Isolation and Characterization of a New Mr 18,000 Protein with Calcium Vector Properties in Amphioxus Muscle and Identification of Its Endogenous Target Protein*

A new Ca2+-binding protein, called CaVP, has been detected in muscle of the cephalochordate amphioxus and purified to electrophoretic homogeneity. The M. 18,000 protein (PI = 4.9) binds 2 Ca2+ atoms in a noncooperative way with an intrinsic binding constant of 8.2 x 10‘ M-’. Ca2+, but not Mg2+, induces a 10% increase in a-helical content in the metal-free protein. CaVP does not interact with chlorpromazine, but forms a Ca2+-dependent complex with melittin. In situ, CaVP forms a high affinity Ca2+-dependent complex with an M. 36,000 protein present in muscle extracts of amphioxus. This complex has been purified by gel filtration and ion exchange chromatography, and the target protein further purified after dissociation of the complex in the presence of Ca2+-chelating agents and 6 M urea. The nearly pure M, 36,000 protein also forms a Ca2+-dependent complex with calmodulin which, how-ever, is less stable during electrophoresis than the CaVP-M, 36,000 protein complex. Amphioxus CaVP does not substitute for calmodulin in a specific enzyme assay nor for troponin C in restoring Ca2+ sensitivity to skinned muscle fibers. Its polyclonal antibody does not cross-react with the latter two activators. No im- munological cross-reacting counterpart of CaVP was found

A new Ca2+-binding protein, called CaVP, has been detected in muscle of the cephalochordate amphioxus and purified to electrophoretic homogeneity. The M. 18,000 protein (PI = 4.9) binds 2 Ca2+ atoms in a noncooperative way with an intrinsic binding constant of 8.2 x 10' M-'. Ca2+, but not Mg2+, induces a 10% increase in a-helical content in the metal-free protein.
CaVP does not interact with chlorpromazine, but forms a Ca2+-dependent complex with melittin. In situ, CaVP forms a high affinity Ca2+-dependent complex with an M. 36,000 protein present in muscle extracts of amphioxus. This complex has been purified by gel filtration and ion exchange chromatography, and the target protein further purified after dissociation of the complex in the presence of Ca2+-chelating agents and 6 M urea. The nearly pure M, 36,000 protein also forms a Ca2+-dependent complex with calmodulin which, however, is less stable during electrophoresis than the CaVP-M, 36,000 protein complex. Amphioxus CaVP does not substitute for calmodulin in a specific enzyme assay nor for troponin C in restoring Ca2+ sensitivity to skinned muscle fibers. Its polyclonal antibody does not cross-react with the latter two activators. No immunological cross-reacting counterpart of CaVP was found in organs of fish and rat. Its relative abundance in amphioxus muscle indicates that CaVP must underlie an important new limb of Ca2+ regulation in this particular muscle. Two types of low molecular weight acidic Ca2+-binding proteins have been found in muscles of different animal species. Type 1 is presumably freely floating in the sarcoplasm (1) and is called parvalbumin in vertebrates and SCP' in invertebrates (for review, see Ref. is still not known. The second type, type 2, of small acidic muscular Ca2+-binding proteins is more directly involved in the stimulus-contraction coupling and is mainly composed of troponin C and calmodulin. They are less abundant (10-100 mg/kg of wet muscle), have a quite low affinity for Ca2+ (Kcaz+ from lo5 to lo6 "I), and are rather well conserved in evolution. Some of these Ca2+ vectors are permanently associated with the contractile machinery; others interact in a reversible, Ca2+-dependent manner with enzymes that modulate the contractile force. Recently, it was shown that hydrophobic interactions are important for the action of these Ca2+ vectors and that they all interact with hydrophobic matrices such as immobilized phenothiazines (4), small amphiphilic peptides (5, 6) and/or with phenyl-Sepharose (7). It should be noted that some high molecular weight Ca2+-binding proteins with Ca2+ vector properties, such as calcimedins and actin-interacting factors, are also present in muscle; they typically belong to the second type and interact with hydrophobic matrices (8,9). Finally, calpains, neutral proteases involved in protein turnover in muscle (lo), also interact in a Ca2+-dependent manher with phenyl-Sepharose (11) and presumably exert their activity only after protein-protein interaction at the cellular membrane level (12).
