- Email: [email protected]
Analysis of band 3 cytoplasmic domain phosphorylation and association with ankyrin
ARCHIVES OF BIOCHEMISTRY AND BIOPHYSICS Vol. 254, No. 2, May 1, pp. 509-517,1987
Analysis of Band 3 Cytoplasmic Domain Phosphorylation and Association with Ankyrin’ CHU-JING Department
SOONG, PAO-WEN
of Biological
Chemistry,
Received
University
July 21,1986,
LU,
of Illinois,
and in revised
AND
MARIANO
College
of Medicine,
form
December
TAO2 Chicago,
IUinois
60612
30,1986
The phosphorylation of the cytoplasmic domain of band 3 by the human erythrocyte membrane kinase and casein kinase A has been investigated. The cytoplasmic domain of band 3 was released from erythrocyte vesicles by treatment with a-chymotrypsin and isolated as a 43,000-Da peptide. Both the membrane kinase and casein kinase A catalyzed the incorporation of about 1 mol of phosphate per mole of the band 3 fragment. The phosphorylation of the band 3 fragment by both kinases was not additive, suggesting that the two enzymes might recognize the same phosphorylation sites. Also in support of this notion was the observation that the phosphopeptide maps of the band 3 fragment phosphorylated by the two kinases were identical. Phosphoamino acid analysis of the band 3 fragment phosphorylated by casein kinase A revealed the presence of approximately equal amounts of phosphoserine and phosphothreonine and, to a lesser extent, phosphotyrosine. The interaction between the 43,000-Da peptide with ankyrin and the effect of phosphorylation on this interaction have been examined. The band 3 fragment was found to form two different types of complexes, termed Cl and C2, with ankyrin in a saturable manner. The Cl and C2 complexes contained about 1.7 and 0.43 mol of band 3 fragment per mole of ankyrin, respectively. Interestingly, these binding stoichiometries were found to be reduced by half by the phosphorylation of ankyrin but not by the phosphorylation of the band 3 fragment. The results suggest that the structure and dynamics of the erythrocyte membrane cytoskeletal network may be regulated by phosphorylation. 0 1987 Academic Press, Inc. Studies on the interactions between intrinsic and extrinsic proteins of the erythrocyte membrane indicate that the membrane cytoskeletal network is attached to the membrane bilayer, in part, by the association of spectrin with ankyrin. Ankyrin, in turn, is bound to the cytoplasmic domain of band 3, an integral membrane protein that also functions as an anion transporter. The cytoplasmic domain of band 3 has been obtained as a 43,000-Da fragment by mild digestion of inside-out vesicles depleted of spectrin and actin with cu-chymotrypsin (l-3). This fragment has i This work was supported tional Institutes of Health Chicago Heart Association. 2 To whom correspondence
by grants (AM23045) should
been purified in relatively large quantities (1,2). In addition to ankyrin, studies have shown that this domain contains binding sites for the glycolytic enzymes, aldolase, phosphofructokinase, and glyceraldehyde3-phosphate dehydrogenase (4). A binding site for hemoglobin has also been identified (5). The association between ankyrin and band 3 has been investigated using two different approaches. One of these involves the measurement of the association of ankyrin with inverted vesicles depleted of ankyrin, spectrin, actin, 4.1, 4.2, and 6 (3). Bennett and Stenbuck (3) showed that the ankyrin bound to band 3 in the inverted vesicles with an affinity constant (KD) of 4.6 X lo-’ M. Hargreaves et al. (6) have similarly examined the reassociation of an-
from the Naand from the be addressed. 509
0003-9861/87 Cowriaht All rights
$3.06
0 1987 bv Academic Press. Inc. of reproduction in any form reserved.
