Human S100A3 Tetramerization Propagates Ca2+/Zn2+ Binding States☆

Biochimica et Biophysica Acta 1833 (2013) 1712–1719

Contents lists available at SciVerse ScienceDirect

Biochimica et Biophysica Acta

journal homepage: www.elsevier.com/locate/bbamcr

Human S100A3 tetramerization propagates Ca2+/Zn2+ binding states☆

Kenji Kizawa a,⁎, Yuji Jinbo b, Takafumi Inoue a, Hidenari Takahara c,d, Masaki Unno d, Claus W. Heizmann e, Yoshinobu Izumi b a Innovative Beauty Science Laboratory, Kanebo Cosmetics Inc., 5-3-28 Kotobuki-cho, Odawara 250-0002, Japan b Graduate Program of Human Sensing and Functional Sensor Engineering, Yamagata University, 4-3-16 Jo-nan, Yonezawa 992-8510, Japan c Department of Applied Biological Resource Sciences, School of Agriculture, Ibaraki University, 3-21-1 Chuou, Ami 300-0393, Japan d Frontier Research Center for Applied Atomic Sciences, Ibaraki University, 162–1 Shirakata, Tokai, Naka 319-1106, Japan e Department of Pediatrics, Division of Clinical Chemistry and Biochemistry, University of Zürich, Steinwiesstrasse 75, CH-8032 Zürich, Switzerland article info abstract

Article history: The S100A3 homotetramer assembles upon citrullination of a specific symmetric Arg51 pair on its homodimer Received 23 May 2012 interface in human hair cuticular cells. Each S100A3 subunit contains two EF-hand-type Ca2+-binding motifs Received in revised form 16 July 2012 2+ and one (Cys)3His-type Zn -binding site in the C-terminus. The C-terminal coiled domain is cross-linked to Accepted 19 July 2012 the presumed docking surface of the dimeric S100A3 via adisulfide bridge. The aim of this study was to deter- Available online 28 July 2012 mine the structural and functional role of the C-terminal Zn2+-binding domain, which is unique to S100A3, in homotetramer assembly. The binding of either Ca2+ or Zn2+ reduced the α-helix content of S100A3 and mod- Keywords: fi 2+ 2+ Calcium ulated its af nity for the other cation. The binding of a single Zn accelerated the Ca -dependent 2+ 2+ EF-hand tetramerization of S100A3 while inducing an extensive unfolding of helix IV. The Ca and Zn binding affin- Citrullination ities of S100A3 were enhanced when the other cation bound in concert with the tetramerization of S100A3. S100 protein Small angle scattering analyses revealed that the overall structure of the S100A3 tetramer bound both Ca2+ Small angle scattering and Zn2+ had a similar molecular shape to the Ca2+-bound form in solution. The binding states of the Ca2+ or Zinc Zn2+ to each S100A3 subunit within a homotetramer appear to be propagated by sensing the repositioning of helix III and the rearrangement of the C-terminal tail domain. This article is part of a Special Issue entitled: 12th European Symposium on Calcium. © 2012 Elsevier B.V. All rights reserved.

1. Introduction The binding of Ca2+ to the S100A12 dimer and the S100B tetramer in- terfaces, in addition to the EF-hand motifs of each subunit, facilitates the The S100 protein family is the largest sub-group of EF-hand-type assembly of the hexamer [9] and octamer [5], respectively. Ca2+-binding proteins [1]. This family comprises more than 20 In addition to Ca2+, other metal ions, such as Zn2+ and Cu2+,bind multigenic members, which are characterized by two highly conserved to several S100 proteins with high affinities. Although characterization Ca2+-binding domains: a classical C-terminal EF-hand motif with a ca- of the Zn2+-andCu2+-binding sites has been hampered due to the ab- nonical Ca2+-binding loop and an S100-specific N-terminal EF-hand sence of a conserved motif, recent analyses of the crystal and solution motif. Most members of the S100 protein family, except for the mono- structures have identified their locations. While His and Asp/Glu resi- meric S100G, form a non-covalently associated anti-parallel dimer or- dues coordinate a common Zn2+-binding site at the homodimer inter- ganized into an eight helix bundle, even in the absence of Ca2+. faces of S100A7 [10],S100A12[11,12] and S100B [13] and most likely at However, several S100 proteins have been reported to assemble into a a similar site in the S100A8/S100A9 heterodimer [14], Cys/His residues variety of biologically active multimeric forms larger than a dimer coordinate the extremely high affinity Zn2+-binding sites within each [2–5]. The assembly of the S100A3 tetramer [6], S100A8/S100A9 subunit of S100A2 [15] and S100A3 [16–18].TheZn2+-binding sites heterotetramers [4,7,8] and S100A12 hexamer [9] strictly depends on are also coordinated between the dimer components of the S100A2 tet- Ca2+, while the S100B tetramer is stable in the absence of Ca2+ [5]. ramer [15] and S100A12 hexamer [12]. Although binding of Zn2+ to some S100 proteins has been reported to influence their Ca2+-affinities [12,15,19,20], neither the molecular mechanism of how the binding of onecationmodulatestheaffinity for another, nor the functional role Abbreviations: CD, Circular dichroism; PAD3, type III isozyme of peptidylarginine of Zn2+ in assembling S100 multimers, is understood. deiminase; SAXS, small angle X-ray scattering; WT, wild type S100A3 is highly expressed in the cuticular cells of hair follicles and ☆ This article is part of a Special Issue entitled: 12th European Symposium on organizes into the mature hair cuticles that function as a physical and Calcium. ⁎ Corresponding author. Tel.: +81 465 34 6116; fax: +81 465 34 3037. chemical barrier [21,22]. S100A3 is unique among S100 proteins in E-mail address: [email protected] (K. Kizawa). that it has the highest cysteine content (10%) [23].AsshowninFig. 1,

