Heat treatment could affect the biochemical properties of caldesmon.

Smooth muscle caldesmon (CaD) exhibits apparent heat stability. A widely used purification procedure of CaD involves extensive heat treatment (Bretscher, A. (1984) J. Biol. Chem. 259, 12873-12880). CaD thus purified co-sediments with actin, inhibits actomyosin ATPase activity, and interacts with Ca2+/calmodulin, similarly to the unheated protein. On the other hand, heat-treated CaD binds to actin filaments in a tether-like fashion, whereas lengthwise binding dominates in vivo (Mabuchi, K., Lin, J. J.-C., and Wang, C.-L. A. (1993) J. Muscle Res. Cell Motil. 14, 54-64), suggesting that differences do exist between heat-purified CaD and the native protein. We have isolated, without heat treatment, full-length recombinant chicken gizzard CaD overexpressed in insect cells (High-Five™) using a baculovirus expression system. We found that such unheated CaD interacts with calmodulin 10 times stronger than does the heated CaD; its inhibitory action on actomyosin ATPase is reversed by a much lesser amount of calmodulin. Moreover, electron microscopic examination indicated that actin binding at the N-terminal region is more frequent in the unheated CaD, resulting in more lengthwise binding. These findings point to the fact that CaD is not entirely heat-stable; the C-terminal CaM-binding regions and the N-terminal actin-binding region are possibly affected by heat treatment.

Caldesmon (CaD) 1 is an actin-binding protein found in smooth muscle and many non-muscle cells (for reviews, see Marston and Redwood (1991) and Matsumura and Yamashiro (1993)). It also interacts with myosin, tropomyosin, and calmodulin (CaM). In vitro CaD inhibits the actomyosin ATPase activity, and this inhibition is reversible by Ca 2ϩ /CaM. An inhibitory role of CaD in vivo was also implicated by the observation that an 18-residue CaM-binding CaD peptide (GS17C), which also binds actin in a manner similar to the intact protein, is able to induce force in a permeabilized smooth muscle cell, presumably by nudging off the endogenous CaD from its inhibitory position (Katsuyama et al., 1992). It was proposed that CaD regulates smooth muscle contraction by a troponin-like mechanism (Marston et al., 1994), but direct evidence for such a hypothesis has not been established.
CaD, when free in solution, is extremely sensitive to proteolysis. Because of its apparent heat stability, a widely used purification procedure of CaD involves extensive heat treatment (Bretscher, 1984;Lynch and Bretscher, 1986). Upon boiling for 10 -15 min most of the proteins in the tissue homogenate, including proteases, precipitate, and are thereby easily removed by centrifugation, resulting in a protein preparation that is relatively stable in the absence of Ca 2ϩ . Since the secondary structure of CaD undergoes reversible helix-coil transition upon heating (Graceffa and Jancsó, 1993;, it is generally accepted that thermally unfolded CaD would return to its "native" conformation upon cooling. CaD thus purified retains its capacity to co-sediment with actin, to interact with Ca 2ϩ /CaM, as well as the ability to inhibit the actomyosin ATPase activity. On the other hand, heat-purified CaD binds to reconstituted actin filaments in a tether-like fashion, whereas lengthwise binding appears to dominate in the native thin filament (Lehman et al., 1989;Mabuchi et al., 1993), suggesting subtle differences between heat-purified CaD and the native protein.
In this work we have isolated full-length recombinant chicken gizzard CaD from insect (High-Five TM ) cells without heat treatment. We were able to do this rather easily, probably because of the much lower proteolytic activities present in these cells. We found that such unheated CaD binds CaM with an affinity that is at least an order of magnitude higher than that of CaD purified with heat treatment. More interestingly, electron microscopy indicates stronger actin binding at the N-terminal region, in comparison to the heated CaD. This finding indicates possible existence of an additional, heatlabile, actin-binding site in the N-terminal region of CaD, and also raises the possibility that the difference previously observed in the binding modes between the native and the reconstituted systems can be explained, at least partly, by alterations of CaD by heating.

