S100A11 is a dimeric EF-hand calcium-binding protein. Calcium binding to S100A11 results in a large conformational change that uncovers a broad hydrophobic surface used to interact with phospholipid-binding proteins (annexins A1 and A2) and facilitate membrane vesiculation events. In contrast with other S100 proteins, S100A10 is unable to bind calcium due to deletion and substitution of calcium-ligating residues. Despite this, calcium-free S100A10 assumes an ‘open’ conformation that is very similar to S100A11 in its calcium-bound state. To understand how S100A10 is able to adopt an open conformation in the absence of calcium, seven chimaeric proteins were constructed where regions from calcium-binding sites I and II, and helices II–IV in S100A11 were replaced with the corresponding regions of S100A10. The chimaeric proteins having substitutions in calcium-binding site II displayed increased hydrophobic surface exposure as assessed by bis-ANS (4,4′-dianilino-1,1′-binaphthyl-5,5′disulfonic acid, dipotassium salt) fluorescence and phenyl-Sepharose binding in the absence of calcium. This response is similar to that observed for Ca2+–S100A11 and calcium-free S100A10. Further, this substitution resulted in calcium-insensitive binding to annexin A2 for one chimaeric protein. The results indicate that residues within site II are important in stabilizing the open conformation of S100A10 and presentation of its target binding site. In contrast, S100A11 chimaeric proteins with helical substitutions displayed poorer hydrophobic surface exposure and, consequently, unobservable annexin A2 binding. The present study represents a first attempt to systematically understand the molecular basis for the calcium-insensitive open conformation of S100A10.
- calcium-binding protein
- NMR spectroscopy
- phospholipid-binding protein
The S100 proteins comprise a group of 25 members within the EF-hand calcium-binding protein family. Most of these dimeric proteins have been shown to act as calcium-signalling molecules, in a manner similar to that of calmodulin or troponin C, whereby the protein binds calcium and undergoes a conformational change [1–3]. Each S100 protein can interact with a diverse array of target proteins, giving rise to a wide range of biological responses [4,5]. For example, S100B undergoes a calcium-sensitive interaction with tubulin and GFAP (glial fibrillary acidic protein) in order to modulate cytoskeletal architecture through disassembly of these components [6,7]. Likewise, S100A6 binds to members of the annexin protein family (A2, A5 and A6) resulting in membrane localization and fusion events [8–10], or associates with SIP (Siah1-interacting protein), a component of a Skp1/Cullin/F-box-style E3 ligase in the ubiquitination pathway [11,12].
S100A11 is proposed to facilitate membrane vesicle-fusion events such as endocytosis and membrane repair, through calcium-dependent recruitment of annexins A1, A2 [13–16] and A6 . The three-dimensional structures of apo-S100A11 (calcium-free S100A11) and Ca2+−S100A11 are representative of most other S100 proteins, showing that each protomer consists of four α-helices that make up the two calcium-binding sites [18,19] (Figure 1). EF-hand I comprises helix I, a non-canonical 14-residue calcium-binding loop and exiting helix II, whereas EF-hand II is formed by helix III, a canonical 12-residue calcium-binding loop and exiting helix IV. The dimer interface is formed by helices I and IV from each subunit, first observed in S100A6 . In S100A11, calcium binding causes a conformational change forming an ‘open’ annexin-binding surface  (Figure 1A). In contrast, S100A10 is a unique member of the family, unable to bind calcium because of substitutions and deletions of calcium-co-ordinating residues in both EF-hands. Despite this deficiency, calcium-free S100A10  has a similar open conformation of other S100 proteins in their calcium-bound forms, particularly Ca2+−S100A11  (Figure 1B). Thus S100A10 forms a tight complex with annexin A2 and a variety of other proteins in vivo in a calcium-independent manner. To date, S100A10 is the only S100 protein to display this property [4,22].
It is not clear how S100A10 is able to adopt an open structure in the absence of calcium binding, although sequence differences with other S100 proteins probably contribute to its altered conformation. Attempting to understand the connection between the sequences, interactions and the conformation of EF-hand proteins has been a goal of researchers for many years. For example, sequence conservation and three-dimensional structures have been used to engineer calcium-binding sites into lysozyme , CD2 [24,25] and kinase-inducible domains . Furthermore, site-directed mutagenesis experiments have identified residues in calbindin D9k and troponin C that contribute most towards calcium affinity [27–29]. However, fewer examples exist where the overall conformation of an EF-hand calcium-binding protein has been altered. Bunick et al.  showed that substitution of approx. 20% of the protein sequence in calbindin D9k produced a modified protein (calbindomodulin) that was folded and adopted a more open calmodulin-like structure having a larger exposed hydrophobic surface upon binding of calcium, not present in the parent protein. More recently, substitutions in S100P have allowed this protein to interact with one of its targets (ezrin) in the absence of calcium . However, this work examined only one set of substitutions, making the contributions of different residues and regions towards the required open conformation and target binding difficult to assess.
In the present study, we have taken a systematic approach to identify the regions of S100A10 most responsible for stabilizing its open conformation, allowing it to interact with target proteins in the absence of calcium binding. Using the sequences of S100A10 and S100A11, and a structural analysis of these proteins, we substituted regions of S100A10 into the sequence of S100A11. The resulting chimaeric S100A11 proteins were examined for their abilities to bind calcium, their degree of hydrophobic surface exposure, proper folding and interactions with the known S100A11 target protein annexin A2. The present study provides the first details of the regions in S100A10 most responsible for its open conformation and constitutes an important step towards re-engineering a calcium-free protein into a constitutively active form.