0167-4889/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://doi.http://dx.doi.org/10.1016/j.bbamcr.2012.07.009 K. Kizawa et al. / Biochimica et Biophysica Acta 1833 (2013) 1712–1719 1713 four out of the 10 cysteines in S100A3 are capable of forming two disul- 2.2. Fluorescence measurement fide bridges [18]. Cys30 and Cys68 connect two Ca2+-binding loops (SS1), and Cys99 in the C-terminal coil interacts with Cys81 in the The Ca2+-induced reorientation of helix III in S100A3 was moni- helix IV (SS2). Furthermore, affinity for Zn2+ of S100A3 has been tored by measuring single Trp45 fluorescence with Zn2+.Therecombi- reported to be variable, depends on the adopted purification and bind- nant WT-S100A3 and its mutated proteins (2 μM each) were premixed ing assay conditions. Initially reported Kd of the aerobically purified re- with up to 5 molar equivalents (eq.) of ZnSO4 per monomer in 50 mM combinant protein was ~10 μM [24]. Adoption of anaerobic condition Tris–HCl buffer containing 150 mM KCl and 1 mM DTT (pH 7.5). The 2+ 2+ for S100A3 purification enhanced the Zn -affinity (Kd =1.5 μM) emission fluorescence intensities at 340 nm of Zn -loaded S100A3 ex- [16].TheKd determined by dialysis experiment in the presence of gluta- cited at 295 nm were recorded after the addition of serial concentra- 2+ thione was 4 nM [17]. The tetrahedral binding site for a single Zn was tions of CaCl2 as has been described previously [6,16,24]. Alternately, predicted to be coordinated with minor movements of Cys83, Cys86 the fluorescence intensities of S100A3 preloaded with Ca2+ were and His87; and recruitment of Cys93 within a subunit chain [17,18]. recorded after the cumulative addition of ZnSO4. When only partial re- We previously reported that the citrullination of Arg-51 in S100A3 by duction was achieved with Zn2+ in the Ca2+-free form, the full reduc- the type III isozyme of peptidylarginine deiminases (PAD3) remarkably tionwasaccomplishedbyaddingCaCl2 at concentrations of up to 2+ increased its very low Ca affinity (Kd =13 to 2.4 mM) concomitant 10 mM. with the assembly of the homotetramer [6]. Although the crystal structure of the properly folded S100A3 revealed that the C-terminal 2.3. CD spectrometry Zn2+-binding domain is exposed on the presumed docking surface [18], its structural and functional roles in the assembly of a The secondary structural change of S100A3, which has been previ- homotetramer are still unknown. In this study, we analyzed S100A3 ously reported for helix IV to occur in a Zn2+-dependent manner [17], structures in solution by fluorescence and circular dichroism spectrom- was monitored by far-UV CD spectrometry with the addition of either etry and small angle X-ray scattering (SAXS) in the presence of either Ca2+ alone or both Ca2+ and Zn2+. The CD spectra of recombinant Ca2+ alone or both Ca2+ and Zn2+ to elucidate their corporative roles WT-S100A3 and R51A (15 μM of each) were recorded in 20 mM Tris– in the assembly of the homotetramer. HCl buffer containing 150 mM KCl and 0.2 mM DTT (pH 7.5) using a J-720 spectropolarimeter (Jasco, Japan) equipped with a rectangular quartz cell (1-mm path length). The spectrum data were collected four 2. Materials and methods times at a bandwidth of 2 nm, a scan speed of 100 nm/min, and a re- sponse time of 4 s and combined. Considering that the crystal structures 2.1. Human recombinant S100A3 proteins of apo-dimers of WT-S100A3 (PDB code: 3NSI) and R51A (3NSK) [18] comprise four helices per monomer (13.8 residues per helix) but not a The wild type (WT)-S100A3, Arg51 substituted mutant (R51A) and β-sheet, the overall α-helix content was calculated with the following

SS2-deficient mutant (C81A+C99A) used in this study were produced equation: α-helix content (%)=[θ]222/−32100×100 [26]. in Sf21 cells by inoculating the cells with recombinant baculoviruses followed by protein purification, as described previously [18,25].The 2.4. SAXS measurement and data analyses protein concentrations were determined using a molar extinction coefficient of 14,500 M−1 cm−1 at 280 nm. The synchrotron radiation X-ray scattering data were collected using an instrument for SAXS measurement installed at BL-10C of the 2.5 GeV Photon Factory storage ring in KEK (Tsukuba, Japan) [27,28]. The purified recombinant S100A3 and R51A, which were serially diluted to 100, 200, 300 and 400 μMin50mMTris–HCl buffer containing N N 150 mM KCl and 10 mM DTT (pH 7.5), were combined with various

concentrations of CaCl2 and/or ZnSO4 solution (9:1). The protein samples or the supernatants collected after centrifugation (in the case of Zn2+-supplemented samples) were transferred to a mica cell with a vol- − I ume of 40 μL. A range of momentum transfer equal to 0.02bs/Å 1 b0.35 I (s=4πsinθ/λ, where 2θ is the scattering angle and λ=1.488 Å is a se- 87 lected X-ray wavelength) was covered. The calibration of the equipment 86 II was determined by observing peaks from dried chicken collagen. The II 83 forward scattering data were collected for 10 min at 25±0.1 °C; X-ray IV fi EF1 irradiation did not affect the scattering pro les for up to 30 min. The IV four data sets collected for the lowest concentration (i.e., ~100 μM) of SS1 S100A3/R51A and the two data sets collected for the higher concentra- III 93 III tions were combined, and the solvent scattering values were subtracted. SS2 The scattering data were extrapolated to infinite dilution. The values EF2 of I(0)/K,whereI(0) is the forward scattering intensity at an angle of C C zero and K is a constant determined using calmodulin as the standard protein, were estimated by the Guinier approximation [29].Theexper- Fig. 1. Representative structure of apo-S100A3 dimer. Two subunits (in green and imentally determined I(0)/K for S100A3/R51A under metal-free condi- – brown) comprise four helices each (I IV), based on a crystal structure of properly tions and in the presence of either Ca2+ alone or both Ca2+ and Zn2+ folded S100A3 (PDB code, 3NSI). SS1 connects Cys30 in the N-terminal S100 specific EF-hand type Ca2+ binding loop (EF-1) and Cys68 in the C-terminal classical EF-hand indicated the dimer weight (MD) and the average weight of a dimer/ (EF-2) is labeled in the green subunit. SS2 attaches Cys99 in the C-terminal coil struc- tetramer mixture (Mw), respectively. The weight fraction of the tetra- ture to Cys81 in the helix IV is labeled in the brown subunit. SS1 and SS2 in each other mer is given by: subunit locate behind the translucent helices. An indole group of Trp45 (an intrinsic fluorescent probe) and a guanidinium group of Arg51 (a target residue of PAD3 enzyme) are illustrated in the brown subunit. The residue numbers of Cys83, Cys86, 2+ M −M Cys93 and His87 that coordinate a single Zn binding site of the brown subunit are ¼ w D : ð Þ wT − 1 indicated in yellow and blue circles. MT MD 1714 K. Kizawa et al. / Biochimica et Biophysica Acta 1833 (2013) 1712–1719

2+ In Eq. (1), MT =2MD +8MCa for the Ca -bound form and MT = A 2+ 2+ 100 2MD +8MCa+4MZn for the Ca /Zn -bound form. The mole fraction 2+ of the tetramer is given by: [Zn ]/[WT] 0 80 ζ ¼ wT MT ¼ wTMT : ð Þ 0.25 T ðÞ− þ 2 1 wT MD wTMT MW 0.5 60 1 The particle forward scattering function and the radius of gyration 2 for the tetramer (I(s)T, Rg,T) were derived subtracting the values de- 40 5 termined experimentally for the dimer (I(s)D, Rg,D) from those of the mixture (I(s), Rg) as follows: 20