510
SOONG,
LU,
kyrin with band 3 in erythrocyte membranes and in lipid vesicles. They found that ankyrin reassociated at high affinity (K. 5-8 X lo-* M) with a limited number of protease-sensitive sites located on the cytoplasmic side of the erythrocyte membrane. On the other hand, the binding of ankyrin to band 3 reconstituted into phosphatidylcholine vesicles was found to exhibit high and low affinity binding characteristics. An alternate approach to determining the interaction between ankyrin and band 3 is by measurement of the direct association between ankyrin and the ‘%Ilabeled 43,000-Da fragment in solution. The binary complex formed was coupled to antiankyrin antibodies and the IgG-ankyrin43,000-Da fragment complex was adsorbed to protein A-bearing staphylococci and collected by centrifugation (3). These studies indicate that ankyrin and the 43,000Da fragment associated with a KD of about 8 nM and an estimated stoichiometry of 0.85 mol of fragment per mole of ankyrin (3). However, the Scatchard plot of the binding data was curvilinear and appeared to reflect a more complicated mode of interaction between the two components. The association of the cytoplasmic domain of band 3 with ankyrin in solution has also been examined using nondenaturing polyacrylamide gradient gel electrophoresis (7). In this study, a Scatchard plot of the binding of the 43,000-Da fragment to ankyrin revealed a single class of high affinity sites corresponding to a Ko of 8 X lo-* M and a binding stoichiometry of 1 mol of band 3 fragment per mole of ankyrin. The nature of the discrepancies among these studies in terms of the different binding affinities as well as association constants remains to be determined. Both ankyrin and band 3 are phosphoproteins. Based on data obtained from membrane autophosphorylation, ankyrin was found to be phosphorylated by both cyclic AMP-dependent and -independent protein kinases, whereas the phosphorylation of band 3 was due primarily to the activity of the cyclic AMP-independent enzyme (8). The effect of phosphorylation of ankyrin by either the membrane-bound
AND
TAO
cyclic AMP-independent protein kinase or a cytosolic casein kinase on its interaction with spectrin has been examined recently by our laboratory (8). We showed that the phosphorylated ankyrin exhibited a lower affinity for either phosphorylated or unphosphorylated spectrin tetramer but not spectrin dimer. In contrast, phosphorylation of spectrin had no effect on its binding to ankyrin. Bennett (9) has examined the effect of dephosphorylation of ankyrin and band 3 by phosphatases and concluded that the interaction between these two membrane proteins is not affected by their states of phosphorylation. In this report, we have examined the phosphorylation of the cytoplasmic domain of band 3 using protein kinases purified from the erythrocyte membrane and cytosol. Our results indicate that the affinity of band 3 fragment for ankyrin is not affected by phosphorylation. However, the stoichiometry of binding of band 3 fragment to ankyrin appears to decrease by the phosphorylation of ankyrin. EXPERIMENTAL
PROCEDURES
Muterids. Human blood was obtained from either the University of Illinois Hospital blood bank or the United Blood Services and used within 2 weeks of the drawing date. [y-**P]ATP and ‘l-labeled BoltonHunter reagent were purchased from Amersham Corp. cY-Chymotrypsin and aprotinin were supplied by Sigma Chemical Co. Leupeptin and pepstatin A were obtained from Fluka Chemical Corp. and Transformation Research Inc., respectively. All other reagents were of analytical grade. Preparation of the cytoplusmic domain of band 3. Human erythrocyte membranes were prepared as described earlier by Lu et al (8). The cytoplasmic domain of band 3 was obtained from treatment of acid stripped, spectrin-depleted vesicles with a-chymotrypsin according to the procedure described by Bennett (2). The fragments released from the membranes were further purified by DEAE-cellulose (DE-52) chromatography and by gel filtration on Sephacryl S-200 column. The purified preparation was analyzed by SDS’-polyacrylamide gel electrophoresis. Over 80% of the fragment was a polypeptide of Mr 43,000 and
‘Abbreviation
used: SDS, sodium
dodecyl
sulfate.
