presented in Figs. 2-5 all indicate that LipL32 D163-168A has lost the ability to bind calcium. Furthermore, the observations that the intrinsic fluorescence spectrum and thermal denaturation profile is similar to that observed for the apo form of the wild-type protein indicates that the mutant can still adopt a native-like fold.

Ca 2ϩ Binding to LipL32 Is Not Required for Its Interaction with Plasma Fibronectin, Type IV Collagen, and Human
Plasminogen-It has been suggested that Ca 2ϩ binding to LipL32 is important to mediate interaction with the plasma fibronectin F30 fragment (16). We therefore tested the ability of the LipL32 Histag mutants to bind F30, full-length plasma fibronectin, collagen type IV, and human plasminogen using ELISA-based assay. Before discussing these results, it is important to note that in all the ELISA-based assays described below, Ca 2ϩ was always present in the primary and secondary antibody binding steps because the overall signal is significantly reduced if Ca 2ϩ is removed at this step (see supplemental Figs. S1 and S2 for details). Furthermore, MgCl 2 was present in all steps of the ELISA-based assays at a concentration of 1 mM, within the normal physiological range for Mg 2ϩ in the extracellular space (23).
The LipL32-ECM binding steps in the ELISA-based assays were performed in the presence or absence of 1 mM CaCl 2 . All proteins were observed to bind to ECM components as well as to human plasminogen in a dose-dependent manner ( Fig. 6 and supplemental Figs. S3-S5). Interestingly, the curves for wildtype LipL32 His tag binding to F30 in the presence and absence of Ca 2ϩ were essentially the same (Fig. 6A), thus indicating that this divalent cation is not required for LipL32 interaction with F30. Calculation of the apparent dissociation constants from these binding curves gave values of 0.15 and 0.17 M for Ca 2ϩ -and apo-LipL32, respectively (Table 3). We also observed little or no Ca 2ϩ -dependent differences for the binding of wild-type LipL32 His tag with full-length fibronectin, collagen type IV, and plasminogen (Table 3 and supplemental Figs. S3A, S4A, and S5A). Consistent with these findings, the LipL32 D163-168A_His tag mutant, which does not bind calcium, bound to all ECM components and to plasminogen with affinities similar to those observed for wild-type LipL32 His tag both in the presence and absence of calcium (Table 3, Fig. 6C, and supplemental Figs. S3C, S4C, and S5C). Similar results were observed for the LipL32 Q67A_His tag and LipL32 S247A_His tag proteins (Table 3, Fig. 6, B and D, and supplemental Figs. S3, B and  D, S4, B and D, and S5, B and D). Finally, binding measurements were carried out using the recombinant LipL32 185-272_His tag FIGURE 3. Calcium binding by LipL32. Calcium-induced reduction in intrinsic tryptophan fluorescence was used to accompany calcium binding using a Ca 2ϩ -EGTA buffer system in which the free Ca 2ϩ concentration was varied between the nanomolar and millimolar ranges (pCa ϭ 9 to 3). Conditions were 2 M protein in 10 mM Tris-HCl (pH 7.0), 50 mM KCl, 0.5 mM EGTA. Excitation wavelength was 285 nm and emission wavelength was 338 -342 nm. We observed clear transitions for wild-type LipL32 (A) and for mutants LipL32 Q67A (B) and LipL32 S247A (C) between pCa ϭ 5 and 6, indicating that these mutants bind calcium ions with affinities between 1 and 10 M. For the mutant LipL32 D163-168A , no clear Ca 2ϩ -dependent transition was observed (data not shown). Error bars present S.D. of at least 3 independent experiments. fragment (12). This C terminus fragment of LipL32 consistently presented significantly greater affinity than the full-length protein (lower K D app ) for plasminogen and all three ECM compo-nents tested (Table 3, Fig. 6E, and supplemental Figs. S3E, S4E, and S5E). The unrelated Leptospira rLIC11030 protein did not present significant binding to ECM components or plasmino-

DISCUSSION