IsðÞ−ðÞ1−ζ IsðÞ Trp Fluorescence Reduction (%) IsðÞ ¼ T D ð3Þ T ζ 0 T 0.1 1 10 100 Ca2+ (mM) R2−ðÞ1−ζ R2; 2; ¼ g T g D : ð Þ B Rg T ζ 4 T 100 [Zn2+]/[R51A] To elucidate the overall molecular shape of the proteins, the pair 80 0 distribution function P(r) was calculated by the indirect Fourier 0.1 transformation package GNOM (freely available at http://www. 0.25 embl-hamburg.de/ExternalInfo/Research/Sax/gnom.html) [30].The 60 0.5 maximum dimension (Dmax) was estimated from the P(r) curve, 1 which becomes zero at values of r equal to or greater than the 40 2 Dmax of the particle. Dummy atom models for the WT-S100A3 and R51A dimers were generated using the program DAMMIN (freely 20 available at http://www.embl-hamburg.de/ExternalInfo/Research/

Sax/dammin.html) with a molecular point symmetry constraint to Trp Fluorescence Reduction (%) P2 and oblate shape constraint and for both tetramers with a molec- 0 0.1 1 10 ular point symmetry constraint to P22 but no shape constraint [31]. Ca2+ (mM) Five ab initio models were averaged, and a low resolution structure was computed. C [Zn2+]/[S100A3] 0.1 1 10 100 3. Results Ca2+ WT (mM) 0 3.1. Allosteric Ca2+/Zn2+-binding modulation of S100A3 80 1 5 Trp45 fluorescence has been utilized for evaluation of both Ca2+ and 60 R51A 2+ Zn -induced conformational changes of S100A3 [6,16,18,24].The 0 fluorescence intensity at concentrations below 10 μMZn2+ was only 40 0.2 ~5% of wild type and ~25% of R51A compared to the reduction induced 1 by Ca2+ (not shown). This finding indicates that the binding of Zn2+ to 20 its specific binding site, which is preformed in the C-terminus [17],rare- 2+

ly induces helix III reorientation. The relatively small reduction by Zn Trp Fluorescence Reduction (%) allowed us to estimate the Ca2+ affinity of S100A3 from the fluorescent 0 0.1 1 10 100 titration curve created in the presence of Zn2+. Concurrent binding of Zn2+ (μM) Zn2+ shifted both sigmoid curves of the Ca2+-induced conformational change in WT-S100A3 and R51A to similar extents, but compared to Fig. 2. Cooperative Ca2+- and Zn2+-binding modulation of S100A3. The effects of Zn2+ 2+ WT-S100A3, the affinity of R51A for Ca apparently increased at a on Ca2+-dependent helix III reorientation in S100A3 were monitored using single tryp- lower Zn2+ concentration (Fig. 2A and B). tophan fluorescence. The fluorescence intensities of WT-S100A3 (A) and R51A μ 2+ The fluorescent titration curves for Ca2+-free WT-S100A3 and R51A (B) (2 M each) at preloaded serial molar ratios of Zn were recorded after the cumu- 2+ fl 2+ 2+ lative addition of Ca . The uorescence intensities of Ca -loaded WT-S100A3 and measured at various concentrations of Zn (Fig. 2C) are both bi-phasic, R51A were measured after the cumulative addition of Zn2+ (C). Overall, the S100A3 as has been previously reported [24]. Previously reported spectroscopic Zn2+-affinities were indirectly estimated from these titration curves. The relative investigations strongly suggest that there are two binding sites for Zn2+ rates of each measured point versus the maximum reduction are plotted. with different affinities [16]. Although the high affinity Zn2+ binding site is presumed to comprise three residues (Cys83, Cys86 and His87) preformed on helix IV and one cysteine (Cys93) recruited within a 2+ 2+ 2+ subunit chain [18], the location of the low affinity Zn binding site in presence of 1 mM Ca ,theZn affinity of R51A (Kd=1.5 μM) was the properly folded S100A3 is still vague. The first gentle slope likely cor- apparently higher than that of WT-S100A3 (Kd =7.4 μM). We previous- 2+ responds to the Zn binding to the C-terminal (Cys)3His-type specific ly reported that disruption of the disulfide bridge between Cys81 and 2+ site [18]. Although the first slope of R51A was apparently larger than Cys99 also increased the Ca affinity (Kd=3.5 mM for C81A+C99A that of WT-S100A3, the Zn2+ affinities estimated from the half maxi- and 0.85 mM for R51A+C81A+C99A), which was most likely because mum points were almost comparable to those previously reported the relocation of the C-terminal coil structure exposed on the docking 2+ (Kd=~10 μM) [24].Thefluorescence intensities of S100A3 were fully surface made another dimer accessible [18].TheCa affinities reduced by Zn2+ in conjunction with sub- to milli-molar concentrations of these mutants are affected by an equivalent molar of Zn2+ to a 2+ of Ca , and the titration curves became sigmoid (Fig. 2C). In the lesser extent (Kd =2.1 mM for C81A+C99A and 0.48 mM for K. Kizawa et al. / Biochimica et Biophysica Acta 1833 (2013) 1712–1719 1715

Wavelength (nm) Ca2+ (mM) 200 210 220 230 240 210 220 230 240 210 220 230 240 0102030 0 60

) D -1 -5000 50 1 mM Ca2+ + [Zn2+]/[R51A] 4 [Zn2+]/[R51A]

decimol 2+ 2 Ca -10000 4 40 1 mM Ca2++Zn2+ (mM) 2 2 20/30 1

1 -Helix Content (%)

-15000 10 0.5 α 30 ] (deg cm 5

θ 0.5 0 [ 0 AC0 B 01234 -20000 , [Zn2+]/[R51A]

Fig. 3. Reduction of the α-helix content in S100A3. Shown are the far-UV spectra of the R51A mutant (15 μM) recorded after cumulative addition of Ca2+ (A) or Zn2+ (B) and cu- 2+ 2+ 2+ 2+ mulative addition of Zn following 1 mM Ca (C). (D) Effect of Ca and Zn on the secondary structure. The α-helical content in R51A calculated from [θ]222 (see Materials and methods) is plotted against either Ca2+ concentration (open circle) or the Zn2+-molar ratio (closed circle and triangle). The inﬂection point of the biphasic curve obtained with Zn2+ combined with 1 mM Ca2+ is indicated with an arrow.

