Optimal Mutant Model of Human S100A3 Protein Citrullinated at Arg51 by Peptidylarginine Deiminase Type III and Its Solution Structural Properties.

S100A3 protein, a member of the EF-hand-type Ca2+-binding S100 protein family, undergoes a Ca2+-/Zn2+-induced structural change to a tetrameric state upon specific citrullination of R51 in human hair cuticular cells. To elucidate the underlying mechanism, we prepared recombinant mutant S100A3 proteins, including R51A, R51C, R51E, R51K, and R51Q, as potential models of post-translationally modified S100A3 and evaluated their biophysical and biochemical properties relative to wild-type (WT) S100A3 and WT citrullinated in vitro. Size exclusion chromatography (SEC) showed that R51Q formed a tetramer in the presence of Ca2+, while Ca2+ titration monitored by Trp fluorescence indicated that R51Q had Ca2+-binding properties similar to those of citrullinated S1003A. We therefore concluded that R51Q is the optimal mutant model of post-translationally modified S100A3. We compared the solution structure of WT S100A3 and the R51Q mutant in the absence and presence of Ca2+ and Zn2+ by SEC-small-angle X-ray scattering. The radius of gyration of R51Q in the metal-free state was almost the same as that of WT; however, it increased by ∼1.5-fold in the presence of Ca2+/Zn2+, indicating a large expansion in molecular size. By contrast, addition of Ca2+/Zn2+ to WT led to nonspecific aggregation in SEC analysis and dynamic light scattering, suggesting that citrullination of S100A3 is essential for stabilization of the Ca2+-/Zn2+-bound state. These findings will lead to the further development of structural analyses for the Ca2+-/Zn2+-bound S100A3.

Introduction

The S100 protein family, with more than 20 members, constitutes the largest subfamily of EF-hand-type Ca2+-binding proteins.[1] S100 proteins are characterized by their N-terminal S100 protein-specific and C-terminal canonical Ca2+-binding loops.[2,3] In addition, some S100 proteins have been reported to bind other divalent cations, such as Zn2+ and Cu2+.[4−10] Whereas most family members are noncovalently linked to form homo- or heterodimers, some S100 proteins form assemblies of tetramers or higher-order oligomers that undergo structural changes upon binding of divalent cations such as Ca2+.[11−17]

S100 proteins are expressed in various tissues, and some are distributed in human hair follicles.[18] For example, S100A3 is highly expressed in human hair follicles and retained in hair cuticular cells.[19,20] Although S100A3 has recently been reported to interact with the retinoid receptor,[21] its functional role in hair cuticular cells remains to be clarified. S100A3 consists of 101 amino acids, including four arginine residues (R3, R22, R51, and R77) per monomer.[22] These arginine residues in S100A3 are converted to citrulline by peptidylarginine deiminases (PADs) that are coexpressed in the hair follicle.[23] Citrullination alters both intramolecular and intermolecular ionic and/or hydrophobic interactions and is involved in various physiological functions.[24] In particular, R51 of S100A3 is selectively citrullinated by the isozyme PAD type III (PAD3) that is colocalized with S100A3 in hair cuticle cells. Upon citrullination, the binding affinity of S100A3 for Ca2+ and Zn2+ increases cooperatively.

Although S100A3 is usually present as a dimer, it has been suggested that PAD3-mediated citrullination of R51 in the S100A3 dimer causes concomitant Ca2+-dependent assembly to a homotetramer.[25] It has been also reported that there is a correlation between hair damage and citrullination of S100A3.[26] That study suggested that structural changes associated with citrullination of S100A3 have an important role in the maturation of hair cuticles.[26] Interestingly, it has been revealed that S100A3 and PAD3 are evolutionally related to the emergence of mammalian hair.[27]

In the present study, therefore, we have used structural biology techniques to investigate the mechanism underlying the increased affinity for Ca2+ and Zn2+ and the conformational change of S100A3 that follows citrullination by PAD3. Although the dimeric structure of S100A3 without metal ions has been reported,[4,28] the structure of the Ca2+-/Zn2+-bound citrullinated S100A3 homotetramer has not been clarified in detail for two reasons: it is ethically difficult to obtain large quantities of the citrullinated form of S100A3 from human tissues; and in vitro modification of S100A3 by PAD3 will mainly convert R51, but it may also partially convert other arginine residues to citrulline.[23] For this reason, the highly homogeneous citrullinated form of S100A3 is difficult to prepare in large quantities. We therefore decided to use an artificial model of post-translationally modified S100A3.

Because there is no codon for citrulline, it is difficult to prepare citrullinated proteins by current genetic technology. Therefore, we prepared R51-substituted mutants as mimics of citrullinated S100A3. Based on a previous investigation of the citrullination of myelin basic protein (MBP) by PAD,[29] in which arginine residues were converted to glutamine, which is structurally similar to citrulline, we first prepared an R51Q variant. For comparison, we also prepared R51 variants substituted with noncharged alanine (R51A), cysteine with a thiol group (R51C), negatively charged glutamic acid (R51E), positively charged lysine (R51K), and hydrophobic leucine (R51L). To verify which mutant provided the best mimic of S100A3 citrullinated by PAD3, we compared the biophysical and biochemical properties of wild-type (WT) S100A3, the citrullinated form, and the mutated proteins. In addition, we analyzed the solution structure of WT and R51Q as a mimic of citrullinated S100A3 by using small-angle X-ray scattering (SAXS) in the presence or absence of Ca2+ and/or Zn2+.

