Calcium-dependent Conformational Changes in the 36-kDa Subunit of Intestinal Protein I Related to the Cellular 36-kDa Target of Rous Sarcoma Virus Tyrosine Kinase*

Protein I from intestinal epithelium is biochemically and immunologically related to the fibroblast 36-kDa substrate of the Rous sarcoma virus-encoded tyrosine protein kinase (Gerke and Weber (1984) EMBO J. 3, 227-233). Protein I is a Ca2+-binding protein contain- ing two copies each of a 36- and IO-kDa subunit. Denaturation/renaturation experiments show that the 36-kDa subunit is a monomer, whereas the 10-kDa subunit forms a dimer. Mixing of the subunits leads to reconstituted protein I. Physicochemical properties of protein I and its isolated subunits reveal a Ca2+-de- pendent conformational change in the 36-kDa subunit which involves the exposure of 1 or more tyrosine residues to a more aqueous environment. This change points to a Ca2+ binding constant of about lo4 M” in the presence of 2 mM M$+ and induces the ability of protein I and the 36-kDa subunit to bind in vitro to F- actin and nonerythroid spectrin. The same high Ca” requirement has been reported for the in vitro tyrosine phosphorylation of a 35-kDa protein from A-431 carcinoma cells by the epidermal growth factor receptor kinase (Fava and Cohen (1984) J. Biol. Chern. 259, 2636-2645). Here we show that this 35-kDa substrate is biochemically and immunologically related to the 36-kDa subunit of protein I, which in turn corresponds to the substrate of the its Difference in the way, different concentrations CD measurements were made on a Jobin-Lyon Mark V Autodicro-graph spectropolarimeter equipped with signal processor and a range 300 to nm. instrument calibrated with A E at 218 nm. Measurements were in path length cells with concentrations rang-ing from 0.05 subunit), and 2.6 (10-kDa subunit). The values obtained were in good agreement with protein concentration derived from amino acid analysis or calculated from the absorbance at 230 nm. Chemical cross-linking of the renatured 10-kDa subunit was performed by adding 5, 10, or 20 p1 of freshly prepared dimethyl suberimidate in 0.01 M 2-mercaptoethanol, 0.1 M Na phosphate, pH 8.0 (10 mg/ml), to 30 pl of the 10-kDa polypeptide (0.6 mg/ml) dialyzed into the same buffer. Incubation was carried out for 2 h at room temperature. The reaction was stopped by addition of 30 pl of 1 M Tris acetate, pH 8.0. After 10 min, protein was precipitated by trichloroacetic acid, washed with acetone, and dissolved in sample buffer for SDS-gel electrophoresis. Fodrin from pig brain and porcine muscle actin were isolated as before. Their CaZ+-dependent interaction with protein I and its subunits was analyzed as previously described (13).


Calcium-dependent Conformational Changes in the 36-kDa Subunit of
Intestinal Protein I Related to the Cellular 36-kDa Target of Rous Sarcoma Virus Tyrosine Kinase* (Received for publication, August 30, 1984) Volker GerkeS and Klaus Weber From the Max Planck Institute for Biophysical Chemistry, 0-3400 Goettingen, Federal Republic of Germany Protein I from intestinal epithelium is biochemically and immunologically related to the fibroblast 36-kDa substrate of the Rous sarcoma virus-encoded tyrosine protein kinase (Gerke and Weber (1984) EMBO J. 3,[227][228][229][230][231][232][233]. Protein I is a Ca2+-binding protein containing two copies each of a 36-and IO-kDa subunit. Denaturation/renaturation experiments show that the 36-kDa subunit is a monomer, whereas the 10-kDa subunit forms a dimer. Mixing of the subunits leads to reconstituted protein I. Physicochemical properties of protein I and its isolated subunits reveal a Ca2+-dependent conformational change in the 36-kDa subunit which involves the exposure of 1 or more tyrosine residues to a more aqueous environment. This change points to a Ca2+ binding constant of about lo4 M" in the presence of 2 mM M$+ and induces the ability of protein I and the 36-kDa subunit to bind in vitro to Factin and nonerythroid spectrin. The same high Ca" requirement has been reported for the in vitro tyrosine phosphorylation of a 35-kDa protein from A-431 carcinoma cells by the epidermal growth factor receptor kinase (Fava and Cohen (1984) J. Biol. Chern. 259,[2636][2637][2638][2639][2640][2641][2642][2643][2644][2645]. Here we show that this 35-kDa substrate is biochemically and immunologically related to the 36-kDa subunit of protein I, which in turn corresponds to the substrate of the Rous sarcoma virus kinase. The protein of A-43 1 cells exists not only as a monomer but also as a dimer. The latter fraction contains a 10-kDa polypeptide immunologically related to the corresponding subunit of protein I. Given past results on the A-431 system, we speculate that the monomer rather than the dimer is the preferred in vitro substrate €or the epidermal growth factor receptor kinase. Thus, the 10-kDa subunit, which induces dimerization of the phosphorylatable large subunit, may act as an inhibitor.