R51A+C81A+C99A) than the properly folded S100A3 and R51A pro- experimentally determined Mw and Rg of the WT-S100A3 and R51A teins (Fig. S1). homodimers at an inﬁnite dilution on Zimm plots were consistent

with the theoretical Mw (23,250 and 23,080; Fig. 4B) and previously 3.2. α-Helical rearrangement within the S100A3 homotetramer reported Rg values for S100A1 and S100B homo- and heterodimer mix- tures derived from bovine brain (19.5–20.2 Å; Fig. 4C), respectively To evaluate changes in the secondary structure, we utilized far-UV [33].AsshowninFig. 5, the weight and mole fractions of the R51A tet- CD spectrometry to analyze WT-S100A3 (not shown) and the R51A mu- ramer, which were calculated by Eqs. (1) and (2) (in section 2.4.), tant loaded with either Ca2+ alone or both Ca2+ and Zn2+ (Fig. 3). The increased with lower Ca2+ concentrations compared to that of WT. calculated α-helix contents of the apo-forms of WT-S100A3 (56%) and This ﬁnding suggests that both conformational change in helix III R51A (54%) were consistent with those of their crystal structures (Fig. 2A and B) and reduction of helical structure (Fig. 3A) were induced 2+ (55%; PDB codes, 3NSI for WT and 3NSK for R51A), assuming that the during Ca -dependent S100A3 tetramerization. The Rg value and Mw seven disordered residues all adopt into a random coil conformation. of WT-S100A3 increased to 25.8 Å and 1.9-fold at 100 mM CaC12,and 2+ Ca -bound S100A6 has slightly higher α-helix content compared to those of R51A increased to 24.7 Å and 1.6-fold at 30 mM CaCl2,respec- its apo-form, mainly due to the elongation of helix IV (60% in tively. The overall structural parameters computed from the SAXS data 2+ apo-S100A6:1K9P vs. 64% in the Ca -bound form:1K9K). In contrast, are summarized in Table 1.BoththeMw and Rg of WT-S100A3/R51A WT-S100A3 (not shown) and R51A (Fig. 3A) apparently underwent a at enriched Ca2+ levels increased depending on the ﬁnite protein con- small reduction in their α-helix content when loading Ca2+ alone. centrations (Fig. 4B and C), which were determined from the second

The higher extent of the helix reduction in R51A compared to that in virial coefficient (A2) and interparticle interference (Bif) values. S100A1 previously reported [32] plateaued (i.e., ~9 residues) at concen- These results suggest that Ca2+ loading alone cannot complete the con- trations higher than 10 mM of CaCl2 (Fig. 3D). version from dimer to tetramer, and some populations may undergo The binding of Zn2+ to a specific site, which is preformed in the Ca2+-dependent conformational changes, remaining in the dimeric C-terminal domain of S100A3, was previously reported to result in the form at dilute protein concentrations while supra-physiological Ca2+ partial unfolding of helix IV [17]. The addition of excessive Zn2+ (an concentrations (>10 mM) tend to aggregate S100A3 into oligomers equivalent molar ratio of 4) reduced the α-helix contents of R51A to larger than the tetramer. ~28% (i.e., 26 total helical residues not restricted to the helix IV) The remarkable heterotrophic allosteric Ca2+-binding modulation (Fig. 3B and D). In the presence of 1 mM Ca2+, the extent of of S100A3 by Zn2+ (Fig. 2) led us to examine the effect of Zn2+ on Zn2+-dependent reduction in α-helix content in R51A was less than tetramerization. Because excessive Zn2+ precipitated S100A3 in the 2+ 2+ that of Zn alone (34% with an equivalent molar ratio of 4 Zn ) presence of greater than millimolar levels of CaCl2, SAXS data could be (Fig. 3C and D). The bi-phasic curve of α-helix contents in R51A collected only for R51A. Both Mw and Rg values for R51A were increased inflected at an equivalent molar ratio of Zn2+ as the fluorescent titra- at higher molar ratios of Zn2+ in combination with 1 mM Ca2+, tion curve shift of R51A ceased (Fig. 2B), suggesting that the binding af- although Zn2+ supplementation caused large deviations in the SAXS finity of the single Zn2+ is enhanced by 1 mM Ca2+. Considering that data, as shown in the Zimm plots (Fig. 4B and C). These data fluctuations the plateaus of CD spectra appear at 1 molar eq. of Zn2+ combined might result from impaired loading of Zn2+ into sites other than the with 1 mM Ca2+, the binding of Ca2+ induced a helix IV partial specific binding sites of R51A. The calculated molar fraction of unfolding similar to that induced by the binding of Zn2+ to the high af- the R51A tetramer reached 100% in the presence of one molar equiva- finity site. The helical region that unfolded with excessive Ca2+ alone lent of Zn2+ and 1 mM Ca2+ (Fig. 5). This result suggests that the (~9 residues) appeared to overlap with that induced at 1 eq. of Zn2+ incorporation of a single Zn2+, which most likely occurs at the 2+ combined with 1 mM Ca (~12 residues). Likewise, the α-helix con- tetrahedral (Cys)3His site in each S100A3 subunit [18],isessential tent of WT-S100A3 was decreased to a similar extent by the addition for tetramerization. The Rg values of the extracted tetramer compo- of excessive Zn2+ in combination with Ca2+ (data not shown). nents bound to either Ca2+ alone or both Ca2+/Zn2+, which were calculated by Eq. (4) (in section 2.4.), are similar (~25 Å; Table 1). 3.3. Zn2+ is required for efficient S100A3 tetramerization Under the Ca2+/Zn2+ combined conditions, S100A3 efficiently assembled a homotetramer regardless of the protein concentrations. These re- To directly monitor changes in the molecular sizes of WT-S100A3 sults suggest that both Ca2+ and Zn2+ cooperatively regulate S100A3 and R51A (i.e., the dimer to tetramer ratio), SAXS measurements were tetramerization. Although S100A3 assembles as a homotetramer in a performed using synchrotron radiation. The SAXS data, which were col- Ca2+-dependent manner similar to the S100A8/S100A9 heterotetramer lected at four protein concentrations and extrapolated to an infinite di- [7,8] and S100A12 homohexamer [9] in vitro,Zn2+ appears to be essen- lution, were evaluated by the Guinier approximation (Fig. 4A). The tial for S100A3 tetramerization in vivo. 1716 K. Kizawa et al. / Biochimica et Biophysica Acta 1833 (2013) 1712–1719 A B C 14 6 12 WT WT Ca2+ 2+ 2+ Ca +Zn R51A R51A (mM) 13 5 100 Ca2+ 10 (mM) A = ) − 1 0 2.5 4

) + A 8 12 30 − 1 ) 2 30 mol g Å − 5 2 3 6 Ca2++Zn2+ 11 1.5 (10 0 R51A 2 g

1.2 R 100 30 2 4 Ca2++Zn2+ 0 ln[ I ( s )/ K ] (g mol 10 0.3 Kc / I (0, c ) (10 0 100 Ca2+ 0 1 2

(mM) WT 9

0 0 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 0 1 2 3 4 5 6 s2 (10−3 Å−2) S100A3 (mg/ml) S100A3 (mg/ml)

Fig. 4. SAXS analyses of S100A3. Forward scattering data from WT and R51A were collected in the absence or presence of Ca2+ and R51A with 1 mM Ca2+ plus 1 molar eq. of Zn2+. (A) Guinier plots of the extrapolated data. Each plot was shifted for clarity with the indicated A values. The straight lines were delineated by the least-squares method. The ﬁlled symbols indicate the Guinier region. The data at ﬁnite protein concentrations are represented by Zimm plots. The concentration-dependence of KC/I (0, c) (B) and Rg (C) is shown. Only the experimental error bars that were larger than the symbols were designated.