Results and Discussion

Preparation of Recombinant S100A3 Samples

We expressed the recombinant WT and mutant S100A3 proteins in the Escherichia coli SHuffle T7 strain. For WT, approximately ∼27 g of cells was harvested from a culture grown in 4 L of Luria–Bertani (LB) medium, and ∼1.2 mg of the recombinant protein was purified from these cells. The purity of S100A3 after each chromatographic step was confirmed by using 15% N-[tris(hydroxylmethyl)-methyl]glycine (Tricine) sodium dodecyl sulfate polyacrylamide electrophoresis (SDS-PAGE). The S100A3 mutants were expressed and purified by the same procedure used for WT. The R51A, R51C, and R51E mutants were obtained at a yield of ∼1.1 mg and R51K and R51Q at ∼2.6 and ∼4.0 mg, respectively. Unfortunately, the amount of R51L mutant obtained was too small to be used in subsequent experiments. Post-translationally citrullinated S100A3 was prepared by a reaction with PAD3 in vitro. Approximately 0.6 mg of the citrullinated form (hereafter R51Z) of S100A3 was obtained by reacting 1.2 mg of WT S100A3 with 0.25 mg PAD3.

Comparison of Ca2+/Zn2+-Dependent Tetramerization of S100A3 Evaluated by SEC

To confirm whether the tetramerization of S100A3 is dependent on the presence or absence of metal ions, SEC analysis of the protein was performed in solution with and without Ca2+ and/or Zn2+. In the absence of metal ions, a clear single peak of absorption at 280 nm was detected in SEC analysis (Figure A). Previous studies have reported that, in the absence of Ca2+ or Zn2+, citrullinated S100A3 exists as a dimer in both crystal and solution structures.[4,25,28] Therefore, we assumed that the detected single peak, which eluted earlier than chymotrypsinogen A (25 kDa) in the absence of Ca2+ and Zn2+, corresponds to dimeric S100A3. The reason why the S100A3 dimer, which has a molecular mass of ∼24 kDa, seemed larger than 25 kDa (chymotrypsinogen A) in SEC might be explained by the molecular shape of S100A3, which is not a standard sphere but rather a hemispheric structure.

Figure 1

Ca2+- or Ca2+/Zn2+-dependent tetramerization of S100A3. WT, R51Z, and various mutants (0.1 mM each) premixed in elution buffer or elution buffer containing 5.0 mM CaCl2 and 0.1 mM Zn(CH3COO)2 were loaded on a Superdex 75 10/300 GL column for SEC analysis. (A) Metal-free elution profile. (B) 5.0 mM Ca2+ elution profile. (C) 5.0 mM Ca2+ plus 0.1 mM Zn2+ elution profile. d indicates the elution profile of dimeric WT. t indicates the elution profile of putative tetrameric R51Z. Arrows indicate the elution volume of ovalbumin (43 kDa) and chymotrypsinogen A (25 kDa) standards.

In the absence of Ca2+ and Zn2+, the peak positions of R51E, R51K, R51Q, and citrullinated S100A3 (R51Z) differed slightly from that of WT in the SEC analysis. Among these mutants, R51E and R51Q were detected at an elution volume similar to that of R51Z (Figure A). In the presence of Ca2+ alone (Figure B) or Ca2+ and Zn2+ (Figure C), more than half of R51Z was present as the tetramer, whereas the majority of WT remained as the dimer. The ratio of tetramerization of R51Q was similar to that of R51Z (Figure B,C). The tetramerization state of R51E was different from that of R51Z: in the presence of Ca2+/Zn2+, most of the R51E formed a tetramer (Figure C). The elution profile of Ca2+/Zn2+-loaded R51C was clearly different from that of WT and R51Z, with a blurred elution peak (Figure C). This suggested that the oligomerization of R51C differed from that of WT S100A3. Collectively, the results of SEC analysis suggest that the oligomeric state of R51Q is very similar to that of R51Z in both the absence and presence of Ca2+ and/or Zn2+.

Comparison of Ca2+-Binding Affinity by Using Trp Fluorescence Titration

The single Trp residue (Trp45) in S100A3, which is located between helix II and helix III,[28] has the ability to emit fluorescence when excited, and the intensity of this fluorescence is sensitive to the hydrophobic environment around the Trp residue. It is assumed that helix III undergoes a substantial reorientation upon Ca2+-binding to S100A3;[11−17] thus, Trp45 fluorescence has previously been used to monitor Ca2+-binding and the concomitant large conformational change of S100A3.[30]

Observation of Trp fluorescence showed that the conformation of R51Z and the mutated S100A3 proteins clearly changed at lower Ca2+ concentrations as compared with WT (Figure ). In other words, R51Z and the mutated S100A3 proteins had higher Ca2+-binding affinities relative to WT S100A3. The shapes of the Ca2+ titration curves for R51A, R51C, and R51Q were similar to that for R51Z, whereas those for R51K and R51E were different from those for WT and R51Z (Figure ). This implies that R51K and R51E bind Ca2+ but undergo structural changes that are significantly different from those of R51Z, R51C, R51Q, and R51A. Furthermore, the titration curves indicated that the Ca2+-binding affinity of R51A and R51C was higher than that of WT but lower than that of R51Z. The titration curve of R51Q showed a substantial overlap with R51Z, suggesting that R51Q has almost the same Ca2+-binding properties as R51Z and also that the Ca2+-binding affinity of R51Q is almost the same as that of R51Z.

