THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 19% by The American Society of Biological Chemists, Inc.

VOl. 260, No. 28, Issue of December

Phosphorylation of Ankyrin Decreases Its Affinity for Spectrin Tetramer*

(Received for publication, April 1, 1985)

Pao-Wen Lu, Chu-Jing Soong, and Mariano Tao From the Department of Biological Chemistry, University of ZUinois ut Chicago, College of Medicine, Chicago, Illinois 60612

The effects of phosphorylation on the interaction between spectrin and ankyrin were investigated. Spec- trin and ankyrin were phosphorylated using purified human erythrocyte membrane and cytosolic (casein kinase A) kinases. These two kinases have similar properties as well as activities toward spectrin and ankyrin. Both kinases catalyzed the incorporation of about 2 mol of phosphate/mol of spectrin and about 7 mol of phosphate/mol of ankyrin. These phosphates were incorporated primarily into seryl and threonyl residues of the proteins. The phosphopeptide maps of ankyrin phosphorylated by the membrane kinase and casein kinase A were identical.

Binding studies indicate that ankyrin exhibits dif- ferent affinities for spectrin dimers (KD = 2.5 f 0.9 x lo-‘ M) and tetramers (KO = 2.7 f 0.8 X lo-’ M). These dissociation constants were not appreciably affected by the phosphorylation of spectrin. On the other hand, phosphorylation of ankyrin was found to significantly reduce its affinity for either phosphorylated or un- phosphorylated spectrin tetramers (KD = 1.2 f 0.1 X lo-‘ M) but not spectrin dimers (KO = 2.5 f 0.4 X M). The same results were obtained using either the membrane kinase or casein kinase A as the phospho- rylating enzyme. The above observation suggests that apkyrin phosphorylation may provide an important mechanism for the regulation of the erythrocyte mem- brane cytoskeletal network.

Underlying the erythrocyte membrane are a group of pe- ripheral proteins which aggregate to form a cytoskeletal net- work. This cytoskeletal network is thought to play an impor- tant role in controlling the shape and deformability of eryth- rocytes (1,2). The major constituents of the cytoskeleton are spectrin, band 4.1, and actin. Spectrin is a heterodimer con- sisting of an a (M, = 240,000) and a @ (Mr = 220,000) subunit (1,2). Spectrin dimers can further associate to form tetramers and other higher molecular aggregates (3). However, there is evidence to suggest that the predominant form of spectrin in the membrane is the tetramer (4, 5). The cytoskeleton is linked to the membrane, at least in part, by the association of ankyrin with spectrin and band 3, an integral membrane protein (6). Ankyrin, also referred to as band 2.1, is a high molecular weight protein (Mr = 210,000) constructed of a single polypeptide chain.

Studies from several laboratories have shown that both

* This work was supported by Grant AM 23045 from the National Institutes of Health and by a grant-in-aid from the Chicago Heart Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisernent” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

spectrin and ankyrin are phosphoproteins. The phosphoryla- tion of these proteins can be demonstrated in intact cells as well as in membrane preparations (7-9). Subsequent studies have identified several erythrocyte protein kinases which are capable of utilizing spectrin as a substrate i n uitro (10). However, only two of these kinases, one isolated from the membrane (membrane kinase) and the other from the cytosol (casein kinase A), are found to yield phosphopeptide patterns of spectrin similar to that of spectrin phosphorylated in intact cells (10).

Studies on the functional significance of spectrin phospho- rylation have yielded conflicting results (11-13). Birchmeier and Singer (11) initially reported that shape changes in eryth- rocyte ghosts were related to the phosphorylation of the B subunit of spectrin. Later on, Pinder et al. (12) showed that phosphorylated spectrin caused a dramatic increase in G- actin polymerization. Phosphorylation, however, did not af- fect the dimer-tetramer equilibrium of spectrin (14). In con- trast, Brenner and Korn (13) showed that the binding of F- actin to spect& was independent of spectrin phosphorylation and that purified spectrin, irrespective of its phosphorylation states, did not bind G-actin. This latter observation is in agreement with the finding of Cohen and Branton (15). Fi- nally, Anderson and Tyler (16) examined the turnover of spectrin-bound phosphates in intact cells and concluded that there was no correlation between spectrin phosphorylation or dephosphorylation and shape changes.

In this work, we have examined the phosphorylation of spectrip and ankyrin by the membrane kinase and casein kinase A in purified preparations. The results indicate that both spectrin and ankyrin contain multiple phosphorylation sites. The phosphorylation of ankyrin appears to affect its binding to spectrin tetramer but not to the dimer. In contrast, the phosphorylation of spectrin does not appreciably affect its binding to either phosphorylated or unphosphorylated ankyrin.

EXPERIMENTAL PROCEDURES - Muteriub-Human blood was obtained from the University of

Illinois Hospital blood bank and used within 2 weeks of the drawing date. [-p32P]ATP and ‘%I-labeled Bolton-Hunter reagent were pur- chased from Amersham Corp. Leupeptin, pepstatin A, aprotinin, and DFPl were purchased from Sigma. All other reagents were of analyt- ical grade.

