Information transfer in the penta-EF-hand protein sorcin does not operate via the canonical structural/functional pairing. A study with site-specific mutants.

Sorcin is a typical penta-EF-hand protein that participates in Ca2+-regulated processes by translocating reversibly from cytosol to membranes, where it interacts with different target proteins in different tissues. Binding of two Ca2+/monomer triggers translocation, although EF1, EF2, and EF3 are potentially able to bind calcium at micromolar concentrations. To identify the functional pair, the conserved bidentate -Z glutamate in these EF-hands was mutated to yield E53Q-, E94A-, and E124A-sorcin, respectively. Limited structural perturbations occur only in E124A-sorcin due to involvement of Glu-124 in a network of interactions that comprise the long D helix connecting EF3 to EF2. The overall affinity for Ca2+ and for two sorcin targets, annexin VII and the ryanodine receptor, follows the order wild-type > E53Q- > E94A- > E124A-sorcin, indicating that disruption of EF3 has the largest functional impact and that disruption of EF2 and EF1 has progressively smaller effects. Based on this experimental evidence, EF3 and EF2, which are not paired in the canonical manner, are the functional EF-hands. Sorcin is proposed to be activated upon Ca2+ binding to EF3 and transmission of the conformational change at Glu-124 via the D helix to EF2 and from there to EF1 via the canonical structural/functional pairing. This mechanism may be applicable to all penta-EF-hand proteins.

the cell types where the protein is expressed constitutively, namely the ryanodine receptor (Ryr) and the ␣ 1 subunit of L-type calcium channels in muscle cells (6,7), presenilin 2 in human brain (8), and annexin VII in adrenal medulla (9), differentiating myocytes (10) and red blood cells (11).
The EF-hand, a structural motif characterized by a helixloop-helix structure, is used by a large number of proteins to bind Ca 2ϩ with high affinity. In the "canonical" motif, Ca 2ϩ is coordinated by a 12-amino-acid-long interhelical loop sequence with pentagonal bipyramidal symmetry. In this arrangement, four conserved side chains provide the five equatorial ligands (Y, Z, -Y, -Z) since the glutamate residue in position -Z establishes a bidentate interaction with the metal. The two apical ligands are furnished by the side chain of an acidic group (X) and by a water molecule (-X). Changes in the loop sequence occur such that the EF-hands may not bind Ca 2ϩ or may not be recognized in the protein primary structure. In the PEF family, the extensively modified EF4 and EF5 loops do not bind Ca 2ϩ , and the N-terminal motif designated EF1 was not recognized by sequence analysis since the loop is one amino acid shorter than the canonical one. The existence of the EF1 hand was revealed only by the x-ray crystal structure of the calpain subunits, the family prototype (1,2).
In a typical manner, EF-hands occur in pairs that are coupled structurally and functionally. The structural association takes place through a short two-stranded ␤ sheet arrangement that gives rise to stable domains. Moreover, binding of Ca 2ϩ elicits a change from a "closed" to an "open" conformation that leads to exposure of hydrophobic surfaces for target interaction. In the PEF family, all the EF-hands are clustered in the Cterminal domain that is characterized by a very similar, compact fold in all the proteins, as exemplified by the sorcin structure depicted in Fig. 1. The EF1-EF2 and EF3-EF4 pairs are associated in the canonical manner, although EF5 is unpaired in the monomer but interacts with the same motif of a second monomer in the native form of the molecule, a dimeric assemblage (2,(12)(13)(14). An intriguing peculiarity of the Ca 2ϩ binding domain is the unusual presence of two long (six turns) ␣-helices that are shared by EF2 and EF3 (D helix) and by EF4 and EF5 (G helix). At least in principle, this direct structural link between the two canonical EF-hand pairs may represent the natural means for transferring functional information between them. The compact structure of the C-terminal Ca 2ϩ binding domain contrasts the flexibility of the N-terminal domain, which is of variable length in the different PEF proteins but is always rich in glycine and proline residues. Because of its nature, the N-terminal domain is not visible in the available x-ray crystal structures of the whole proteins, namely those of sorcin (15), grancalcin (13,16), and calpain (17,18).