Figure 2

Fluorescence titration curves of Ca2+-induced conformational changes. The fluorescence intensities of WT, R51Z, and mutated S100A3 proteins (2 μM each) were recorded in the presence of increasing Ca2+ concentration. The maximum fluorescence reduction (100%) was defined as the reduction observed on the addition of 100 mM CaCl2. Each plot represents the mean value of five independent experiments.

Comparison of Zn2+-Binding Affinity by Using Circular Dichroism (CD) Spectrometry

It is well known that S100A3 changes the secondary structure, expected to helix IV, in a Zn2+-dependent manner.[4] To evaluate Zn2+-binding affinity of S100A3, we utilized far-ultraviolet circular dichroism (UV CD) spectrometry to analyze secondary structural changes of WT, R51Z, and mutated S100A3 proteins loaded with Zn2+ (Figure ). The observed secondary structure of S100A3 showed that R51Z and the mutated S100A3 proteins, except for R51K, significantly changed at the same Zn2+ concentrations as compared with WT (Figure ). Although the secondary structures of WT and R51Z (Figure A,B) in the presence of 0.5 molar equivalent Zn2+ were not much changed from the metal-free form, all of the mutated S100A3 showed a change in the secondary structure (Figure C–G). Among the mutants, R51Q showed the smallest change in the secondary structure at 0.5 molar equivalent Zn2+ as compared with the metal-free form (Figure G). Additionally, the profile of R51Q by Zn2+ titration was very similar to that of R51Z. These are significantly different from that for WT. As a consequence, it is suggested that the structural change of R51Q is the most similar to that of R51Z when Zn2+ is bound (Figure ).

Figure 3

Reduction of the α-helix content in S100A3. The far-UV CD spectra show the secondary structural changes of (A) WT, (B) R51Z, and mutated S100A3 [(C) R51A, (D) R51C, (E) R51E, (F) R51K, and (G) R51Q]. Also seen are the far-UV CD spectra recorded after cumulative addition of Zn2+. The concentration of proteins was 15 μM in all of the cases. The secondary structure of S100A3 was analyzed to a concentration of 4 molar equivalent Zn2+.

The calculated α-helix contents of WT, R51Z, and the mutated S100A3 are described in Table S1 in the Supporting Information. The α-helix content of WT (52.5%) was roughly consistent with that of the crystal structures (55%; PDB code: 3NSI for WT). R51Z and the mutated S100A3 proteins, except for R51K, did not show large fluctuations in α-helix contents by citrullination and site-directed mutagenesis of R51. By contrast, the α-helix content of R51K was apparently different from that of WT, R51Z, and the other mutated S100A3 proteins (Figure A,B,F, and Supporting Information Table S1). The far-UV CD spectra of R51K showed a much lower α-helix content as compared with that of R51Z. This indicates that R51K is unsuitable for a model of R51Z.

Collectively, the results of SEC analysis (Figure ), Trp fluorescence titration (Figure ), and CD spectrometry (Figure ) suggest that the optimal mutant for use as a model of the citrullinated S100A3 is R51Q.

Determination of an Appropriate Mimic of Post-Translationally Modified S100A3

A previous study on the citrullination of recombinant murine MBP demonstrated that the genetic substitution of arginine by glutamine had a similar effect on the MBP structure and function as enzymatic deimination of arginine to citrulline.[29] That investigation indicated that a mutant protein containing glutamine instead of citrulline was effective as a mimic. In the present study, we also tried to identify an appropriate mimic for citrullinated S100A3. The above comparison of the biophysical and biochemical properties of the S100A3 mutants, WT, and post-translationally citrullinated S100A3 indicated that R51Q shows the most similar Ca2+-binding, Zn2+-binding, and Ca2+-/Zn2+-dependent tetramerization properties to those of R51Z. Considering only the surface charge of the protein, however, it might be assumed that R51A would be appropriate as a model of the S100A3 protein citrullinated by PAD3. Indeed, a previous study reported R51A as pseudo-citrullinated S100A3 based on the results of Trp fluorescence titration and SEC analysis, which demonstrated that, under the same Ca2+-conditions, R51A shows similar Ca2+-binding and tetramerization properties to those of R51Z. However, the present results indicated that R51Q is more suitable as a model of citrullinated S100A3 than R51A, suggesting that not only the surface charge but also the conformation and size of the side chains are important factors in determining an appropriate mimic. To confirm this hypothesis, structural information on the S100A3 tetramer at the atomic level will be required.