Here, we describe the isolation and characterization of a novel 18-kDa Ca2+-binding protein from amphioxus muscle. This animal, commonly called lancet fish, belongs to the phylum of the cephalochordates, at the junction between vertebrates and invertebrates. It possesses a dorsal chord, and its morphological organization is a prototype of that of fish (13). However, the total genome of amphioxus is more related to the higher deuterostome invertebrates than to the craniates (14). Previously, we observed that type 1 Ca2+-binding proteins are present in amphioxus muscle: they show polymorphism and resemble much more the SCPs of invertebrates than parvalbumins found in vertebrates (3, 15). The new 18-kDa Ca2+-binding protein described here is apparently of the second type, although its exact role is not known yet. We tentatively called it CaVP, for Ca2+ vector protein. As part of a study aimed at complete characterization and elucidation of the function of CaVP, we report here on its purification, its most salient molecular properties, and its interaction with amphiphilic model compounds. We also report on its interaction with an endogenous 36-kDa protein.  (17). Bovine brain calmodulin and calmodulinfree phosphodiesterase were prepared as previously described (6). Gizzard muscle myosin light chain and myosin light chain kinase were prepared by the method of Hathaway and Haeberle (18) and Adelstein and Klee (19), respectively. Melittin was purified by the method of Maulet et al. (20) and immobilized as in Ref. 6. Chlorpromazine, phenylmethanesulfonyl fluoride, diisopropyl fluorophosphate, and soybean trypsin inhibitor were from Sigma Chemical Co. Leupeptin was from Peninsula Laboratories, Belmont, CA, and pepstatin A from Protein Research Foundation, Osaka, Japan. [ring-3H]Chlorpromazine was from New England Nuclear, Boston, MA. All other chemicals were of analytical grade or "suprapure."

Methods
Electrophoresis-Polyacrylamide disc gel electrophoresis was carried out according to the method of Laemmli (21) in the presence or absence of sodium dodecyl sulfate (SDS) in the upper buffer, and 1 mM EGTA or 1 mM CaCI2 in the gels and in both upper and lower buffer. Isoelectric focusing was carried out on agarose isoelectric focusing according to the method recommended by Pharmacia, Uppsala, Sweden. Complex formation of CaVP with melittin was monitored as previously described (5). Complex formation involving the purified 36-kDa target protein was carried out as follows: 100 pl of -0.2 mg/ml 36-kDa protein, dissolved in 5 mM Tris-HC1, pH 7.5, 2 M urea, 10 mM mercaptoethanol was mixed with 5 pg of the Ca2+binding protein (from concentrated stock solution) to be tested, and the samples were microdialyzed for 1 h against 10 mM mercaptoethanol in water at pH 7.5. Electrophoreses in the presence of CaC1, or EGTA were carried out as described previously (5). Gel scanning was done with a Zeiss PMQII spectrophotometer. Peak area integration was done by weighing.
Metal-binding Measurements-Binding of Ca2+ to CaVP was measured by equilibrium dialysis at 25 "C starting with the Ca2+-saturated protein. Samples (0.5 ml) containing 6 mg/ml protein were dialyzed against 100 ml of 60 mM TES-NaOH buffer, pH 7.0, 150 mM NaC1, 1 mM EGTA, 2000 cpm of "Ca/ml, pepstatin and phenylmethylsulfonyl fluoride, and increasing amounts of CaC12. After equilibrium was reached (48 h with one change of dialysis buffer), inside and outside Ca2+ concentrations were measured by flame atomic absorption (Perkin-Elmer Model 2380) and liquid scintillation counting. The protein concentration was measured spectrophotometrically using a specific extinction coefficient of of 0.75 (see below). The amount of metal bound to the protein and the free metal concentrations were calculated as previously described (22). The results obtained with atomic absorption and with '%a scintillation counting were statistically indistinguishable.