INTERACTION
BETWEEN
the remainder was that of M, 41,000. Data obtained from immunological studies indicate that the 41,000Da polypeptide is probably a subfragment of the larger polypeptide (3). No further attempt was made to separate these two fragments and the preparation will be referred to as the 43K peptide. Preparation of ankyrin and protein kinosea Ankyrin was isolated and purified from human erythrocyte membranes as previously described (8). The ankyrin preparation was judged to be homogeneous based on SDS-polyacrylamide gel electrophoresis. The human erythrocyte membrane cyclic AMP-independent protein kinase (10) and the cytosolic cyclic AMP-independent protein kinase or casein kinase A (11) were purified to homogeneity based on procedures recently developed in our laboratory (8). The kinases were assayed as described earlier; and one unit of kinase activity is defined as that amount of enzyme which catalyzes the incorporation of one nanomole of phosphate into casein per minute (8). Phosphoqlation assay. The phosphorylation of ankyrin and 43K peptide was conducted at 37°C for 1 h in a reaction mixture containing 50 mM Tris-HCl, pH 7.5, 7.5 mM MgCla, 0.2 mM [T-~‘P]ATP, 0.4 mg/ml ankyrin or 43K peptide, 3 units/ml kinase, 40 mM (for ankyrin) or 5 mM (for 43K peptide) KCI, and 10 pg/ ml each of the protease inhibitors (leupeptin, pepstatin A, and aprotinin). The incorporation of phosphate into ankyrin and 43K peptide was analyzed by SDS-polyacrylamide gel electrophoresis (10). The radioactivity incorporated was determined by excising the protein band from the dried gel and counting in a liquid scintillation spectrometer. *e51-Labeling of &%‘K peptide. The 43K peptide was radioiodinated at 0°C for 90 min in a reaction mixture containing 1 mCi of ‘?-labeled Bolton-Hunter reagent, 1 mg of 43K peptide, 0.16 M of Na-borate, pH 8.5, and 250 mM of NaCl, in a final volume of 0.4 ml. The unbound ‘=I was removed by dialysis against a buffer (binding buffer) containing 20 mM Tris-HCI, pH 7.5, 130 mM KCI, 20 mM NaC1, 0.5 mM 2-mercaptoethanol, 20 @g/ml phenylmethylsulfonyl fluoride, and 2 pgg/ml each of the protease inhibitors, followed by repeated washing with the same buffer in an Amicon microconcentrator. The final concentration of stock ‘?-labeled 43K peptide was 1 to 1.5 mg/ml with specific activity of 4-10 X 10” cpm/pg. The phosphorylation of ?-labeled 43K peptide with unlabeled ATP was conducted as described in the preceding section. The excess unreacted ATP was removed by repeated washing with the binding buffer in an Amicon microconcentrator. Binding assay. The binding of ‘ZI-labeled 43K peptide to ankyrin was determined by nondenaturing gel electrophoresis on a 2 to 4% polyacrylamide gradient gel slab according to the method described by Weaver et al. (7).
BAND
3 AND
ANKYRIN
511
Varying amounts of phosphorylated and unphosphorylated ?-labeled 43K peptide were incubated for 30 min on ice with 16 pg of phosphorylated or unphosphorylated ankyrin in 60 ~1 of the binding buffer according to Weaver et al. (7). Immediately prior to electrophoresis, 10 cl1 of a 40% sucrose was added to the incubation mixture and all 70 pl was loaded onto the gel. Electrophoresis was performed at 60 V for 20 h in a cold room. After the run, the gel was stained with Coomassie brilliant blue R-250, dried under vacuum, and exposed to X-ray films. The radioactive bands were also excised from the gel and counted in a gamma counter. RESULTS
Phosphorylation of .&‘K peptide. The phosphorylation of band 3 was initially reported in membrane preparations. The labeling of band 3 in membrane preparations incubated with [y-32P]ATP was found to occur via a membrane-bound cyclic AMPindependent protein kinase-catalyzed reaction (12). Subsequent studies using purified kinase preparations and dimethylmaleic anhydride-extracted membranes indicate that band 3 is a substrate of both the membrane cyclic AMP-independent protein kinase and casein kinase A isolated from the cytosolic fraction (11,12). The dimethylmaleic anhydride-extracted membranes contained mainly the integral membrane proteins, band 3, and the sialoglycoproteins, and were devoid of any kinase activity as measured by membrane autophosphorylation. In light of these observations, the possibility that the cytoplasmic domain of band 3 may serve as substrate of the membrane and cytosolic kinases was examined. Figure 1 shows the SDS-polyacrylamide gel electrophoretic profile and the phosphorylation pattern of the cytoplasmic domain of band 3. The phosphorylation was conducted using casein kinase A. As shown in the figure, radioactivity is incorporated into both the 43K and the 41K peptides. The same result was obtained with the membrane kinase (data not shown). The phosphorylation of the 43K peptide exhibited a broad pH activity profile. Maximal incorporation of phosphates into the 43K peptide was observed between pH 7
512
SOONG,
“j@g q& Y ,p2,,-mi
LU,
.p.,qr7 .pip; I,>, ,F’J1’*‘i.7, ,P?Ji3>3
;$ ? ; ,= . T $’ 9. .a -, 3’
‘,& pig I ! = :*-g:a:;,; “‘c”aY, *
A
0
+29K
FIG. 1. SDS-polyacrylamide gel electrophoresis of the 43K peptide phosphorylated by casein kinase A. The electrophoresis of 43K (6 pg), phosphorylated as described under Experimental Procedures, was conducted on a 5% polyacrylamide gel slab. The specific activities of [r-S”P]ATP used for the phosphorylation of 43K was 420 cpm/pmol. (A) stained gel; (B) radioautogram.