Expression and Purification of Recombinant
CaD-500 ml of High-Five TM cells (Invitrogen) at a density of 4 ϫ 10 5 cells/ml were grown in suspension in a 1-liter spinner (CYTOSTIR, Knotes) in Ex-cell 405 medium (JRH Bioscience) at 28°C for 24 h with stirring at 60 rpm. The cells were infected with pVLCaD viral stock (a gift from Dr. J. Bryan) at a multiplicity of infection of 5. The infected cells were harvested at maximal expression (48 h after infection). Cells were pelleted by centrifugation, resuspended in 50 ml of buffer A (100 mM KCl, 20 mM Tris-HCl, pH 7.5, 50 mM DTT, 0.5 mM EDTA, 0.25 mM PMSF, 5 M leupeptin), frozen and thawed twice, and centrifuged at 180,000 ϫ g for 40 min. The supernatant was, after addition of CaCl 2 to a final concentration of 2 mM, loaded onto a CaM-Sepharose 4B column. The column was washed with buffer B (100 mM KCl, 20 mM Tris-HCl, pH 7.5, 1 mM DTT, 1 mM CaCl 2 , 0.25 mM PMSF, 5 M leupeptin), followed by elution with 5 mM EDTA in buffer B. Fractions containing CaD, as determined by SDS-PAGE, were pooled and loaded onto a fast protein liquid chromatography mono Q column (Pharmacia Biotech Inc.). Proteins were eluted with a 0.1-0.6 M NaCl gradient in buffer C (0.1 mM EDTA, 20 mM Tris-HCl, pH 7.0, 1 mM DTT, 0.25 mM PMSF, 5 M leupeptin). The * This work was supported by National Institutes of Health Grant P01-AR41637. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Other Proteins-Recombinant chicken brain CaM was purified from Escherichia coli (BL21/DE3) cells by phenyl-Sepharose column chromatography (Dedman and Kaetzel, 1983). Skeletal actin was prepared from rabbit skeletal muscle according to Spudich and Watt (1971).
Fluorescence Titrations with CaM-Binding of CaM to both heated and unheated recombinant CaD was studied by adding aliquots of the CaM stock solution (190 M) to 0.6 ml of a solution containing 1-3 M CaD in 50 mM KCl, 1 mM CaCl 2 , 20 mM PIPES, pH 6.9, and monitoring the tryptophan emission ( ex ϭ 295 nm; em ϭ 320 nm) of CaD on a Perkin-Elmer MPF-4 fluorimeter. The titration data were fitted to a binding equation as described previously (Zhuang et al., 1995).
ATPase Assay-The Mg-ATPase activities of the actomyosin complex were measured as described previously (Zhuang et al., 1995). Briefly, phosphorylated chicken gizzard myosin (ϳ2 M), rabbit skeletal F-actin (12 M), chicken gizzard tropomyosin (1.7 M) were mixed in a solution containing the following reagents: 60 mM NaCl, 20 mM Tris-HCl (pH 7.5), 0.4 mM DTT, 2 mM ATP, 4 mM MgCl 2 , and 0.2 mM CaCl 2 ; 0.5-2 M heated or unheated recombinant CaD, or conventionally prepared gizzard native CaD was added to achieve inhibition, and 4 -24 M CaM was used to restore the activity. Total reaction mixture was 250 l. The reaction was allowed to proceed for 40 min at 25°C before termination with an equal volume of 10% trichloroacetic acid. After centrifugation, the amount of inorganic phosphate in the supernatant was determined (Fiske and Subbarow, 1925).
Electron Microscopy-Heated or unheated CaD (50 g/ml, or ϳ0.6 M) was first labeled with monoclonal anti-CaD (C18, 75 g/ml, a gift of Dr. J. J.-C. Lin), in a solution containing 0.1 M ammonium acetate, pH 7.0, 15% glycerol and 0.2 mM DTT. This antibody reacts with the N-terminal 28-kDa fragments and is specific to CaD (Lin et al., 1991). Labeled CaD was diluted into an F-actin solution (final concentrations: 60 nM actin, 6 nM CaD, 2 mM MgCl 2 , 0.1 mM DTT, 0.1 to 0.25 M ammonium acetate, and 30% glycerol). The proteins were adsorbed to freshly cleaved mica surface and processed for rotary shadowing by a method described previously (Mabuchi, 1991), in which the mica was floated on a protein solution to adsorb the proteins without the use of spraying. The specimens were observed with a Philips EM 300 electron microscope at 60 kV.