[3H]Chlorpromntine-binding Measurements-Possible complex formation between CaVP and chlorpromazine was monitored by equilibrium dialysis in the cold room. Samples (0.4 ml) containing 1 mg/ml protein were dialyzed against 100 ml of 50 mM Tris-HC1 buffer, pH 7.3, 150 mM NaC1, 15 mM mercaptoethanol, 20 pM [3H] chlorpromazine (15,000 cpm/nmol), and 100 p M CaC12 or 1 mM EGTA. Samples of calmodulin and amphioxus SCP were dialyzed in identical conditions to serve as controls. After 48 h of dialysis, inside and outside chlorpromazine concentrations were measured by liquid scintillation counting. The protein concentration was measured by the Coomassie staining method (23) with the pure proteins as standards.
Circular Dichroism-Spectra were recorded at room temperature on a Jasco J-20A spectropolarimeter with 1-nm slit as described by Cox and Stein (24). The instrument was calibrated with d-10-camphorsulfonate (Eastman, Rochester). The protein was dissolved in 50 mM PIPES, pH 7.5, 150 mM NaCl to a final concentration of 0.9 mg/ ml in a 0.05-cm cell. The mean residue M, was calculated from the amino acid sequence' and found to be 113.4.
Immunochemical Techniques-For the production of a polyclonal antiserum, a rabbit was injected subcutaneously with 0.2 mg of CaVP in 0.75 ml of 0.145 M NaC1, emulsified with 0.75 ml of Freund's complete adjuvant. Blood was processed further as described by Harboe and Ingild (25). The pure immunoglobulin fraction was passed over a column of Sepharose 4B-conjugated amphioxus SCP I1 (3)  instructions of Bio-Rad Laboratories.
In the screening experiments for the occurrence of CaVP in vertebrates, different organs (muscle, brain, kidney, heart, liver, lung) were excised either from adult rat or from goldhh, homogenized in 20 mM Tris-HC1, pH 7.5, 10 mM mercaptoethanol, 15 p~ CaC12, and the protease inhibitors as mentioned under "Results." The homogenates were centrifuged for 1 h at 45,000 X g and the supernatants subjected to immunodiffusion and Western blotting.

RESULTS
Isolation of the Ca2+-binding Protein-Extraction of the sarcoplasm of 200 animals, lyophilization, and Sephadex G-100 chromatography were carried out as described previously ( E ) , except that all buffers contained the following inhibitors: 0.1 mg/ml pepstatin A, 1 mg/liter leupeptin, 2 mg/liter diisopropyl fluorophosphate, 10 mg/liter N-tosylphenylalanine chloromethyl ketone, 70 mg/liter phenylmethanesulfonyl fluoride. The breakthrough of the column (see Fig. 1 of Kohler et al. (15)) was heated for 2 min at 70 "C, dialyzed against water containing the protease inhibitors and 10 mM mercaptoethanol, lyophilized, and passed on a DEAE-52-cellulose column equilibrated in 20 mM Tris-HC1, pH 7.5, 10 mM mercaptoethanol, 15 pM CaCl,, and the protease inhibitors as mentioned above. The elution profile ( Fig. 1) shows Ca2+ associated with two fractions: a doublet peak emerging at low conductivity and corresponding to the isoforms of SCP which were sequenced recently (3) and a peak eluting at 9 milliohms". Endogenous calmodulin eluted in a peak distinctly behind CaVP (at 12.5 milliohms") as monitored by its activator activity. CaVP was further purified by repeated Sephacryl S-200 and DEAE-52 chromatography. The purity of the final product was checked by disc gel electrophoresis in the presence and absence of CaC12 or of SDS (Fig. 2). The purification was carried out three times starting with 40 up to 1cO g of wet amphioxus muscle; the overall yield was 15 mg of pure CaVP per 100 g of muscle. Uncertainties in the recovery during the extraction and different chromatographic steps, as well as complex formation with the 36-kDaprotein (see below) FIG. 1. DEAE-52-cellulose chromatography of a partially purified protein fraction of amphioxus muscle extract. The preliminary steps consisted of muscle extraction, dialysis, and lyophilization of the supernatant, Sephadex chromatography as described in (15), heat step on the breakthrough fractions of the latter column, dialysis, and lyophilization. The DEAE-52 column (2 X 35 cm) was equilibrated in 20 mM Tris-HC1, pH 7.5, 10 mM mercaptoethanol, 15 p~ CaC12, and eluted with a linear KC1 gradient of 0-500 mM. Absorbance (-) and protein-bound Ca2+ concentration (---) are shown. The latter was obtained from determination of total Ca2+ concentration with atomic absorption minus the background Ca2+ concentration, i.e. 15 p~. The arrow represents the elution position of endogenous calmodulin. precluded a reliable estimate of the total amount of protein present.