and 8.5. In this study, the reaction was routinely conducted at pH 7.5. Mgz+ was required for the reaction and a sharp Mgz+ optimum was observed at about 7.5 II1M. KCl, at 5 mM, was found to be slightly stimulatory; whereas at concentrations greater than 75 mM, inhibition of phosphorylation was observed. The above reaction parameters were the same for both the membrane kinase and casein kinase A. Figure 2 shows the time course of phosphorylation of the 43K peptide by the two kinases under optimal phosphorylation conditions. The results show that each kinase can catalyze the incorporation of a maximum of about 1 mol of phosphate per mole of 43K peptide. The kinetics of phosphorylation of the 43K peptide by the membrane kinase and casein kinase A have been examined using varying concentrations of ATP and 43K peptide. Figure 3 shows the kinetics of the reaction catalyzed by casein kinase A as analyzed by double reciprocal plots of the initial velocity versus the concentrations of ATP at different fixed levels of 43K peptide, and vice versa. The data suggest that the reaction mechanism of the kinase is
AND
TAO
consistent with a sequential bireactant reaction kinetics involving the formation of a ternary complex consisting of the kinase, ATP, and the 43K peptide (13). This reaction mechanism may be typical for reactions catalyzed by protein kinases (14). From replots of the slope and intercepts of the double reciprocal plots shown in Fig. 3, Km values of 2.5 PM and 80 pg/ml were obtained for ATP and the 43K peptide, respectively. Basically the same results were also obtained for the membrane kinasecatalyzed reaction. The phosphoamino acids obtained from acid hydrolysis of 32P-labeled 43K peptide were analyzed by high voltage electrophoresis at pH 1.9 and 3.5 as described earlier (15). The results indicate that a large proportion of the 32P label is attached primarily to seryl and threonyl residues. The 32P label was equally distributed between these two amino acids. Upon closer examination, however, we have also detected trace phosphorylation of tyrosyl residue (16). In order to establish the relationship, if any, between the membrane kinase and casein kinase A, we have continued to com-
::* 0.1) f R 2 0.4 P 0
40
80
.__c
120
MINUTES
FIG. 2. Time course of phosphorylation of the 43K peptide by the membrane kinase and casein kinase A. The phosphorylation of the 43K peptide was conducted in 200 pl of a reaction mixture containing 60.2 ng of the 43K peptide, 0.2 mM [T-~P]ATP (460 cpm/pmol), 50 mM Tris-HCl, pH 7.5, 7.5 mM MgCls, 5 mM KCl, and 4 units/ml of either the membrane kinase (0) or casein kinase A (0). The reaction mixture was incubated at 3’7’C; and at the time indicated, an aliquot (20 ~1) was withdrawn for determination of =P incorporation.
INTERACTION
BETWEEN
BAND
3 AND
513
ANKYRIN
A . i
e
.8 -
1/43K(mg/mlbl
I
1
0
I
.2
I
.4
1 /ATP(rM)-’
FIG. 3. Double reciprocal plots of initial velocity versus substrate concentration. The reactions were carried out in the presence of 4 units/ml casein kinase A. The incubation was conducted at 3’7°C for 4 min. (A) Double reciprocal plots of initial velocity versus 43K concentration at different fixed levels of ATP (0, 5 pM; 0,lO j.tM; A, 25 pM; and A, 50 HIM). (B) Double reciprocal plots of the initial velocity versus ATP concentration at different fixed levels of 43K peptide (0,0.04 mg/ml; 0, 0.08 mg/ml; A, 0.16 mg/ml; and A, 0.32 mg/ml).