The present understanding of the relationships between structure and function in the PEF family is far from complete. The x-ray structures show that three EF sites are potentially able to bind Ca 2ϩ in the physiological range of concentrations, namely the unusual EF1 motif and EF2 and EF3, whose structural features resemble those of EF-hands in regulatory Ca 2ϩbinding proteins. However, binding of only two Ca 2ϩ /monomer suffices to trigger the conformational change that exposes hydrophobic regions and leads to interaction with the respective targets (3,19,20). The available data indicate that functional coupling does not involve the same EF-hands in the different proteins, suggesting that despite the high structural similarity there is no common mode of information transfer within the PEF family. Furthermore, the mechanism of communication between Ca 2ϩ -binding sites has not been elucidated. In sorcin, the characterization of a protein fragment (90 -198) lacking the first two EF-hands pointed to EF1 and EF2 as the physiological pair (19). In grancalcin, EF1 and EF3 that are not linked structurally are coupled functionally; EF2 is unable to bind Ca 2ϩ due to the presence of alanine in place of the canonical, bidentate -Z glutamate (3). In ALG-2, studies with site-specific mutants indicate that, as in grancalcin, the EF1 and EF3 motifs form the functionally relevant Ca 2ϩ binding pair (4), whereas in calpain all the first three EF hands contribute to Ca 2ϩ binding, with EF3 having the highest affinity (2).
The molecular mechanism that triggers the Ca 2ϩ -signaling process and leads to a change in the subcellular localization of PEF proteins through the specific interaction with protein targets is still obscure. The Ca 2ϩ -induced conformational changes manifest in the x-ray crystal structures are unexpectedly small and are limited to the EF1 region (1,2,13,17), and it is not known how these small changes are amplified to the extent required for target protein interaction.
In the present work sorcin was used to address these aspects of the structure-function relationships in PEF proteins. Sorcin and its isolated Ca 2ϩ binding domain have been characterized extensively in solution (19,(21)(22)(23). Moreover, a model for the Ca 2ϩ -dependent processes in the full-length protein has been proposed on the basis of the x-ray crystal structure of the Ca 2ϩ binding domain (12). Ca 2ϩ binding is thought to weaken the interactions between the N-and C-terminal domains, thereby permitting their reorientation, which in turn facilitates interaction with the target proteins at or near membranes. It is relevant in this connection that sorcin has the ability to interact with its partners by means of both the N-and C-terminal domain, as exemplified by annexin VII and the ryanodine receptor, respectively (23,19).
Site-specific mutants of EF1, EF2, and EF3 have been designed in which the canonical glutamate in the -Z position was changed into Gln or Ala. The respective mutant proteins, E53Q, E94A, E124A, have been characterized by determining their structural properties in solution and their affinity for Ca 2ϩ . Thereafter, the effect of the site-specific mutation on the capacity to interact with annexin VII and the ryanodine receptor was assessed. Substitution of the -Z glutamate in the first three EF-hands produces structural alterations of very limited extent. They can be appreciated only in the E124A mutant due to the involvement of Glu-124 in a network of hydrogen bonding and hydrophobic interactions that comprise the D helix. All the substitutions introduced affect the Ca 2ϩ binding properties of sorcin. The effect is largest when EF3 is modified and decreases in the order EF2 and EF1. These findings indicate that EF3 and EF2 represent the functional pair and suggest that binding of Ca 2ϩ to EF3 is the first step in sorcin activation. It is conceivable that Ca 2ϩ binding at the EF3 site changes the conformation of Glu-124 and that this change is transmitted to EF2 via the long D helix. The subsequent step in the pathway could occur in the conventional manner by taking advantage of the canonical structural coupling between EF2 and EF1. This unusual mode of functional linkage may be operative in all PEF proteins given their high structural similarity and, hence, may represent a general property of the family.

MATERIALS AND METHODS
Cloning of Sorcin Mutants-The cDNA of Chinese hamster ovary sorcin, kindly provided by Dr. M. B. Meyers, was amplified by PCR using two oligonucleotides, one of which was designed to generate a novel NdeI restriction site at the 5Ј end at the place of NcoI without altering the sequence within the gene. The amplified DNA thus obtained was digested with the restriction enzymes NdeI and HindIII and was inserted subsequently in a pET22 expression vector (Novagen, Madison, WI) that had been digested with the same enzymes. Sitedirected mutagenesis was carried out by the PCR overlap extension mutagenesis method according to Higuchi et al. (24) using the proofreading enzyme Pfu DNA polymerase to avoid the addition of 3Ј-overhanging residues by Taq polymerase. The following oligonucleotides were used: N forward, 5Ј-GCGAAATTAATACGACTCACTATAGGG-3Ј; C reverse, 5Ј-CAAGCTTTTAGACGGTCATGACAC-3Ј; E53Q forward, 5Ј-CAAATTGATGCTGATCAGTTGCAGAGATGTCTAAC-3Ј; E53Q reverse, 5Ј-GTTAGACATCTCTGCAACTGATCAGCATCAATTTG-3Ј; E94A forward, 5Ј-CACCATGGGATTCAATGCATTTAAAGAGCTC-3Ј; E94A reverse: 5Ј-GAGCTCTTTAAATGCATTGAATCCCATGGTG-3Ј; E124A forward, 5Ј-GGAACGGTGGATCCCCAGGCACTGCAGAAG-GCTCTG-3Ј; E124A reverse, 5Ј-CGTCAGAGCCTTCTGCAGTGCCTG-GGGATCCACCGT-3Ј. The complete mutated genes were digested with NdeI and HindIII and cloned into a pET22 expression vector.