The pAED4-S100A11 vector (pET-derived) used to express and clone all S100A11 proteins, as well as the GST (glutathione transferase)-fusion vector pGEX-6P1 for expression of S100A10 and annexin A2 peptide were gifts from Dr Michael Walsh (Department of Biochemistry and Molecular Biology, University of Calgary, Calgary, Alberta, Canada). Seven chimaeric S100A11–S100A10 proteins were engineered using site-directed mutagenesis based on the two-stage QuikChange™ PCR protocol . As a template, the protocol utilized a modified form of S100A11, using a C9S substitution to prevent protein oxidation  (referred to as S100A11 in the present paper), and complementary primers encoding the desired sequence corresponding to human S100A10. Briefly, S100A11 (100 ng) was combined with 2 μM of each particular primer (forward and reverse primer separately) in a 50 μl reaction mixture. For the extension reactions, 2.5 units of Pfu Turbo (Stratagene) and the following PCR conditions were used: preheating at 95 °C for 30 s, and five cycles of 95 °C for 30 s, 55 °C for 1 min and 68 °C for 7 min. After cooling reaction mixtures to 4 °C, 50 μl of the forward and reverse extended reactions were mixed and subjected to 16 cycles of PCR extension as described above. The product was incubated with 10 units of DpnI (Roche) before transformation into Escherichia coli strain BL21(DE3) for bacterial expression. More detailed cloning information for specific S100A11 proteins is available from G.S.S. on request.
Two different sets of S100A11 chimaeras were designed including substitutions and/or deletions into either the calcium-binding loops or the helical regions of S100A11 (Figure 2). S100A11LI, having the N-terminal calcium-binding loop from S100A10, included three amino acid deletions (ΔK23, ΔG25 and ΔV28) and three substitutions (H26K, S27G and T29Y) made by PCR-based methods  and three additional substitutions (S31T, T33E and E34D) made using the two-stage QuikChange™ protocol (described above). S100A11N66C,E75S was constructed using a similar approach and utilized as the base sequence to produce S100A11LII where the canonical C-terminal EF-hand loop was replaced with that in S100A10 (L65NSDGQLDFQE75→Q65CRDGKVGFQS75). This S100A11LII plasmid served as a template to create the S100A11LI,LII protein where both calciumbinding sites were replaced as in S100A10 (A19GDKGYLTKED29, Q65CRDGKVGFQS75). Proteins S100A11H2,H3 (F35LAFMNTE42 and P53GVLDRMMKKL63→L35RVLMEKE42 and P53LAVDKIMKDL63) and S100A11H3,H4 (P53GVLDRMMKKL63 and F76LNLIGGLAV85→P53LAVDKIMKDL63 and F76FSLIAGLTI85) used an S100A11 construct containing the helix III sequence as in S100A10 (P53LAVDKIMKDL63) and introduced either helix II or helix IV substitutions. S100A11H2−L−H3 was generated through substitution in the linker region of the S100A11H2,H3 containing (F35LAFMNTELAAFTKNQKDPGVLDRMMKKL63→L35RVLMEKEFPGFLENQKDPLAVDKIMKDL63) plasmid.
GST–annexin A2 was constructed by inserting the N-terminal sequence of annexin A2 (STVHEILSKLSLEG) into a GST-fusion vector (pGEX-6P1). PCR primers spanning the peptide sequence (27 bp overlap) using ApaI and XhoI restriction sites were used. PCR products were digested and ligated into the vector. All sequences were verified by DNA sequencing at the Robarts Research Institute (London, Ontario, Canada).
Protein expression and purification
All S100A11 proteins (wild-type and mutants) were expressed in E. coli strain BL21(DE3) at 37 °C. Bacteria were grown to a D600 of 0.7–0.8 and treated with 0.4 mM IPTG (isopropyl β-D-thiogalactopyranoside) for approx. 4 h to induce protein expression. Cells were harvested by centrifugation at 8000 g for 10 min and lysed using a French pressure cell. Lysates were centrifuged at 38000 rev./min for 1 h at 4 °C using a Beckman 50Ti rotor, and the supernatants were subjected to a variety of purification schemes depending on the properties of the protein. Wild-type S100A11, S100A11LI, S100A11H2,H3 and S100A11H2−L−H3 were purified using a phenyl-Sepharose CL-4B matrix (GE Healthcare) as described previously . S100A11H2,H3 and S100A11H2−L−H3 were purified further by gravity flow on a Sephadex G-75 column (GE Healthcare) in 25 mM Tris/HCl, 100 mM NaCl and 5 mM DTT (dithiothreitol) (buffer A) with the addition of 5 mM EGTA (pH 7.5). Purification of S100A11H3,H4 was achieved using a phenyl-Sepharose FF column (1 ml, GE Healthcare) equilibrated in buffer A with 5 mM CaCl2 and eluted with 25 mM Tris/HCl, 5 mM DTT and 5 mM EGTA (pH 7.5). For chimaeric proteins where the second calcium-binding site was modified, cell extracts were loaded on to a HiTrap SP-XL column (GE Healthcare) equilibrated with either 25 mM Mes (S100A11N66C,E75S; pH 5.8) or 25 mM Hepes (S100A11LII, S100A11LI,LII; pH 7.2), 1 mM EDTA and 1 mM DTT and eluted using a linear NaCl gradient. All purified proteins were extensively dialysed against 3 mM KCl and freeze-dried for storage. The purity and identity of each protein was confirmed by SDS/PAGE (16.5% gels) and electrospray ionization MS (Biological Mass Spectrometry Laboratory, University of Western Ontario). For uniformly 15N-labelled proteins, cells were grown in M9 minimal medium with 15NH4Cl as the sole nitrogen source and purified as described above.