Avian sarcoma viruses encode transforming proteins that appear to function as tyrosine-specific protein kinases (for review see Ref. 1). Among the cellular targets of these kinases is a relatively basic protein whose polypeptide molecular weight is around 34-39,000, designated below as 36,000 (2-11). This protein which can be isolated in small amounts from chick embryo fibroblasts (4,12) is phosphorylated i n vitro by Rous sarcoma virus protein kinase (4).
We have shown that a 36-kDa subunit, immunologically related to the fibroblast 36-kDa kinase substrate, can be * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore he hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ Supported by predoctoral fellowship from Max Planck Society.
readily isolated in large quantities from porcine intestinal epithelium (13). This finding was recently confirmed by others (14,15) who demonstrated the presence of high levels of the 36-kDa tyrosine kinase substrate in intestinal epithelial cells by immunological criteria. In these highly differentiated cells, the 36-kDa polypeptide was localized by means of immunofluorescence microscopy in the terminal web that underlies the microvilli (13,15). The 36-kDa distribution in intestinal epithelial cells as well as in fibroblasts (7-11, 13, 15) is similar to that of nonerythroid spectrins (16)(17)(18)(19). When isolated from intestinal epithelium, the 36-kDa polypeptide occurs as a complex with a 10-kDa polypeptide (13). The whole complex, referred to as protein I, contains two copies of the 36-kDa subunit. Protein I binds in vitro to F-actin and nonerythroid spectrin in a Ca2+-dependent manner, once the free concentration of the bivalent cation exceeds 50 ~L M (13).
One approach to understanding the molecular mechanism of sarcoma virus transformation is to begin to study the functions of the target proteins of the tyrosine-specific viral protein kinase. In the case of the 36-kDa protein from chicken fibroblasts, this was difficult since only limited amounts can be isolated. The ability to isolate large quantities of protein I from intestinal epithelium now allows a detailed analysis. Here we characterize some biochemical properties of the 36and 10-kDa subunit of protein I in the separated state as well as in the complex. Reconstitution experiments show that the dimeric character of the 36-kDa polypeptide is only maintained in the presence of the 10-kDa subunit. The 10-kDa subunit forms a dimer under native conditions. Exposure to Ca2+ induces a conformational change both in protein I and in its 36-kDa subunit. This molecular change involves the exposure of 1 or more tyrosine residues to an aqueous environment. We show also that a 35-kDa protein, recently identified as a target of the EGF' receptor tyrosine kinase in human epidermoid carcinoma A-431 cells (20), seems closely related but is not identical to protein I.