To evaluate in more detail whether R51Q is suitable as a model of R51Z, we conducted Ca2+-titration by using Trp fluorescence in the presence of Zn2+ (Figure A,C,E). In addition, Zn2+-titration in the presence of Ca2+ was carried out and analyzed by utilizing far-UV CD spectrometry for WT, R51Z, and R51Q (Figure B,D,F). All of the three proteins showed significant increase of the Ca2+-binding affinity (Figure A,C,E). However, WT showed 1 order lower Ca2+-binding affinity than R51Z and R51Q. In the presence of Ca2+, R51Z and R51Q showed almost the same changes in the secondary structure by Zn2+-titration (Figure D,F). These results support the fact that R51Q is suitable for the model of R51Z. The α-helix contents calculated based on these spectra are listed in Table S2 in the Supporting Information.

Figure 4

Properties for cooperative increase of the Ca2+/Zn2+-binding affinity of S100A3. Ca2+-titration by Trp fluorescence analysis of (A) WT, (C) R51Z, and (E) R51Q in the absence or presence of Zn2+. The relative rates of each measured point versus the maximum reduction are plotted. The plotted points in the horizontal axis of (A) are between 1 and 100 mM Ca2+, whereas those for (C) and (E) are between 0.1 and 20 mM. Reduction of the α-helix contents when Zn2+ is added in the presence of Ca2+ for (B) WT, (D) R51Z, and (F) R51Q. “Metal free” showed CD spectra in the absence of Ca2+ and Zn2+, and “0” showed CD spectra in the presence of 1 mM Ca2+ and in the absence of Zn2+.

In terms of the production of recombinant S100A3 in the SHuffle T7 strain, the yield of WT was only ∼1.2 mg from cells cultured in 4 L of LB medium. After reaction with PAD3 in vitro, the yield of S100A3 was reduced to approximately half (∼0.6 mg). By contrast, the yield of R51Q under the same conditions was ∼4.0 mg (i.e., approximately 3–4-fold more abundant than WT). Among the mutants prepared in the study, R51Q was obtained in the highest yield and was found to be the optimal model mutant for use as citrullinated S100A3. This artificial protein will be suitable for further structural biological analyses, which generally require a large quantity of the protein sample. Taking all of the observations together, we concluded that the R51Q S100A3 protein is the best model of WT S100A3 that has been post-translationally citrullinated by PAD3.

Ca2+-/Zn2+-Dependent Oligomerization of S100A3 Validated Using Dynamic Light Scattering (DLS)

Based on the above findings, we further examined changes in the molecular diameters of WT, R51Z, and R51Q in the absence or presence of Ca2+/Zn2+ using dynamic light scattering (DLS). In the metal-free state, the diameter of all samples at 0.5 mM was approximately 40 Å. Thus, the molecular diameter of S100A3 was not altered by replacement of R51 with citrulline or glutamine residues. In the presence of Ca2+/Zn2+, however, the protein samples showed significant differences. The molecular diameter of WT increased to approximately 600 Å in 1.0 mM CaCl2 and 0.5 mM Zn(CH3COO)2 (Figure A), and the Ca2+-/Zn2+-loaded WT sample was precipitated during measurement. By contrast, in the presence of a low concentration of Ca2+ (less than 1.0 mM) at 1 molar equivalent of Zn2+, no significant changes in the molecular diameter of WT were observed (Figure S1A in the Supporting Information). Previous experiments have shown that S100A3 at concentrations higher than 0.4 mM is often precipitated after the addition of Ca2+/Zn2+. (The concentrations of the S100A3 protein used in the SEC and Trp fluorescence titration were lower at 100 μM and 2 μM, respectively.) This aggregation probably results from nonspecific binding of Ca2+ and Zn2+ at high sample concentrations.

Figure 5

DLS analysis of changes in the molecular diameters of WT, R51Z, and R51Q S100A3 (0.5 mM each) in the absence or presence of Ca2+/Zn2+. Each peak represents the mean value obtained from 10 independent experiments. (A) Molecular diameter of WT increased significantly after the addition of 1 mM Ca2+ and 0.5 mM Zn2+. (B, C) Molecular diameters of R51Z (B) and R51Q (C) increased approximately 1.5-fold after the addition of 2 mM Ca2+ and 0.5 mM Zn2+.

In contrast to WT, the molecular diameter of R51Z did not change and no precipitation occurred in the presence of 1 mM CaCl2 and 0.5 mM Zn(CH3COO)2 (data not shown). On the other hand, the molecular diameter of R51Z increased to approximately 57 Å in the presence of 2 mM CaCl2 and 0.5 mM Zn(CH3COO)2, (Figure B). Similarly, the molecular diameter of R51Q increased to approximately 61 Å in the presence of 2 mM CaCl2 and 0.5 mM Zn(CH3COO)2 (Figure C). The difference between the diameters of R51Z and R51Q was within the range of measurement error. Thus, the molecular diameter of R51Q was almost the same as that of R51Z at the same Ca2+ and Zn2+ concentrations. Collectively, these results suggested that binding of Ca2+ and Zn2+ to S100A3 and tetramerization were stabilized by either citrullination of R51 or its mutation to glutamine.

Furthermore, we analyzed the molecular diameters of R51Z and R51Q with various concentrations of Ca2+ utilizing DLS. As a result, although R51Z and R51Q in the presence of Ca2+ showed small changes in the molecular diameter compared with the metal-free forms, the diameters did not increase to more than 55 Å (Figure S1B,C in the Supporting Information). Considering these results, efficient tetramerization of R51Z and R51Q did not occur when binding Ca2+ alone. Although the structures of R51Z and R51Q were changed by Ca2+-binding, the tetramer was not stabilized. Therefore, it is presumed that both Ca2+- and Zn2+-binding to S100A3 is suggested to be required to stabilize the tetramer.