GST–S100A10 and GST–annexin A2 were expressed from bacteria grown in 1 litre of 2YT medium [1.6% (w/v) tryptone, 1% (w/v) yeast extract and 0.5% NaCl] at 37 °C to a D600 of 0.6 and induced by the addition of 0.4 mM IPTG. Bacteria were lysed using a French pressure cell in 140 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4 and 1.8 mM KH2PO4 (pH 7.3). The lysates were centrifuged at 38000 rev./min for 1 h at 4 °C using a Beckman 50Ti rotor and the supernatant was loaded on to a GSTPrep FF 16/10 column (GE Healthcare) equilibrated with lysis buffer. After extensive washing, GST–S100A10 was eluted in 50 mM Tris/HCl (pH 8) and 10 mM glutathione. Fractions containing GST–S100A10 were pooled and dialysed overnight against 50 mM Tris/HCl (pH 7.0), 150 mM NaCl, 1 mM EDTA and 1 mM DTT. Cleavage of the GST tag was carried out using PreScission Protease (GE Healthcare) according to the manufacturer's directions. After ensuring complete GST cleavage by SDS/PAGE (16.5% gels), S100A10 was purified via the GSTPrep FF 16/10 column collecting the flow-through material. Fractions containing S100A10 were pooled and dialysed overnight against 50 mM Tris/HCl (pH 7.5), 0.2 mM EDTA and 0.5 mM TCEP [tris-(2-carboxyethyl)phosphine], flash frozen in liquid nitrogen and stored at –80 °C.
Folding of all proteins was analysed using a Jasco J-810 spectropolarimeter. Typical samples comprised 7 μM protein (monomer) in 20 mM Tris/HCl (pH 7.2), 10 mM NaCl, 0.5 mM TCEP and 5 mM EDTA (10 mM CaCl2 for Ca2+−S100A11). For each protein, five scans from 260 to 190 nm were recorded using a 1-mm-pathlength cell at 24 °C, averaged and buffer-subtracted. Protein concentrations were quantified by using the peak volumes of alanine and leucine residues from duplicate amino acid analyses (Alberta Peptide Institute).
Sensitivity-enhanced 1H-15N-HSQC (heteronuclear single-quantum coherence) spectra  were performed on a Varian INOVA 600 MHz spectrometer equipped with a pulsed field gradient triple resonance probe at 35 °C. Protein samples were typically 200 μM monomer in 20 mM Mops, 50 mM KCl, 5 mM DTT, 90% 1H2O/10% 2H2O and either 5 mM EDTA (apo) or 10 mM CaCl2 (Ca2+-bound) at pH 7.25. Sequential assignments were made from HNCA , HNCACB  and CBCA(CO)NH  experiments, processed with NMRPipe  and analysed using NMRView  software.
Bis-ANS (4,4′-dianilino-1,1′-binaphthyl-5,5′disulfonic acid, dipotassium salt) emission spectra were measured using a Fluorolog-3 steady-state fluorimeter. Protein samples (4 μM monomer) in 20 mM Tris/HCl (pH 7.2), 10 mM NaCl, 8 μM TCEP and either 100 μM EDTA (apo) or 1 mM CaCl2 (Ca2+-bound) were saturated with 20 μM bis-ANS and their emission spectra were recorded from 430 to 600 nm using an excitation wavelength of 388 nm. All measurements used five averaged scans and were buffer-subtracted.
Chimaeric proteins were evaluated for their capacities to bind to phenyl-Sepharose in the absence and the presence of calcium using a batch method. Protein samples (40 μg of monomer) in 20 mM Tris/HCl (pH 7.2), 10 mM NaCl, 0.5 mM TCEP and excess EDTA (or excess CaCl2) were incubated with phenyl-Sepharose beads (200 μl of a 50% slurry) at room temperature (20 °C) for 1 h with gentle agitation. Proteins that did not interact with the matrix were recovered in the supernatant following centrifugation at 700 g for 5 min. Proteins bound to the matrix were eluted by boiling the beads in SDS/PAGE sample buffer. The degree of protein interaction was quantified by measuring the band intensities using Image J 1.410 software (NIH; http://rsb.info.nih.gov/ij/).
GST–annexin A2 (620 μg) was immobilized on glutathione–Sepharose beads (600 μl of a 50% slurry, GE Healthcare) and incubated for 30 min with 40 μg of S100A10 or S100A11 proteins in 20 mM Tris/HCl (pH 7.2), 50 mM NaCl, 0.5 mM TCEP and 5 mM EDTA (10 mM CaCl2 for Ca2+−S100A11). Beads were spun down at 700 g for 5 min and washed (3×1 ml) with the appropriate buffer. Proteins bound to the beads were eluted using 20 mM glutathione, resolved by Tris/Tricine SDS/PAGE (16.5% gels) and detected using Coomassie Blue stain. Concentrations were quantified using a series of BSA standards. All assays were conducted in duplicate with errors found typically within the 10% range.
S100A11 undergoes a large calcium-induced conformational change involving the reorientation of helix III in calcium-binding site II . This ‘opening’ of site II forms a hydrophobic cleft consisting of residues from helices III, IV and the helix II–III linker, and helix I′ from the partner protomer essential for interaction with annexins A1 and A2 (Figure 1). In contrast, calcium-free S100A10 adopts an open conformation in the absence of calcium. The regions that contribute to and stabilize the open conformation of S100A10 in the absence of calcium were investigated using a series of chimaeric S100A11 proteins. In particular, S100A11 was chosen as a template because (i) the conformation of Ca2+−S100A11 strongly resembles that of S100A10 [rmsd (root mean square deviation) 0.85 Å (1 Å=0.1 nm)] (Figure 1), (ii) both proteins share at least one identical target (annexin A2) with similar binding sites [19,21] (Figure 1), and (iii) the two proteins share 41% sequence identity (Figure 2).