Preparation of Erythrocyte Membrunes-Human erythrocytes were washed as described earlier (17). The washed erythrocytes were sedimented three times at 1 X g in 3 volumes of a buffer containing 160 mM NaCl, 5 mM Na phosphate, pH 7.5, and 0.75% (w/v) dextran 500 (18). Erythrocyte ghosts were prepared by hypotonic lysis in 4 volumes of a pH 7.5 buffer containing 7.5 mM Na phosphate, 1 mM EDTA, 20 pg/ml PMSF, 0.4 mM DFP, and 2 pg/ml each of the protease inhibitors (leupeptin, pepstatin A, and aprotinin) and

The abbreviations used are: DFP, diisopropyl fluorophosphate, PMSF, phenylmethylsulfonyl fluoride; SDS, sodium dodecyl sulfate.

14958

Phosphorylation of Ankyrin and Spectrin 14959

washed free,of hemoglobin with 7.5 mM Na phosphate buffer, pH 7.5, containing 1 - m ~ EDTA.

Preparation of Ankyrin-The solubilization of ankyrin from eryth- rocyte ghost membranes was conducted essentially as described by Bennett and Stenbuck (18) except that the buffers used were supple- mented with 0.4 mM DFP, 0.02% NaN3, and 2 pg/ml each of the protease inhibitors. The solubilized ankyrin (from 600 ml of ghosts) was dialyzed overnight against a pH 7.5 buffer (Buffer B) containing 7.5 mM Na phosphate, 1 mM EDTA, 0.02% NaN3, 0.2 mM dithio- threitol, 20 pg/ml PMSF, 0.4 mM DFP, and 2 pg/ml each of the protease inhibitors. The dialyzed extract was applied to a QAE- Sephadex column (3.2 X 20 cm) which had been equilibrated with Buffer B supplemented with 150 mM KC1. The column was washed with 150 mM KC1 in Buffer B and eluted with a linear KC1 gradient of 0.2-0.6 M in a total volume of 1.2 liters. Peak fractions ( A z s o ~ ) were analyzed by SDS-polyacrylamide gel electrophoresis (17). Those fractions which contained ankyrin were pooled and concentrated in an Amicon ultrafiltration cell equipped with a YM-10 membrane. Ankyrin was further purified by sedimentation in a linear 5 2 0 % sucrose gradient prepared in Buffer B containing 150 mM KCl. The centrifugation was conducted at 38,000 rpm, 4 "C, in a Beckman SW 41 rotor for 20 h. Fractions of 0.5 ml each were collected from the bottom of the tube. Those fractions which contained the least amount of impurities were pooled and applied onto a hydroxylapatite column (1.5 X 3 cm) which had been equilibrated with 20 mM K phosphate, pH 6.7, 0.5 M KC1, 0.2 mM dithiothreitol, 1 mM EDTA, 0.02% NaN3, 0.4 mM DFP, 20 pg/ml PMSF, and 2 pg/ml each of the protease inhibitors. A linear phosphate gradient (total volume of 60 ml) of 20- 300 mM was used to elute the ankyrin. Fractions containing ankyrin were pooled, dialyzed against a buffer containing 20 mM Tris-HC1, pH 7.5, 150 mM KCl, 1 mM EDTA, 0.2 mM dithiothreitol, 0.02% NaN3, 20 pg/ml PMSF, 0.4 mM DFP, and 2 pg/ml each of the protease inhibitors and stored at -20 "C. The purity of the final ankyrin preparation was analyzed by SDS-polyacrylamide electrophoresis and found to be about 96% pure.

Preparation of Spectrin-Spectrin was extracted from human erythrocyte membranes at 37 "C for 30 min with 0.1 mM EDTA, pH 8.0, containing 2 pg/ml each of the protease inhibitors and 0.4 mM DFP, as described by Marchesi (19) with slight modifications. The extract which contained primarily spectrin and actin was purified by gel filtration through a Sepharose CL-GB column. The elution of the column was conducted with 10 mM Tris-HC1, pH 7.5, 0.5 M NaC1, 1 mM EDTA, 20 pg/ml PMSF, and 0.02% NaN3. This procedure was repeated once, and the purity of each fraction collected from the second passage was analyzed by SDS-polyacrylamide gel electropho- resis. Those fractions containing only spectrin were pooled and stored at -20 "C at concentrations of less than 1.5 mg/ml in the elution buffer containing 2 pg/ml each of the protease inhibitors. The con- centration of spectrin was determined by absorbance at 280 nm, using the value = 8.8 (5), and by the methods described by Bradford (20) and by Lowry et ul. (21), using bovine serum albumin as a standard. The spectrin used in the binding assay was stored in a buffer (binding buffer) containing 20 mM Tris-HC1, pH 7.5, 130 mM KCl, 20 mM NaC1,0.5 mM 2-mercaptoethanol, 20 pg/ml PMSF, and 2 pg/ml each of the protease inhibitors. Under the storage conditions described above, no precipitation of spectrin was observed.