LipL32 is the most abundant outer membrane surface-exposed antigen from Leptospira (3). This lipoprotein is highly conserved among pathogenic Leptospira species, but is not present in the nonpathogenic saprophytic Leptospira biflexa (24). LipL32 is expressed at high levels on the bacterial surface during both cultivation and natural infection (24). The protein is immunogenic and some studies have implicated LipL32 as an extracellular matrix-binding protein, interacting with collagen type IV, plasma fibronectin, and laminin (11,12). Recently, Vieira et al. (13) showed that LipL32 is able to bind to human plasminogen, the inactive precursor of the extracellular proteinase plasmin. To better understand the biological role and structural requirements for the function of this important lipoprotein, the crystal structure of recombinant LipL32 was initially determined in the apo state (14,15). Our previous observation that many of the proteins with the closest structural homology to LipL32 bind both calcium ions and extracellular matrix proteins led us to demonstrate for the first time that LipL32 binds Ca 2ϩ , but not other divalent metals such as Mg 2ϩ , Zn 2ϩ , and Cu 2ϩ (14). Attempts by our group to obtain a Ca 2ϩ -LipL32 structure by co-crystallizing LipL32 and Ca 2ϩ or by soaking apo-LipL32 crystals in Ca 2ϩ solutions were unsuccessful. We therefore produced three LipL32 mutants in which residues in two candidate calcium-binding sites (14) were modified ( Fig. 1): LipL32 Q67A and LipL32 S247A from one of the candidate Ca 2ϩ -binding sites and LipL32 D163-168A from the second candidate site (in this last mutant, residues 163-168 were changed from DDDGDD to AAAGAA). During the preparation of this work, Tung et al. (16) published the crystal structure of calcium-bound LipL32 in which they showed that the calcium ion is coordinated with an octahedral geometry by Asp-132, Thr-133, Asp-164, Asp-165, and Tyr-178 ( Fig. 1A; in their numbering scheme, these residues are Asp-113, Thr-114, Asp-145, Asp-146, and Tyr-159). Therefore, the calcium-binding site does in fact involve two of the aspartates mutated in the LipL32 D163-168A mutant: Asp-164 and Asp-165. The fluorescence and CD studies performed with the LipL32 mutants show that calcium binds to and stabilizes LipL32 Q67A and LipL32 S247A to the same extent as that observed for the wildtype LipL32 protein. However, in the case of LipL32 D163-168A no differences were observed in intrinsic fluorescence spectra and thermal denaturation profiles obtained in the absence or presence of calcium.
Several pathogenic bacteria interact with ECM components (11,12,(25)(26)(27)(28)(29)(30)(31)(32) or other host molecules such as plasminogen (13,(33)(34)(35) to facilitate host colonization. Accordingly, leptospires employ surface-exposed proteins to bind to a wide range of host ligands, including laminin, collagen, cellular and plasma fibronectin, elastin, proteoglycans, fibrinogen, and plasminogen (13,30,31,36,37). LipL32 binds to plasma fibronectin, collagen types I, IV, and V, laminin, and plasminogen (11)(12)(13). Mapping assays using different portions of LipL32 protein have shown that its C terminus is responsible for mediating interaction with ECM proteins (11,12) and is recognized early in the course of infection (12). The observation that LipL32 binds Ca 2ϩ (14,16) raised the question of the role of Ca 2ϩ in LipL32 interactions with its target proteins. Extracellular calcium is required for the binding of other human pathogens such as Acanthamoeba to collagen IV, laminin, and fibronectin (26), and the yeast Candida albicans to collagen I and fibronectin (25). The adherence of C. albicans to ECM proteins seems to be mediated by calcium-dependent glycoproteins present on the surface of the yeast (25). Studies conducted by Lin et al. (36) with Leptospiral immunoglobulin-like protein B (LigB) showed that it binds calcium and that metal-binding modulates fibronectin binding.