3.4. Molecular shapes of the S100A3 dimer and tetramer in solution between the average molecular conformations of the WT and R51A dimers. A similar particle scattering data representation as that described The subtracted particle scattering function of components of the by Kratky [34] (Fig. 6A) and P(r) curves (Fig. 6B) for the apo-dimers R51A tetramer bound to Ca2+ alone (Fig. 6A) or both Ca2+ and of WT-S100A3 and R51A were obtained by indirect Fourier transfor- Zn2+ (not shown), which was calculated using Eq. (3) (in section mation of the SAXS data using the GNOM program. The DAMMIN pro- 2.4.), produced similar P(r) curves (Fig. 6B). The DAMMIN program gram generated similar overall shapes (Dmax =52 Å) for the apo-WT generated an overall rod-like elongated shape with two concave seg- and R51A dimers, consistent with their substantially identical crystal ments for the R51A tetramers bound to Ca2+ alone and Ca2+/Zn2+ backbone structures (PDB codes, 3NSI and 3NSK) [18]. The dummy (Figs. S2B, S2C and 7A). The signature change in the P(r) curve corre- atom model showed that the C-terminal coil structure exposed from lated with a shorter dimension of the front section of the Ca2+/ the presumed tetramerization surface consists of helices III and IV in Zn2+-bound form (~51 Å) compared to the Ca2+-bound form solution (Figs. S2A and 7A). Because the S100A3 dimer most likely (64 Å). The maximum dimension of the tetramers (Dmax =~75 Å) collides with other dimers in solution, the electric repulsion between was signiﬁcantly larger than the usual dimeric organization of S100 two WT dimers and the high occupancy of two intramolecular disul- proteins, suggesting that the original subunits were organized in a ﬁde bridges might explain the slightly smaller Rg (19.2 Å) of S100A3 stretched conformation. compared to R51A (19.7 Å). There might be a slight difference

4. Discussion [Zn2+]/[R51A] The multimeric forms of several S100 proteins have been previ- 0 0.5 1 1.5 2 60 ously demonstrated. In most cases, their Ca2+-dependent dimeric structures are conserved across different arrays of the crystal units R51A [5,8,9,35]. While it is necessary to conﬁrm the structures of S100 olig- (1 mM Ca2++Zn2+) 50 omers in a non-crystal state, little information is available regarding

) their structures in solution. We previously reported that S100A3 -1 WT MT spontaneously assembles as a homotetramer in hair cuticular cells due to the regulation of post-translational citrullination by PAD3 [6]. g mol

3 40 R51A In the current study, we obtained a low resolution structure of S100A3 tetramers bound to Ca2+ alone and both Ca2+ and Zn2+ in

M w (10 solution by SAXS analyses. The overall assembled shape of both 30 S100A3 homotetramers presented here is different from the previously reported crystal structures of Zn2+-bound homotetramers of S100A2 [15] and S100A12 [11] and Ca2+-bound S100A8/S100A9 M D heterotetramers [8]. The large conformational changes, both in helix 20 0 20 40 60 80 100 III and the C-terminal Zn2+-binding domain, which is unique to , Ca2+ (mM) S100A3, revealed by the fluorescent titration and CD spectra (Figs. 2 and 3) prevented us from manipulating any of the known structures 2+ 2+ Fig. 5. Ca and Zn -dependent increase in the size of S100A3. The Mw of R51A in- of the apo-, Ca2+- and/or Zn2+-bound S100 proteins that fit into creased at a lower Ca2+ concentration compared to that of WT. The M did not reach w the SAXS envelope. the M at Ca2+ concentrations up to 100 mM but exceeded the M when a molar equiv- T T 2+ 2+ 2+ 2+ Combined Zn have been reported to increase the Ca -affinities of alent of Zn was added in combination with 1 mM Ca . MD and MT indicate the dimer and tetramer masses, respectively. S100B and S100A12 [19,20] while decreasing that of S100A2 [15]. K. Kizawa et al. / Biochimica et Biophysica Acta 1833 (2013) 1712–1719 1717

Table 1 SAXS parameters of S100A3 collected under various Ca2+/Zn2+ concentrations.

2+ 2+ −3 a b 5 c 13 d Ca Zn 10 Mw Rg Rg, T Dmax 10 A2 10 Bif (mM) (mol eq.) (g/mol) (Å) (Å) (Å) (mol cm3/g2) (cm5/g)

WT 0 0 23.2±0.2 19.2±0.2e – 52 −17 3 WT 100 0 42.8±0.4 25.8±0.3 26.2±0.3 75 −77 −67 R51A 0 0 23.1±0.2 19.7±0.2e – 52 −81 R51A 30 0 37.7±0.4 24.7±0.2 25.8±0.2 75 −57 −32 R51A 1 1 45.2±2.2 25.0±1.0 25.1±1.0 73 −13 −9

a Experimentally determined Rg at inﬁnite dilution. b Calculated Rg for tetramer component. c Second virial coefﬁcient. d Parameter of interparticle interference. e Equal to Rg,D.

Neither modulation mechanisms that propagate the Ca2+/Zn2+-binding among reduced glutathione and high sulfur keratin associated pro- states nor their association with oligomerization have been reported for teins and to some degree binds to S100A3. The citrullinated S100A3 any S100 proteins. In this study, we showed that the Ca2+-affinity of the complexes with Zn2+ at sub-μM levels in the presence of Ca2+ at ap- EF-hand-type motif was enhanced in the presence of an equivalent proximately the mM level (Fig. 2C), which fully activates the respon- molar ratio of Zn2+.Inturn,theZn2+-binding affinity for the tetrahedral sible PAD enzymes. It has also been reported that the Zn2+-affinity of 2+ (Cys)3His site increased upon binding of Ca . This heterotrophic allo- S100A3 remarkably increases (Kd =4 nM) in the presence of gluta- steric effect coincided with the reorientation of helix III, estimated from thione [17]. Thus, the free Zn2+ concentration may reach these levels the Trp45 fluorescence reduction (Fig. 2), and the collapse of ~3 turns in helix IV, deduced from the far UV-CD spectra (Fig 3), within the homotetramer. Fig. 7B shows the postulated conformations of the A S100A3 subunit in dimers and tetramers assembled in the presence of ei- 2+ Ca (mM) R51A ther Ca2+ or Zn2+. In the docking center of the apo-dimer, Cys99 revers- 3 0 ibly cross-links with Cys81 on the helix IV via adisulfide bridge (SS2). The rigid C-terminal proline-rich domain exposed on the presumed docking surface hinders self-association of S100A3 dimers. Compared )

to the WT dimer, the R51A dimer (i.e., citrullinated S100A3) would col- -2

Å 2 30 lide more frequently with other dimers as a result of the loss of electric -3 repulsion. In addition, the formation of SS2 might be required to main- tain the homotetramer assembly because the low occupancy of SS2 in I ( s ) (10 2