Solution Scattering Profile of Ca2+-/Zn2+-Binding S100A3 Using SEC-SAXS

We performed SEC-SAXS measurements to elucidate the solution structure of the R51Q mutant, which was determined as the best model of S100A3 that has been post-translationally citrullinated by PAD3. The chromatogram of forward scattering intensities, I(0), of the metal-free R51Q mutant showed a single peak, and the radius of gyration was almost constant (Figure A), indicating that the metal-free R51Q solution was monodisperse. Figure B shows that the scattering profile of metal-free WT (blue line) was well superimposed on that of R51Q (black line), indicating that the solution structure of S100A3 was not changed by the mutation of R51 to glutamine. Furthermore, the theoretical scattering profile calculated from the WT crystal structure (PDB ID: 3NSI; red line) was identical to the experimental scattering of the R51Q mutant, indicating that the solution structure of the dimer of the S100A3 protein is essentially the same as the crystal structure.

Figure 6

(A) SEC-SAXS analysis of metal-free R51Q. Solid line and circles indicate, respectively, intensities at zero angles (left axis) and radius of gyration (right axis) from Guinier approximation. (B) Scattering profiles derived from SAXS experiments. Blue and black lines show metal-free WT and metal-free R51Q, respectively. The red line shows the theoretical scattering profile of (PDB ID: 3NSI) determined by CRYSOL. The χ2 value is 0.925.

Next, SEC-SAXS analysis of the Ca2+-/Zn2+-bound S100A3 proteins was carried out. As described above for DLS, addition of Ca2+/Zn2+ to high concentrations of the WT sample led to precipitation. Therefore, Ca2+-/Zn2+-loaded samples were diluted to 0.23 mg/mL (20 μM) in the buffer containing 50 mM tris-(hydroxy-methyl)aminomethane (Tris)–HCl buffer (pH 7.6), 150 mM NaCl, 1.0 mM dithiothreitol (DTT), 2.5 mM CaCl2, and 20 μM Zn(CH3COO)2 and then concentrated to 5.8 mg/mL, which solved the problem of aggregation. The chromatogram of I(0) showed a minor shoulder to the left of the major peak (Figure A). To avoid this minor component, the right side of the major peak was selected as the region for extrapolation to infinite dilution of the Ca2+-/Zn2+-bound protein. Figure B shows the dimensionless Kratky plots of metal-free (black line) and Ca2+-/Zn2+-bound (red line) R51Q. Addition of Ca2+/Zn2 led to a drastic change in the scattering profile of R51Q. The Rg value calculated from Guinier analysis increased from 19.0 ± 0.4 to 30.3 ± 1.4 Å, and the bell-shaped peak of Ca2+-/Zn2+-bound R51Q was shifted to high Q × Rg as compared with the metal-free R51Q, indicating that the size of R51Q was enlarged by binding Ca2+/Zn2+.

Figure 7

(A) SEC-SAXS analysis of the Ca2+/Zn2+ complex of R51Q. Solid line and circles indicate, respectively, intensity at zero angles (left axis) and radius of gyration (right axis) from Guinier approximation. (B) Dimensionless Kratky plots of R51Q. Black and red lines show metal-free and Ca2+-/Zn2+-bound R51Q, respectively.

Evaluation of the Solution Structure of S100A3

In this study, we used a series of biophysical analyses to evaluate the character of S100A3 in solution. In DLS analysis, R51Z and R51Q were not precipitated at concentrations of 0.5 mM and both showed a ∼1.5-fold increase in molecular diameter in the presence of 2.0 mM Ca2+ and 0.5 mM Zn2+, whereas WT was aggregated under conditions with similar (but lower) metal ion concentrations. In SEC analysis, WT showed a small absorbance peak at 280 nm corresponding to the oligomeric form in the presence of Ca2+/Zn2+ (Figure C), indicating that it also undergoes partial nonspecific aggregation even at a concentration of 100 μM. In a previous study, it was assumed that efficient tetramerization of S100A3 requires both citrullination of R51 and the presence of Ca2+ and Zn2+.[25] It has also been reported that S100A3 shows a citrullination-associated increase in binding affinity for Ca2+ and Zn2+, which is also associated with structural changes involving these metal ions.[23]

In terms of Ca2+-/Zn2+-dependent oligomerization, we found that the molecular size of R51Q observed by DLS and SEC-SAXS was ∼1.5-fold larger in the presence of Ca2+ and Zn2+ than in a metal-free solution. Thus, the same changes in molecular size were observed in different experiments under similar conditions. It seems that citrullinated S100A3 is regularly oligomerized in the presence of 4–5 molar equivalents of Ca2+ and 1 molar equivalent of Zn2+. In addition, the present data indicate that citrullination of S100A3 plays an important role in stabilizing the Ca2+-/Zn2+-bound state. Under the same Ca2+ and Zn2+ conditions, the molecular sizes of R51Z and R51Q were increased and the Ca2+-/Zn2+-binding states were more stable relative to WT.