Molecular Weight and Electrophoretic Properties-The apparent molecular weight was estimated from Sephacryl S-200 chromatography in the presence of 5 FM Ca" with the same protein markers as previously indicated (24), and equals 28,000 2,000. During polyacrylamide disc gel electrophoresis in 0.1% SDS, CaVP migrates faster in the presence of EGTA (Fig. 2) which is opposite in different other Ca2+-binding proteins (24). The apparent molecular weight equals 23,500 in the presence of 1 mM Ca'+, and 20,000 in the presence of 1 mM EDTA. Since the determination of the amino acid sequence revealed a molecular weight of 18,324,' the much higher apparent value obtained by gel filtration indicates that the protein is either highly asymmetrical, which is then reminiscent of calmodulin and troponin C (17), but not of SCPs (2), or behaves as a dimer with a slight tendency to dissociation. The electrophoretic mobility of CaVP in the absence of SDS is also calcium-dependent (Fig. 2), apparently since the Ca'+-free protein is more acidic. This behavior is very similar to that of calmodulin.
UV Spectrum, Specific Extinction Coefficient, and Isoelectric Point-The UV spectrum of pure CaVP (Fig. 3) shows, besides the main peak at 280 nm, a peak at 290 nm and a shoulder at 275 nm which are characteristic of a high content of Trp. The phenylalanine bands at 253, 260, 265, and 270 nm are also visible. The purest preparations have an AZRO ",,,/ AZs0 , , , , , ratio slightly above 2.0. The specific extinction coefficient based on weight determination by amino acid analyses, equals 7.5. The isoelectric point equals 4.9.
Ca" Content and Binding-In different final chromatographic steps of the purification procedure of CaVP, the amount of bound Ca'+/unit of optical density at 280 nm at micromolar amounts of free Ca'+ equalled 130 PM, which corresponds to 2 mol of Ca'+/mol of CaVP. Fig. 4 depicts the saturation curve of Ca'+ for CaVP and shows that the two sites have the same intrinsic affinity, Kca-= 8.2 X lo6 M", without cooperativity between the sites.
Ca2+-dependent Structural Changes-The circular dichroic spectra below 250 nm (data not shown) show an important difference in ellipticity between Ca2+-saturated and metalfree CaVP (data not shown). The mean residue ellipticity at 222 nm, [0]222 ", , , , is -10,000 degrees cm'/dmol" for Ca'+- P-form, and disordered form has been attempted from these data, but the difference spectrum is indicative of a 10% increase in the a-helical content, i.e. about 18 amino acid residues in the molecule would become incorporated in ahelical segments upon saturation of the protein by Ca'+. M e has only a minor influence on the secondary structure of CaVP ([0]222,,,,, = -7270 degrees cm'/mol"), which is an indication but not a proof that CaVP contains only so-called Ca'+-specific sites.
Near UV circular dichroic spectra and Trp fluorescence after excitation at 280 nm did not display any difference between the metal-free and Ca'+-saturated protein (results not shown). In contrast to other Trp-containing calciumbinding proteins (24,26), the Trp residues in amphioxus CaVP are not only unperturbed by divalent cations, but also show a maximal light emission at 350 nm, which is similar to that of free Trp (27). The reason for the Ca2+ insensitivity of these optical properties is linked to the fact that the aromatic chromophores tyrosine and Trp are all located in the Nterminal half of CaVP, which does not contain functional Ca2+-binding domains.2 Ca2+-dependent Complex Formation between CaVP and Chlorpromazine, and Ca VP and Melittin-Ca2+-dependent enzyme activators, such as calmodulin and troponin C, which form a complex with their target, interact with phenothiazines (28) and also characteristically form high affinity complexes with positive amphiphilic a-helical peptides, such as melittin (5,29). Parvalbumins and invertebrate SCPs do not show this behavior (4, 5). We therefore investigated the behavior of CaVP in this respect. Table I shows that, after equilibrium dialysis against 20 PM free radiolabeled chlorpromazine in the presence as well as absence of Ca2+, CaVP binds insignificant amounts of the drug; as expected, amphioxus SCP also does not bind chlorpromazine. A control experiment carried out with brain calmodulin indicates that, a t 20 PM free chlorpromazine, this protein binds 1.8 mol of the drug in a Ca2+dependent way. For the latter protein, a value of 3.3 has been reported a t 20 PM free chlorpromazine (30), but in a buffer of much lower ionic strength.
