four times in 40 vol of 10mM Tris-HCl (ph 7.4) at 00C. ATPinduced under isotonic conditions (12). A spectrin-actin complex was

Transcription

1 Proc. Natl. Acad. Sci. SA Vol. 77, No. 6, pp , June 198 Biochemistry Phosphorylation and dephosphorylation of spectrin from human erythrocyte ghosts under physiological conditions: Autocatalysis rather than reaction with separate kinase and phosphatase (autophosphorylation/casein kinase/spectrin preparation/endocytotic inside-out vesicles/photoaffinity labeling) BEAT A. IMHOF, HANS J. ACHA-ORBEA, TOWIA A. LIBERMANN, BERNHARD F. X. REBER, JAKOB H. LANZ, KASPAR H. WINTERHALTER, AND WALTER BIRCHMEIER Laboratorium fur Biochemie, Federal Institute of Technology (ETH-Z), CH-892 Zurich, Switzerland Communicated by V. Prelog, March 17,198 ABSTRACT The mechanism of phosphorylation and dephosphorylation of spectrin from human erythrocyte membranes has been examined under closely physiological conditions. The results suprt the hypothesis tat spectrin is an autophosphorylating and dephosphorylating system. (i) Extraction from ghosts of up to 85% of the kinase (casein kinase) suggested to catalyze the reaction [see Fairbanks, G., Avruch, J, Dino, E. J. & Patel, V. P. (1978)J. SupramoL Struct 9, J only slightly reduced spectrin component 2 phosphorylation and did not affect ATP-induced changes in the ghosts' shapes. (il) A spectrin-actin complex isolated from endocytotic inside-out vesicles under hypertonic conditions contained virtually no casein kinase activity and still exhibited a largely intact phosphorylation machinery. (iii) Photoaffinity labeling experiments indicated that spectrin component 2 fulfills the necessary prerequisite of the hypothesis-i.e., it contains its own An-binding site. (iv) nder various conditions, spectrin phosphorylation and dephosphorylation seem to be tightly coupled. The implications of these findings for the understanding of spectrin function and the maintenance of erythrocyte shape are discussed. The erythrocyte spectrin-actin complex, the "red cell cytoskeleton," has been identified as an actomyosin-like system linked to the cytoplasmic face of the lipid bilayer membrane (1-3). Its reversible phosphorylation is of critical importance for the maintenance of the disc shape of the cell (4, 5). In erythrocyte ghosts, phosphorylation reactions have been studied under a wide variety of conditions (see ref. 6 for a review). (i) At physiological ATP concentration (- 2 mm) and isotonicity, spectrin component 2 is stoichiometrically and -reversibly phosphorylated and represents the main radioactively labeled species of the ghost membrane (4). (ii) At micromolar ATP concentration and isotonic conditions, spectrin is only minimally phosphorylated, and other components become labeled to a similar extent. (iii) Phosphorylation under hypotonic conditions needs cyclic AMP, and proteins other than spectrin are modified. Fairbanks and coworkers (6-9) and Hosey and Tao (1, 11) have studied phosphorylations under the latter two conditions and found that they are catalyzed by two types of enzyme(s), one with a casein kinase-like (cyclic AMP-independent) activity and the other with a histone kinase-like (cyclic AMP-dependent) activity, respectively. No specific information, however, is available about the mechanism of component 2 phosphorylation under closely physiological (i.e., shape change) conditions (4). The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18. S. C solely to indicate this fact EXPERIMENTAL PROCEDRES Erythrocyte Membrane Preparations and Isolation of Two Different Spectrin-Actin Complexes. Ghosts from human erythrocytes were prepared as described (4, 12) and washed four times in 4 vol of 1mM Tris-HCl (ph 7.4) at C. ATPinduced inside-out vesicles were obtained from these ghosts under isotonic conditions (12). A spectrin-actin complex was purified by: (i) preextraction of inside-out vesicles (5-1 mg of membrane protein per ml) at C for 1 hr with 2 vol of 1 M NaCl/1 mm phenylmethylsulfonyl fluoride/7.1 mm 2- mercaptoethanol/1 mm Tris-HCl, ph 7.4, followed by centrifugation in a Sorvall SS-34 rotor for 2 min at 18, rpm; and (ii) extraction at 37C for 2 min with 4 vol of the same buffer, followed by centrifugation as described. Hypotonic spectrin-actin was extracted from ghosts in.1 mm EDTA (13) and