Implication for the Tetramerization Mechanism and Biological Function of S100A3

In this work, we showed that R51Q is the best mimic of R51Z (the citrullinated S100A3 at R51). The mechanism by which modification of R51 induces tetramerization of S100A3 was not elucidated; however, our work and the previous various reports gave us insight into the mechanism of the unique structural change.

From the X-ray structure of S100A3 that we determined previously (PDB code: 3NSO), R51 forms a salt bridge with D54 of helix III and hydrogen-bonding interaction with the main chain carbonyl O of C93 at the C-terminal tail (Figure A and Supporting information Figure S2A).[28] The role of R51 seems to stabilize the compact structure. This stabilized state cannot bind low concentrations of Ca2+ and hardly induces reorientation of helix III. Two R51 residues of this S100A3 dimer are placed at the surface of the molecule. The dimer–dimer interface would be caused repulsion by positively charged if the two dimers of S100A3 formed a homotetramer. Therefore, two dimers would not efficiently form a homotetramer, which easily dissociates to go back to the dimers (Figure S3 in the Supporting Information). When R51 is citrullinated, the salt bridge with D54 should be lost (Figure B). This structural modification will release the C-terminal tail or helix III from the position of R51. The Ca2+-binding induces helix III reorientation. These two hypotheses are consistent with the results in this work. R51Z had higher affinity for Ca2+ than WT (Figure ). Additionally, R51Z showed induction of the tetramerization, whereas WT retained the dimer when Ca2+ is bound to the proteins (Figure B). The Zn2+-binding disrupts the secondary structure; the α-helix content is reduced (Figure ). This further enhances the structural change and increases the Ca2+-binding affinity cooperatively (Figure ). In fact, without secondary structural change, the complete Ca2+-bound model of S100A3 could not be constructed from the WT S100A3 and Ca2+-bound S100A4 structures (Figure S4 in the Supporting Information) (PDB codes: 3NSO and 3C1V).[31] The Ca2+-/Zn2+-bound R51Z, in which the α-helix content was reduced and the helix III was reoriented, could form the homotetramer efficiently (Figure C,D,B).

Figure 8

One of the roles of R51 is to stabilize the structure of S100A3. (A) R51 forms a salt bridge with D54 and also forms a hydrogen-bonding interaction with C90. This was found in PDB 3NSO. (B) When R51 is substituted, the interactions are disrupted. This is observed from a putative model constructed based on 3NSO. This structural relaxation probably causes the increase in metal binding and enhances the structural change (and further tetramerization).

From our results, R51Q showed similar properties to R51Z. When R51 was substituted with glutamine, the length of the side chain is shortened (Figure S2B in the Supporting Information). The putative model that was constructed from the WT S100A3 structure (PDB code: 3NSO) showed that R51Q should impair the interaction with D54 or the C-terminal tail (Figure S2B in the Supporting Information). This small structural change would make the molecule to be relaxed. This might be why R51Q increases the Ca2+-binding affinity (Figure ) and induces tetramerization efficiently (Figure B). In the presence of Zn2+, the increase in the Ca2+-binding affinity and the efficiency of homotetramerization are further enhanced (Figures E,F and 5C). These phenomena were very similar to those found in R51Z. Further experiments including near-UV CD, dynamic quenching fluorescence, fluorescence spectra using 8-anilino-1-naphthalene sulfonate of all of the mutants, WT, and R51Z under a variety of conditions are required to elucidate their detailed structures.

The citrullinated S100A3 is a major component of the hair cuticle. The biological role of citrullination of R51 in S100A3 is not clearly determined; however, the content of citrullinated S100A3 was reported to be related to the rigidity of endocuticle of aged hair.[26] Unfortunately, whether R51Q behaves in a similar way to R51Z in terms of biological function was not elucidated in this work. In future works, the biological behavior of R51Q should be validated to make this study definitive. For instance, if recombinant R51Q is expressed in hair cuticular cells to create “R51Q mutant hair cuticle”, it can be compared with natural hair cuticle. Furthermore, if transgenic mice that express R51Q can be created in the future, we can further discuss whether R51Q is suitable as a model of citrullinated S100A3.