For a statistical analysis of CaD binding to F-actin, anti-CaD molecules on or very near actin filaments (within 20 nm, or 3 mm at a magnification of ϫ 150,000; judged as linearly bound), as well as those appeared within ϳ70 nm from the filament (tethered), were counted as "bound" CaD/anti-CaD complexes. The ratio of bound CaD to actin was estimated by tallying the number of bound CaD/anti-CaD complexes on a given length of actin filament using a value of 14 actin subunits per 37 nm of the filament. For the histogram (see Fig. 6), distances between the far edge of the anti-CaD molecules and the nearest actin filaments were measured and plotted (using KaleidaGraph, v. 3.04; Abelbeck).
During the course of this studies we noticed that the densitiy of the recombinant CaD on electron micrographs was far less than expected based on the optical density or the staining of SDS-gel bands. This discrepancy was due to the presence of large aggregates which precipitated upon incubation with anti-CaD; we also found that the use of 50 mM DTT during purification reduced aggregate formation. However, oligomerization at smaller scales was difficult to prevent, particularly for the heated preparations. To check the quality of CaD molecules such as density or aggregates, the specimen was examined at higher concentrations (0.25 M) of ammonium acetate (see Fig. 4).

RESULTS AND DISCUSSION
General Properties of Recombinant Full-length CaD-The properties of cell lines are important in works involving baculovirus expression vectors. The cell that is most commonly used with AcMNPV-based vectors is Sf-9 cell line. Full-length recombinant chicken gizzard CaD has previously been overexpressed in Sf-9 cells and purified by conventional procedures that includes a heating step; the insect cell-expressed CaD exhibited identical biochemical properties as CaD purified from gizzard (Wang et al., 1994). In this report we took advantage of the High-Five TM cell line (BTI-TN-5B1-4) that has recently become commercially available. Compared to Sf-9 cells, High Five TM cells have shorter doubling time and offer high expression levels for many recombinant proteins (Phillips et al., 1994).
The insect cell-expressed CaD, purified without heat treatment by a two-step procedure (see "Materials and Methods"), was reasonably pure on SDS-gel electrophoresis (Fig. 1), and analytical ultracentrifugation showed a single Gaussian peak with a sedimentation coefficient s 20,w 0 ϭ 2.9 S (data not shown), corresponding to a monomeric molecular mass (M app ) of ϳ88,000. Overexpressed CaD inside the insect cells appeared to be in aggregated forms, which could be broken up by treatment with high concentrations (e.g. 50 mM) of DTT in the extraction buffer and 1 mM DTT in the subsequent buffers.
Binding of CaD to CaM-CaD was originally identified as a CaM-binding protein (Sobue et al., 1981). Subsequent studies on the CaD-CaM interactions by various methods yielded an affinity at around M range (Malencik et al., 1989;Pritchard and Marston, 1989;Shirinsky et al., 1988;Smith et al., 1987). In particular, since binding is accompanied by a conformational change in CaD that increases its tryptophan fluorescence intensity, by monitoring the tryptophan fluorescence change, the binding constant was determined to be 1.8 ϫ 10 6 M Ϫ1 (Shirinsky et al., 1988). A C-terminal synthetic peptide, GS17C, also binds CaM with a similar affinity (Zhan et al., 1991). Such a weaker affinity relative to other CaM targets (e.g. MLCK, K ass ϭ 10 9 M Ϫ1 ) (see Blumenthal et al., 1985) has raised concerns whether or not there is enough CaM in the cell to allow CaM-CaD interactions to be physiologically relevant (Marston and Redwood, 1991). These concerns have led some laboratories to yet inconclusive searches for other Ca 2ϩ -binding proteins that would exhibit higher affinities for CaD (Pritchard and Marston, 1988). On the other hand, despite its lower association constant, CaM's regulatory action may still be accomplished by a relatively rapid on-rate of binding to CaD, as demonstrated by kinetic studies (Kasturi et al., 1993).