concentrated against Aquacide III. Elution and Determination of Casein Kinase Activity. For elution of the enzyme, washed packed ghosts were incubated for the times indicated with.5 M NaCl/1.4 mm 2-mercaptoethanol on ice (9). This was followed by lysis in 4 vol of 1 mm Tris-HCI (ph 7.4) and centrifugation as described above. Casein kinase activity was assayed in.3 mg of casein per ml, 1 mm NaCl, 1 mm Tris-HCI,.4% Triton X-1, 4 mm MgSO4,5 um ["y-32p]atp, and 4.ul of packed ghosts per ml or the equivalent volume of extracted membranes or supernatant, all at ph 7.4. After incubation for 2 min at 37C, the protein was precipitated with a 1-fold volume of 1.5% perchloric acid/2.5 mm ATP/5 mm MgSO4/2 mm sodium pyrophosphate/1 mm sodium phosphate. The mixture was filtered on Millipore filters and radioactivity was measured (9). Phosphorylation of Erythrocyte Membranes and Isolated Spectrin-Actin Complexes. Fresh, packed ghosts and those extracted with.5 M NaCl were incubated with 2 mm ['y- 32P]MgATP/14 mm NaCl/2 mm KCI/1 mm Tris-HCl, ph 7.4, at 37C (4). Endocytotic inside-out vesicles (12), at 1 mg of membrane protein per ml, and the two different spectrinactin preparations, at.3 mg of protein per ml, were phosphorylated under similar conditions. The labeled samples were analyzed by NaDodSO4 gel electrophoresis as described (4, 12) Ṗhotoaffinity Labeling of Erythrocyte Ghosts with MgATP. Packed fresh ghosts were incubated with 7 MM [a- 32P]ATP/14 mm NaCl/2 mm KCI/2 mm MgSO4/1.4 mm 2-mercaptoethanol/1 mm Tris1HCl, ph 7.4, on ice and illuminated under argon atmosphere with a Philips TV 15 W lamp (254 nm) from a distance of less than 1.5 cm (14, 15) followed by NaDodSO' gel electrophoresis.

2 Biochemistry: Imhof et al. Table 1. Extraction of casein kinase activity from erythrocyte ghosts with.5 M NaCl at C followed by hypotonic.. lysis and centrifugation Activity in ghost pellets, Conditions % of total Incubation in 1 mm Tris-HCl (ph 7.4) (2 min) 87.4 Extraction with.5 M NaCl 1 min min min 27. Extraction twice with.5 M NaCl* (2 min) 15.9 * After 2 min of first extraction followed by hypotonic lysis. RESLTS Effect of Casein Kinase Elution upon ATP-Induced Shape Changes and Spectrin Phosphorylation in Erythrocyte Ghosts. Extraction of human erythrocyte ghosts with.5 M NaCl at C resulted in the progressive elution of up to 85% of the membrane-associated casein kinase activity (Table 1; see also refs. 6 and 9). However, extraction of the bulk of this enzyme from the membrane did not significantly affect rate and extent of ATP-induced disc cell formation (Fig. 1) and decreased concomitant spectrin component 2 phosphorylation by only 35-45% (Fig. 2).* Phosphorylation Capacity and Casein Kinase Activity in Preparations of Isolated Spectrin-Actin Complexes. Preextraction of endocytotic inside-out vesicles (12) with 1 M NaCl at C resulted in a quantitative release of casein kinase activity (Table 2). At 37C, however, a virtually casein kinase-free spectrin-actin complex was eluted from the preextracted vesicles. Incubation of this hypertonically extracted spectrin-actin complex with [y-32p]mgatp under shape change conditions resulted in the specific phosphorylation of component 2 (Fig. 3, slots c). "Water-extracted" spectrin-actin (13), however, exhibited little phosphorylation capacity and, furthermore, both subunits were modified (Fig. 3, slots b). The kinetics of phosphate incorporation into the new spectrin-actin complex were similar to those obtained with intact endocytotic inside-out vesicles (12) and intact ghosts (4)-i.e., a maximum was quickly reached at 15 min of incubation (Fig. 4). At the different ATP concentrations tested, a molar ratio of phosphate to component 2 of up to 6% of that of salt-treated membranes (which was 4% compared to fresh ghosts) could be reached. * The reduction of the spectrin phosphorylation capacity by --35% after treatment with.5 M NaCI (Fig. 2) is likely due to a nonspecific salt effect; it could even be observed after less than 1 min of extraction-i.e., before casein kinase was eluted to a significant extent (Table 1). It could further be interpreted as a decrease in turnover because dephosphorylation was slightly reduced as well (see Fig. 6). Table 2. Extraction of a spectrin-actin complex free of casein kinase activity from endocytotic