Expression and Purification of Wild-type Sorcin and Its Mutants-Chinese hamster ovary recombinant sorcin was expressed in Escherichia coli BL21 (DE3) cells and was purified as previously described (21). The protein concentration was determined spectrophotometrically at 280 nm using a molar extinction coefficient on a monomer basis of 29,400 (21). The same purification procedure and the same extinction coefficient were used for the mutant proteins.
Circular Dichroism Spectra-CD spectra were recorded on a Jasco J-710 spectropolarimeter in the far UV (195-240 nm) and in the near UV (250 -350 nm) region. The experiments at pH 7.5 were carried out at 20°C in 5 mM Tris-HCl buffer, and those at pH 6.0 were carried out in 0.1 M sodium acetate buffer at 40°C. The ␣-helical content was calculated from the ellipticity value at 222 nm according to Chou and Fasman (25).
Determination of Ca 2ϩ Affinity-Indirect titrations were carried out in a Fluoromax spectrofluorimeter in 0.1 M Tris-HCl buffer at pH 7.5 and 25°C in the presence of the fluorescent calcium chelator Quin2, according to Bryant (26). The excitation wavelength was 339 nm (slit width, 0.5 nm); the increment of emission intensity due to calcium binding to Quin2 was followed at 492 nm. Special care was taken to reduce Ca 2ϩ contamination to 0.5-1.0 M by treating the protein solutions and the glassware with Chelex 100, as recommended by André and Linse (27). For each mutant four sets of independent experiments were carried out, which included control titrations of Quin2 alone and of native sorcin. At the end of each titration, the fluorescence intensity corresponding to zero free Ca 2ϩ concentration was determined by the addition of 5 mM EGTA. The fluorescence intensity of Quin2 at high calcium concentrations was taken as the higher asymptote. To estimate the Ca 2ϩ affinity constants of native sorcin and its mutants, each experimental set was fitted with the program CaLigator (27); the choice between one and two binding constants was made on the basis of the 2 values.
Interaction of Wild-type Sorcin and Its Mutants with Annexin VII in Overlay Assay Experiments-Aliquots of purified wild-type sorcin and of the mutants were subjected to electrophoresis on a 15% polyacrylamide gel under denaturing conditions (28) and transferred to polyvinyl difluoride membranes (PVDF, Problott, Applied Biosystems, Foster City, CA) in transfer buffer (25 mM Tris-HCl, 192 mM glycine, 20% methanol, pH 8.3) for 45 min at 100 mA (29). For Western blotting analysis, the PVDF membranes were incubated at room temperature with annexin VII (5 g/ml) in 1% gelatin in TBST buffer (20 mM Tris-HCl, 0.5 M NaCl, 0.05% Tween 20, pH 7.5) containing 10 M or 500 M CaCl 2 . Subsequently, the membranes were incubated with antiannexin VII polyclonal antibody (dilution 1:3000) in 1% gelatin in TBST buffer. The blots were developed by incubation with alkaline phosphatase conjugate monoclonal anti-mouse IgG (Sigma) in 1% gelatin in TBST. Control experiments ruled out the existence of cross-reactivity between sorcin and its mutants and the anti-annexin VII antibody (data not shown).

Interaction of Wild-type Sorcin and Its Mutants with Annexin VII in Surface Plasmon Resonance (SPR) Measurements-SPR experiments
were carried out using a BIACORE X system (Biacore AB, Uppsala, Sweden). Two N-terminal peptides of annexin VII were synthesized by Sigma-Genosys (Cambridge, UK); to improve the peptide solubility and binding to the sensor chip, three lysine residues were added to the first 20 amino acids of annexin VII (MSYPGYPPTGYPPFPGYPPA). Peptide P1 contained three lysine residues at the N terminus of the sequence, whereas peptide P2 contained three lysine residues at the C-terminal end. The peptide purity was checked by matrix-assisted laser desorption ionization time-of-flight mass spectrometry. The calculated isoelectric point of both P1 and P2 is 9.90, based on the program pI-tool (30).