As for the interaction of CaVP with amphiphilid peptides, Fig. 5 shows that, upon electrophoresis of mixtures of CaVP and increasing ratios of melittin, one new protein band appears with a mobility distinctly lower than that of pure CaVP. Since free melittin does not enter the gels in this electrophoretic system (6), it is assumed that the newly formed band corresponds to a CaVP-melittin complex, as such a behavior was already observed previously with calmodulin and different model peptides (5,6,31). No such complex is formed in the Ca:+ + + + + ---w- absence of Ca2+ (lanes 6 and 7). At variance with the interaction between calmodulin and melittin (5), complex formation with CaVP (a) is abolished when 6 M urea is present in the gels even in the presence of Ca2+, and (b) is only completed a t a ratio of melittin/CaVP 2 2; furthermore, the zone between the band of the complex and of CaVP shows significant Coomassie staining, indicating that during the electrophoresis the complex may partly dissociate. This precludes a reliable estimation of the stoichiometry of the CaVP-melittin comlex by electrophoresis.
Lately, we found out that CaVP binds to Sepharose 4Bconjugated melittin (see Ref. 6) equilibrated in 20 mM Tris-HCl, pH 7.5, 50 mM NaCl, 10 pM CaCb and is eluted upon replacement of CaC12 by 1 mM EDTA (data not shown). This property will be exploited in future for fast and efficient purification of CaVP.
Complex Formation between CaVP and an Endogenous 36-kDa Protein-Initially during this work, the CaVP-rich fractions of the first Sephadex G-100 column (15) were found in the column breakthrough; after heat treatment, dialysis against water, and lyophilization, CaVP eluted with a much smaller Stokes radius, suggesting an irreversible disruption of an endogenous CaVP-containing complex. When the most concentrated breakthrough fractions were directly chromatographed on a Sephacryl S-200 column equilibrated in 5 PM free Ca2+ (Fig. 6), two Ca2+-containing peaks emerged soon after the void volume a t positions corresponding to relative molecular weights of -65,000 (peak b) and 28,000 (peak a), respectively. Electrophoresis in the presence of SDS showed that peak a is mainly pure CaVP (not shown), whereas peak b is composed of CaVP plus a protein band of lower mobility (Fig. 6, inset). Using marker proteins, the apparent molecular weight of the slowly migrating band was evaluated as 35,600 (36 kDa). Electrophoresis under non-denaturing conditions in the presence of Ca2+ revealed that, in peak b, no protein band corresponding to CaVP was present, but a band of low mobility (RF = 0.17) could be seen at the top of the gel (Fig.  6, tnset, lane 4 as compared to lane 3). We assume that this band corresponds to the complex between CaVP and the 36-  kDa protein. Upon electrophoresis in the presence of EGTA, the intensity of this protein band strongly decreases, whereas free CaVP becomes apparent (Fig. 6, inset, lane 2 as compared to lane I). The presence of this endogenous complex can also be demonstrated, and its purification achieved, by passing the Sephadex G-100 breakthrough fractions through a column of DEAE-52-cellulose equilibrated in 20 mM Tris, pH 7.0, 5 PM CaCI2. The elution profile (Fig. 7) shows bound Ca2+ associated with three out of the four major protein peaks. The first Ca"-binding protein elutes a t low conductivity, and electrophoreses showed that it is identical with SCP (3). The electrophoretic profiles of the second and third Ca2+-binding peaks are very similar to those of the corresponding peaks of Fig. 6: peak b is mainly composed of the complex CaVP-36-kDa protein, whereas peak a consists of pure CaVP (results not shown). Upon storage of the complex-containing fractions of Figs. 6 and 7, there is a rapid tendency for insoluble aggregates to form, which could be prevented by addition of a t least 2 M urea.