In the first study to address the role of Ca 2ϩ in LipL32-ECM interactions, Tung et al. (16) concluded that Ca 2ϩ modulates FIGURE 6. Calcium is not required for LipL32 binding to F30 fragment of fibronectin. F30 binding to LipL32 proteins and to the unrelated rLIC11030 protein (negative control) were assayed by ELISA. LipL32 proteins or rLIC11030 (0 -4 M) were added to immobilized F30 and allowed to attach in the absence of Ca 2ϩ or presence of Ca 2ϩ . After washing away unbound LipL32 proteins or rLIC11030 using the corresponding incubation buffers, all subsequent steps were carried out in the presence of CaCl 2 (see "Experimental Procedures" for details). Each point represents the mean absorbance value at 492 nm Ϯ S.D. of three independent experiments, each performed in duplicate. Binding to F30 by wild-type LipL32 His tag (A), LipL32 Q67_His tag (B), LipL32 D163-168A_His tag (C), LipL32 S247A_His tag (D), LipL32 185-272_His tag (E), and rLIC11030 His tag (F). the binding of LipL32 to the fibronectin F30 fragment. Using an ELISA-based assay, they reported that the dissociation constant for the LipL32 interaction with F30 decreases 4-fold from 1.15 to 0.29 M (16). They also used a CD-based assay to monitor the binding of the cationic carbocyanine dye Stains-all to LipL32 in the presence of F30 Ϯ Ca 2ϩ and observed that F30 could displace the dye in the presence of Ca 2ϩ but not in its absence. This was interpreted as evidence for stronger F30 binding in the presence of Ca 2ϩ , even though another interpretation is possible, namely that Ca 2ϩ reduces the affinity of LipL32 for the dye. In that work, the Ca 2ϩ dependence of Stains-all binding to LipL32 was not determined. However, previous studies have shown that Ca 2ϩ binding eliminates the interaction of Stainsall with some proteins (parvalbumin and ␤and ␦-crystallin) and modifies its interactions with others (calmodulin and troponin C) (38 -40). Therefore, the evidence that Ca 2ϩ binding to LipL32 modulates its interaction with F30 can be considered tenuous at best. The case is weakened even further by observations by two independent research groups that fragments derived from the C terminus of LipL32, which do not contain the Ca 2ϩ binding site, interact with several ECM proteins (11,12).
We therefore decided to characterize the LipL32 Q67A , LipL32 S247A , and LipL32 D163-168A mutants in terms of their abilities to interact with host proteins previously characterized as LipL32 targets: full-length fibronectin and its 30-kDa fragment (F30), type IV collagen, and plasminogen (11)(12)(13). Binding studies based on an ELISA were performed in the presence or absence of Ca 2ϩ using wild-type and LipL32 mutants as well as a fragment derived from the LipL32 C terminus (LipL32 185-272_His tag ) as a positive control and the unrelated leptospiral protein rLIC11030 as a negative control (12). The K D app values obtained for wild-type LipL32 binding with different substrates (F30, full-length fibronectin, collagen type IV and plasminogen) changed very little or not at all upon addition of Ca 2ϩ . The same was observed for the Q67A and S247A mutants, both of which retain the ability to bind Ca 2ϩ . Furthermore, the D163-168A mutant and LipL32 C-terminal fragment, both of which lack an intact Ca 2ϩ -binding site, interacted with target molecules with affinities equal or greater (in the case of the C-terminal fragment) than that observed for the wildtype protein (Table 3). These observations lead us to conclude that calcium ions do not influence binding of LipL32 to ECM components or plasminogen.
Our results do not reproduce those reported by Tung et al. (16) who observed a 4-fold decrease in affinity for the LipL32-F30 interaction in the absence of Ca 2ϩ . We believe that this is due to the special attention we paid to control Ca 2ϩ levels and avoid Ca 2ϩ -dependent artifacts at all stages of the ELISA protocol. The following three observations illustrate this point. 1) The results in Fig. 6, Table 3, and supplemental Figs. S1-S5 show that if Ca 2ϩ is present in the antibody binding stages, no significant differences in apparent affinities are observed in experiments in which Ca 2ϩ was present or absent in the initial LipL32-ECM incubation and associated washes. 2) We observed that if Ca 2ϩ is removed from the primary and secondary antibody binding steps, the overall signal is significantly reduced, independent of whether or not Ca 2ϩ is present during the ECM-LipL32 incubation stage (supplemental Figs. S1 and S2). 3) Furthermore, this Ca 2ϩ -dependent behavior associated with the antibody binding stages of the protocol was also observed for the D163-168A mutant and the LipL32 185-272_His tag C-terminal fragment, both of which do not bind Ca 2ϩ (supplemental Figs. S1 and S2). These results show that special attention should be paid to control Ca 2ϩ levels at all stages of the ELISA-based protocols employed to measure LipL32 interactions with ECM components.
Our results lead us to conclude that whereas Ca 2ϩ binding contributes to the conformational stability of LipL32, calcium binding does not play a crucial role in mediating or modulating its interactions with host extracellular matrix proteins. As Ca 2ϩ is abundant in extracellular host fluids at concentrations well above the Ca 2ϩ binding constants measured in this study, it is most likely that Ca 2ϩ plays a structural role in LipL32 function or contributes to other uncharacterized biological LipL32 activities.  ) values for the interaction of LipL32 wild-type, LipL32 mutants, and C terminus LipL32 in the presence or absence of Ca 2؉ calculated by ELISA from data presented in Fig. 6