pare the substrate specificity of these enzymes. Our earlier study with ankyrin indicates that the two kinases may recognize the same phosphorylation sites on this protein (8). In this study, we likewise examined the activity of both kinases toward the 43K peptide. First, we found that the degree of phosphorylation of the 43K peptide in the presence of both kinases was the same as in the presence of either kinase alone (Fig. 2). Since the phosphorylation is not additive, it suggests that the kinases catalyze the phosphorylation of the same
sites on the 43K peptide. The data obtained from the phosphopeptide maps of 43K peptide phosphorylated by the membrane kinase and casein kinase A appear to support this possibility. We found no discernible differences between the phosphopeptide maps of the 43K peptide phosphorylated by the two kinases. E3ect.s of phosphorglation on the interaction between the .@K peptide and ankgriti The binding of ankyrin to both band 3 and spectrin has been widely believed to represent the major mechanism by which the
514
SOONG,
LU,
cytoskeletal network is attached to the membrane. The possibility that these interactions may be modified by phosphorylation has been investigated in light of the common knowledge that band 3, ankyrin, and spectrin are all phosphoproteins. Previous studies indicate that phosphorylation of ankyrin reduces significantly its affinity for spectrin tetramer (8). Therefore, it is of interest to determine whether phosphorylation also affects the interaction between band 3 and ankyrin. The association between the 43K peptide of band 3 with ankyrin has been investigated by the method of Weaver et al. (7) using nondenaturing gradient gel electrophoresis. In order to monitor the formation of complexes, the 43K band 3 fragment was labeled with ‘%I. As shown in Fig. 4, the mobilities of the 43K peptide and ankyrin on the 2-4% gradient gel are distinctly different and that the two polypeptides are clearly resolved on the gel. It is of interest to note that the phosphorylated ankyrin (Fig. 4, lane 2) appears to migrate further to the anode on the gel than the unphosphorylated form (lane 1). This increase in mobility is not due to proteolysis of ankyrin during incubation. Phosphorylated and unphosphorylated ankyrin exhibit the same mobility on SDS-polyacrylamide gel electrophoresis. Since ankyrin was phosphorylated to the extent of about 7 mol of phosphate per mole of ankyrin, these added negative charges could conceivably account for the observed increase in mobility on the gradient gel. Somewhat surprisingly we found that incubation of the 43K peptide with ankyrin resulted in the formation of two different complexes, Cl and C2, which were resolved on the gel (Fig. 4, lanes 5-8). One of these complexes, C2, has about the same mobility as the corresponding phosphorylated or unphosphorylated ankyrin. It should be noted that in the absence of band 3, phosphorylated and unphosphorylated ankyrin migrated as a single protein band on the gel as shown in Fig. 4. The relative distribution of proteins in Cl and C2 has been determined by scanning the Coomassie blue stained gel and estimated to be about
AND
TAO
12
345676
'*, +@ 12
3-
4
5
6
7
6
WC. 4. Nondenaturing gradient gel electrophoretic analysis of the interaction between ‘?-labeled 43K peptide and ankyrin. The labeling of the 43K peptide with ‘%I, the phosphorylation of the ‘zI-labeled 43K peptide and ankyrin, and the binding assay were as described under Experimental Procedures. Lanes 1 and 2 contain only ankyrin (15.6 pg) and phosphoankyrin (15.8 pg), respectively. Lanes 3 and 4 contain 15 rg of unphosphorylated and phosphorylated ‘=I-labeled 43K peptide, respectively. Lane 5 contains an incubation mixture of ankyrin (15.6 pg) and ‘?-labeled 43K peptide (4.2 pg); whereas lane 6 contains ankyrin (15.6 ng) and phosphorylated ‘=I-labeled 43K peptide (4.2 gg). Lane 7 contains an incubation mixture of pbosphoankyrin (15.8 rg) and ‘l-labeled 43K peptide (4.2 ng); whereas lane 8 contains phosphoankyrin and phosphorylated ‘l-labeled 43K peptide. (A) Coomassie blue stained gel; (B) radioautogram. The Cl and C2 complexes correspond to the top and lower bands of the complexes between the 43K peptide and ankyrin or their phosphorylated forms.
1:5. As shown in Figure 4B, both Cl and C2 contained ‘l-labeled 43K peptide. The formation of the Cl and C2 complexes was investigated in the presence of varying amounts of the 43K peptide. The radioactivities in the Cl and C2 complexes were analyzed by excising the protein bands and counted in a gamma counter. Figure 5 shows that ankyrin binds the 43K peptide in a saturable manner. Although
INTERACTION
BETWEEN
BAND
3 AND
ANKYRIN
515
no significant effect on its binding to phosphorylated and unphosphorylated ankyrin to form the Cl and C2 complexes; and the binding isotherm was in general agreement with that shown in Fig. 5. DISCUSSION
40 43K
’
60
(rrg/ml)
FIG. 5. Binding of 1251-labeled 43K peptide to unphosphorylated and phosphorylated ankyrin. Ankyrin was phosphorylated with ATP in a reaction catalyzed by casein kinase A as described under Experimental Procedures. The associations of the 43K peptide with unphosphorylated (A, A) and phosphorylated (0, 0) ankyrin in the Cl and C2 complexes were determined by excising and counting the radioactive bands after separation of the complexes by gel electrophoresis as described in Fig. 4.