Experimental Section

Protein Expression and Purification

The cDNA of human S100A3 was inserted between the NdeI and XhoI sites of a pET-41a (+) (Novegen) transfer vector. The E. coli SHuffle T7 strain (New England Biolabs) transformed with the resulting plasmid was grown in LB medium at 30 °C. At OD600 = 0.8–1.0, the expression of S100A3 was induced by adding 0.1 mM isopropyl β-d-1-thiogalactopyranoside, followed by culture at 27 °C overnight. Cells were harvested by centrifugation at 13 000g for 30 min, washed with phosphate-buffered saline, and sonicated in 0.1 M Tris–HCl buffer (pH 7.6) containing 1.0 mM DTT, 1.0 mM ethylenediamineteraacetic acid (EDTA), and 1.0 mM phenyl-methylsulfonyl fluoride. The cell lysate was centrifugated at 48 000g for 45 min. The supernatant was loaded onto a HiTrap Q FF column (GE Healthcare), which was pre-equilibrated and washed with buffer A [20 mM Tris–HCl buffer (pH 7.6), 1.0 mM DTT, and 1.0 mM EDTA]. S100A3 was eluted with a NaCl linear gradient (0–0.45 M). The eluted fractions were applied to Tricine SDS-PAGE, and fractions containing S100A3 were collected.[32] (NH4)2SO4 was added to the pooled fractions to a final concentration of 1.5 M. After centrifugation at 12 000g for 30 min, the resultant supernatant was loaded onto a HiTrap Butyl HP column (GE Healthcare) pre-equilibrated with buffer A containing 1.5 M (NH4)2SO4. To remove contaminants, the column was washed with a linear gradient of (NH4)2SO4 (1.5–0.3 M). Bound S100A3 was eluted with a linear gradient of (NH4)2SO4 (0.3–0 M). The eluted fractions were applied to Tricine SDS-PAGE, and fractions containing S100A3 were again collected. After dialysis against buffer A, the sample was applied to a Mono Q 5/50 GL column (GE Healthcare) pre-equilibrated, and washed with buffer A. S100A3 was eluted with NaCl gradient (0–0.18 M), and fractions containing S100A3 were pooled and concentrated to approximately 3.0 mL by centrifugal filtration. Lastly, the sample was passed through a HiLoad 16/60 Superdex 75 prep-grade column (GE Healthcare) pre-equilibrated with 50 mM Tris–HCl buffer (pH 7.6) containing 150 mM NaCl, 1.0 mM DTT, and 10 μM EDTA.

Production of Mutated S100A3

Site-directed mutagenesis of Arg51 of human S100A3 was done by polymerase chain reaction (PCR) using plasmid pET-41a(+)-S100A3 as a template and QuikChange II (Agilent) according to the instruction manual. The following mutagenic primer sets were used with the modified bases underlined: for R51A, 5′-ACCTGGACCCCGACTGAGTTTGCGGAATGTGACT-3′ and 5′-AGTCACATTCCGCAAACTCAGTCGGGGTCCAGGT-3′; for R51C, 5′-CTGGACCCCGACTGAGTTTTGCGAATGTGACTACAACAAAT-3′ and 5′-ATTTGTTGTAGTCACATTCGCAAAACTCAGTCGGGGTCCAG-3′; for R51E, 5′-TAGTCACATTCCTCAAACTCAGTCGGGGTCCAGGTGG-3′ and 5′-CCACCTGGACCCCGACTGAGTTTAAGGAATGTGACTA-3′; for R51K, 5′-TAGTCACATTCCTTAAACTCAGTCGGGGTCCAGGTGG-3′ and 5′-CCACCTGGACCCCGACTGAGTTTAAGGAATGTGACTA-3′; for R51L, 5′-GACCCCGACTGAGTTTCTGGAATGTGAC-TACAACA-3′ and 5′-TGTTGTAGTCACATTCCAGAAACTCAGTCGGGGTC-3′; for R51Q, 5′-GACCCCGACTGAGTTTCAGGAATGTGACTACAAC-A-3′ and 5′-TGTTGTAGTCACATTCCTGAAACTCAGTCGGGGTC-3′. Transformation, expression, and purification were carried out as described for WT S100A3.

Post-Translational Modification of S100A3 by the PAD3 Enzyme in Vitro

Recombinant PAD3 enzyme was prepared in E. coli, and its specific activity (where the amount that citrullinated 1 μmol benzoyl-l-arginine ethyl ester per hour at 55 °C was defined as 1 unit) was determined as previously reported.[33] Conventionally, 1.0 μg of recombinant S100A3 was reacted with 25 milliunits of PAD3 enzyme in 20 μL of 100 mM Tris–HCl buffer (pH 7.5) containing 10 mM CaCl2 and 5.0 mM DTT at 37 °C. Modified S100A3 was isolated by size exclusion chromatography (SEC) using Superdex 75 10/300 GL (GE Healthcare) in 50 mM Tris–HCl buffer (pH 7.6) containing 150 mM NaCl, 1.0 mM DTT, and 1.0 mM EDTA. Modification of S100A3 was confirmed by 2-dimensional PAGE. The 2-dimensional PAGE gel is shown in Figure S5 in the Supporting Information.

Tryptophan Fluorescence Titration

Ca2+-dependent structural changes in S100A3 were observed by Trp fluorescence. Ca2+ titration into recombinant WT, R51Z, and mutated S100A3 proteins (2 μM each) was carried out in 50 mM Tris–HCl buffer (pH 7.6) containing 150 mM NaCl and 1.0 mM DTT, as previously described.[30,34] The emission fluorescence intensity at 340 nm (excitation, 295 nm) was recorded at 60 s after each cumulative addition of CaCl2 using an F-4500 fluorospectrometer (Hitachi) and a quartz cell (1.0 cm × 1.0 cm, Jasco). In addition, to confirm the effect of Zn2+ on Ca2+-dependent structural change in S100A3, Ca2+ titration was carried out in the absence or presence of Zn2+.