R51A resulted in slightly lower frequency of transformation to a tetramer s 2+ compared to the WT in the presence of excessive Ca alone (Fig. 5). 1 Therefore, we suggest that the middle part of helix IV neighboring Cys81 is collapsed in the Ca2+-bound S100A3 homotetramer. Because the following Cys83 coordinates a specificZn2+-binding site [17],the 2+ Ca -dependent helical rearrangement is likely extended to the tail 0 parts of helix IV in the Ca2+/Zn2+-bound R51A tetramer but not to 0 0.1 0.2 other helices. This larger rearrangement might result in a tighter associ- s (Å-1) ation compared to the Ca2+-bound form. The binding of a single Zn2+ to fi fi B the speci c(Cys)3His site can ef ciently reorient SS2 [18],andthusthe 120 C-terminal domain would be excluded from the docking face. The helix IV is reported to be essential for maintaining the dimeric association of R51A tetramer S100 proteins [36]. The C-terminal domain rearrangement may also 100 Ca2+-complex force extension of the original S100A3 dimer component within a Ca2+/Zn2+-complex homotetramer, considering that loading of S100A3 with Zn2+ alone 80 causes it to dissociate into monomers [17] in the decreased ionic strength [37].TheZn2+ seems to be a co-factor essential for S100A3 2+ 60 tetramerization under physiological intracellular Ca levels. The P (r) concave–convex docking surface of the S100A3 dimer is likely transformed by helices III and IV in a Ca2+ and Zn2+-dependent manner. 40 Both S100A3 and PAD3 form homodimers in the absence of metal Apo-dimer 2+ ions [38]. Considering that Zn inhibits PAD3 activity (data not 20 WT shown), the responsible dimeric PAD enzyme is thought to R51A citrullinate target arginine residues in dimeric S100A3 in the presence 2+ 0 of Ca alone. Subsequently, citrullinated S100A3 assembles into a 0 20 40 60 80 homotetramer in the epithelial cellular compartments in the pres- r (Å) ence of both Ca2+ and Zn2+. The free Zn2+ concentration in eukary- otic cells is tightly controlled below the nM level [39], although its Fig. 6. SAXS scattering data profile. (A) Kratky plots of the scattering function of R51A 2+ total level when considering Zn2+ bound to various Zn2+-finger in the absence (open circles) or presence (open triangles) of 30 mM Ca . The particle scattering function of the Ca2+-bound R51A tetramer (closed circles) was calculated and ring-finger proteins and metallothioneins reaches up to a several from the experimental data collected with the apo-dimer and the dimer/tetramer mix- hundred nM to micromolar concentration [40]. In premature cuticu- ture as described in Experimental procedures. (B) P(r) curves for apo-WT and R51A di- 2+ lar cells, the majority of intracellular Zn is most likely labile mers and Ca2+ (and Zn2+)-bound R51A homotetramers. 1718 K. Kizawa et al. / Biochimica et Biophysica Acta 1833 (2013) 1712–1719

A Apo Ca2+-complex Ca2+/Zn2+-complex

Ca2+ Zn2+

Z Z

Z Z Z Z Z Z Z Z

Z Z

B N N N I I II I II C II SS1 C SS2 86 Z Z 87 83 93 83 IV Z III IV IV 86 III 93 83 III 87 93 IV C 87 86

Fig. 7. Cooperative roles of Ca2+ and Zn2+ in assembling an S100A3 homotetramer. (A) The molecular shapes of S100A3 dimer and tetramer Dummy atom models of the apo-R51A dimer and tetramer bound to Ca2+ alone and to both Ca2+ and Zn2+ were computed using the DAMMIN software. An S100A3 subunit within each model is shown in green. The different views refer to supplemental Fig. S2. (B) Predicted S100A3 subunit conformations in a dimer and tetramer. After the symmetric pair of Arg51 in an S100A3 dimer was converted to citrulline (purple circle labeled with Z) by the PAD3 enzyme, the C-terminal domains exposed on the docking surface still prevent its association with another dimer. The Ca2+- and Zn2+-dependent conformational changes of each subunit allow the assembly of a homotetramer. The Ca2+-bound S100A3 subunits take advantage of the conformation of the distorted helix III and partially unfolded helix IV (~9 residues in the middle part) within a homotetramer, where each Z51 is predicted to be in contact with another at the docking interface. The additional binding of the single Zn2+ to the preformed site (Cys83, Cys86 and His87) of each subunit by recruiting Cys93 allows its efﬁcient tetramerization, which is most likely concomitant with the extensive unfolding of helix IV (~12 residues from the middle to tail parts). These residues are indicated with yellow (Cys) and blue (His) circles, respectively. The occupancies of two disulﬁde bonds (SS1 and SS2) are indicated with dotted (low) or solid (high) lines. The brown and gray balls rep- resent Ca2+ and Zn2+. The citrulline residues (Z) and cation balls located on the back side are faintly indicated.