Design of S100A11 chimaeric proteins
To help to identify the sequences and regions responsible for the open conformation of S100A10, a combination of Cα distances, residue–residue contacts, hydrophobic surface area exposure and three-dimensional structure examination was used (see Supplementary Figure S1 at http://www.BiochemJ.org/bj/434/bj4340037add.htm). The key observations included: (i) the first EF-hand loop in S100A10 is three amino acids shorter than the N-terminal loop in S100A11 (Figure 2A), (ii) the calcium ligands Asn66 and Glu75 in the C-terminal loop in S100A11 are replaced by Cys61 and Ser70 in S100A10, (iii) Leu43 and Phe46 in the linker region contact amino acids in helices III and IV in S100A11; in S100A10, Phe38 (Leu43 in S100A11) shows interactions mainly with residues in helix IV and Phe41 (Phe46 in S100A11) is more exposed with a different orientation, leading to alternative interactions with residues in helices III and IV, (iv) a network of hydrogen bonds involving residues in helix III (Asp52, Lys53) with those in site I (Lys27), helix II (Arg31) and the linker (Asp47) are present in S100A10, but absent from apo-S100A11, (v) S100A10 is the only protein that has a lysine residue at the –y position in site II, where its side chain forms a unique hydrogen bond with Tyr24 in site I, and (vi) the N-terminus of helix III in apo-S100A11 has multiple contacts with the C-terminus in helix IV, absent from S100A10.
On the basis of these observations, we designed seven chimaeric proteins that incorporated distinct regions of the S100A10 sequence into S100A11. To identify the structural effects of altering the calcium-binding sites of S100A11, four different chimaeric proteins were constructed (Figure 2B). The first protein, S100A11LI, replaced the 14-residue calcium-binding loop in site I with the shorter loop (A19GDKGYLTKED29) found S100A10. A second chimaera, (S100A11N66C,E75S) replaced two calcium ligands in site II (Asn66 and Glu75) with residues found in S100A10 (Cys61 and Ser70). As with the S100A11LI protein, the second of these replacements (Glu75) occurred at the −z position (an invariant glutamate residue in S100 proteins) for calcium co-ordination. Replacing the entire site II loop in S100A11 (L65NSDGQLDFQE75) with that found in S100A10 (Q65CRDGKVGFQS75) was also examined (S100A11LII). Within this region, hydrophobic interactions are observed between Phe73 in site II and Leu13 and Ile14 of helix I in Ca2+−S100A11 (Phe68 with Met12 in S100A10), and an increased network of hydrogen bonds exists that are not present in apo-S100A11. Furthermore, structural analyses (S100A2, PDB code 2RGI; S100A3, PDB code 1KSO; S100A6, PDB codes 2CNP and 1K9K; and S100B, PDB codes 1B4C and 1UWO), reveals an inversion of the geometry of the site II loop upon calcium binding (Figure 2C), also present in calbindin D9k  and S100A10 , but not apo-S100A11 . The co-operative effect of the substitutions in S100A11LI and S100A11LII was tested using a fourth chimaeric protein (S100A11LI,LII).
To evaluate the importance of the helices towards stabilizing the open form of S100A10, three chimaeric proteins were engineered, focusing on helix III where the largest structural differences exist between apo-S100A11 and S100A10. The contribution of the packing of helices II/III was assessed using a chimaeric S100A11 protein with helices II and III from S100A10 (S100A11H2,H3). Analyses using Cα distance measurements showed that the C-terminus of helix III is proximal to the N-terminus of helix II (Figure 2D) in both Ca2+−S100A11 and S100A10, resulting in the burial of Ser37, Asn40, Thr41 and Glu42 (S100A11). The helix III–IV interaction was tested using the chimaeric protein S100A11H3,H4 containing residues from helix III and the first ten residues of helix IV in S100A10. The aim of the S100A11H3,H4 chimaera was to disrupt interactions between residues at the N-terminus of helix III (Val55, Leu56) and residues in helix IV (Leu83, Ala86, Cys87) that occur in apo-S100A11, but not S100A10 (Figure 2E). A final chimaeric protein (S100A11H2−L−H3) where the entire helix II–linker–helix III section of S100A11 was replaced by the corresponding section in S100A10 was made because Ala44, Ala45, Phe46, Thr47 and Asp52 become more exposed upon calcium binding in S100A11 and are involved in target interactions with annexin A1 [18,19].
Characterization of chimaeric S100A11 proteins
The structural integrity of the chimaeric proteins was evaluated since several of these proteins had large sections of their sequences replaced that could potentially disrupt their folding. Dynamic light scattering and sedimentation equilibrium experiments showed that the molecular masses for all chimaeric proteins were within 10% of their calculated dimeric molecular masses (results not shown), indicating that all proteins remained as dimers, similar to S100A11 and S100A10.
CD spectropolarimetry was used to probe the proper folding and secondary structure of apo-S100A11, Ca2+−S100A11, S100A10 and the seven chimaeric proteins (see Supplementary Figure S2 at http://www.BiochemJ.org/bj/434/bj4340037add.htm). The CD spectrum of apo-S100A11 had θ222/θ208=0.93 that decreased in magnitude and θ222/θ208=0.87 upon calcium binding. This trend is similar to that reported for apo-S100B  and results from a change in helix–helix interactions. Although structurally similar, S100A10 displayed an altered θ222/θ208 (1.05) and increased magnitude compared with Ca2+−S100A11. This difference probably arises from the calcium-co-ordination in S100A11 not present in S100A10. A comparison of the chimaeric proteins containing substitutions in the calcium-binding sites showed a maximum 13% deviation for S100A11LI,LII and apo-S100A11. For the chimaeric proteins that had modifications in the helices, the difference in the CD spectra between the individual substituted proteins was less than 1%. These results indicated that all chimaeric proteins studied maintained the same high degree of α-helical structure even when large numbers of substitutions were introduced (18 substitutions in S100A11H2−L−H3).