MATERIALS AND METHODS
Isolation of Protein I and Separation of Subunits-Protein I was isolated from porcine intestinal epithelium as described previously (13). Protein I was dissociated into subunits by dialysis against 9 M urea, 100 mM NaCl, 1 mM EGTA, 2 mM NaN3,3 mM DTT, in 20 mM Tris-HC1, pH 7.5, and subjected at room temperature to gel filtration through Sephacryl S-200 (Pharmacia, 0.9 X 100 cm) equilibrated in the same buffer. Fractions containing the separated 36-and 10-kDa polypeptide were pooled and individually dialyzed at 4 "C against 100 mM NaCl, 1. stored in renaturation buffer at 4 "C for at least 3 weeks. Tryptic Peptide Analysis of IO-kDa Subunit-The 10-kDa subunit was dialyzed against 0.1 M ammonium bicarbonate, lyophilized, and oxidized with performic acid. Protein was then suspended in 0.1 M ammonium bicarbonate at a concentration of 0.5-1 mg/ml. Tosylsulfonylphenylalanyl chloromethyl ketone-treated trypsin (Serva) was added to an enzyme/substrate ratio of 2% (w/w) and digestion was performed for 6 h at 37 "C. The reaction was stopped by lyophilization. The lyophilized mixture was dissolved in pH 6.5 buffer (10% pyridine, 0.5% acetic acid), clarified by centrifugation, and applied to Whatman No. 3MM paper for electrophoresis in the same buffer. Descending chromatography was performed in the second dimension using butan-1-ol/acetic acid/pyridine/water (15:3:12:10). Peptides were visualized by staining with 0.0005% fluorescamine in acetone, eluted with pH 6.5 buffer, and dried under vacuum. Aliquots were hydrolyzed in 6 M HC1 containing 0.05% 2-mercaptoethanol for 22 h and processed for amino acid composition. Amino-terminal residues were obtained by the dansyl technique. Partial sequences were obtained by a modified Edman technique (21). Molecular Weight Estimation by Gel Chromatography-Analytical gel filtration under native conditions was performed at 4 "C in a column (0.9 X 90 cm) packed with Sephacryl S-200 (Pharmacia). The column was equilibrated in 100 mM NaC1,2 mM NaN3, 0.5 mM DTT, 1 mM EGTA, 20 mM imidazole-HC1, pH 7.4, and calibrated with different proteins of known Stokes radii. The Stokes radii of protein I subunits were calculated from a standard curve of (kd)"3 uersus Rs as in Siege1 and Monty (22), where kd is the partition coefficient. Analytical gel filtration was also performed under denaturation conditions in 0.1 M 2-mercaptoethanol, 6 M guanidine HCl, pH 6.5, using an S-200 column (Pharmacia, 0.9 X 100 cm) calibrated with different polypeptides according to Fish et al. (23). The polypeptide molecular mass of the 10-kDa subunit was obtained from a plot of (kd)'I3 versus (molecular (23). Spectroscopic Methods-UV adsorption spectra of protein I and its subunits were recorded on a Cary 118 spectrophotometer. Difference spectroscopy was performed from 250 to 330 nm using matched tacdem cuvettes (Hellma). Protein samples were dialyzed into 80 mM NaCl, 0.1 mM DTT, 2 mM MgCl,, 0.1 mM EGTA, 50 mM Tris-HC1, pH 7.5, and adjusted to a concentration of approximately 1 mg/ml with the same buffer. Identical 0.75-ml aliquots of these solutions were placed on one side of each cuvette. In the reference cell 0.75 ml of buffer plus EGTA was used in the second side of the cuvette, whereas in the sample cell 0.75 ml of buffer plus 2 mM CaC12 were present. After a base-line was obtained, the solutions were mixed well by inverting. The cuvettes were placed into the spectrophotometer, and the difference spectrum was recorded. Ca2+ titration experiments were performed in the same way, except that different free Ca2+ concentrations in the sample cell were obtained by means of balanced Ca2+/EGTA buffers (24).
CD measurements were made on a Jobin-Lyon Mark V Autodicrograph spectropolarimeter equipped with signal processor and control unit over a range from 300 to 200 nm. The instrument was calibrated with testosterone, A E = 10.7 at 218 nm. Measurements were performed in 0.2-cm path length cells with protein concentrations ranging from 0.05 to 0.4 mg/ml. Stock solutions of the protein samples were dialyzed against 10 mM NaCl, 0.1 mM DTT, 10 mM Tris-HC1, pH 7.8, diluted into 0.2 mM EGTA or 1 mM CaClz in the same buffer, and centrifuged prior to use.
Fluorescence measurements were made on an SLM 8000 S photon counting spectrofluorometer (Urbana). The excitation wavelength was either 280 or 297 nm. Each emission spectrum was scanned from 285 to 400 nm or from 300 to 400 nm. Stock solutions of protein I and its subunits were dialyzed against 80 mM NaCl, 0.1 mM DTT, 50 mM Tris-HC1, pH 7.5. Prior to fluorescence measurements, samples were diluted to a concentration of approximately 0.05 mg/ml in 0.2 mM EGTA or 1 mM CaC1, in the same buffer.
Growth of A-431 Carcinoma Cells-A-431 cells were obtained from the American Type Culture Collection and were grown in lQO-mm plastic dishes or roller bottles containing Dulbecco's modified Eagle's medium (GIBCO) supplemented with 15% fetal calf serum and nonessential amino acids (GIBCO).