Phosphorylation Assays-The phosphorylation of spectrin and an-

mM Tris-HC1, pH 7.5, 5 mM MgCl2, 0.2 mM [y-32P]ATP, 0.25-0.4 kyrin was carried out a t 37 "C in a reaction mixture containing 50

mg/ml spectrin or 0.4 mg/ml ankyrin, 3 4 units/ml kinase, 100 mM NaCl (for spectrin) or 40 mM KC1 (for ankyrin), and 10 pg/ml each of the protease inhibitors. One unit of kinase activity was defined as that amount of enzyme which catalyzed the incorporation of 1 nmol of phosphate into casein/min (17).

The incorporation of 32P into spectrin and ankyrin was analyzed by SDS-gel electrophoresis. Following SDS-polyacrylamide gel elec- trophoresis, the radioactivity incorporated was determined by excis- ing the protein band from the dried gel and counting in a liquid scintillation spectrometer.

Preparation of '25Z-Ankyrin-Ankyrin was radioiodinated at 0 'C for 90 min in a reaction mixture containing 1 mCi of lZ5I-labeled Bolton-Hunter reagent, 700 pg of ankyrin, 0.16 M Na borate, pH 8.5, and 250 mM NaCl in a final volume of 0.2 ml. The reaction mixture was then diluted with the binding buffer and dialyzed for 3 h against this buffer. Unbound lZ5I was further removed by repeated washing with the binding buffer in a microconcentrator (Amicon) and finally concentrated to yield lZ5I-ankyrin concentrations of 1.1-1.6 mg/ml.

The phosphorylation of 'T-ankyrin with unlabeled ATP using either the membrane kinase or casein kinase A was conducted as described in the preceding section. The excess, unreacted ATP was removed in the reaction mixture by dialysis and by repeated washing in a micro- concentrator.

Binding Assay-The binding of lZ5I-ankyrin to spectrin was ana- lyzed by nondenaturing density gradient gel electrophoresis according to the method described by Weaver et al. (22). Varying amounts of phosphorylated and unphosphorylated lZ5I-ankyrin were incubated for 1 h at 0 "C with 40.5 pg of phosphorylated or unphosphorylated spectrin in 60 pl of the binding buffer. The binding of '%I-ankyrin to spectrin was analyzed by electrophoresis in a 2-4% acrylamide gra- dient slab gel at 60 V for 40 h in a cold room. After the electrophoresis was terminated, the gel was stained with Coomassie Brilliant Blue, dried under vacuum, and exposed to x-ray films. The radioactive bands were also excised from the dried gel and counted in a y-counter.

Preparation of Protein Kinase-The human erythrocyte membrane cyclic AMP-independent protein kinase was extracted from ghosts at 0 "C for 30 min with 0.5 M NaCl, prepared in a buffer containing 5 mM Na phosphate, pH 7.5, 1 mM EDTA, 15 mM 2-mercaptoethanol, 20 pg/ml PMSF, 0.4 mM DFP, and 2 pg/ml each of the protease inhibitors (17). The NaCl concentration of the extract was adjusted to 0.3 M with the above buffer and applied to a phosphocellulose column. The kinase was eluted from the column with a linear (0.3- 1.0 M) NaCl gradient.

The peak fractions containing kinase activity, as assayed using casein as substrate (17), were pooled, concentrated by Diaflo ultra- filtration, and applied to a Sephacryl S-200column. The column was equilibrated and eluted with 0.5 M KC1 in a buffer (Buffer C) con- taining 20 mM Tris-HC1, pH 7.5, 1 mM EDTA, 15 mM Z-mercapto- ethanol, 20 pg/ml PMSF, 0.4 mM DFP, and 2 pg/ml each of the protease inhibitors. The active kinase fractions obtained from the Sephacryl S-200 column were pooled, concentrated, and diluted with 4 volumes of Buffer C. The diluted sample was applied to a casein- Sepharose 4B affinity column. The column was first washed with the above buffer containing 0.1 M KC1 followed by a second wash in the same buffer but containing 0.3 M KCl. The membrane kinase was eluted with 1 M KCl, concentrated by Diaflo ultrafiltration, dialyzed against Buffer C containing 0.15 M KC1 and 50% glycerol, and stored at -20 "C. No significant loss of kinase activity was observed during storage for a period of a t least 6 months. The enzyme preparation was judged to be homogeneous based on SDS-polyacrylamide gel electrophoresis.

The cytosolic cyclic AMP-independent protein kinase described earlier by Simkowski and Tao (23) was purified from the hemolysate using essentially the same procedures described above for the mem- brane kinase. The final enzyme preparation also appeared to be homogeneous as analyzed by SDS-polyacrylamide gel electrophoresis. This enzyme is tentatively identified as casein kinase A since it exhibits a high specificity for ATP as a phosphoryl donor.