Partial Purification of the 36-kDa Protein-DEAE-52-cellulose chromatography in the presence of 1 mM EDTA and 6 M urea, under conditions otherwise identical with those of Fig. 7, leads to dissociation of the CaVP-36-kDa protein complex: the 36-kDa protein eluted a t 2.3 mmho and CaVP at 7.2 mmho (results not shown). The 36-kDa protein was further purified by repeating the DEAE-52-cellulose chromatography in 2 M urea. Fig. 8, far right, shows the electrophoretic pattern of the final product in SDS-containing gels: the protein is pure for about 90%. Its spectrum in the presence of 2 M urea is shown in Fig. 3 and is typical for a protein without a UV absorbing prosthetic group. As could be expected from its early elution from anion exchange columns, the 36-kDa protein does not enter significantly in gels in the absence of SDS: its relative mobility is 0.06 in this system (Fig. 8, lane 5 ) . Fig. 8 also shows that, in the presence of Ca2+, the 36-kDa protein forms a complex with CaVP ( l a n e I as compared to lane 2: 85% of CaVP involved in complex forn fraction no.  (15). The column (1.5 X 15 cm) was equilibrated with 20 mM Tris-HC1, pH 7.5, 5 pM CaC12, 10 mM mercaptoethanol, and the protese inhibitors as indicated under "Results," and eluted with a linear KC1 gradient (0-0.5 M). 2-ml fractions were collected. Absorbance (-) and total Caz+ concentration (---) are shown. Peak7 a and 6 were labeled in analogy to ,those of Fig. 6 and show the same electrophoretic profile as shown in Fig. 6, inset. The first Ca"-containing peak corresponds to contaminating amounts of SCP, the bulk of which was already removed in the preceding Sephadex G-100 column (3,15). mation) and to a lesser degree with brain calmodulin ( l a n e 3 as compared to lane 4: 53% of calmodulin involved in complex formation). Neither complex is formed in the presence of EGTA (lanes [6][7][8][9]. The complex itself has a mobility of 0.19 as in Fig. 6, inset, lane 4. The Ca2+-dependent interaction of the 36-kDa protein with calmodulin was confirmed by chromatography on an immobilized calmodulin-Sepharose 4B column equilibrated in 20 mM Tris, pH 7.5,lO p M CaC12: the 36-kDa protein is selectively eluted upon replacing Ca2' by 1 mM EGTA (results not shown).
Functional Properties of CaVP-Preliminary immunolocation experiments carried out by Dr. C. Heizman, Zurich, Switzerland, clearly showed that CaVP is a genuine muscle protein. Disc gel electrophoresis of mixtures of CaVP and rabbit or crayfish troponin I under conditions suitable for complex formation does not allow detection of any complex formation between the two proteins. Furthermore, tensiometric measurements performed on skinned rabbit adductor magnus fibers selectively depleted of all their troponin C by EDTA showed that CaVP is incapable of restoring tension (experiments kindly performed for us by Drs. P. K. Hoar and G. Kerrick at the University of Miami, Miami, FL). Amphioxus CaVP does not stimulate calmodulin-free phosphodiesterase and myosin light chain kinase, even at concentrations 50-fold higher than those of bovine brain calmodulin that yield half-maximal activation of the enzyme.
Immunological Properties of CaVP-Partly purified antibodies against CaVP allow unambiguous detection of the protein in crude amphioxus muscle extracts and subsequent chromatographic steps. They also react with the CaVP-36-kDa protein complex, but a much longer time is needed for the formation of the precipitation arcs (data not shown). Western blotting experiments with the antibodies showed no cross-reactivity with the two major forms of amphioxus SCPs, with rabbit skeletal troponin C, and with bovine brain calmodulin, indicating that these proteins do not possess common immunogenic determinants with CaVP. Furthermore,

A New Ca2+ Vector
Protein in Protochordates the antibodies do not cross-react with any component in extracts of different organs from rat and fish. Hence, if this apparently new Ca2+ vector protein exists in vertebrates, its concentration must be much lower, or its immunogenic profile is quite different from the protein in invertebrates.