Miscellaneous Procedures-Analytical ultracentrifugation studies were carried out as described before (25). Immunoblotting of proteins separated in SDS-polyacrylamide gel electrophoresis and electrophoretically transferred to nitrocellulose paper was performed according to Burnette (26). Instead of radioiodinated protein A, peroxidasecoupled swine anti-rabbit IgGs (DAKO) were used and visualized by 4-chlor-1-naphthol and H,O,. The specificity of the rabbit antibodies to protein I has been documented. These antibodies react with both subunits (13). Protein concentrations were determined from the adsorption at 280 nm. The specific absorbances at 280 nm) for protein I and its subunits were estimated from the total number of tyrosine and tryptophan residues, according to Cantor and Schimmel (27) to be 6.5 (protein I), 7.0 (36-kDa subunit), and 2.6 (10-kDa subunit). The values obtained were in good agreement with protein concentration derived from amino acid analysis or calculated from the absorbance at 230 nm. Chemical cross-linking of the renatured 10-kDa subunit was performed by adding 5, 10, or 20 p1 of freshly prepared dimethyl suberimidate in 0.01 M 2-mercaptoethanol, 0.1 M Na phosphate, pH 8.0 (10 mg/ml), to 30 pl of the 10-kDa polypeptide (0.6 mg/ml) dialyzed into the same buffer. Incubation was carried out for 2 h at room temperature. The reaction was stopped by addition of 30 pl of 1 M Tris acetate, pH 8.0. After 10 min, protein was precipitated by trichloroacetic acid, washed with acetone, and dissolved in sample buffer for SDS-gel electrophoresis. Fodrin from pig brain and porcine muscle actin were isolated as before. Their CaZ+dependent interaction with protein I and its subunits was analyzed as previously described (13).

RESULTS
Separation of the Subunits of Protein I-We have previously demonstrated that protein I from intestinal epithelial cells contains two different subunits with apparent molecular weights of 36,000 and 10,000 when analyzed by SDS-polyacrylamide gels. Given the molecular weight difference, we attempted to separate the two subunits in urea-containing buffers by gel filtration on Sephacryl S-200. No separation was obtained in 6 M urea, indicating a relatively strong interaction. However, raising the urea concentration to 9 M led to a complete dissociation of protein I, and the two polypeptides were easily separated (Fig. 1). Dialysis of the separated subunits against urea-free buffers led to renaturation as judged by the different physicochemical properties described below. Subunits were characterized by their amino acid composition. The results in Table I represent the average of several amino acid analyses and differ slightly from our preliminary calculations which were based on only one analysis (13). The lack of tryptophan in the 10-kDa subunit was confirmed by the UV absorption spectrum which is dominated by the phenylalanine bands ( Fig. 2) since the 10-kDa polypeptide contains 8 phenylalanine and only 2 tyrosine residues ( Table I). The UV spectra of protein I and the 36-kDa subunit show characteristic tryptophan and tyrosine absorptions which obscure the phenylalanine signals (Fig. 2).  actin-bound 36 kDa was 35% upon low speed and 42% upon high speed centrifugation assays. These numbers are appreciably lower than the 94 and 95% found when the same assays were performed with intact protein I (Table 11). When protein I was reconst.ituted from its isolated subunits, the F-actin binding activity reached the same high values documented previously for native protein I (Fig. 5 and Tahle 11). SDS-gel electrophoresis showed that the harvested complexes contained as expected both the 36-kDa and the 10-kDa poluypeptide. Electron micrographs of mixtures consisting of the X-kDa suhunit and F-actin showed some hundles of F-actin filaments. These were, however, thinner t.han those observed with either native or renatured protein I used at the same total concentration of the 36-kDa polypeptide (not shown). As previously demonstrated for intact protein I (13), the interaction of F-actin with the isolated 36-kDa subunit and the renatured protein I always required Ca2+ in concentrations ahove 50 p~ and was consequently suppressed in EGTAcontaining buffers (Fig. 5 ) .