The sensor chips (CM5, Biacore AB) were activated chemically by injection of 35 l of a 1:1 mixture of N-hydroxysuccinimide (50 mM) and N-ethyl-NЈ-(3-dimethylaminopropyl)carbodiimide (200 mM) at a flow rate of 5 l/min. The N-terminal peptides of annexin VII were immobilized on the activated sensor chip via amine coupling. The reaction was carried out in 20 mM sodium acetate at pH 6.0, and the remaining ester groups were blocked by injecting 1 M ethanolamine hydrochloride (35 l). In control experiments, the sensor chip was treated as described above in the absence of peptide. The interaction of the immobilized annexin VII peptides with wild-type sorcin and its mutants was detected through mass concentration-dependent changes of the refractive index on the sensor chip surface. The changes in the observed SPR signal are expressed as resonance units (RU). Typically, a response change of 1000 RU corresponds to a change in the surface concentration on the sensor chip of about 1 ng of protein per mm 2 (40). The experiments were carried out at 25°C in 10 mM HEPES, pH 7.4, 0.15 M NaCl, and 0.005% surfactant P-20 (HBS-P buffer), treated with Chelex 100 to eliminate contaminating calcium, and degassed. For the experiments as a function of Ca 2ϩ concentration, calcium chloride or EGTA were added to the buffer. Measurements were performed at a flow rate of 20 l/min with an immobilization level of the annexin VII N-terminal peptides corresponding to 200 -1200 RU. Values of the plateau signal at steady state (R eq ) were calculated from kinetic evaluation of the sensorgrams using the BIAevaluation 3.0 software. Scatchard analysis of the dependence of R eq on the concentration of wt sorcin was also performed to assess the equilibrium dissociation constant at 25 M Ca 2ϩ . The amine coupling kit, the surfactant P-20, and the CM-5 sensor chip were purchased from Biacore AB; all the other reagents were high purity grade.
Interaction of Wild-type Sorcin and Its Mutants with the Ryanodine Receptor-Aliquots of purified wild-type sorcin, EF1, EF2, and EF3 sorcin mutants were transferred to PVDF membranes as described above for the sorcin-annexin VII interaction. Incubation of the membranes with terminal cisternae vesicles from rabbit skeletal muscle enriched in Ryr (5 g/ml) was likewise carried out in TBST buffer containing 10 M or 500 M CaCl 2 as described above. The blots were incubated with an anti-Ryr monoclonal antibody (INALCO) and were developed by incubation with alkaline phosphatase conjugate monoclonal anti-mouse IgG. Control experiments ruled out the existence of cross-reactivity between sorcin and its mutants and the anti-Ryr antibody (data not shown).

RESULTS
Structural Characterization of the Site-specific Sorcin Mutants-Far and near UV CD spectroscopy was used to assess the possible occurrence of structural changes in the mutated apoproteins. The experiments were carried out at pH 7.5 and at pH 6.0 since the x-ray crystal structure of the Ca 2ϩ binding domain was obtained at the latter pH value (12). Although sorcin operates in a medium containing about 1 mM Mg 2ϩ , this cation was not included in the buffers used throughout this study since previous work showed that Mg 2ϩ has no effect on the structural and functional properties of the protein (21). The far UV CD spectra of all the site-specific mutants under study are indistinguishable from those of native sorcin both at pH 6.0 and at pH 7.5 (data not shown).
In the near UV CD region the spectrum of wild-type sorcin is negative and is characterized by two sharp bands at 262 and 268 nm attributed to phenylalanines, by a weaker band at 283 nm due to the tyrosyl fine structure (0 -0 transition), and by a fairly intense, well resolved peak at 292 nm due to tryptophan residues (22). The two tryptophan residues, Trp-99 and Trp-105, which are located on the D helix (Fig. 1, B and C), both contribute to the observed signal. However, the contribution of Trp-99 is likely to be smaller since this residue is more exposed to solvent than Trp-105 (12). The spectra of the mutant proteins are very similar to those of wild-type sorcin at both pH values with the exception of E124A, albeit only over the wavelength range 272-300 nm (Fig. 2). Thus, in the E124A mutant, the phenylalanine bands are unchanged, whereas the bands of tyrosine and tryptophan residues occur at the same wavelength as in the wild-type protein at both pH values but have a smaller amplitude. The difference is particularly significant for the tyrosine contribution.
Sorcin precipitates upon binding of 2 Ca 2ϩ /monomer in the absence of target proteins due to the exposure of hydrophobic surfaces (22). The same phenomenon was observed also for the site-specific mutants under study. Therefore, the effect of calcium on the near UV CD spectra was not studied.
Calcium Affinity of the Site-specific Sorcin Mutants-As for the wild-type protein (19), the affinity for calcium could not be determined in direct fluorescence experiments since the fluorescence intensity is practically unchanged upon binding of 2 eq of Ca 2ϩ /monomer, and higher amounts of calcium lead to precipitation.
Calcium affinity was assessed in indirect fluorescence titration experiments in the presence of the calcium chelator Quin2. The results obtained at pH 7.5 are presented in Fig. 3 in terms of the degree of saturation of Quin2 as a function of total calcium concentration. Simple inspection of the data indicates that in the mutants calcium affinity decreases with respect to the wild-type protein in the order E53Q Ͼ E94A Ͼ E124A.