the overall radioactivity found in the C2 complex was significantly greater than that in Cl, the Cl complex contained more radioactivity per microgram of protein than the C2 complex. Based on our preliminary estimates, the Cl complex was found to contain approximately 1.7 mol of 43K peptide per mole of ankyrin; whereas the C2 complex was found to contain about 0.43 mol of 43K peptide per mole of ankyrin. These values are based on rough estimates of the amount of protein in each band by densitometric scanning of the gel. The error limits between duplicate experiments were 3-10s .Interestingly, phosphorylation of ankyrin significantly reduced its ability to bind band 3 and the amounts of 43K peptide bound to the phosphorylated ankyrin in the Cl and C2 complexes were estimated to be 0.8 and 0.28 mol per mole, respectively. The effect of phosphorylation of the 43K peptide on its binding to phosphorylated and unphosphorylated ankyrin was also investigated. As shown in Fig. 6, phosphorylation of the 43K peptide exhibited
Due to the difficulty in working with the integral membrane band 3 protein in solution in the absence of detergent and the propensity of the protein to form irreversible molecular aggregates (1’7), studies on the cytoskeletal function of band 3 have focused on the cytoplasmic domain of the protein. The band 3 cytoplasmic domain can be obtained by digestion with cu-chymotrypsin and has been isolated in highly purified form (1, 2). The 43K proteolytic fragment appears to retain the binding properties of band 3 (3,4,7). In this report, we have shown that the 43K peptide can also serve as substrate of the membranebound and cytosolic casein kinases. Although preliminary estimates indicate that approximately 1 mol of phosphate is incorporated per mole of 43K peptide, this
FIG. 6. Binding of phosphorylated ‘?-labeled 43K peptide to unphosphorylated and phosphorylated ankyrin. The experimental details are essentially as described under Fig. 5, except that the 43K peptide is phosphorylated with casein kinase A. A, A, binding to unphosphorylated ankyrin (15.6 pg/ml); 0, l , binding to phosphorylated ankyrin (15.8 pg/ml).
516
SOONG.
LU,
value may represent the sum of phosphates incorporated into several phosphorylation sites. That there may be more than one phosphorylation site on the band 3 fragment has been deduced from phosphoamino acid analysis. Phosphoserine, phosphothreonine, and phosphotyrosine have been detected in the acid hydrolysates of band 3 fragment phosphorylated by casein kinase A. Approximately equal amounts of 3ZP-labeled phosphoserine and phosphothreonine were recovered from the acid hydrolysate. The presence of phosphotyrosine was detected only when the 32P-labeled peptide was subjected to partial acid hydrolysis (16). Thus it can be concluded from these observations that there are at least three different phosphorylation sites on the 43K fragment of band 3. Whether all of these phosphorylation sites are localized within the lO,OOO-Da region of the NH2 terminal remains to be determined (18). That band 3 may contain multiple phosphorylation sites has been suggested earlier by Drickamer (19). One of these sites has been identified as tyrosine 8 located in the NH2 terminal cytoplasmic domain. This residue is found to be phosphorylated by a membrane-bound tyrosine kinase (l&20). It is of interest to note that the solubilized chymotryptic fragment representing the cytoplasmic domain can also serve as substrate for the tyrosine kinase (20). In an effort to gain insight into the role of protein phosphorylation in the regulation of membrane cytoskeletal function and dynamics, we also examined the effect of phosphorylation on the interaction between the 43K fragment and ankyrin. The binding studies employing nondenaturing gel electrophoresis have yielded an unexpected and interesting result. Our data indicate that two different complexes are formed between ankyrin and the 43K peptide. The results suggest that there may be two different species of ankyrin exhibiting a difference in their binding stoichiometry for the 43K peptide. This interpretation is based on the observation (Fig. 5) that the formation of the complexes is found to reach saturation in the presence of varying