The interaction o f t he isolated and renatured subunits with pig hrain fodrin was studied hy gel filtration on Sepharose 4R as descrihed before with protein I (13). Coelution of the 36-kIla suhunit and fodrin indicative of a direct interaction was rrduced to approximately 307;. of the value obtained with native or renatured protein I (Fig. 6). On the other hand, fodrin and the purified IO-kDa subunit were clearly separated without any indication o f interaction when the 36-kDa polypeptide was absent (Fig. 6). As seen previously with protein I, binding of the 36-kDn subunit required Ca" in concentra- Calcium-inducrd Conformntionnl Chnnpv of 1'rotr.in I nnd its 96-kDn Suhunit-LJV difference spectroscopy o f protrin I revealed a strong calcium-dependent change in tbr environment of aromatic amino acids (Fig. 7 A 1. Exposure t o Ca'" Ird to a decrease in ahsorhance in the range brtwren 270 and 293 nm. According to Donavan ( 2 8 ) such a change usually indicates an increased exposure of the aromatic amino acids tyrosine and tryptophan t o the aqueous solvent in t h r prrsence of the ligand. A similar Ca"-indrlcerl L I ' difference spectrum was observed with the isolntrd 36-kI)n srrhunit (Fig.  7 B ) , while the 10-kDa subunit failed t o exhihit any change (not shown). The characteristic negative spectral diffrrrncrs were used to monitor Ca" hinding in order t o (~~a l u a t r calcium binding constants of protein I and its 3fi-kI)a sr~hunit. In hoth cases, tit.ration of the Ca-binding site(s). as shnwn in Fig. 7  sine and tryptophan. To strengthen this h-ypothesis, we recorded the fluorescence emission spectra of protein I and its suhunits. As seen in Fig. 8. protein I and the 36-kDa subunit showed very similar fluorescence emission when excited at 280 nm. The two emission spectra revealed almost identical Ca"-induced changes. Addition of 1 mM Ca'" shifted the emission maximum of bot h, protein I and the 36-kDa subunit, from 328 to 320 nm. In order to evaluate the contribution of tr-yptophan to the fluorescence emission, we used an excitation wavelength of 297 nm. Llnder these conditions, the fluorescence of protein I and its larger subunit changed from 342 nm in the absence to 334 nm in t.he presence of Ca2+, clearly indicating that. t,he environment, oft he sole tryptophan residue of the 36-kDa suhunit is changed upon the binding of calcium (not. shown). The IO-kDa subunit, however, did not show Ca"+-induced changes in fluorescence emission. The emission maximum appeared at 308 nm when an excitation wavelength of 280 nm was used (Fig. 8C). Excitation at 297 nm did not cause any fluorescence, in line with the lack of t,r-yptophan in the 10-kL)a subunit ( Table I).
Rclntionship of I'rotcin I to n 35-kIIn Protein from A 4 3 1 Cclls-After we completed the characterization of protein I and its subunits, Fava and Cohen (20) described the purification of a 35-kDa protein from A-431 human epidermoid carcinoma cells, which serves as a suhst.rate for the EGF receptor tvrosine kinase. Although the isolation procedure of this protein ( 2 0 ) resemhled that of protein I (13), i.c. extraction in EGTA-containing buffers, gel filtration, DEAE-, and CM-chromatography, some differences seemed obvious. Thus, the ,4431 35-kDa protein behaved as a monomer in gel filtration and was not precipitated with an anti-34-kDa antiserum, raised against a protein of the 34-39-kDa family from chicken fihrothsts which serves as a substrate for the sarcoma virus-encoded tyrosine kinase (4). In order to compare protein kDa monomer. In addit ion, howrvcr, t hr wmc pol>T)ept irlc was also found in the dimer position. Inln~rrnr~t~lottinr: w i t h protein I antibodies exhihifed staining nf t hr :!rl-kl)a h n r l in hoth fractions, indicating that this pr)I.\-prpt idr exists in n monomeric as well as n dinlrric form (Pig. ! I ) . 'T'hr protrin I antibodies also decorated a 10-kDa polypeptide which was present only in the dimeric 35-kDa eluted from the G-100 column (Fig. 9).
The combined results suggest that A-431 cells contain a 35-kDa protein immunologically related to the 36-kDa subunit of protein I. Whereas the majority of this protein (>60%) seems to exist as monomer in A-431 cells, there is also some dimeric species which contains the 10-kDa subunit as does protein I. In contrast, in intestinal epithelial cells more than 90% of the 36-kDa subunit is recovered in the dimer and less than 10% in the monomer position. Thus, in spite of their extensive similarities (see "Discussion"), the two proteins are not identical.