A series of 1H-15N-HSQC spectra were used to verify that all proteins were properly folded (see Supplementary Figure S3 at http://www.BiochemJ.org/bj/434/bj4340037add.htm). In the absence of calcium, all chimaeric proteins showed well-dispersed spectra with resonances of similar linewidth to apo-S100A11 and no obvious line-broadening or doubling of peaks, suggestive of multiple conformations. The chemical shift differences observed between the chimaeric S100A11 proteins and apo-S100A11 were used to pinpoint whether substitutions simply affected the environment of residues within close proximity to the substitution points or had a more global effect on the structures. For example, a superposition of the 1H-15N-HSQC spectra for apo-S100A11 and S100A11N66C,E75S (see Supplementary Figure S3A) showed that most of the peaks in helices I (Thr4, Glu5, Glu7, Arg8, Ile10 and Leu13), III (Gly54 and Val55), the C-terminus of helix IV (Glu89, Ala94 and Ala95) and portions of calcium-binding site I (Gly22 and Gly25) for S100A11N66C,E75S remained in similar positions compared with apo-S100A11, indicating that the two-residue substitution had little influence on the environments of these residues. In addition, residues in calcium-binding site II (Lys62, Leu63, Gly69 and Gln74) close in sequence to the substitution site, and residues near the C-terminus of site I (Thr29, Ser31, Thr33 and Phe35) underwent expected shifts due to close proximity to the substitutions. Surprisingly, chemical-shift changes were noted in S100A11N66C,E75S for more remote residues in the linker (Leu43, Ala44, Ala45 and Lys48) and in the central portion of helix IV (Gly81, Leu83, Val85 and Ala86). This indicated that replacement of Asn66 and Glu75 in apo-S100A11 affected the environments of residues in these two regions via another means, possibly from a structural change in the chimaeric protein. Similar allosteric effects have been noted previously for calbindin D9k . Unlike S100A11N66C,E75S, other chimaeric proteins with more substitutions resulted in large numbers of chemicalshift changes that made it difficult to discern between local environment changes due to the substitutions and alterations in the protein conformation. However, these NMR observations indicated that all of the chimaeric S100A11 proteins adopted folded α-helical dimeric species.
Most chimaeric Ca2+−S100A11 proteins can adopt an open conformation
Calcium binding to apo-S100A11 leads to a conformational change that is easily monitored by 1H-15N-HSQC spectroscopy since the positions of many residues in Ca2+−S100A11 differ from those in the apo state (Figure 3A). Using this approach, we compared spectra of the S100A11 chimaeric proteins in the absence and presence of calcium to determine whether these proteins could bind calcium and adopt a more open conformation as in Ca2+−S100A11.
Substitutions in the second calcium-binding site of S100A11 (S100A11N66C,E75S, S100A11LII) showed several minor chemical-shift changes upon calcium addition (Figure 3B and see Supplementary Figure S4 at http://www.BiochemJ.org/bj/434/bj4340037add.htm), suggesting that co-ordination of calcium to site II was abolished. In particular, there was little change in the position of Gly69, which typically shifts to ~10.5 p.p.m. in Ca2+−S100A11 (Figure 3A). However, chemical-shift changes were observed near site I (Phe17, Gln18, Gly22, Leu30 and Thr33) in S100A11N66C,E75S, indicating that this site binds calcium and affects other more remote residues (Leu43, Ala44, Lys48, Gly81 and Ala86). The chimaeric protein S100A11LI,LII showed near identical 1H-15N-HSQC spectra in the absence and presence of calcium, showing that this protein has lost its ability to chelate calcium (see Supplementary Figure S4). Mutations of the low-affinity calcium-binding site alone (S100A11LI, Figure 3C) perturbed calcium binding to the higher-affinity site, since residues diagnostic of the calcium-bound state in site II (Leu65, Gly69, Leu71 and Asp72) did not shift to the characteristic regions found in Ca2+−S100A11. Overall, these results indicated that chimaeric proteins with substitutions in calcium-binding sites I or II (S100A11LI, S100A11LI,LII and S100A11LII) had strongly attenuated calcium-binding ability much like S100A10.
Addition of calcium to the chimaeric proteins having altered helices (S100A11H2,H3, S100A11H2−L−H3 and S100A11H3,H4) resulted in 1H-15N-HSQC spectra (Figure 3D and see Supplementary Figure S4) that were typical of Ca2+−S100A11 (Figure 4A). For example, calcium binding to S100A11H3,H4 resulted in large chemical-shift changes for residues in both sites I (Gly22, Thr29, Leu30 and Ser31) and II (Leu65, Gly69 and Leu71), the linker (Leu43 and Ala44), helix III (Asn49, Val55 and Met60) and helix IV (Ala86 and His88). This implies that large sequence changes in the helices of S100A11 do not prevent it from binding calcium or undergoing a large conformational change.
Evidence for a calcium-independent hydrophobic surface in site II chimaeric proteins
Two biochemical assays were used to evaluate the exposed hydrophobic surface area in apo-S100A11, Ca2+−S100A11, S100A10 and the seven chimaeric proteins. The first assay utilized the extrinsic fluorophore bis-ANS which exhibits weak fluorescence in solution, but shows greatly enhanced fluorescence upon binding to a hydrophobic protein surface such as those exhibited by the calcium-bound states for S100B , troponin C  and calmodulin . In the presence of apo-S100A11, bis-ANS showed weak fluorescence near 520 nm (Figure 4A), whereas S100A10 or Ca2+−S100A11 caused a dramatic increase in fluorescence, accompanied by a blue shift to ~500 nm, indicative of bis-ANS binding to a exposed hydrophobic surface in each protein. The fluorescence for S100A10 was approx. 40% higher than for Ca2+−S100A11, consistent with its greater exposed hydrophobic surface area (~200 Å).