We have examined the interactions between CaM and recombinant CaD by monitoring the tryptophan fluorescence of CaD. Both the heated and unheated CaD were titrated with CaM, and the fluorescence changes were analyzed by curve fitting to a binding equation (Morris and Lehrer, 1984). The binding constant for the heated CaD was (2.3 Ϯ 0.3) ϫ 10 6 M Ϫ1 , in agreement with previous studies of heated gizzard CaD (Shirinsky et al., 1988); other parameters include the apparent stoichiometry n ϭ 0.76 Ϯ 0.08, and the overall fluorescence enhancement (F/F 0 ) max ϭ 3.05 Ϯ 0.04 (Fig. 2). The binding constant for the unheated CaD, on the other hand, was found to be (2.0 Ϯ 0.9) ϫ 10 7 M Ϫ1 (n ϭ 0.69 Ϯ 0.10; (F/F 0 ) max ϭ 2.18 Ϯ 0.03; see Fig. 2). Thus CaD purified without heat treatment interacts with CaM one order of magnitude more strongly than that prepared by boiling. The simplest explanation is that CaM-binding of CaD is weakened by heating; such an effect may result from a heat-induced, apparently irreversible, conformational change in the C-terminal portion of the CaD molecule. This would also suggest that the CaM-CaD interaction has a better chance to play a physiologically significant role.
It should be pointed out that several earlier studies also utilized CaD prepared without heating (Bretscher, 1984;Ngai and Walsh, 1984;Sobue et al., 1985), although quantitative measurements of the affinity for CaM have been lacking. The only detailed report using unheated CaD (from sheep aorta) yielded a binding affinity of 0.8 ϫ 10 6 M Ϫ1 (Smith et al., 1987); but since the reported purity of this preparation by a rather lengthy procedure was only 70 -80% (Smith and Marston, 1985), it can not be ruled out that the observed weaker CaD-CaM interaction was due to proteolysis during purification. It is known that CaD is extremely susceptible to proteolysis; in fact one of the major gains of the heat treatment was to rapidly remove the proteases, so that a better yield could be achieved. Apparently High-Five TM cells contain much less proteases than gizzard smooth muscle cells, thus allowing us to purify CaD without heating.

Effect of CaD on the Actomyosin ATPase Activity-Unheated
recombinant CaD inhibited the actin-activated ATPase activity of phosphorylated smooth muscle myosin more effectively than both the heat-treated recombinant CaD and chicken gizzard CaD (Fig. 3A). For the same amount of added CaD, the unheated protein consistently resulted in a greater inhibition than the heated proteins. The inhibition caused by either heated or unheated CaD was reversed by Ca 2ϩ /CaM, but the efficiency of such reversal appeared to be different between the two cases. A mere 1-fold excess of total CaM resulted in about 70% recovery of the inhibited ATPase activity caused by unheated CaD, whereas a 4-fold excess of CaM was needed to achieve the same level of deinhibition when the heated CaD was used (Fig. 3B), as shown by much of the earlier reports. That CaM is able to reverse the inhibition induced by unheated CaD at a lower concentration is consistent with its observed higher affinity toward the unheated CaD.
It has become apparent that the molecular mechanism involved in the inhibitory effect of CaD on the actomyosin interaction and its reversal by Ca 2ϩ /CaM is more complicated than Smooth curves were drawn only to show the trend. 100% ATPase activity corresponds to 35 nmol of P i /mg of myosin/min. the previously suggested "flip-flop" model (Sobue et al., 1982). There are experimental data indicating that even an excess amount of CaD fails to completely inhibit the smooth muscle actomyosin ATPase activity (Nomura and Sobue, 1987;Sobue et al., 1985), and that the pattern of CaM-induced deinhibition is not always parallel to the binding of CaM to CaD (Pritchard and Marston, 1989). While a more satisfactory mechanism for the deinhibition is still to emerge, the true picture may have been obscured by the varying quality of the purified CaD itself. From this study it is clear that, compared to the conventionally prepared CaD, unheated CaD is more effective in both inhib-iting the actomyosin ATPase activity and being de-inhibited by CaM, indicating that either heating itself or the cooling process following the heat treatment may cause alterations in the functional properties of CaD.