AND
TAO
amounts of 43K peptide and a constant amount of ankyrin. If all ankyrin species were alike, one would suspect that complex C2 would be converted to Cl at increasing concentrations of 43K peptide. As a result, there would be a decrease in the C2 complex and an increase in the Cl complex as the concentration of the 43K peptide was increased; and the result obtained would be quite different from that shown in Fig. 5. The possibility of microheterogeneity in the ankyrin preparation and the presence of multiple forms of ankyrin is presently under investigation in our laboratory. The complex Cl, which represents the minor component, contains about 2 mol of 43K peptide per mole of ankyrin. On the other hand, complex C2 exhibits a binding stoicbiometry of about 0.5 mol of 43K peptide per mole of ankyrin. The phosphorylation of the 43K peptide did not affect its interaction with ankyrin and the same binding stoichiometric values were obtained for the Cl and C2 complexes. In contrast, phosphorylation of ankyrin appears to decrease significantly the ratios of 43K peptide to ankyrin in the Cl and C2 complexes. The reason for this change in the binding property of ankyrin is not known but preliminary data indicate that this does not appear to be due to a change in the affinity constants of ankyrin for the 43K peptide (C-J. Soong, unpublished work). These data, together with our previous observations (8, 21), strongly suggest that pbosphorylation of ankyrin and band 4.1 may play an important role in the regulation of membrane cytoskeletal structure and dynamics. REFERENCES 1. APELL, K. C., AND Low, P. S. (1981) J. Biol Chem. 256,11104-11111. 2. BENNETT, V. (1983) in Methods in Enzymology (Fleischer, S., and Fleischer, B., Eds.), Vol. 96, pp. 313-342, Academic Press, Orlando, FL. 3. BENNETT, V., AND STENBUCK, D. J. (1980) J. Biol Chem 255,6424+X32. 4. BENNETP, V. (1985) Annu Rev. Biochem. 54,273304. 5. WALDER, J. A., CHATl’ERJEE, R., STECK, T. L., LOW, P. S., Musso, G. F., KAISER, E. T., ROGERS, P. H.,
INTERACTION
AND ARNONE, 10238-10246. 6. HARGREAVES,
A. (1984)
W. R., GIEDD,
AND BRANTON, 11965-11972.
J. Biol
BETWEEN
Chem
K. N., VERKLEIJ,
BAND
259, A.,
14.
D. (1980) J. Biol. Chem 255,
7. WEAVER, D. C., PASTERNACK, G. R., AND MARCHESI, V. T. (1984) J. BioL Chem. 259,61’70-61’75. 8. Lu, P.-W., SOONG, C. J., AND TAO, M. (1985) J. Biol. Chem 260,14958-14964. 9. BENNE~, V. (1982) J. Cell. B&hem l&49-65. 10. TAO, M., CONWAY, R., AND CHETA, S. (1980) J. BioL Chem. 255,2563-2568. 11. SIMKOWSKI, K. W., AND TAO, M. (1980) J. BioL Chem 255,6456-6461. 12. HOSEY, M. M., AND TAO, M. (1977) J. Sup-raw& Stmct. 6, 61-75. 13. CLELAND, W. W. (1970) in The Enzymes (Boyer,
15. 16. 17. 18. 19. 20. 21.
3 AND
ANKYRIN
517
P. D., Ed.), Vol. 2, pp. l-65, Academic Press, New York. MATSUO, M., CHANG, L., HUANG, C.-H., AND VILLAR-PALASI, C. (19%) $%Bs htt. 87,77-79. YAN, T.-F., AND TAO, M. (1982) J. BioL C&m. 257, 7044-7049. Lu, P.-W., AND TAO, M. (1986) Biochem Biophys. Res. Commun 139,855-860. Yu, J., AND STECK, T. L. (1975) J. BioL Chem. 250, 9170-9175. DEKOWSKI, S. A., RYBICKI, A., AND DRICKAMER, K. (1983) J. BioL Chem. 253,2750-2753. DRICKAMER, L. K. (1976) J BioL Chem. 251,55155123. MOHAMED, A. H., AND STECK, T. L. (1986) J. BioL Chem 261,2804-2809. EDER, P., SOONG, C.-J., AND TAO, M. (1986) Biochemistry 25,1764-1770.