DISCUSSION
Protein I is a major protein of the intestinal epithelium (13) and resembles in its immunological and many biochemical properties the 36-kDa protein which serves as a substrate for the ROUS sarcoma virus-encoded tyrosine-specific protein kinase (2-4, 7-9, 12). Although this substrate can be isolated from fibroblasts (4, 12), only the large amount of protein I available from porcine intestine allows detailed biochemical studies. Protein I of enterocytes is a tetrameric molecule containing two copies each of a 36-kDa and a 10-kDa subunit. It shows Ca2+-dependent binding to F-actin and nonerythroid spectrin in vitro (13). The physiological importance of this in vitro interaction is not understood since it occurs only a t Ca2+ concentrations exceeding 50 PM, whereas the microfilament organization of the enterocyte brush border is thought to be particularly sensitive to elevated Ca2+ levels (for review see Ref. 30). Nevertheless, as discussed below, these Ca2+ concentrations induce a pronounced conformational change in protein I and its 36-kDa subunit. They are also necessary for the in vitro phosphorylation of a 35-kDa protein of A-431 cells by the EGF receptor kinase. As described below, this protein is immunologically and biochemically related to the 36-kDa subunit of protein I.
Here we have used denaturation/renaturation experiments to characterize the biochemical properties of the two subunits. Gel filtration in the presence of 9 M urea resulted in the separation of the 36-kDa and the 10-kDa polypeptide. Dialysis against urea-free buffers led to renatured subunits. Gel filtration and sedimentation equilibrium data show that the renatured 36-kDa subunit alone is a monomer while the renatured 10-kDa subunit forms a dimer. Addition of renatured 10-kDa dimer to the 36-kDa subunits allows the reconstitution of protein I, which in all properties tested closely resembles native protein I.
Although the fibroblast 36-kDa protein was known for some time to be a dimeric molecule (4), the presence of an additional small subunit was only recently found (13,31). There is now good agreement that the small subunit is necessary to establish the dimeric character of the 36-kDa polypeptide. Erikson et al. (31) described two fractions of the 36-kDa protein (34 kDa under their gel electrophoresis conditions) upon hydroxylapatite fractionation. The minor fraction lacking the small polypeptide showed a considerably lower sedimentation value in glycerol gradient centrifugation than the major fraction which contained an associated small polypeptide. The latter complex exhibited a molecular weight of approximately 70,000, corresponding to a 36-kDa dimer. The apparent molecular masses reported for the small polypeptide associated with the 36-kDa dimer differ between 6,000 for chicken fibroblasts (31) and 10,000 for porcine enterocytes (13). We feel that this difference is most likely due to different gel conditions rather than to very variable proteins. Antibodies against the chicken fibroblast protein recognize in immunoblotting experiments the small subunit of protein I at 10,000 under our conditions (13). Similarly, antibodies against protein I immunoprecipitate a 36-and a 10-kDa subunit in extracts of rat fibroblasts. In addition, the discrepancy between the molecular weights of the small subunits is not due to the different species (avian or mammalian) because we were able to purify protein I from chicken intestine which also contained a small polypeptide migrating at 10,000 under our gel conditions (not shown).
We have now determined the molecular weight of the small subunit of protein I by two independent methods. Gel filtration in 6 M guanidine HC1 as well as amino acid composition combined with tryptic peptide analysis point to a molecular weight between 9,000 and 10,000. Under native conditions, the 10-kDa polypeptide seems to form a dimer, as judged by gel filtration and chemical cross-linking studies. The dimeric 10-kDa subunit is most likely a linker between the two 36-kDa subunits present in the intact or renatured protein I complex.
The Ca2+-dependent F-actin and fodrin-binding properties of protein I are already displayed by the isolated 36-kDa subunit, although dimerization of the 36-kDa polypeptide by adding the 10-kDa dimer enhances the binding considerably. The monomeric 36-kDa subunit not only binds but also bundles F-actin filaments. Thus, it seems that the 36-kDa subunit contains at least two actin-binding sites as a presupposition for bundling. Several other basic proteins are reported to cross-link actin filaments and induce gelation (32). Although this situation differs from bundle formation, proteins such as aldolase or lysozyme also need to have two or more actin-binding sites when their cross-linking ability is considered. In the case of the 36-kDa subunit, the total number as well as the diameter of the actin-36-kDa bundles increases markedly upon dimerization. If the 36-kDa subunit itself already contains two actin-binding sites, it seems possible that the F-actin-36-kDa equilibrium shifts towards complex formation when dimerization of the 36-kDa subunit is induced in the presence of the 10-kDa dimer.