An important consequence of the occurrence of precipitation upon saturation of sorcin with 2 eq of Ca 2ϩ /monomer is that higher Ca 2ϩ /protein ratios cannot be scrutinized. Hence, one cannot establish whether more than two high affinity Ca 2ϩbinding sites are present on the sorcin molecule and is forced to analyze the titration data in terms of a two-site model. In the framework of this minimum scheme and on the basis of the statistical parameters obtained from the fitting procedure, the behavior of the wild-type protein can be described with two apparent dissociation constants in the micromolar range, namely If one assumes that the site-specific mutations studied introduce local structural perturbations, the observed changes in Ca 2ϩ affinity can be taken to indicate that glutamic acid in the -Z position of the EF3 hand is more important than the corresponding residue of the EF2 and of the EF1 hands in determining the Ca 2ϩ -dependent conformational change that permits interaction with the target proteins. To prove this contention, binding of the mutant proteins with two physiological sorcin partners was assessed, namely annexin VII and Ryr, which are known to interact with the N-and C-terminal sorcin domains, respectively (23,19).

Interaction of Wild-type Sorcin and Its Site-specific Mutants with Annexin VII and the Ryanodine Receptor in Immunoblot
Experiments-In a first set of experiments wild-type sorcin and its mutants were transferred to PVDF membranes and incubated with annexin VII at pH 7.5 in buffer containing 10 or 500 M calcium. Polyclonal anti-annexin VII antisera were used to detect complex formation. At the lower calcium concentration, wild-type sorcin and the E53Q and the E94A mutant appear to interact similarly with annexin VII, whereas no interaction occurs in the case of the E124A mutant. At the higher calcium concentration (500 M) all the proteins interact with annexin VII in a similar manner (Fig. 4).
A second set of experiments was carried out to test the interaction with Ryr under the same experimental conditions. Wild-type sorcin and its mutants were blotted on PVDF membranes and incubated with Ryr-enriched terminal cisternae vesicles from rabbit skeletal muscle, and monoclonal anti-Ryr antibodies were used to detect complex formation. In accordance with the results obtained with annexin VII, at about 10 M calcium native sorcin, the E53Q and the E94A mutants form a complex with the ryanodine receptor, whereas no interaction occurs in the case of E124A. At the higher calcium concentration, sorcin and the mutants all interact to a similar extent with Ryr (Fig. 5). It should be mentioned that no inter- action of sorcin and its mutants with annexin VII or with Ryr was observed in control experiments carried out in EGTAcontaining buffer (data not shown).

Interaction of Wild-type Sorcin and Its Site-specific Mutants with the N-terminal Peptide of Annexin VII in SPR Experiments-SPR
was used to obtain quantitative information on the interaction between sorcin and annexin VII, which is known to involve the N-terminal domains of both proteins (9,23).
During the course of previous experiments, the lifetime of chips with immobilized annexin VII was found to be limited to a few days (23). To increase the chip stability, the use of a synthetic peptide corresponding to the annexin VII N terminus (amino acids 1-20) was explored. The use of the annexin N terminus has a further advantage. Because this peptide does not bind Ca 2ϩ , a simplified experimental picture is obtained by annulling the contribution of Ca 2ϩ binding to the target protein. The peptide was rendered soluble by introducing three lysine residues before amino acid 1 (peptide P1) or at the end of the sequence (peptide P2). Both P1 and P2 are immobilized efficiently on CM5 chips and react similarly with wild-type sorcin in the presence of calcium at micromolar concentrations, although P1 yields consistently higher RU values. Like immobilized full-length annexin VII (23), the immobilized peptides regain full binding activity after treatment with EGTA-containing buffer, which leads to complete dissociation of bound sorcin. Verzili et al. (23) established that at pH 7.5 and 6 M calcium the interaction of immobilized full-length annexin VII with sorcin is significant (K D ϭ 0.63 M). Under similar conditions, the immobilized P1 peptide yields apparent K D values in the order of 1 M, in good agreement with the results obtained with the full-length protein. The P1 peptide, therefore, was chosen for the systematic study of the interaction with sorcin and its mutants.
The Ca 2ϩ dependence of the interaction with the N-terminal peptide of annexin VII was assessed at constant (300 nM) concentration of native or mutant sorcin. Typically, the sensorgrams display three phases (Fig. 6); the initial increase in RU from the base line corresponds to association of sorcin or its mutants to the immobilized annexin peptide, the plateau region represents the steady-state phase of the interaction, where the rate of sorcin association to the immobilized peptide is balanced by the rate of its dissociation from the complex, and the decrease in RU corresponds to the dissociation phase during buffer flow at the end of the sample injection. Simple inspection of the sensorgrams depicted in Fig. 6, A-D, shows that in the mutants both k on and k off increase with respect to native sorcin, whereas the RU values in the plateau region and, hence, the amount of bound analyte decrease at any given calcium concentration. Moreover, for each protein the R eq values increase when the calcium concentration in the medium increases. Fig. 6E summarizes the R eq values obtained from the data depicted in Fig. 6, A-D, and in another independent set of experiments. The R eq values show that the affinity for the annexin N terminus decreases in the order wt sorcin Ͼ E53Q Ͼ E94A Ͼ E124A.