The chimaeric S100A11 proteins were examined in the absence of calcium to determine whether the substitutions might liberate a greater hydrophobic surface area. For proteins with substitutions in the calcium-binding sites, a wide range of responses was noted (Figure 4B). S100A11LI showed a similar fluorescence spectrum to that of apo-S100A11, suggesting that changes to calcium-binding site I had little effect on the hydrophobicity of the protein. On the other hand, the site II chimaeric proteins (apo-S100A11N66C,E75S, apo-S100A11LII and apo-S100A11LI,LII) showed 3–5-fold increases in fluorescence compared with apo-S100A11. These observations suggest a greater ability to occupy an open conformation in these proteins, marked by the exposure of a greater hydrophobic surface. Although apo-S100A11LII had the largest increase in fluorescence, approx. 70% of this was achieved with only two substitutions (apo-S100A11N66C,E75S), indicating that positions Asn66 and Glu75 are potentially key residues for the structural difference in S100A10 compared with apo-S100A11. Substitutions in both calcium-binding loops (apo-S100A11LI,LII) had a moderating effect, since this protein displayed an intermediate fluorescence between apo-S100A11LI and apo-S100A11LII, showing that substitutions to both EF-hands are not simply additive. Since this protein has lost its ability to bind calcium, no change was noted in its bis-ANS fluorescence spectrum upon calcium addition. However, the remaining calcium-binding site chimaeras all exhibited increased fluorescence intensities ranging from ~4-fold for S100A11LI to ~15% for S100A11LII upon calcium addition.
Substitutions of helices II, III or IV in S100A11 resulted in more modest fluorescence changes (Figure 4C) compared with the calcium-binding loop substitutions. For example, both the apo-S100A11H3,H4 and apo-S100A11H2,H3 chimaeric proteins had increased fluorescence intensities, although these changes were consistently lower than observed for the site II-substituted proteins (apo-S100A11LII and apo-S100A11N66C,E75S). The apo-S100A11H2−L−H3 protein had a fluorescence spectrum that was more similar to apo-S100A11, suggesting a minimally exposed hydrophobic surface. Upon addition of calcium, all of the chimaeric helix proteins exhibited >50% increases from their apo states, indicating a greater hydrophobic surface as a result of the calcium-induced conformational change. Overall, these assays showed that the largest contribution to the increased hydrophobic surface observed in S100A10 is produced from substitutions in its second calcium-binding loop (S100A11N66C,E75S and S100A11LII).
To identify whether the increased bis-ANS fluorescence resulted from a structural change similar to that observed upon calcium binding to S100A11, the capacity for binding of the S100A11 chimaeric proteins to a hydrophobic phenyl-Sepharose matrix (Figure 5) was evaluated. Many EF-hand proteins, including S100A11, preferentially bind to phenyl-Sepharose in the calcium-bound form using an exposed hydrophobic surface to interact with the matrix. In the absence of calcium, S100A11 did not bind to the phenyl-Sepharose beads (Figure 5B). In contrast, Ca2+−S100A11 and S100A10 both bound strongly to the beads, with a minimal amount of material found in the supernatant. For S100A11, this indicated that the protein had an exposed hydrophobic surface consistent with adopted an open conformation in the calcium-bound state. The chimaeric apo-S100A11 proteins showed a range of binding efficiencies to the matrix, although all of the chimaeric proteins (Figures 5C and 5D) exhibited lower phenyl-Sepharose binding compared with either Ca2+−S100A11 or S100A10. Three chimaeric proteins (S100A11LI, S100A11H2−L−H3 and S100A11H3,H4) bound poorly to phenyl-Sepharose in the absence of calcium, but, like S100A11, the interaction was increased upon calcium binding. The apo-S100A11LI,LII chimaera showed weak phenyl-Sepharose binding that was not altered upon calcium addition, consistent with the compromised calcium-binding sites in this protein. In contrast, when residues in site II of S100A11 were replaced with those in S100A10 (S100A11N66C,E75S and S100A11LII), a significant increase in bound protein was observed in the calcium-free states (Figure 5C), consistent with each of these proteins having a greater hydrophobic surface than apo-S100A11. Upon calcium addition, the proportion of these proteins bound to the beads was enhanced, indicating that calcium binding was able to alter the exposed hydrophobic surface further. The S100A11H2,H3 protein behaved in a similar fashion, the only S100A11 chimaera having substitutions in the helical regions that also showed an increase in bis-ANS fluorescence. These results indicate that, although both bis-ANS fluorescence and phenyl-Sepharose binding are good indicators of exposed hydrophobic surface, subtle differences exist between the two methods. Nevertheless, apo-S100A11N66C,E75S, apo-S100A11LII and apo-S100A11H2,H3 consistently showed that these chimaeric proteins had a greater exposed hydrophobic surface than apo-S100A11, probably due to an increased ability of these proteins to occupy an open conformation, similar to that observed upon calcium binding to S100A11 or in S100A10.
S100A11LII interacts with annexin A2
The exposed hydrophobic surface in S100A11LII and S100A11N66C,E75S, observed from bis-ANS and phenyl-Sepharose experiments, may be adequate for target protein binding to these proteins in the calcium-free state. We tested this for several of the chimaeric proteins using a GST construct containing the N-terminus of annexin A2, a phospholipid-binding protein shown to bind to both Ca2+−S100A11  and S100A10 [48,49]. As expected, apo-S100A11 did not interact with GST–annexin A2, but showed tight binding in the presence of calcium (Figure 6), a similar effect to that noted for S100A10 in the absence of calcium. Of the four chimaeric S100A11 proteins that showed the highest hydrophobic surface area by bis-ANS and phenyl-Sepharose experiments in the absence of calcium, three proteins (S100A11N66C,E75S, S100A11H2,H3 and S100A11H3,H4) repeatedly showed no detectable binding to annexin A2. This indicated that the increased hydrophobicity for these proteins was not sufficient for annexin A2 recognition. Conversely, apo-S100A11LII displayed a reproducible interaction with annexin A2, albeit weaker than either S100A10 or Ca2+−S100A11. This result indicates that the site II loop region provides important contributions for the increased hydrophobicity of S100A10 and its interaction with annexin A2.