inside-out vesicles of erythrocyte membrane at 1 M NaCl Preextraction Extraction Components released at C, % at 37C, %* Spectrin components 1 and 2t Casein kinase activity The values determined with intact vesicles are taken as 1%. * After removal of the C extract. t From scanning of Coomassie blue-stained material after NaDodSO4 gel electrophoresis. Proc. Nati. Acad. Sci. SA 77 (198) ~ 8 4. c 4 - E FIG. 1. ATP-induced conversion of crenated erythrocyte ghosts to the disc form: Effect of preextraction with.5 M NaCl at C. After 2 min of preextraction (Table 1; control cells were kept under hypotonic conditions on ice) the ghosts were lysed. This was followed by a second incubation with 2 mm MgATP under shape change conditions at 37 C (see Experimental Procedures and ref. 4). Disc cell formation was examined by phase-contrast microscopy. At least 3 ghosts were counted on photographs to obtain a data point., Not extracted; *, extracted. V Light-Induced Photoaffinity Labeling of Erythrocyte Ghosts with MgATP. Erythrocyte membranes incubated under shape change conditions with [a-32p]mgatp and illuminated at 254 nm followed by NaDodSO4 gel electrophoresis yielded mainly two labeled proteins (Fig. 5, slots a): spectrin component 2 and a protein migrating between components 4 and 5. Two proteins of the erythrocyte membrane that were expected to bind ATP-i.e., actin (component S; refs. 2 and 3) and glyceraldehyde 6-phosphate dehydrogenase (component 6; ref. 16)-were in fact weakly modified. The main protein of the erythrocyte membrane, band 3, however, was not labeled. Furthermore, the additional presence of adenylyl imidodiphosphate significantly reduced the incorporation of radioactive ATP into the two main labeled species (Fig. 5, slots b). cs Q8 o 6 E. E C o E. M.2. FIG. 2. Kinetics of ATP-induced phosphorylation of spectrin component 2 in erythrocyte ghosts: Effect of one cycle (2 min) or two cycles (twice for 2 min each) of preextraction with.5 M NaCl at C. Ghosts were incubated with 2 mm [y-32p]mgatp. Triton X-1 (.5%) was added in order to ensure full accessibility of ATP to the phosphorylation sites. The detergent slightly increased the overall phosphorylation of component 2 (4). Similar results were obtained, however, in the absence of the detergent. The same amount of protein was placed in each gel slot. Scanning of Coomassie blue-stained gels (12) indicated that extraction with NaCl did not significantly change the relative amounts of the five major membrane proteins. Component 6 was lost during extraction (6)., Not treated; *, one cycle of preextraction; *, two cycles of preextraction.

3 3266 Biochemistry: Imhof et al. Proc. Natl. Acad. Sci. SA 77 (198) #p db. 46 Alb., I w VW b,s FIG. 3. ATP-induced phosphorylation of isolated spectrinactin complexes: Comparison with intact endocytotic inside-out vesicles. [y-32p]mgatp was present at.75 mm; the time of incubation was 15 min under isotonic conditions. (Left) Coomassie bluestained gel; (Right) autoradiogram. Slots a, endocytotic insideout vesicles; slots b, spectrin-actin extracted with.1 mm EDTA (13); slots c, new spectrin-actin complex extracted from endocytotic inside-out vesicles (12) under hypertonic conditions. Radioactivity on top of the gel was observed depending upon the batch of labeled ATP used. However, no protein comigrated with this probably nonspecific material. In a reevaluation of the photoaffinity labeling method, we also tested well-characterized, soluble proteins for ATP binding. We found that skeletal muscle actin, the heavy chain of skeletal muscle myosin, human hemoglobin, and pyruvate kinase gave positive reactions and pancreatic DNase I, brain tubulin, and aspartate aminotransferase from pig heart gave negative reactions. Tubulin, however, was positive for GTP binding, whereas actin was negative (data not shown). Analysis of Turnover of Spectrin-Bound Phosphate. In isotonic ghosts prephosphorylated on spectrin component 2 the bulk of spectrin-bound label is hydrolyzed within less than 2 min at 37 C (4). This dephosphorylation reaction was only slightly affected by previous extraction with.5 M NaCI (Fig. 6). Turnover of spectrin phosphate under isotonic conditions was further examined with endocytotic inside-out vesicles. If -r I- FIG. 5. Affinity labeling of erythrocyte ghosts with [a-32p]- MgATP. Ghost-ATP complexes were illuminated at 254 nm and C. This was followed by Na- DodSO4 gel electrophoresis. Autoradiograms of incubations for 45 and 12 min are shown. Slots a, radioactive ATP alone; slots b, addition of 2 mm adenylyl imidodiphosphate. Radioactivity on top of the gel is probably due to some labeled protein crosslinked during illumination. spectrin component 2 was prephosphorylated in these vesicles with unlabeled 2 mm MgATP for 2 min, further phosphorylation, as analyzed by addition of radioactive tracer ATP, was significantly suppressed (Fig. 7A; see also ref. 12). Similarly, the rate of dephosphorylation was also reduced under these conditions (Fig. 7B). Furthermore, ADP suppressed both residual phosphorylation and dephosphorylation and thus did not seem to act as an acceptor of spectrin-bound phosphate. c6 t.2. C hypertonic spectrin 4w c. E E Q ~~~~hypotonic spectrin FIG. 4. Kinetics of phosphorylation of the two isolated spectrin-actin complexes. Incubations were both isotonic and as in Fig. 3, except that 2 mm [7y-32P]MgATP was present. Incorporation of radioactivity into component 2, as measured after NaDodSO4 gel electrophoresis, is given. n FIG. 6. Dephosphorylation of spectrin component 2 of erythrocyte ghosts preextracted with.5 M NaCl. Fresh ghosts were first labeled for 1 min with [y-32p]mgatp under shape change conditions at 37C followed by hypotonic lysis. Casein kinase activity was extracted with.5 M NaCl at C for 1 min (A) or 6 min (). (, Not extracted.) The final incubation, as plotted in the graph, was at 37 C under isotonic conditions and in the presence of 5 mm MgSO4. Radioactivity incorporated into component 2 was analyzed after Na- DodSO4 gel electrophoresis (4, 12).

4 Biochemistry: e Q3 B E C. ue Cx C I 2 E.1 Imhof et al FIG. 7. Turnover of spectrin-bound phosphate in endocytotic inside-out vesicles. (A) Vesicles isolated under isotonic conditions (12) and containing an ATP-regenerating system (2 mm phosphoenolpyruvate and 4,ug of pyruvate kinase per ml) were first preincubated for 2 min with unlabeled 2 mm MgATP at 37C. At zero time radioactive tracer [y-32p]atp (and 5 mm ADP as indicated) were added and incubation was continued at 37C. (B) Vesicles as above were preincubated for 2 min with 2 mm [y-32p]mgatp at 37C, followed by washing twice with 14 mm NaClI1 mm Tris-HCl, ph 7.4, on ice. At zero time incubation was continued at 37C with 5 mm MgSO4 in the presence and absence of 5 mm ADP. Radioactivity of component 2 in A and B was analyzed after NaDodSO4 gel electrophoresis (4, 12). DISCSSION The present investigation addresses the question of possible mechanisms of spectrin phosphorylation and dephosphorylation in human erythrocyte ghosts occurring under closely physiological conditions. The known features of these reactions can be summarized as follows. (i) Spectrin component 2 is stoichiometrically phosphorylated. Thus, at 2 mm MgATP, molar ratios of phosphate to protein of.4-.8 are easily obtained (ref. 4 and Fig. 2). (ii) Physiological spectrin phosphorylation is specific.t The phosphorylated residue could be identified as phosphoserine and has been further localized on a 19,-dalton CNBr fragment, on a 48-dalton tryptic peptide, and on an 8-dalton fragment located at either the NH2 or COOH terminus of the component 2 molecule (4, 19, 2). (iii) Spectrin phosphorylation is accompanied by continuous dephosphorylation (ref. 4 and Fig. 6).t The actual steady-state level of phosphorylated spectrin is thus regulated by a combination of these reactions, the rates of both being dependent on the concentration of MgATP (for further details see ref. 4 and below). Specific and reversible phosphorylation of spectrin component 2, (i.e., of a molecule with nearly 2 amino acid residues) t It has been determined that more than one phosphorylated site exists on spectrin component 2 (17, 18). In the present study, however, we confine ourselves to the one specific site exhibiting rapid turnover and thus becoming stoichiometrically labeled within the time of shape change experiments in ghosts (i.e., within less than 3 min; see ref. 4). It was claimed that this activity is due to a cytoplasmic enzyme (21). The cytoplasmic phosphatase, however, although highly active on isolated spectrin, does not seem to catalyze spectrin