A quantitative estimate of the equilibrium dissociation constant could be obtained only for wt sorcin by measuring the dependence of the R eq values on sorcin concentration at constant Ca 2ϩ (25 M). Scatchard analysis of the data yielded an asymptotic value of about 3000 RU, corresponding to saturation of the immobilized annexin VII N-terminal peptide with wt sorcin, and an apparent K D value of about 45 nM describing this interaction (data not shown). Under these experimental conditions, the R eq values for the mutants are much lower than for wt sorcin, pointing to much higher dissociation equilibrium constants, such that the K D values could not be assessed with accuracy. However, if the same asymptotic value of 3000 RU is assumed for wt sorcin and its mutants, the data of Fig. 6E can be used to estimate the strength of the sorcin-annexin peptide interaction. In fact, the sorcin concentration that yields an R eq value of 1500, i.e. 50% saturation of the immobilized annexin peptide, corresponds to the K D value. The existence of a linkage between the immobilized peptide-sorcin equilibrium and the binding of Ca 2ϩ demands that the lower the affinity of the sorcin mutant for the immobilized annexin VII peptide, the higher the Ca 2ϩ concentration has to be to achieve the same R eq value. On this basis, an apparent K D of about 300 nM is obtained at about 18 M Ca 2ϩ for native sorcin and at increasingly higher Ca 2ϩ concentrations for the EF1, EF2, and EF3 mutants, namely 100 M for E53Q, 600 M for E94A, and about 5 mM for E124A. DISCUSSION The present data provide insight into the molecular details of the events that start with the Ca 2ϩ binding step, lead to activation of sorcin, and ultimately result in its interaction with protein targets at or near cell membranes. The behavior of the three site-specific sorcin mutants studied shows that the EF3 hand is the site endowed with the highest affinity for Ca 2ϩ . Because the structurally associated EF4 site is defective and does not bind the metal, the typical EF-hand structural/functional pairing cannot operate (32). It is conceivable that information of Ca 2ϩ binding to EF3 is transferred to the rest of the molecule by taking advantage of the specific hydrogen-bonding interactions established by the EF3 Ca 2ϩ binding loop and of the long D helix that connects EF3 to EF2. The proposed mechanism may be applicable to all members of the PEF family since these structural properties are common to all these proteins.
In sorcin, only the first three EF-hands are potentially capable of binding Ca 2ϩ at micromolar concentrations. In EF4, the loop is non-canonical (only 11 residues long), the metal-coordinating residue in position Y is lacking, and the bidentate -Z glutamic acid is substituted by the shorter aspartic acid. EF5 is likewise defective since the loop is two residues shorter than in canonical EF-hands and there is no glutamic acid in position -Z. A comparison of the sequences of EF1 through EF3 predicts that the EF3-hand should be endowed with the highest affinity for calcium due to the presence of four acidic residues in the metal coordination positions. Indeed, EF3 is the most conserved EF-hand within the PEF family in terms of Ca 2ϩ -coordinating ligands (33). The unusual EF1 site has a gap in the Y coordination position and lacks an acidic residue at position X. In EF2, which is canonical in term of length and similar in sequence to EF3, the affinity for Ca 2ϩ should be diminished relative to EF3 by the substitution of the canonical aspartate in position -X by a glycine residue. EF1 and EF2 are associated structurally through a canonical short two-stranded ␤ sheet arrangement in which Met-90 and Gly-91 establish hydrogen bonds with Gln-48 and Ile-49, respectively. It is of interest that a methionine residue (Met-90) can take the place of the highly conserved isoleucine and valine residues in position 8 of the EF2 Ca 2ϩ binding loop without disrupting the interaction with the partner EF1, at variance with observations on calmodulin mutants (34,35). However, these sequence-based arguments do not allow one to predict how different the Ca 2ϩ affinities of the first three sites will be.
To arrive at an experimental assessment of this property, the highly conserved glutamate at position -Z has been chosen for site-specific mutagenesis in the first three Ca 2ϩ binding motifs, with the expectation that removal of the bidentate coordination with the metal would abolish the capacity of the mutated site to bind Ca 2ϩ , as has been observed for site-specific m-calpain mutants (36). Accordingly, the assumption made in the interpretation of the results is that any given mutation influences the structural and functional properties of the mutated site, leaving those of the other EF motifs essentially unaltered. It follows that only those mutations involving a physiologically relevant site are expected to alter the properties of the mutant significantly relative to the native protein.