DISCUSSION
A new 18-kDa calcium-binding protein was isolated from amphioxus muscle using classical purification procedures, and the electrophoretically pure product was characterized. Structurally, the protein is quite different from the other wellknown or newly discovered Ca2+-binding proteins; this aspect will be more specifically examined in a forthcoming study.2 CaVP binds 2 Ca2+ atoms with the rather low dissociation constant of 1.2 X M. At present, it is not clear whether these Ca2+-binding sites can also accommodate Mg2+ (socalled Ca2+-Mp2+ mixed sites, see Ref. 32), or whether the sites are specific for Ca2+. The lack of significant changes in a-helical content upon addition of M e to the metal-free protein is reminiscent of the behavior of calmodulin, in which high Mg2+ concentrations decrease moderately the affinity for Ca2+ according to a non-competitive at tern.^ A detailed study on divalent ion interaction with CaVP is needed to settle the question of how responsive CaVP is to a Ca2+ transient which in muscle is thought to go from -100 nM (resting muscle) to -2 PM (contracting muscle). Ca2+ binding, even in the presence of 2 mM free [Mg2+], leads to an important reorganization of the secondary structure, which may be instrumental for its interaction with cellular components.
Like calmodulin and troponin C, CaVP interacts with melittin in a Ca2+-dependent manner, but the interaction CaVPmelittin does not display the same characteristics as that of the former proteins with melittin: in electrophoresis experiments, the CaVP-melittin complex is not fully stable; in the presence of urea, the complex is not formed at all; finally, CaVP is released from immobilized melittin by 1 M NaClcontaining buffers even in the presence of Ca2+ (data not shown). The fact that the newly isolated protein possesses Ca2+ vector characteristics is most convincingly proven by the isolation of its complex with the 36-kDa endogenous protein. The latter complex is stable during the purification steps in the presence of micromolar concentrations of Ca2+. Chelation of Ca2+ by EGTA leads to full dissociation of the complex and to precipitation of the 36-kDa protein. The latter phenomenon can be prevented by addition of 2 M urea. Based on this information, we have developed two different chromatographic procedures for the purification of the 36-kDa protein to above 90% electrophoretic purity. Preliminary experiments revealed that in the near future further purification will be possible using immobilized calmodulin chromatography. In the present study, about 0.5 mg of 90% pure 36-kDa protein could be obtained starting with 100 animals; it is anticipated that the yield may be significantly better when chromatographic procedures are modified to cope with our present knowledge, since upon SDS electrophoresis on the whole sarcoplasm of amphioxus, the Coomassie stain intensity of the 36-kDa protein band is roughly equal to that of CaVP (not shown).
Although CaVP has many characteristics of an information transducer in muscle, no information could be gathered about its function in the animal. Functional assays for calmodulin or troponin C activity were clearly negative. A more detailed knowledge of the 36-kDa target protein as well as immunolocalization experiments with our specific CaVP antibodies should help us to clarify this point. One of the difficulties encountered during the search on the importance of this potentially new signal transducer resides in the fact that amphioxus is becoming rare and its supply is very limited. Therefore, we have screened vertebrate organs for immunological cross-reaction with polyclonal antibodies against amphioxus CaVP, but the outcome was negative. Nonetheless, the immunogenic determinants of the protein might be poorly conserved in distant phyla.
The relative abundance of CaVP, which is comparable to that of calmodulin in vertebrate brain, and of its endogenous target protein, as well as the Ca2+ sensitivity of CaVP and of its interaction with the 36-kDa target indicate that this may be a new limb of Ca2+ control of the contraction-relaxation cycle in amphioxus, in addition to the existing troponin C limb of control. Although amphioxus muscle is quite unusual with respect to excitation-contraction coupling (33), it is likely that the above described Ca2+ vector system is also present in other animals.