dephosphorylation in situ (6). D Proc. Natl. Acad. Sci. SA 77 (198) 3267 appears to be catalyzed by constituents that are intrinsic to the erythrocyte membrane and, furthermore, probably belong to the erythrocyte cytoskeleton. This is strongly indicated by the fact that extensive washing of ghosts with hypotonic (4), isotonic (12), and hypertonic buffers or even with nonionic detergents (Fig. 2 and unpublished data) does not significantly affect the rates of the reactions. Physiological spectrin phosphorylation and dephosphorylation could thus either be catalyzed by strongly membrane-bound kinase(s) and phosphatase(s) or one or both reactions could be autocatalytic (i.e., catalyzed by spectrin itself). Based on the evidence presented, however, we must favor the latter possibility. Thus, erythrocyte ghosts, after elution of the bulk of casein kinase activity, which had earlier been implicated in the reaction (6), showed virtually unimpaired shape changes and a component 2 phosphorylation capacity reduced by only 35-45% (Figs. 1 and 2). The apparent discrepancy with the earlier reports (6, 9) is likely due to the fact that ATP was used there at micromolar concentration (i.e., much below the physiological value). nder those conditions we could in fact confirm the earlier findings (data not shown). It thus seems that at micromolar concentration of ATP, when the phosphorylation capacity of spectrin is generally minimal (4), the casein kinase-like reaction (Km <.1 mm) becomes apparent (6). With increasing ATP concentration, however, physiological phosphorylation (Km >1 mm) reaches stoichiometric level, being then by far the main reaction. Further evidence supporting the second hypothesis was obtained by examination of the catalytic capacities of isolated spectrin-actin complexes. The largely casein kinase-free spectrin-actin preparation eluted from endocytotic inside-out vesicles (12) under hypertonic conditions exhibited a virtually intact phosphorylation machinery (Figs. 3 and 4). In contrast, conventional spectrin-actin eluted from the membrane at low ionic strength (13) has almost lost the phosphorylation capacity (refs. 11 and 21; Figs. 3 and 4). These latter findings were previously interpreted to indicate effective segregation of both kinase and phosphatase from their substrate during extraction. An alternative explanation, however, is suggested by the present data: the "water-extracted" spectrin is partially denatured and thus not capable of native catalysis. An autophosphorylating spectrin must have its own ATPbinding site. nder shape change conditions we have here specifically identified such a site on component 2 by direct irradiation of complexes of ghosts and MgATP in order to achieve covalent affinity labeling (Fig. 5). This technique has also been successfully used on several soluble protein-nucleotide complexes (see Results). Examples are cyclic AMP receptors, puromycin-ribosome complexes, ATP bound to histone 4, and an ATP-aminoacyl-tRNA synthetase complex (see refs. in refs. 14 and 15). For ribonuclease A, proper covalent attachment of the nucleotide to the enzyme of known three-dimensional structure could be demonstrated (15). Haley and Hoffmann (22) have also reported such an investigation using a photoaffinity analog of ATP and erythrocyte ghosts under conditions slightly different from ours. Nevertheless, they found incorporation of the label into spectrin component 2, besides strong labeling of components 3 and 5. It remains to be experimentally shown, however, whether the specific site of component 2 phosphorylation and the site of ATP binding identified here are in fact interrelated. Preliminary studies have indicated that this might be the case because a common tryptic peptide of component 2 containing both probes could be identified (data not shown). Further indication for an autocatalytic spectrin was obtained by the finding that phosphorylation and dephosphorylation