RESULTS

Phosphorylatwn of Spectrin-Fig. 1 shows the SDS-gel electrophoretic pattern of the spectrin preparations used in this study. The spectrin preparations are free of any contam- inating membrane proteins and degradation products.

The effect of various agents on the phosphorylation of ankyrin and spectrin by the kinases has been examined in order to establish conditions for maximal incorporation of phosphates into these two membrane proteins. The phospho- rylation of spectrin exhibits an optimum Mg2+ concentration of about 5 mM. Since a moderate concentration of salt was needed to maintain the solubility of spectrin (24), the effect of NaCl on the phosphorylation was also examined. NaCl, in the range of 50-100 mM, was found to have no appreciable effect on the rate of spectrin phosphorylation. At concentra- tions greater than 150 mM, however, inhibition of the phos- phorylation reaction was observed. This inhibition could be attributed, in part, to the interference of high salt concentra- tions on the interaction between the kinase and spectrin as suggested earlier by Conway and Tao (25). The phosphoryla- tion of spectrin exhibits a broad pH activity profile. No significant difference in incorporation was observed between

14960 Phosphorylation of Ankyrin and Spectrin

bwr

A B A B Spectrin Ankyrin

FIG. 1. SDS-polyacrylamide gel electrophoresis of spectrin and ankyrin phosphorylated by casein kinase A. The electro- phoresis of spectrin (6 pg) and ankyrin (7 pg), phosphorylated as described under “Experimental Procedures,” was conducted on a 5% polyacrylamide gel slab. The specific activities of [T-~~P]ATP used for the phosphorylation of spectrin and ankyrin are 750 and 350 cpm/ pmol, respectively. A , stained gel; B, radioautogram.

pH 6 and 8.5. In this study, the reaction was routinely con- ducted at pH 7.5.

Fig. 1 shows that the incorporation of phosphate occurs only on the /3 subunit of spectrin. However, under extreme conditions, such as at pH 9, a small degree of phosphorylation of the LY subunit was also detected. The phosphorylation assays were conducted in the presence of protease inhibitors. No detectable degradation of spectrin was observed during incubations. An analysis of the acid hydrolysis products of 32P-labeled spectrin reveals the presence of 32P labels in seryl and threonyl residues (data not shown). The ratio of 32P labels in seryl and threonyl residues was estimated to be about 3:2. The membrane kinase and casein kinase A exhibit no signif- icant differences in their activities toward spectrin. Under optimum condition, both kinases catalyze the incorporation of 1.7-2.0 mol of phosphate/mol of /3 subunit. This value is somewhat greater than that estimated previously by Harris and Lux (26) and by Tao et al. (10). This difference, however, is not due to a difference in the endogenous phosphate con- tents of the isolated spectrin. The spectrin preparation used in this study had been routinely analyzed for protein-bound phosphates using the malachite green method of Kallner (27) and found to be about 4-5 mol/mol of spectrin. This value is similar to that reported earlier for the spectrin preparation employed by Harris et al. (26) in their study.

PhosphoTylation ofdnkyrin-The procedure of Bennett and Stenbuck (18) for the preparatin of ankyrin has been modified in our laboratory. We have employed QAE-Sephadex in place of DEAE-cellulose for the initial fractionation step as this appears to give us a better separation of ankyrin from other membrane proteins. An additional step involving hydroxyl- apatite chromatography is introduced in order to remove minor degradation products of ankyrin. Especial precaution has also been taken to prevent proteolysis of ankyrin during fractionation. By including the various protease inhibitors such as DFP, aprotinin, leupeptin, and pepstatin A in the buffers, we have been able to minimize proteolysis; and the ankyrin preparation obtained can be stored for a prolonged period of time with no detectable degradation. An SDS-gel electrophoretic pattern of ankyrin is shown in Fig. 1 (Anlzyrin,

lane A). The gel contains a barely visible minor protein contaminant migrating in the region of about 155,000 daltons. This same protein component is also reported in the ankyrin preparation of Bennett and Stenbuck (18). Our purification procedure generally yielded 3-5 mg of ankyrinlunit of blood.

Since there is a paucity of information concerning the phosphorylation of ankyrin by purified kinases, in this study we have investigated in greater detail the activity of the membrane kinase and casein kinase A toward ankyrin. Fig. 1 (Ankyrin, lane B ) shows that the phosphorylation of ankyrin with either enzyme results in the labeling of only the protein band corresponding to ankyrin. The phosphorylation of an- kyrin exhibited a pH optimum of about 7.5 and a M$+ optimum of 5 mM. KCl, at 40 mM, was found to be slightly stimulatory; whereas at concentrations greater than 0.1 M, inhibition of phosphorylation was observed. These parameters are the same for both the membrane kinase and casein kinase A.