in the zone proximate to the Ca2+-enriched microenvironment enzymes that mature hair cuticles [1,6] but also suggests that S100 pro- where PAD enzymes function. teins are associated with Zn2+ homeostasis in the superficial epithelium. In the absence of Ca2+, S100A3 has been reported to be capable of Zn2+-deficiency has been implicated to cause hair loss, thickening and binding 2.4–3.0 Zn2+ per monomer [17].PreviouslyreportedUV–vis dif- hyperkeratinization of the epidermis [41] and the formation of esophage- ference spectra strongly suggest that S100A3 possesses two Zn2+ binding al squamous cell carcinoma [42]. Because the S100A3 subunit induced sites with tetrahedral geometry [16]. In the previously proposed model, large conformational changes within the homotetramer, it was difficult 2+ in addition to the single Zn -binding site coordinated by (Cys)3His res- to construct atomic coordination models for its tetramers in the present idues [18], Cys81 and Cys99 participate in the binuclear GAL4 study. The transformed S100A3 dimer seems to be capable of interacting 2+ (Zn )2-like cluster. Our results reveal that the properly folded with other dimers via a larger area than the conservative dimer structures S100A3, which carries a disulfide bridge between Cys81 and Cys99 of previously reported crystal forms of tetramers and higher oligomers, (SS2), is also capable of binding more than one molar equivalent of such as the S100A12 hexamer [4] andtheS100Boctamer[5]. This unique Zn2+.TheZn2+ cluster may be coordinated between two subunits or as- tetrameric architecture, which acts as a Ca2+/Zn2+-signal transducer, semblies because higher order oligomers than the tetramer seem to be should be further analyzed at a higher resolution to elucidate the under- formed under such circumstances (Fig. 5). The binding of Zn2+ to low af- lying molecular mechanism that propagates Ca2+/Zn2+ binding states at finity site(s) largely reduced the α-helix content (Fig. 3B), but this was the atomic level. partially abrogated by Ca2+ (Fig. 3C). These results suggest that the bind- Supplementary data to this article can be found online at http:// ing of the second Zn2+ to S100A3 apparently has different effects from dx.doi.org/10.1016/j.bbamcr.2012.07.009. that of the firstion.ItispossiblethattheZn2+ cluster formation releases Ca2+ from S100A3 tetramers, which is recycled for PAD activation. Acknowledgements The heterotrophic allosteric Ca2+/Zn2+-binding modulation of S100A3 in concert with its tetramerization sustains not only its previous- The small-angle X-ray scattering measurements were approved by ly proposed role as a Ca2+-modulator required for the protein-modifying the Photon Factory Advisory Committee, KEK, Tsukuba, Japan (Proposal K. Kizawa et al. / Biochimica et Biophysica Acta 1833 (2013) 1712–1719 1719