The behavior of the site-specific mutants under study supports these ideas. The structural markers used, namely the far and near UV CD spectra, indicate that in the mutant proteins both the global fold of wild-type sorcin and the environment of the aromatic residues are essentially unperturbed. The only exception concerns the environment of tyrosine and tryptophan residues in the E124A mutant as manifest in the near UV CD spectrum. In particular the tyrosine band at 283 nm has a significantly smaller amplitude than in the native protein and in the E53Q and E94A mutants (Fig. 2). The observed change can be attributed to participation of the Glu-124 carboxylate in a network of hydrogen-bonding and hydrophobic interactions that reaches Tyr-67 located on the loop connecting EF1 to EF2. In Ca 2ϩ -free sorcin the Glu-124 carboxylate is not solventexposed, as is usually the case for the -Z ligand in regulatory Ca 2ϩ -binding proteins (12,15), but is hydrogen-bonded to Phe-112, located at the D helix C terminus (Fig. 1C). This phenylalanine residue is part of an extended hydrophobic core that includes Phe-70, Phe-156, Phe-159, the two Trp residues (99 and 105) on the D helix, and the aromatic ring of Tyr-67. In turn, the Tyr-67 OH group is hydrogen bonded to Asp-113 (OD2), which occupies the X position in the EF3 Ca 2ϩ binding loop. Tyr-67, due to this hydrogen-bonding interaction and to coupling with the electronic -* transitions of the neighboring aromatic side chains, should contribute significantly to the strength of the tyrosine CD-band (37). If this holds true, disruption of the interaction network should result in a significant decrease of the band amplitude as is indeed observed in the E124A mutant. The concomitant decrease in the tryptophan band at 292 nm is of importance for its functional implications. It indicates that the loss of the hydrogen-bonding interactions established by Asp-113 and Glu-124 of the EF3 Ca 2ϩ binding loop is sensed also by Trp-99 and Trp-105, located on the long and relatively rigid D helix that connects EF3 to EF2 (Fig. 1C). The D helix, therefore, appears capable of transmitting the structural perturbations originating in the EF3 Ca 2ϩ binding loop far from the site of mutation. This role of the D helix is reminiscent of that played in calmodulin by the long Leu-69 -Phe-92 E-helix that transmits the occurrence of Ca 2ϩ binding between the two Ca 2ϩ binding domains. In calmodulin, alteration of a single hydrogen bond, like that established between glutamic acid (Glu-82) of the E helix with Tyr-138, has the potential to disrupt the structural coupling between the two domains (38), as observed for the E124A sorcin mutant.
The changes in the near UV CD spectrum just discussed are observed uniquely in the E124A mutant (Fig. 2). This fact taken together with the far-reaching effects of the Glu-124 mutation, points to EF3 as the ideal trigger of the pathway that leads from Ca 2ϩ binding to sorcin activation. A first, qualitative indication that this view is correct is provided by the close similarity of the near UV CD spectra measured in native sorcin (22) and the isolated Ca 2ϩ binding domain (data not shown) during the first stages of Ca 2ϩ titrations with that of the E124A mutant. It may be envisaged that in the native protein the first metal ion is coordinated by the EF3 Ca 2ϩ binding loop with loss of the H-bonds established by Asp-113 and Glu-124 just as in the E124A mutant. There is, however, another requirement for this picture to be true, namely EF3 should be endowed with the highest affinity for Ca 2ϩ . The fluorescence titration data in the presence of Quin2 show that it is so (Fig. 3). The overall affinity for Ca 2ϩ follows the order wild-type sorcin Ͼ E53Q Ͼ E94A Ͼ E124A, indicating that disruption of the EF3 site has the largest effect on this property of the protein. Disruption of the EF2 and EF1 sites has progressively smaller effects.
Direct evidence for the major functional role of EF3 is provided by the immunoblot and SPR experiments, which allow a measure of the interaction with physiological targets and of its Ca 2ϩ dependence. Based on the immunoblot data, the E124A mutant requires a significantly higher Ca 2ϩ concentration (500 M) than wild-type sorcin or the E53Q and E94A mutants (about 10 M) for interaction with the binding partners. Figs. 4 -5 show that there is no significant difference between annexin VII and Ryr, which interact with the N-and C-terminal sorcin domains, respectively (23,19). Importantly, all the mutants retain the capacity to be activated by Ca 2ϩ , in accordance with the fact that they are all able to translocate reversibly to E. coli membranes at millimolar concentrations of the cation, as observed during the purification procedure.