CD Spectrometry

Zn2+-dependent secondary structural changes of recombinant WT, R51Z, and mutated S100A3s (15μM) were monitored by far-UV CD spectrometry.[4,25] The CD spectra of S100A3 were recorded in 20 mM Tris–HCl buffer (pH 7.6) containing 150 mM NaCl and 0.2 mM DTT using a J-720 spectropolarimeter (Jasco, Japan) equipped with a rectangular quartz cell (1 mm path length). The spectrum data were collected four times at a bandwidth of 2 nm, a scan speed of 100 nm/min, and a response time of 4 s and combined. In addition, to observe the effect of Ca2+ on Zn2+-dependent secondary structural changes of S100A3, Zn2+-titration was carried out in the absence or presence of 1 mM CaCl2. Considering that the crystal structures of apo-dimers of WT (PDB code: 3NSI)[28] comprise four helices per monomer but not a β-sheet, the overall α-helix content was calculated with the following equation: α-helix content (%) = ([θ]222 + 2340)/-30 300.[35]

SEC and Dynamic Light Scattering

The native apparent molecular mass of WT and mutated S100A3 (0.1 mM) in solution was estimated by SEC using Superdex 75 10/300 GL (GE Healthcare) in 50 mM Tris–HCl buffer (pH 7.6) containing 150 mM NaCl and 1.0 mM DTT at a flow rate of 0.5 mL/min. Measurements of the Ca2+-dependent molecular weight change were performed in the presence of 5.0 mM CaCl2 or 5.0 mM CaCl2 and 0.1 mM Zn(CH3COO)2. In addition, Ca2+-dependent molecular diameter changes in WT, the citrullinated protein, and the mutated proteins (0.5 mM) were observed by DLS with the addition of Ca2+ and Zn2+ in 50 mM Tris–HCl buffer (pH 7.6) containing 150 mM NaCl and 1.0 mM DTT using a Zetasizer Nano ZS90 (Malvern Panalytical) and a ZEN0118 cell (Malvern Panalytical). Both types of experiments were conducted at room temperature.

SEC-SAXS and Data Analysis

SEC-SAXS measurements were performed at BL-10C of the Photon factory (Tsukuba, Japan) with an X-ray wavelength of λ = 1.5 Å.[36] An Acquity UPLC H-Class system (Waters) was operated with Superdex 75 Increase 10/300 GL (GE Healthcare) to isolate oligomers and dimers of S100A3. The solution used for metal-free S100A3 (WT and R51Q) was 50 mM Tris–HCl (pH 7.6), 150 mM NaCl, and 1 mM DTT. In the case of the R51Q mutant, an equivalent of Zn2+ and 2.5 mM Ca2+ (final concentration) were added to the metal-free solution to obtain a more stable oligomeric state.[36] Before each SEC-SAXS measurement, the column was equilibrated with the appropriate buffer for each condition. The initial concentrations of both the metal-free and Ca2+/Zn2+-complex were adjusted to 5.8 mg/mL (0.5 mM), and 170 μL of the sample was injected onto the column. For the SAXS measurement, the sample eluted from the column flowed at 0.05 mL/min through a stainless steel cell with an H1.5 × W3.0 × T1.0 mm aperture covered by a 0.02 mm-thick quartz glass window. At the same time as the SAXS measurement, UV–visible light absorption was monitored to evaluate the sample concentration at the X-ray-exposed position.[37,38]

The X-ray exposure times and number of images for each frame were set to 20 s and 429 images, respectively. Scattering intensities [I(Q)] were measured in the region of 0.0121 Å–1 < Q < 0.6088 Å–1, where Q = 4π sin θ/λ, with a distance of 1.01 m between the sample and the detector. Images were collected on a PILATUS3 2M detector (Dectris). Hundreds of two-dimensional images were azimuthally averaged and converted to one-dimensional profiles by using SAngler.[39] All scattering profiles divided by transmission were subtracted from the background intensities collected in 15 images before sample injection. Intensities were converted to an absolute scale by using water scattering as a standard. Hundreds of subtracted scattering profiles were processed by Serial Analyzer software,[40] which corrects the baseline of the chromatogram on each Q point and ultimately calculates the extrapolated scattering profile at infinite dilution. SAXS data figures were drawn by using Igor Pro ver. 6.34 (Wave Metrics Inc.). The values of the scattering intensity at zero angle [I(Q)] and the radius of gyration (Rg) from Guinier analysis (Q × Rg < 1.3) were estimated by using AUTORG.[41] To compare the solution structure with the dimer in the crystal structure, theoretical scattering profiles of X-ray crystal structures (PDB ID: 3NSI) were calculated by using CRYSOL.[28,42]

Conclusions

In this work, we produced recombinant S100A3 in the E. coli SHuffle T7 strain and obtained S100A3 that was post-translationally citrullinated by PAD3. In addition, mutants in which R51 was replaced with various amino acids were obtained by the same procedure used for WT. Among the various mutants, R51Q was obtained in the highest yield. In addition, a comparison of biophysical and biochemical properties indicated that the R51Q mutant showed the most similarity to R51Z. From these results, we conclude that R51Q is the optimal mutant model of S100A3 that has been post-translationally modified by PAD3. This is consistent with a previous study in which arginine residues were converted to glutamine as a model of MBP citrullinated by PAD. Furthermore, citrullination of S100A3 may be essential for stabilization of the Ca2+-/Zn2+-binding state. The results of this study will lead to the further development of structural analyses for the Ca2+-/Zn2+-bound S100A3.