Fig. 2 shows the time course of phosphorylation of ankyrin by the membrane kinase and casein kinase A under optimum conditions. The data show that each kinase can catalyze the incorporation of about 7 mol of phosphate/mol of ankyrin. Preliminary estimate indicates that each ankyrin contains three endogenous protein-bound phosphates.

The kinetic parameters were determined by measuring the kinase activity at varying concentrations of one substrate in the presence of different fixed levels of the other. The data for the membrane kinase-catalyzed reaction shown in Fig. 3 are presented as double reciprocal plots of the initial velocity versus the concentrations of ankyrin at different fixed con- centrations of ATP, and vice versa. Essentially the same results were obtained for casein kinase A. The data suggest that the reaction mechanism of the kinases is the same as that of the wheat germ kinase (28) and the cyclic AMP- dependent protein kinase (29) and is consistent with a se- quential bireactant reaction kinetics involving the formation of a ternary enzyme-substrate complex (30). From replots of the slopes and intercepts of the double reciprocal plots shown in Fig. 3, K,,, values of 10 PM and 0.18 mg/ml were obtained for ATP and ankyrin, respectively. These values are the same for both the membrane kinase and casein kinase A.

An analysis of the phosphoamino acids (31) obtained from

V . . 0 40 ‘ 8’0 ’ ti0 ’ t i0

MINUTES

FIG. 2. Time course of phosphorylation of ankyrin by the membrane kinase and casein kinase A. Phosphorylation of an- kyrin was conducted in 200 p1 of a reaction mixture containing 80 pg of ankyrin, 0.2 mM [T-~~PIATP (400 cpm/pmol), 50 mM Tris-HC1, pH 7.5,5 mM MgC12,40 mM KC1 and kinase, and 4 units/ml of either the membrane kinase (M) or casein kinase A (A-A). The reaction mixture was incubated at 37 “C; and at the time indicated, an aliquot (20 pl) was withdrawn for determination of 32P incorpo- ration.

Phosphorylation of Ankyrin and Spectrin 14961

1 I I I . I

0 0. I 0.2 l /ATP( vMI"

FIG. 3. Double reciprocal plots of initial velocity uersua substrate concentration. The reactions were carried out in the presence of 4 units/ml membrane kinase. The incubation was con- ducted at 37 "C for 4 min. A, double reciprocal plots of initial velocity uersus ankyrin concentration at different fixed levels of ATP (M, 5 p ~ ; o " 0 , 10 pM; A-A, 25 pM; and A-A, 50 pM). B, double reciprocal plots of the initial velocity uersus ATP concentra- tion at different fixed levels of ankyrin (u, 0.1 mg/ml; & 0,0.2 mg/ml; A-A, 0.3 mg/ml; and A-A, 0.4 mg/ml).

acid hydrolysis of 32P-labeled ankyrin indicates that phospho- rylation occurs primarily on seryl and threonyl residues (data not shown). The possibility that phosphotyrosine may also be formed has been examined by conducting the high voltage electrophoresis at pH 3.5 according to the procedure of Hunter and Sefton (32). Our result failed to reveal the presence of phosphotyrosine (data not shown). The same results were obtained for both the membrane kinase and casein kinase A.

Since both the membrane kinase and casein kinase A can catalyze the incorporation of approximately 7 mol of phos- phate into each mol of ankyrin, it is of interest to determine whether these two kinases phosphorylate the same or different sites on the ankyrin molecule. Experiments in which the phosphorylation of ankyrin was conducted in the presence of both kinases showed that the amount of phosphate incorpo- rated was not additive. Under these conditions, the number of phosphate incorporated was the same as that using either kinase alone (data not shown). A comparison of the phospho- peptide maps of ankyrin phosphorylated by the membrane kinase and casein kinase A also revealed no significant differ- ence in their 32P-labeled patterns (Fig. 4). The results suggest,

-E FIG. 4. Phosphopeptide maps of ankyrin phosphorylated by

the membrane kinase and casein kinase A. Ankyrin was phos- phorylated by the membrane kinase and casein kinase A using [y - 32P]ATP and digested with 50 pg/ml L-1-tosylamido-2-phenylethyl chloromethyl ketone-trypsin (Worthington) in 50 mM ammonium bicarbonate. The digestion was conducted for 24 h at 37 "C after which the mixture was lyophilized. The trypsin-digested sample was dissolved in acetic acid/formic acid/water (2925:946, v/v/v), and aliquots of 5 pl were spotted on cellulose-coated thin layer sheets (20 X 20 cm) for electrophoresis. The electrophoresis was carried out at 300 V, 4 "C, for 4 h. Chromatography in the second dimension was conducted in butanol/acetic acid/pyridine/water (8012:30,40, v/v/ v/v). The 32P-labeled peptides were located by radioautography. A, phosphopeptide map of ankyrin phosphorylated by the membrane kinase; B, phosphopeptide map of ankyrin phosphorylated by casein kinase A. Direction of electrophoresis ( E ) and chromatography (C) are indicated at the origin.

although do not prove, that the two kinases may have the same specificities toward ankyrin and that they recognize the same phosphorylation sites.