No.2007G03). This work was supported in part by Grants-in-Aid for Sci- [19] J. Baudier, N. Glasser, D. Gerard, Ions binding to S100 proteins. I. Calcium- and αα αβ ββ fi zinc-binding properties of bovine brain S100 , S100a ( ), and S100b ( ) enti c Research on Innovative Areas (Structural Cell Biology) from the protein: Zn2+ regulates Ca2+ binding on S100b protein, J. Biol. Chem. 261 Ministry of Education, Culture, Sports, Science, and Technology (1986) 8192–8203. (MEXT), Japan to MU (23121504). We thank Professor Seiu Otomo [20] E.C. Dell'Angelica, C.H. Schleicher, J.A. Santomé, Primary structure and binding properties of calgranulin C, a novel S100-like calcium-binding protein from pig (Ibaraki University, Faculty of Science) for the kind instructions on granulocytes, J. Biol. Chem. 269 (1994) 28929–28936. CD-spectrometry. [21] K. Kizawa, H. Uchiwa, U. Murakami, Highly-expressed S100A3, a calcium-binding protein, in human hair cuticle, Biochim. Biophys. Acta 1312 (1996) 94–98. [22] K. Kizawa, S. Tsuchimoto, K. Hashimoto, H. Uchiwa, Gene expression of mouse References S100A3, a cysteine-rich calcium-binding protein, in developing hair follicle, J. Investig. Dermatol. 111 (1998) 879–886. [1] K. Kizawa, H. Takahara, M. Unno, C.W. Heizmann, S100 and S100 fused-type pro- [23] D. Engelkamp, B.W. Schäfer, M.G. Mattei, P. Erne, C.W. Heizmann, Six S100 genes tein families in epidermal maturation with special focus on S100A3 in mammali- are clustered on human chromosome 1q21: identification of two genes coding for an hair cuticles, Biochimie 93 (2011) 2038–2047. the two previously unreported calcium-binding proteins S100D and S100E, Proc. [2] V. Novitskaya, M. Grigorian, M. Kriajevska, S. Tarabykina, I. Bronstein, V. Berezin, Natl. Acad. Sci. U. S. A. 90 (1993) 6547–6551. E. Bock, E. Lukanidin, Oligomeric forms of the metastasis-related Mts1 (S100A4) [24] U.G. Föhr, C.W. Heizmann, D. Engelkamp, B.W. Schäfer, J.A. Cox, Purification and protein stimulate neuronal differentiation in cultures of rat hippocampal neu- cation binding properties of the recombinant human S100 calcium-binding pro- rons, J. Biol. Chem. 275 (2000) 41278–41286. tein A3, an EF-hand motif protein with high affinity for zinc, J. Biol. Chem. 270 [3] N. Leukert, T. Vogl, K. Strupat, R. Reichelt, C. Sorg, J. Roth, Calcium-dependent tet- (1995) 21056–21061. ramer formation of S100A8 and S100A9 is essential for biological activity, J. Mol. [25] K. Kizawa, M. Unno, H. Takahara, C. W. Heizmann, Purification and characteriza- Biol. 359 (2006) 961–972. tion of the human cysteine-rich S100A3 protein and its pseudo citrullinated [4] J. Xie, D.S. Burz, W. He, I.B. Bronstein, I. Lednev, A. Shekhtman, Hexameric forms expressed in insect cells, in: C. W. Heizmann (Ed.), Calcium-binding Pro- calgranulin C (S100A12) binds to the receptor for advanced glycated end prod- teins and RAGE: from Structural Basics to Clinical Applications, in the series: ucts (RAGE) using symmetric hydrophobic target-binding patches, J. Biol. Chem. Methods Mol. Biol. Springer, Humana Press, in press 282 (2007) 4218–4231. [26] Y.H. Chen, J.T. Yang, K.H. Chau, Determination of the helix and β form of proteins [5] T. Ostendorp, E. Leclerc, A. Galichet, M. Koch, N. Demling, B. Weigle, C.W. in aqueous solution by circular dichroism, Biochemistry 13 (1974) 3350–3359. Heizmann, P.M. Kroneck, G. Fritz, Structural and functional insights into RAGE ac- [27] T. Ueki, Y. Hiragi, M. Kataoka, Y. Inoko, Y. Amemiya, Y. Izumi, H. Tagawa, Y. tivation by multimeric S100B, EMBO J. 26 (2007) 3868–3878. Muroga, Aggregation of bovine serum albumin upon cleavage of its disulfide [6] K. Kizawa, H. Takahara, H. Troxler, P. Kleinert, U. Mochida, C.W. Heizmann, Specific bonds, studied by the time-resolved small-angle X-ray scattering technique citrullination causes assembly of a globular S100A3 homotetramer: a putative with synchrotron radiation, Biophys. Chem. 23 (1985) 115–124. Ca2+ modulator matures human hair cuticle, J. Biol. Chem. 283 (2008) 5004–5013. [28] Y. Izumi, H. Watanabe, N. Watanabe, A. Aoyama, Y. Jinbo, N. Hayashi, Solution [7] S. Teigelkamp, R.S. Bhardwaj, J. Roth, G. Meinardus-Hager, M. Karas, C. Sorg, X-ray scattering reveals a novel structure of calmodulin complexed with a bind- Calcium-dependent complex assembly of the myeloic differentiation proteins ing domain peptide from the HIV-1 matrix protein p17, Biochemistry 47 (2008) MRP-8 and MRP-14, J. Biol. Chem. 266 (1991) 13462–13467. 7158–7166. [8] I.P. Korndörfer, F. Brueckner, A. Skerra, The crystal structure of the human [29] A. Guinier, La diffraction des rayons X aux tres petits angles; application à l'etude (S100A8/S100A9)2 heterotetramer, calprotectin, illustrates how conformational de phenomenes ultramicroscopique, Ann. Phys. 12 (1939) 166–237. changes of interacting α-helices can determine specific association of two [30] D.I. Svergun, Determination on the regularization parameter in indirect-transform EF-hand proteins, J. Mol. Biol. 370 (2007) 887–898. methods using perceptual criteria, J. Appl. Crystallogr. 25 (1992) 495–503. [9] O.V. Moroz, A.A. Antson, S.J. Grist, N.J. Maitland, G.G. Dodson, K.S. Wilson, E. [31] D.I. Svergun, Restoring low resolution structure of biological macromolecules from Lukanidin, I.B. Bronstein, Structure of the human S100A12–copper complex: im- solution scattering using simulated annealing, Biophys. J. 76 (1999) 2879–2886. plications for host-parasite defence, Acta Crystallogr. D59 (2003) 859–867. [32] J. Baudier, D. Gerard, Ions binding to S100 proteins. II. Conformational studies and [10] D.E. Brodersen, J. Nyborg, M. Kjeldgaard, Zinc-binding site of an S100 protein re- calcium-induced conformational changes in S100αα protein: the effect of acidic vealed. Two crystal structures of Ca2+-bound human psoriasin (S100A7) in the pH and calcium incubation on subunit exchange in S100a (αβ) protein, J. Biol. Zn2+-loaded and Zn2+-free states, Biochemistry 38 (1999) 1695–1704. Chem. 261 (1986) 8204–8212. [11] O.V. Moroz, E.V. Blagova, A.J. Wilkinson, K.S. Wilson, I.B. Bronstein, The crystal [33] S. Matsuda, Y. Izumi, N. Matsushima, Small angle X-ray scattering study of S100 structures of human S100A12 in the apo form and in complex with Zinc: new in- proteins, Biochem. Mol. Biol. Int. 30 (1993) 419–424. sights on S100A12 oligomerisation, J. Mol. Biol. 391 (2009) 536–551. [34] O. Glatter, Data treatment, in: Glatter O, O. Kratky (Eds.), Small Angle X-ray Scat- [12] O.V. Moroz, W. Burkitt, H. Wittkowski, W. He, A. Ianoul, V. Novitskaya, J. Xie, O. tering, Academic Press, London, 1982, pp. 119–196. Polyakova, I.K. Lednev, A. Shekhtman, P.J. Derrick, P. Bjoerk, D. Foell, I.B. Bronstein, [35] O.V. Moroz, I.B. Bronstein, K.S. Wilson, The crystal structure of zebrafish S100Z: Both Ca2+ and Zn2+ are essential for S100A12 protein-oligomerization and function, implications for calcium-promoted S100 protein oligomerisation, J. Mol. Biol. 41 BMC Biochem. 10 (2009) 11. (2011) 1072–1082. [13] P.T. Wilder, D.M. Baldisseri, R. Udan, K.M. Vallely, D.J. Weber, Location of the [36] S. Tarabykina, D.J. Scott, P. Herzyk, T.J. Hill, J.R. Tame, M. Kriajevska, D. Lafitte, P.J. Zn2+-binding site on S100B as determined by NMR spectroscopy and site-directed Derrick, G.G. Dodson, N.J. Maitland, E.M. Lukanidin, I.B. Bronstein, The dimeriza- mutagenesis, Biochemistry 42 (2003) 13410–13421. tion interface of the metastasis-associated protein S100A4 (Mts1): in vivo and [14] T. Vogl, N. Leukert, K. Barczyk, K. Strupat, J. Roth, Biophysical characterization of in vitro studies, J. Biol. Chem. 276 (2001) 24212–24222. S100A8 and S100A9 in the absence and presence of bivalent cations, Biochim. [37] N.M. Marlatt, B.L. Boys, L. Konermann, G.S. Shaw, Formation of monomeric S100B Biophys. Acta 1763 (2006) 1298–1306. and S100A11 proteins at low ionic strength, Biochemistry 48 (2009) 1954–1963. [15] M. Koch, S. Bhattacharya, T. Kehl, M. Gimona, M. Vasák, W. Chazin, C.W. Heizmann, [38] M. Unno, K. Kizawa, M. Ishihara, H. Takahara, Crystallization and preliminary P.M. Kroneck, G. Fritz, Implications on zinc binding to S100A2, Biochim. Biophys. X-ray crystallographic analysis of human peptidylarginine deiminase type III, Acta 1773 (2007) 457–470. Acta Crystallogr. F68 (2012) 668–670. [16] G. Fritz, C.W. Heizmann, P.M. Kroneck, Probing the structure of the human Ca2+- [39] W. Maret, Metals on the move: zinc ions in cellular regulation and in the coordi- and Zn2+-binding protein S100A3: spectroscopic investigations of its transition nation dynamics of zinc proteins, Biometals 24 (2011) 411–418. metal ion complexes, and three-dimensional structural model, Biochim. Biophys. [40] I.A. Brand, J. Kleineke, Intracellular zinc movement and its effect on the carbohydrate Acta 1448 (1998) 264–276. metabolism of isolated rat hepatocytes, J. Biol. Chem. 271 (1996) 1941–1949. [17] G. Fritz, P.R. Mittl, M. Vasak, M.G. Grutter, C.W. Heizmann, The crystal structure of [41] W.R. Todd, C.A. Elvehjem, E.B. Hart, Zinc in the nutrition of the rat, Am. J. Physiol. metal-free human EF-hand protein S100A3 at 1.7-Å resolution, J. Biol. Chem. 277 107 (1933) 146–156. (2002) 33092–33098. [42] C. Taccioli, H. Chen, Y. Jiang, X.P. Liu, K. Huang, K.J. Smalley, J.L. Farber, C.M. Croce, L.Y. [18] M. Unno, T. Kawasaki, H. Takahara, C.W. Heizmann, K. Kizawa, Refined crystal Fong, Dietary zinc deficiency fuels esophageal cancer development by inducing a structure of human Ca2+/Zn2+-binding S100A3 protein characterized by two di- distinct inflammatory signature, Oncogene (2011), http://dx.doi.org/10.1038/ sulphide bridges, J. Mol. Biol. 408 (2011) 477–490. onc.2011.592.