A better estimate of the Ca 2ϩ concentration required to activate the mutants is furnished by the SPR experiments that involve measurement of their interaction with the immobilized N-terminal peptide of annexin VII. This peptide provides a suitable model to monitor complex formation between the whole protein and sorcin, as indicated by the data of Brownawell and Creutz (9) and by the control experiments performed. Fig. 6 shows that at any given Ca 2ϩ concentration, the amount of complex formed decreases in the order wild-type sorcin Ͼ E53Q Ͼ E94A Ͼ E124A, which parallels their affinity for Ca 2ϩ (Fig. 3). In wt sorcin an apparent K D of 300 nM is obtained at about 18 M Ca 2ϩ . The E53Q, E94A, and E124A mutants require a 5-, 30-, and 250-fold increase in Ca 2ϩ concentration, respectively, to interact with the immobilized annexin VII peptide with similar K D values. It follows that sorcin activation at micromolar Ca 2ϩ concentrations requires preservation not only of the EF3 but also of the EF2 hand. On this basis EF3 and EF2 can be identified as the physiological pair.
The picture that emerges is that EF3 is the major player in sorcin functionality. It has the highest affinity for Ca 2ϩ and can trigger a Ca 2ϩ -dependent conformational change by means of the hydrogen-bonding interactions established by Asp-113 and Glu-124 in the Ca 2ϩ binding loop. The conformational change, although limited in extent, reaches EF2 and the EF1-EF2 loop through reorganization of the hydrophobic core packing around the D helix, which comprises Phe-95 and residues LWAVL in positions 98 -102, Trp-105, Ile-110, Phe-112, Phe-156, and Phe-159. The canonical pairing of EF2 and EF1 facilitates transfer of information to the latter site. Integrity of the whole transmission pathway is necessary for sorcin activation, as shown by the behavior of the 90 -198 sorcin fragment, which lacks EF1 and EF2. This fragment binds Ca 2ϩ with decreased affinity and is unable to translocate to membranes, a finding that suggested wrongly that EF1 and EF2 represent the physiological pair (19).
The x-ray crystal structures of native sorcin and of the sorcin binding domain support the mechanism just proposed in that the different conformations predicted by the model of sorcin activation are frozen in individual monomers. This implies that the energies involved in the Ca 2ϩ -induced rearrangements of the sorcin molecule are balanced by the lattice energies. In the structure of the whole molecule (15), the D helix and the loop connecting EF1 to EF2 have slightly different conformations in the two monomers forming the dimer. In one monomer Tyr-67 is hydrogen-bonded to Asp-113, and the D helix is rather straight, whereas in the other monomer the hydrogen bond is missing, the phenol moiety points toward the solvent, and the D helix is slightly bent. In the structure of the sorcin Ca 2ϩ binding domain (12), which has a tetramer in the asymmetric unit, monomers belonging to different dimers display relevant differences that are concentrated in the EF3 Ca 2ϩ binding loop, in the D helix, and in the loop connecting EF3 to EF4. It is of interest that the EF3-EF4 loop interacts with the N-terminal peptide in the model proposed by Ilari et al. (12), which depicts the interaction between the C-and the N-terminal domains.
The very minor conformational changes that differentiate the x-ray crystal structures of the Ca 2ϩ -free and Ca 2ϩ -bound forms of calpain (1, 2) and grancalcin (3,13) are likewise consistent with the mechanism proposed. The observed changes are limited to the EF1 region. This region is linked to the flexible N-terminal domain and, hence, must amplify the Ca 2ϩ -induced conformational changes just described to bring about the exposure of hydrophobic residues for interaction with the binding partners. As a matter of fact, the mechanism proposed may be applicable to all members of the PEF family since they share the structural/functional elements necessary for sorcin activation. Thus, EF3 is the Ca 2ϩ -binding site with the highest affinity for the metal, the -Z glutamate and Y aspartate residues in the EF3 Ca 2ϩ -binding loop are not solvent-exposed but are involved in hydrogen-bonding interactions with other amino acids, and the hydrophobic core around the D helix, whose packing rearrangement provides the driving force in the sorcin signal transmission pathway, is conserved.
At this point it appears useful to speculate on the mechanism used by PEF proteins to amplify the small Ca 2ϩ -induced conformational changes just discussed that occur in the Ca 2ϩ binding domain to achieve target protein interaction. This requires exposure of hydrophobic patches resulting from the movement of the N-and C-terminal domains relative to each other (12). Interestingly, the loop connecting EF3 to EF4 contains a stretch of conserved residues, GFRL in sorcin and GYRL in ALG-2, that were proposed to interact with the Nterminal domain in the Ca 2ϩ -free state and to become solventexposed upon Ca 2ϩ binding (12,14). One may speculate that these conserved residues play a role in the amplification of the Ca 2ϩ signal by transmitting the conformational change to the N-terminal domain, and this may be the reason for the sequence conservation.
In conclusion, the unusual functional coupling operative in sorcin appears applicable to all PEF proteins. It may be envisaged that their activation mechanism depends on subtle changes in the packing of hydrophobic residues located around the D helix that links the site endowed with the highest affinity for Ca 2ϩ to the rest of the molecule.