The physicochemical properties of protein I and its subunits confirm our previous view as to the presence of a Ca2+-binding site (13). We have located such a site to the 36-kDa subunit.
Ca2+ binding of the 36-kDa subunit, intact, and reconstituted protein I results in a pronounced conformational change characterized by an altered environment of some of the aromatic amino acid residues. UV difference and fluorescence spectra indicate that tyrosine and tryptophan residues become exposed to a more aqueous environment in the presence of Ca2+. Titration experiments show that the Ca2+-induced conformational change coincides with the Ca2+-induced acquisition of binding to F-actin or fodrin. All three properties become first noticeable a t a free Ca2+ concentration of 50 pM when M$+ is present at 2 mM. Our results do not allow a conclusion as to the total number of such Ca2+-binding sites or to the absence of additional sites not resulting in a conformational change seen by difference spectroscopy.
Currently, it is premature to speculate whether one of the tyrosines of the 36-kDa subunit, which is involved in the Ca2+-dependent conformational change, is the target of the viral tyrosine kinase. However, future amino acid sequence data may be able to assess this possibility since the sole tryptophan residue of the 36-kDa subunit also seems affected by the environmental change. Our own in vitro phosphorylation studies of protein I have been hampered by the poor stoichiometry of phosphate incorporation. Although a tyrosine-specific kinase from Fujinami sarcoma virus-transformed cells (33, 34) is able to phosphorylate protein I, the degree of i n uitro phosphorylation is rather low? Whereas autophosphorylation of the Fujinami virus kinase and phosphorylation of substrates such as casein were markedly reduced in the presence of 500 PM Ca", the phosphorylation of protein I was not inhibited but indeed slightly enhanced by calcium.' This observation indicates that at least in uitro phosphorylation of protein I by the viral kinase may be enhanced by Ca2+.
Recently, Fava and Cohen (20) reported that the in uitro phosphorylation of a 35-kDa polypeptide of A-431 human epidermoid carcinoma cells by the EGF receptor kinase is regulated by Ca2+. Tyrosine phosphorylation was greatly enhanced at Ca'+ concentrations exceeding 25 PM and could under optimal conditions approach 15%. Since this high Ca2+ requirement approximately coincides with the high Ca2+ levels necessary to induce a conformational change in protein I and its 36-kDa subunit, we have performed a preliminary characterization of the 35-kDa acceptor protein of A-431 cells. This protein as well as protein I show a very similar fractionation and purification scheme, including the specific solubilization from cytoskeletal and membrane fractions by EGTA-containing buffers. In addition, the EGTA extract of A-431 cells contains a 35-and a 10-kDa polypeptide cross-reacting with protein I antibodies. By gel filtration, the 35-kDa protein from A-431 cells immunologically related to protein I was found to elute both as a monomer and a dimer. Whereas the monomer fractions are free of a cross-reacting IO-kDa polypeptide, the dimer fractions reveal the immunologically related 10-kDa species. Thus, as in protein I, the A-431 protein can form a dimer in the presence of a 10-kDa polypeptide. Interestingly, Fava and Cohen (20) monitoring their purification of the 35-kDa protein by Ca2+-dependent phosphorylation using a crude EGF receptor kinase found acceptor activity only in the monomer position. The combined data open the possibility that the 35-kDa dimer from A-431 cells is only poorly phosphorylatable in uitro by the EGF receptor kinase. In contrast, the monomeric 35-kDa polypeptide of A-431 cells, which very likely resembles the large protein I subunit, seems a much better substrate. This suggestion is supported by our preliminary experiments on in vitro phosphorylation of protein I and its 36-kDa subunit by membrane vesicles of A-431 cells, which contain EGF receptor kinase (data not shown). It remains, however, unclear why different amounts of the 35to 36-kDa polypeptides exist as dimers in different cell types. Thus, we have found more than 90% dimer in intestinal cells but only up to 40% dimer in A-431 cells. One possibility to account for this difference may lie in slightly different primary structures of the subunits in different cell types. In this respect, we note a reproducible difference in apparent molecular weight of the large subunit, i.e. 36,000 in enterocytes and 35,000 in A-431 cells.