Effects of Phosphorylation on the Interaction between Spec- trin and Ankyrin-Studies have shown that ankyrin contains two important binding sites: one is for spectrin and the other is for band 3 (22, 33). Thus, ankyrin serves to link the membrane cytoskeleton to the overlying membrane. Since both spectrin and ankyrin are phosphoproteins, it was of interest to determine whether phosphorylation could affect their interactions.

The binding of '251-ankyrin to spectrin was investigated using nondenaturing gel electrophoresis according to the method of Morrow and Marchesi (3). This gel system is particularly useful since it resolves spectrin into its various oligomeric forms; therefore, it allows us to measure the bind- ing of ankyrin, not only in the dimeric form, but more impor- tantly also to the tetrameric form of spectrin. As indicated

14962 Phosphorylation of Ankyrin and Spectrin

earlier, spectrin tetramer has been suggested to represent the native form of spectrin in the membrane cytoskeleton. We have confirmed the earlier observation of Morrow and Mar- chesi (3) that the aggregation of spectrin to form tetramers and higher oligomers is a time- and temperature-dependent process.

A representative experiment illustrating the binding of 1251- ankyrin to spectrin is shown in Fig. 5. It can be seen from the figure that ankyrin can bind to the various aggregated forms of spectrin. The extent of binding to each of these forms, however, was dependent on ankyrin concentrations. At low concentrations, nearly all of the ankyrin was found to be associated with the tetramer and higher oligomers, even though the dimeric form was the predominant species present. Increasing the concentration of ankyrin resulted in an in- crease in the binding to the dimeric species. Ankyrin was found to exhibit a greater affinity for spectrin tetramer (KD = 2.7 f 0.8 X M) than for spectrin dimer (KO = 2.5 f 0.9 X 1O"j M). The dissociation constants were calculated from the slopes of double reciprocal plots of ankyrin binding to spectrin dimer and tetramer according to the method of Weaver et al. (22). The values are averaged from five deter- minations. The binding of ankyrin to higher oligomers was not determined due to poor resolution of these oligomers. Our results are in general agreement with those reported earlier by Weaver et al. (22).

In a similar study, we examined the binding of ankyrin to phosphorylated spectrin. Spectrin was phosphorylated by either the membrane kinase or casein kinase A to the extent of about 2 mol/mol of spectrin. It should be noted that phosphorylation did not affect the dimer-tetramer equilib- rium, and the distribution of these molecular species on the gel was the same as the unphosphorylated preparation. Our

T- D-

1 2 3 4 5 6 7 8 9

T- D-

1 2 3 4 5 6 7 8 9 FIG. 5. Binding of "'I-ankyrin to spectrin. Varying amounts

of '251-ankyrin (8 X lo' cpmlpg) were incubated with spectrin as described under "Experimental Procedures." Lanes 1-4 and lanes 6- 8 contain, respectively, the following amounts of '=I-ankyrin: 1.6,4.8, 6.4, and 8 and 9.6,12.8, and 16 pg. Lane 5 contains 40.5 pg of spectrin, whereas lane 9 contains only '251-ankyrin (19.2 pg). The following are the distribution of the various molecular forms of spectrin in lane 5: dimer, 55%; tetramer, 35%; and oligomers, 10%. A, stained gel; B, radioautogram; T, tetramer; D, dimer.

results indicate that phosphorylation of spectrin also does not affect significantly its interaction with ankyrin. The dissocia- tion constants of ankyrin for phosphorylated spectrin dimer and tetramer were approximately the same as those obtained for unphosphorylated spectrin (data not shown). On the other hand, phosphorylation of ankyrin (about 7 mol of phosphate/ mol of ankyrin) was found to affect its affinity for spectrin tetramer but not the dimer (Fig. 6). The KD, based on five

T" D"

1 2 3 4 5 6 7 8

T- D-

1 2 3 4 5 6 7 8 FIG. 6. Binding of phosphorylated '"I-ankyrin to spectrin.

Varying amounts of the phosphorylated '%I-ankyrin (phosphorylated with ATP and casein kinase A) were incubated with spectrin as described under "Experimental Procedures." Lanes 1 3 and lanes 5- 7 contain, respectively, the following amounts of phosphorylated '%I- ankyrin: 4.4, 5.9, and 8.1 and 9.6, 12.6, and 16.3 pg. Lanes 4 and 8 contain only spectrin (40 pg) and '%I-ankyrin (19.2 pg), respectively. Lane 4 contains 59% dimer, 30% tetramer, and 10% oligomers. A, stained gel; B, radioautogram; T, tetramer; D, dimer.