Electron Microscopic Examination of Unheated CaD Molecules-To visualize the molecular shape of unheated CaD, insect cell-expressed CaD, and F-actin were first mixed in high salt (0.25 M ammonium acetate) and examined by electron microscopy. We found that a globular structure appeared at the C terminus (arrows at the end opposite to the bound antibody; Fig.  4A) of many unheated CaD, including those isolated from chicken  (Mabuchi, 1991). The C termini of CaD molecules (arrows) are easily identified by being an opposite end to the triangular antibody-labeled end. Note that the C terminus of many unheated CaD molecules exhibits golobular structure (A) and that the presence of oligomers in heated CaD fraction (asterisks in B). Also note that at 0.25 M ammonium acetate, more heated CaD molecules appear to bind to actin filaments than unheated CaD molecules do. For comparison, native CaD (gift of Dr. P. Graceffa) was extracted from chicken gizzard by high salt and purified without the use of heating, and after antibody labeling, it was mixed with actin filaments in a solution containing 0.15 M ammonium acetate and 30% glycerol (C). Many native CaD molecules also shows globular C termini. Magnification, ϫ 100,000. gizzard (Fig. 4C, arrows), 2 but rarely seen for the heated proteins (Fig. 4B). As there was no detectable difference between heated and unheated CaD in their SDS-gel electrophoretograms, it is unlikely that such a globular structure represents another bound protein. Moreover, since the globular structure was not detected when examined by the spray method (data not shown), which is known to generate strong shearing force, it is quite possible that the C-terminal region of unheated CaD is indeed globular but flexible enough to be deformed by heating or shearing force. The change in the molecular shape of the C-terminal region of CaD may be related to the observed difference in the affinity toward CaM.
Also noted was that many unheated CaD remained unbound and only a few molecules were closely associated with the actin filament (Fig. 4A). There appeared to be more CaD/anti-CaD complexes associated with actin filaments when heated CaD was used (Fig. 4B). This apparently higher degree of binding, however, may have resulted from the fact that heated recombinant CaD easily formed oligomers (asterisks). Because of the aggregation problem, these samples were deemed not suitable for the analysis of actin-binding by electron microscopy, and CaD purified from chicken gizzard with heating was used in-2 P. Graceffa, unpublished results.

FIG. 5. Rotary shadowed electron micrographs of CaD incubated with reconstituted actin filaments.
Insect cell-expressed unheated CaD (A) or conventionally purified (heated) chicken gizzard CaD (B) was first labeled with anti-CaD and then mixed with actin filaments in a solution conatining 0.1 M ammonium acetate and 30% glycerol for visualization of complexes. It was assumed that the presence of anti-CaD on or very near the actin filaments (arrowheads, which were placed without masking any antibodies) indicates the presence of CaD molecules bound to actin filaments at their N termini (see text). The C termini of a few free CaD molecules were marked by arrows. Magnification, ϫ 75,000. stead (see below).
Electron Microscopic Examination of Heating Effect on CaD Binding to F-actin-There also appeared some differences between heated and unheated CaD in actin binding. Such differences mainly occurred in the N-terminal region of CaD. We found that the N-terminal region of unheated CaD exhibited stronger actin binding than that of the heated CaD. To quantitatively describe the difference in actin binding, we have carried out a statistical analysis of the location of the anti-N terminus antibody in the electron micrograph. As the heated, insect-expressed CaD was unsuitable for this analysis owing to oligomer formation (see above and Fig. 4), conventionally purified chicken gizzard CaD was used instead as a comparison. This is not unreasonable in view of the fact that biochemical studies indicated no differences between the two heat-treated CaD preparations (Wang et al., 1994). In doing so, we assumed that antibodies on or very near the actin filaments (Fig. 5, A and B, arrowheads) represent CaD bound at its N terminus, while those slightly away from the filaments, but within a distance corresponding to the length of a CaD molecule (ϳ70 nm), represent CaD bound at its C terminus. Examination of a total of ϳ1400 (unheated recombinant CaD) and ϳ2000 (heated gizzard CaD) antibody/CaD molecules indicated that ϳ60% of unheated CaD, but only ϳ30% of heat purified CaD, bind at its N terminus, as exemplified in Fig. 5, A and B.
To avoid subjective judgment in describing the distribution of anti-CaD, we have measured the distances between the far edge of antibody molecules and the nearest actin filaments in Fig. 5 (magnification of ϫ 150,000) and plotted these distances on histograms (Fig. 6). The results clearly demonstrate that there are more antibodies on or very near the actin filament in the unheated sample (Fig. 6B) than the heated one (Fig. 6A). The abrupt drop of antibody counts at 10 mm (corresponding to ϳ70 nm) in both CaD preparations is reasonable for the distribution of the bound CaD considering its dimension. Since the epitope of this antibody is in the N-terminal region of CaD, the higher degree of close association between the antibody and actin filament indicates that unheated CaD is likely to bind actin in a more lengthwise manner than heated CaD.