Structural effects of substitutions in S100A11LII
Apo-S100A11LII was the only chimaeric protein tested that exhibited both increased hydrophobicity compared with apo-S100A11 and the ability to bind to annexin A2. This suggests that substitutions in site II alter the equilibrium between closed (apo-S100A11) and open (Ca2+−S100A11 and S100A10) states to favour an open conformation in apo-S100A11LII. To identify whether a shift towards the open state has occurred for apo-S100A11LII, we completed its backbone assignment using NMR techniques and used chemical-shift perturbation methods to compare its 1H-15N-HSQC spectrum with that of apo-S100A11 (Figures 7A and 7B). A similar comparison between Ca2+−S100A11 and apo-S100A11 was also completed (Figure 7C). It was anticipated that residues >6 Å from the sites of the substitutions in S100A11LII might exhibit chemical-shift changes, resulting from an altered conformation, similar to those observed between Ca2+−S100A11 and apo-S100A11. As expected, the data showed that many peaks in site II of apo-S100A11LII displayed changes (Leu63, Leu65, Cys66, Gly69, Asp72, Phe73 and Gln70) where the substitutions were made in S100A11, and in site I (Ala21, Thr29, Leu30 and Ser31) that adjoins this region. Most of these residues also experienced chemical-shift changes upon calcium binding to S100A11, although the magnitude of the shifts are more pronounced due to the added electrostatic effect of calcium co-ordination. Several residues experienced larger than average chemical-shift differences between S100A11LII and apo-S100A11 that were >6 Å from the nearest substitution on the basis of the three-dimensional structure of apo-S100A11. These included several residues in helix II (Leu36, Ser37 and Asn40) and helix IV (Gly81, Gly82, Leu83, Ala84, Val85 and His88). A similar comparison between Ca2+−S100A11 and apo-S100A11 (Figure 7C) revealed that the calcium-induced conformational change affected many residues remote from the calcium-binding sites including Ile14 and Val16 (helix I), Leu36 (helix II), Ala44 and Thr47−Lys51 (linker) and Gly82, Val85 and Cys87 (helix IV). This analysis revealed that some similarity existed for perturbed residues in S100A11LII (Figure 7B) and Ca2+−S100A11 (Figure 7C) when compared with apo-S100A11 especially with respect to residues affected in helices II and IV. However, it was clear that a large segment in the linker region (Thr47−Lys51) and the C-terminus of helix I (Ile14 and Val16) showed much larger changes upon calcium binding to S100A11 compared with apo-S100A11LII. Most of these residues also exhibit large chemical-shift changes upon interaction with annexin A2, indicating their importance in Ca2+−S100A11. Overall, this analysis shows that replacing residues in the second calcium-binding site of S100A11 with those from S100A10 affects the environments of residues in helix II and helix IV, but has a smaller impact on residues in helix I and the linker regions that comprise a major portion of the binding site for annexin A2. Combined with hydrophobicity and annexin A2-binding experiments, this would be consistent with an increase in the population of the open state for apo-S100A11LII, which is not fully reached as in Ca2+−S100A11 or S100A10.
In the present study, we used a series of chimaeric S100A11 proteins to provide insights into the most important regions in S100A10 responsible for its constitutively open conformation. Substitutions ranged in complexity from two residues in the second calcium-binding loop (S100A11N66C,E75S) to the complete exchange of helices II, III and their intervening linker in S100A11H2−L−H3. Replacement of helices II, III and IV in S100A11 with the corresponding regions from S100A10 (apo-S100A11H2,H3, apo-S100A11H3,H4 and apo-S100A11H2−L−H3) resulted in smaller changes in exposed hydrophobic surface area that are characteristic of Ca2+−S100A11 or S100A10. Each of these chimaeras utilized helix III from S100A10, the helix that undergoes a large structural change upon calcium binding to S100A11 (Figure 1). Only apo-S100A11H2,H3 showed evidence of phenyl-Sepharose binding, although its bis-ANS fluorescence was consistently lower than substitutions made in calcium-binding site II (see below). However, the chimaeras retained the ability to bind calcium and, on the basis of characteristic chemical-shift patterns, were able to adopt the calcium-bound conformation. These observations suggest that single helix–helix interfaces are not dominant contributors to the stability required to occupy the open conformation in S100A11 in the absence of calcium. However, substitutions in the helices, especially helix IV, have been shown to have dramatic effects on the dimerization propensity (S100A4, S100A1 and S100P) and the interaction with target proteins [50–53].
Site II chimaeras stabilize the open conformation of apo-S100A11
Of the seven S100A11 hybrid proteins designed, only apo-S100A11N66C,E75S and apo-S100A11LII consistently showed significant increases in hydrophobic surface exposure similar to S100A10 and Ca2+−S100A11. However, binding of these two chimaeric proteins to phenyl-Sepharose was less efficient than the wild-type proteins. This might suggest that the chimaeric proteins are adopting a conformation that is midway between the closed and open states observed for apo-S100A11 and Ca2+−S100A11 respectively. However, such an intermediate state has not been demonstrated, even under mild denaturing conditions, where the only intermediate observed for an S100 protein is a less compact dimer. An alternative and more likely explanation is that the substitutions in site II of S100A11 act to stabilize the open conformation relative to the wild-type protein in the calcium-free state, altering the populations of closed and open state observed. This concept has been well established for other EF-hand calcium-binding proteins such as calbindin D9k and calmodulin where the conformations of these proteins have been altered by substitution [28,53–59]. NMR studies of the E104Q/E140Q mutant of the C-terminal domain of calmodulin by Lundstrom and Akke  and Evenas et al. [55–58] have elegantly shown exchange in the calcium-saturated state similar to the closed and open conformations of wild-type calmodulin. Their findings provide evidence of the presence of a small but finite population of calcium-bound protein, even though the closed conformation is favoured [54–58]. Moreover, Ababou et al. [59,60] have found that the calcium-induced opening of the N-terminal domain in calmodulin can be compromised when the polar residues Gln39 (B/C linker) and Lys75 (helix D) are replaced with the corresponding hydrophobic amino acids present in calbindin D9k (Leu39 and Ile75) effectively stabilizing the apo state. In a complementary approach, the Chazin group substituted 15 residues in the interhelical regions of calbindin D9k, allowing this protein to favour a more open calmodulin-like conformation upon calcium binding . In apo-S100A11N66C,E75S and apo-S100A11LII, the substitutions in loop II probably act to stabilize and populate the open state to a greater extent than in apo-S100A11.