/: P - A R /OA/

p I I I I I

0 OA oa 1.2 I /[FREE ANKYRINI( 10=rM)-l

FIG. 7. Double reciprocal plots of the binding of phospho- rylated "'I-ankyrin to unphosphorylated spectrin. The phos- phorylation of '"I-ankyrin with ATP and casein kinase A was con- ducted as described under "Experimental Procedures." The data were plotted according to the relationship [spectrin]/[bound ankyrin] = &/[free ankyrin] + 1 for binding to spectrin dimers (0) and tetra- mers (A).

Phosphorylation of Ankyrin and Spectrin 14963

determinations, for the complex between phosphorylated an- kyrin and spectrin tetramer was increased about 4-fold to a value of 1.2 k 0.1 X M, whereas the KO, 2.5 & 0.4 X M, for the complex between phospho-ankyrin and spectrin dimer was not significantly affected (Fig. 7). The same results were obtained for the complexes between phospho-ankyrin and phosphospectrin. Thus, the interaction between spectrin and phospho-ankyrin was also independent of the phospho- rylation states of spectrin. The results described above are the same irrespective of whether the membrane kinase or casein kinase A was used to catalyze the phosphorylation reaction.

Quantitation of the amounts of ankyrin bound to spectrin dimer and tetramer showed that the tetramer bound about twice the amount of ankyrin as did the dimer, confirming an earlier report by Weaver et al. (22). The binding molar ratio was not affected by phosphorylation of either ankyrin or spectrin, or both. It should also be noted that the labeling of ankyrin with lZ5I did not affect its phosphate accepting capac- ity.

DISCUSSION

Although the phosphorylation of erythrocyte membrane proteins has been widely observed and investigated, the sig- nificance of the phosphorylation reaction remains unknown. Among the major erythrocyte membrane proteins, spectrin, band 3, ankyrin and bands 4.1, 4.5, and 4.8 have all been identified as substrates of protein kinases (7). However, the study of phosphorylation has focused primarily on spectrin in light of the importance of this protein in the cytoskeletal network assembly of the erythrocyte membrane.

Spectrin is a phosphoprotein containing 4-5 mol of bound phosphate. Since a similar amount of protein-bound phos- phate was found in spectrin isolated from both fresh and outdated erythrocytes, it would appear that these endogenous phosphates had low turnover rate. The present study indicates that an additional 2 mol of phosphates can be incorporated into spectrin by either the membrane kinase or casein kinase A. We have preliminary evidence to indicate that the extent of spectrin phosphorylation by the purified kinases is depend- ent upon the salt concentration in the reaction mixture. The amount of phosphate incorporated into spectrin was decreased significantly in the presence of 150 mM or greater of NaCl.

In contrast to spectrin, the isolated ankyrin was found to contain a significantly greater number of phosphorylation sites available for the membrane kinase and casein kinase A. That ankyrin may contain multiple phosphorylation sites has been suggested earlier by the observation of Weaver and Marchesi (34). The phosphopeptide maps of ankyrin phos- phorylated by the membrane kinase and casein kinase A are identical. These data together with the observation that the same amounts of phosphate are incorporated into ankyrin in reactions containing either each of the kinases alone or both kinases suggest that the two kinases exhibit the same speci- ficity toward ankyrin. The phosphates incorporated into an- kyrin are found primarily on seryl and threonyl residues with a somewhat greater amount of labels found to be associated with phosphothreonine. This distribution of phosphate be- tween the two amino acids in ankyrin is contrary to that found in spectrin. It remains to be determined as to whether the membrane kinase or casein kinase A, or both, are also responsible for the phosphorylation of ankyrin in the intact cells. It should be noted, however, that there is considerable evidence to indicate that the membrane kinase and, perhaps, also casein kinase A may play an important role in the phosphorylation of spectrin in viuo (lo,%).

Since both spectrin and ankyrin contained a significant number of phosphorylation sites, it was of interest to deter- mine whether phosphorylation could affect the interaction between these two proteins. Our results clearly show that phosphorylation of ankyrin decreases its affinity for either phosphorylated or unphosphorylated spectrin. Somewhat sur- prisingly, this effect appears to be confined mainly to the interaction between ankyrin and spectrin tetramer. The affin- ity of phosphorylated ankyrin for phosphorylated and un- phosphory1ated.spectrin tetramer was found to decrease by about 4-fold. This observation that only the binding to spec- trin tetramer is affected is particularly interesting in light of the available evidence which suggests that the tetramer is the predominant form of spectrin in the membrane cytoskeletal network in situ. In contrast, the phosphorylation of spectrin appears to have little or no appreciable effect on its interaction with either phosphorylated or unphosphorylated ankyrin. These data lend further support to an earlier report which indicates that spectrin phosphorylation does not affect its binding to cell membrane via ankyrin (16). Hence the phos- phorylation of spectrin appears to have little, if any, role in the regulation of the assembly of the cytoskeletal network and the interaction of the network with the membrane.