On the other hand, when compared with the heated sample (Fig. 6A), the unheated CaD preparation (Fig. 6B) exhibited a greater number of antibody molecules much farther away from the filaments, which obviously are the free CaD molecules, thus indicating a smaller fraction of bound CaD. An estimate of the stoichiometry of bound CaD to actin by counting the average number of anti-CaD molecules on a unit length of actin filament yielded 1 CaD per ϳ14 actin subunits for the heated CaD and 1 CaD per ϳ20 actin subunits for the unheated CaD. Such a difference in the stoichiometry may be explained by the lengthwise binding of the unheated CaD which results in fewer available binding sites along the actin filament. Further studies are needed to verify this point.
There have been three groups showing independently that purified CaD forms tether-like structure when it binds to actin filaments (Katayama and Ikebe, 1995;Mabuchi et al., 1993;Moody et al., 1990). However, while the two earlier reports acknowledged that tethering might be artificial, the more recent one (Katayama and Ikebe, 1995) argued that tethered binding is the only way CaD associates with actin filaments. It is known that freshly prepared thin filaments do not bundle, whereas aged samples do. Since bundling activity could result from the ability of tethered CaD to cross-link actin filaments, it is conceivable that some CaD molecules in the native thin filaments that have been handled improperly or stored too long become tethered form. Similar phenomenon may account for the observations made with the reconstituted filaments using heated CaD.
We have previously shown that both ends of heat-purified CaD could tether to actin filaments in vitro, although the majority of binding occurred at the C-terminal region (Mabuchi et al., 1993). In contrast, native thin filaments isolated from chicken gizzard did not show any sign of tethering of CaD (Mabuchi et al., 1993), indicating that the N-terminal region of CaD in situ is not projecting away from actin filaments. Therefore, it was proposed that the N-terminal portion of intact CaD may actually interact with actin filaments strongly enough to confer a lengthwise binding, instead of tethered binding as seen in the reconstituted system. This difference raises the possibility that the structure of the N-terminal region of CaD may be altered during purification, such as heat treatment. Our finding of the unheated CaD indeed supports this hypothesis. CONCLUSION We have shown that it is possible to purify CaD more readily from insect cells using the baculovirus expression system by a two-step procedure without heat treatment. Compared to the similarly prepared CaD from smooth muscle tissues, the recombinant protein offers much less aggregation, lower proteolysis, and therefore, a much higher yield. More significantly, FIG. 6. Histograms showing the distribution of anti-CaD around actin filaments. Distances between the edge of actin filaments and the far edge of anti-CaD molecules in Fig. 5 were measured (in millimeters at magnification of ϫ 150,000) and their distributions displayed in histograms. A greater number of anti-CaD molecules distribute themselves very near (less than 2.5 mm in the micrograph, or ϳ17 nm in actual length) to actin filaments in unheated CaD (B) than heated CaD (A); the antibody counts in both CaD preparations show abrupt drops at ϳ10 mm (corresponding to ϳ70 nm), a reasonable distribution for tethered CaD molecules. Also a larger population of CaD at more distant locations relative to actin filaments in unheated CaD indicates more free anti-CaD (see text).
such unheated CaD exhibits more lengthwise binding to actin filaments than heat-treated CaD, suggesting that, although the majority of the structural characteristics of CaD is heatresistant, there are certain subtle, heat-labile elements, such as an additional actin-binding site in the N-terminal region of CaD. Also affected by heating is the tertiary folding in the C-terminal domain, which confers a stronger interaction with CaM, thus allowing CaM to function as a physiological regulator for CaD's inhibitory effect. These results suggest that fulllength recombinant CaD purified without heat treatment alleviates complications owing either to improper renaturation after heating or to proteolysis when purified from tissues and not using heating, thus offering a better and more consistent quality of preparations. Furthermore, the difference previously observed in the binding modes between the native and the reconstituted systems can also be explained without invoking another protein component.