5 3268 Biochemistry: Imhof et al. might be coupled (i.e., that one reaction occurs only if concomitantly a possibility for the other exists as well). In ghosts, for instance, both reactions are rapid immediately after the restoration of isotonic conditions and concomitantly slow down after 1-2 min of incubation (ref. 4 and Fig. 7). If the phosphorylation capacity is reduced (e.g., due to spectrin extraction at low ionic strength), dephosphorylation is also inhibited (data not shown). The two-step experiments on the turnover of spectrin phosphate in isotonic inside-out vesicles (Fig. 7) corroborate the above interpretation; both reactions are inhibited after previous phosphorylation and are even further quenched by ADP. However, we do not know why the rate of turnover of spectrin phosphate decreases with the time of isotonic incubation (Fig. 7). At present, we consider these correlative observations as highly suggestive, but not as unambiguous proof, that phosphorylation and dephosphorylation are linked. Based on the evidence obtained and with the ATPase activity of the related myosin (2) in mind, we thus suggest for spectrin phosphorylation the following sequential mechanism as a working hypothesis. (i) ATP first binds to component 2 at a terminal and possibly globular site (ref. 2 and Fig. 5). It would be of interest to test whether this part of the spectrin molecule contains amino acid sequences particularly homologous to the SF1 fragment of muscle myosin. (fi) Hydrolysis of the bound ATP then proceeds via phosphorylation of a neighboring serine residue of component 2 (4, 19). Such a reaction intermediate could be trapped by denaturation of the system with NaDodSO4 or by water extraction (see above). However, no phosphorylated intermediate has been detected in muscle myosin (23). (im) In the native sepectrin molecule hydrolysis of the phosphorylated amino acid residue may then closely followed and Pi would be released. Dephosphorylation of spectrin, however, does not seem to be capable of reforming ATP from ADP (Fig. 7). Many questions concerning the presented working hypothesis remain to be answered. It is not known, for instance, how and at what stage energy derived from ATP could be converted into an altered spectrin conformation (4) or how such conformational changes might effect the interaction of spectrin with other membrane components to produce the shape changes of the membrane (4), e.g., with erythrocyte actin (5, 24-26), ankyrin (27), or component 4.1 (28). However, further comparisons with the related myosin (23) and with other systems capable of autophosphorylation (29-31) will have to direct future efforts. This work was supported by Grant from the Swiss National Science Foundation. Proc. Natl. Acad. Sci. SA 77 (198) 1. Nicolson G. L., Marchesi V. T. & Singer, S. J. (1971) J. Cell Biol. 51, Sheetz, M. P., Painter, R. G. & Singer, S. J. (1976) Biochemistry 15, Tilney, L. G. & Detmers, P. (1975) J. Cell Biol. 66, Birchmeier, W. & Singer, S. J. (1977) J. Cell Biol. 73, Pinder, J. C., Bray, D. & Gratzer, W. B. (1977) Nature (London) 27, Fairbanks, G., Avruch, J., Dino, E. J. & Patel, V. P. (1978) J. Supramol. Struct. 9, Avruch, J. & Fairbanks, G. (1974) Biochemistry 13, Fairbanks, G. & Avruch, J. (1974) Biochemistry 13, Avruch, J., Fairbanks, G. & Crapo, L. M. (1976) J. Cell Physiol. 89, Hosey, M. M. & Tao, M. (1977) J. Biol. Chem. 252, Hosey, M. M. & Tao, M. (1977) Biochemistry 16, Birchmeier, W., Lanz, J. H., Winterhalter, K. H. & Conrad, M. J. (1979) J. Biol. Chem. 254, Marchesi, V. T. (1974) Methods Enzymol. 32, Yue, T. V. & Schimmel, R. (1977) Biochemistry 16, Sperling, J. & Havron, A. (1976) Biochemistry 15, Skin, B. C. & Carraway, K. L. (1973) J. Biol. Chem. 248, Wolfe, L. C. & Lux, S. E. (1978) J. Biol. Chem. 253, Schechter, N. M., Sharp, M., Reynolds, J. A. & Tanford, C. (1976) Biochemistry 15, Wyatt, J. L., Greenquist, A. C. & Shohet, S. B. (1978) Biochem. Biophys. Res. Commun. 79, Anderson, A. J. (1979) J. Biol. Chem. 254, Graham, C., Avruch, J. & Fairbanks G. (1976) Biochem. Biophys. Res. Commun. 72, Haley, B. E. & Hoffman, J. F. (1974) Proc. Natl. Acad. Sci. SA 71, Lymn, R. W. & Taylor, E. W. (1971) Biochemistry 1, Birchmeier, W. & Singer, S. J. (1977) Biochem. Biophys. Res. Commun. 77, Sheetz, M. P. (1979) J. Cell Biol. 81, Cohen, C. M., Jackson, P. L. & Branton, D. (1978) J. Supramol. Struct. 9, Bennett, V. & Stenbuck, P. J. (1979) J. Biol. Chem. 254, ngewickell, E., Bennett, P. M., Calvert, R., Ohanian, V. & Gratzer, W. B. (1979) Nature (London) 28, Rosen,. M. & Ehrlichman, J. (1975) J. Biol. Chem. 25, Chin, Y. S. & Tao, M. (1978) J. Biol. Chem. 253, Ikeda, I. & Steiner, M. (1979) J. Biol. Chem. 254,