Although the significance of membrane protein phospho- rylation remains unknown, there is sufficient evidence to indicate that phosphorylation may play an important role in erythrocyte shape changes and deformability through the regulation of the interactions of the various cytoskeletal net- work components. The results presented in this study could conceivably provide an attractive mechanism to explain the dynamics of the cytoskeletal network. Since phospho-ankyrin binds spectrin tetramer less tightly, the phosphorylation of ankyrin could lead to a weakening of the interaction between the cytoskeletal network and the membrane. As a result, the membrane cytoskeleton could assume a more relaxed and flexible structure. Conversely, dephosphorylation of ankyrin could lead to a more rigid network due to a stronger associa- tion with the cell membrane through ankyrin. The possibility that phosphorylation may affect the binding of ankyrin to spectrin is further strengthened by studies which show that the spectrin-binding site of ankyrin is located within a 32,000- dalton region near the end of the molecule (22). This 32,000- dalton region is shown to be phosphorylated (22). The above proposed mechanism represents a working hypothesis based on the available data. Obviously, more experimentation is needed in order to confirm or reject the validity of the hy- pothesis.

REFERENCES 1. Marchesi, V. T. (1979) Semin. Hematol. 16, 3-20 2. Steck, T. L. (1974) J. Cell Biol. 62, 1-19 3. Morrow, J. S., and Marchesi, V. T. (1981) J. Cell Biol. 88, 463-

4. Liu, S.-C., and Palek, J. (1980) Nature 285, 586-588 5. Kam, Z., Joseph, S. R., Eisenberg, H., and Gratzer, W. B. (1977)

6. Bennett, V., and Stenbuck, P. J. (1980) J. Biol. Chem. 255,6424-

7. Hosey, M. M., and Tao, M. (1976) Biochemistry 15,1561-1568 8. Plut, D. A., Hosey, M. M., and Tao, M. (1978) Eur. J. Biochem.

9. Bennett, V. (1977) Life Sci. 21, 433-440

468

Biochemistry 16,5568-5572

6432

82,333-337

10. Tao, M., Conway, R., Chiang, M.-C., Cheta, S., and Yan, T.-F. (1981) Cold Spring Harbor Conf. Cell Proliferation 8, 1301- 1302

11. Birchmeier, W., and Singer, S. J. (1977) J. Cell Biol. 73,647-659 12. Pinder, 3. C., Bray, D., and Gratzer, W. B. (1977) Nature 270,

752-754

14964 Phosphorylation of Ankyrin and Spectrin

13. Brenner, S. L., and Korn, E. D. (1979) J. Biol. Chem. 254,8620-

14. Ungewickell, E., and Gratzer, W. (1978) Eur. J. Biochem. 88,

15. Cohen, C. M., and Branton, D. (1979) Nature 279,164-165 16. Anderson, J. M., and Tyler, J. M. (1980) J. Biol. Chem. 255,

17. Tao, M., Conway, R., and Cheta, S. (1980) J. Biol. Chem. 255,

18. Bennett, V., and Stenbuck, P. J. (1980) J. Biol. Chem. 255,2540-

19. Marchesi, V. T. (1974) Methods Enzymol. 32B, 275-277 20. Bradford, M. M. (1976) Anal. Biochem. 72,248-254 21. Lowry, 0. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J.

22. Weaver, D. C., Pasterneck, G. R., and Marchesi, V. T. (1984) J.

23. Simkowski, K. W., and Tao, M. (1980) J. Biol. Chem. 255,6456-

8627

379-385

1259-1265

2563-2568

2548

(1951) J. Biol. Chem. 193,265-275

Biol. Chem. 259,6170-6175

6461

24. Fairbanks, G., Avruch, J., Dino, J. E., and Patel, V. P. (1978) J.

25. Conway, R. G., and Tao, M. (1981) J. Biol. Chem. 256, 11932-

26. Harris, H. W., Jr., and Lux, S. E. (1980) J. Biol. Chem. 255,

27. Kallner, A. (1975) Clin. Chim. Acta 59,35-39 28. Yan, T.-F. J., and Tao, M. (1982) J. Biol. Chem. 257,7037-7043

Supramol. Struct. 9,97-112

11938

11512-11520

29. Matsuo, M., Chang, L., Huang, C.-H., andVillar-Palasi, C. (1978) . .

FEBS Lett. 8 7 , 77-79 30. Cleland, W. W. (1970) in The Enzymes (Boyer, P. D., ed) Vol. 11,

31. Yan, J.T.-F., and Tao, M. (1982) J. Biol. Chem. 257, 7044-7049 32. Hunter, T.. and Sefton, B. M. (1980) Proc. Natl. Acad. Sci. U. S.

pp. 1-65, Academic Press, New York

A. 77,1311-1315 '

. .

33. Bennett. V., and Stenbuck, P. J. (1979) Nature 280,468-473 34. Weaver,'D.'C., and Marchesi, V.'T. (1984) J. Biol. Chem. 259,

6165-6169 35. Harris, H. W., Jr., Levin, N., and Lux, S. E. (1980) J. Biol. Chem.

255,11521-11525