The increase in hydrophobicity in apo-S100A11LII and apo-S100A11N66C,E75S and the limited number of substitutions in each protein provides some details about the residues that could have an important role in stabilizing the open state of S100A10 in the absence of calcium. For example, substitution of the highly conserved −z position (Glu75) in the calcium-binding loop in S100A11N66C,E75S and S100A11LII reduces their calcium-binding abilities, similar to other EF-hand proteins. However, even in the presence of a functional −z position, this residue contributes little to the conformational change towards the open state. Upon calcium binding in S100A11, the distance (~6.5 Å) between Glu75 (Cβ) and the calcium ligand Gln70 (CO) changes by less than 0.5 Å. Furthermore, superposition of the C-terminal portion of site II and helix IV (Lys65–Thr79) from S100A10, which resides in the open state, reveals modest differences with the same region of the calcium-free forms (closed) for S100A2 (0.37 Å), S100A3 (0.63 Å), S100A6 (0.63 Å), S100B (0.83 Å) and S100A11 (0.96 Å). This indicates that, despite being a critical residue for calcium binding in most other S100 proteins, Glu75 in S100A11 (Ser70 in S100A10) appears to have little influence on the open/closed conformations. This is in contrast with experiments for calmodulin and troponin C that show substitution at the −z co-ordinating position not only deters calcium binding, but also favours a closed state [55,56,61,62]. The structural comparison described for S100A10 indicates this does not occur in the S100 proteins, providing evidence that the E75S replacement alone is not a dominant factor in favouring the open conformation of S100A10.
Unlike the structurally similar C-terminal portions of the site II calcium-binding loop in the closed (apo-S100A11) and open (S100A10 and Ca2+−S100A11) states, the N-terminus of the loop (Asp59–Asp63) adopts two very different conformations. In S100A10, the conformation is similar to that found in the calcium-bound forms of most S100 proteins. However, structures of calcium-free S100A2, S100A3, S100A11 and S100A6 show that this portion of the loop adopts an alternative orientation (‘inverted’, Figure 2C), centred near the y co-ordinating position, site of the N66C substitution used in both S100A11N66C,E75S and S100A11LII. Upon calcium binding, this residue and the following y+1 residue reorient (Δϕ~30 °,Δψ~33 °;Δϕ~101 °,Δψ~80 ° respectively), leading to the large repositioning of helix III observed in Ca2+−S100A11 (or S100A10) and the exposed hydrophobic surface used for target binding. In the present study, we noted increased binding of S100A11N66C,E75S to bis-ANS and phenyl-Sepharose, suggesting that the N66C substitution might be a contributing factor to the increased hydrophobicity. In apo-S100A11 the y position occupied by Asn66 forms a salt bridge to Lys32 in site I. In contrast, in S100A10, the y position (Cys61) is exposed and the analogous salt bridge (Lys27) forms with Asp52 in helix III, implying that substitution for Asn66 (N66C) in apo-S100A11 might favour the exposure of this residue and contribute to the stabilization of the open conformation as it does in S100A10. It is noteworthy that substitutions for Asn66 and Glu75 of S100A11 were sufficient for the protein to acquire an increased hydrophobic surface, but not sufficient for annexin A2 binding. Furthermore, although five additional substitutions used in S100A11LII enhanced annexin A2 binding in the calcium-free state, target binding was less efficient compared with either S100A10 and Ca2+−S100A11. These findings are similar to those for a modified S100P protein having three substitutions in site II that exhibited phenyl-Sepharose binding in the absence of calcium and partial interaction with ezrin, one of its protein targets . The S100P protein also contained a three-residue deletion in site I. However, it is unclear how this deletion contributes to the open conformation, since a chimaeric protein possessing the site I sequence of S100A10 (S100A11LI) displays little increase in hydrophobicity in the absence of calcium and is unable to adopt the open conformation even in the presence of calcium. A significant outcome of our work is that it shows that residues in site II are important contributors towards the open state in S100A10. Further refinement of these chimaeric proteins should allow the design of an S100A11 protein that favours the open conformation and allow its interaction with target proteins in a calcium-independent manner to be assessed.
Liliana Santamaria-Kisiel designed and performed experiments, analysed results and wrote the paper; Gary Shaw designed the study, analysed results and wrote the paper.
This work was supported by operating grants from the Canadian Institutes of Health Research [MOP grant number 93520] and the Canada Research Chairs Program (G.S.S.) and an Ontario Graduate Scholarship (L.S.K.).
We thank Qin Liu for maintenance of the Biomolecular NMR Facility, Lee-Ann Briere for help with sedimentation equilibrium and dynamic light scattering experiments, and Brian Dempsey and Kathy Barber for careful reading of the paper.
Abbreviations: bis-ANS, 4,4′-dianilino-1,1′-binaphthyl-5,5′disulfonic acid, dipotassium salt; DTT, dithiothreitol; GST, glutathione transferase; HSQC, heteronuclear single-quantum coherence; IPTG, isopropyl β-D-thiogalactopyranoside; rmsd, root mean square deviation; TCEP, tris-(2-carboxyethyl)phosphine
- © The Authors Journal compilation © 2011 Biochemical Society