Analysis of Band 3 Cytoplasmic Domain Phosphorylation and Association with Ankyrin’ CHU-JING Department
SOONG, PAO-WEN
of Biological
Chemistry,
Received
University
July 21,1986,
LU,
of Illinois,
and in revised
AND
MARIANO
College
of Medicine,
form
December
TAO2 Chicago,
IUinois
60612
30,1986
The phosphorylation of the cytoplasmic domain of band 3 by the human erythrocyte membrane kinase and casein kinase A has been investigated. The cytoplasmic domain of band 3 was released from erythrocyte vesicles by treatment with a-chymotrypsin and isolated as a 43,000-Da peptide. Both the membrane kinase and casein kinase A catalyzed the incorporation of about 1 mol of phosphate per mole of the band 3 fragment. The phosphorylation of the band 3 fragment by both kinases was not additive, suggesting that the two enzymes might recognize the same phosphorylation sites. Also in support of this notion was the observation that the phosphopeptide maps of the band 3 fragment phosphorylated by the two kinases were identical. Phosphoamino acid analysis of the band 3 fragment phosphorylated by casein kinase A revealed the presence of approximately equal amounts of phosphoserine and phosphothreonine and, to a lesser extent, phosphotyrosine. The interaction between the 43,000-Da peptide with ankyrin and the effect of phosphorylation on this interaction have been examined. The band 3 fragment was found to form two different types of complexes, termed Cl and C2, with ankyrin in a saturable manner. The Cl and C2 complexes contained about 1.7 and 0.43 mol of band 3 fragment per mole of ankyrin, respectively. Interestingly, these binding stoichiometries were found to be reduced by half by the phosphorylation of ankyrin but not by the phosphorylation of the band 3 fragment. The results suggest that the structure and dynamics of the erythrocyte membrane cytoskeletal network may be regulated by phosphorylation. 0 1987 Academic Press, Inc. Studies on the interactions between intrinsic and extrinsic proteins of the erythrocyte membrane indicate that the membrane cytoskeletal network is attached to the membrane bilayer, in part, by the association of spectrin with ankyrin. Ankyrin, in turn, is bound to the cytoplasmic domain of band 3, an integral membrane protein that also functions as an anion transporter. The cytoplasmic domain of band 3 has been obtained as a 43,000-Da fragment by mild digestion of inside-out vesicles depleted of spectrin and actin with cu-chymotrypsin (l-3). This fragment has i This work was supported tional Institutes of Health Chicago Heart Association. 2 To whom correspondence
by grants (AM23045) should
been purified in relatively large quantities (1,2). In addition to ankyrin, studies have shown that this domain contains binding sites for the glycolytic enzymes, aldolase, phosphofructokinase, and glyceraldehyde3-phosphate dehydrogenase (4). A binding site for hemoglobin has also been identified (5). The association between ankyrin and band 3 has been investigated using two different approaches. One of these involves the measurement of the association of ankyrin with inverted vesicles depleted of ankyrin, spectrin, actin, 4.1, 4.2, and 6 (3). Bennett and Stenbuck (3) showed that the ankyrin bound to band 3 in the inverted vesicles with an affinity constant (KD) of 4.6 X lo-’ M. Hargreaves et al. (6) have similarly examined the reassociation of an-
from the Naand from the be addressed. 509
0003-9861/87 Cowriaht All rights
$3.06
0 1987 bv Academic Press. Inc. of reproduction in any form reserved.
510
SOONG,
LU,
kyrin with band 3 in erythrocyte membranes and in lipid vesicles. They found that ankyrin reassociated at high affinity (K. 5-8 X lo-* M) with a limited number of protease-sensitive sites located on the cytoplasmic side of the erythrocyte membrane. On the other hand, the binding of ankyrin to band 3 reconstituted into phosphatidylcholine vesicles was found to exhibit high and low affinity binding characteristics. An alternate approach to determining the interaction between ankyrin and band 3 is by measurement of the direct association between ankyrin and the ‘%Ilabeled 43,000-Da fragment in solution. The binary complex formed was coupled to antiankyrin antibodies and the IgG-ankyrin43,000-Da fragment complex was adsorbed to protein A-bearing staphylococci and collected by centrifugation (3). These studies indicate that ankyrin and the 43,000Da fragment associated with a KD of about 8 nM and an estimated stoichiometry of 0.85 mol of fragment per mole of ankyrin (3). However, the Scatchard plot of the binding data was curvilinear and appeared to reflect a more complicated mode of interaction between the two components. The association of the cytoplasmic domain of band 3 with ankyrin in solution has also been examined using nondenaturing polyacrylamide gradient gel electrophoresis (7). In this study, a Scatchard plot of the binding of the 43,000-Da fragment to ankyrin revealed a single class of high affinity sites corresponding to a Ko of 8 X lo-* M and a binding stoichiometry of 1 mol of band 3 fragment per mole of ankyrin. The nature of the discrepancies among these studies in terms of the different binding affinities as well as association constants remains to be determined. Both ankyrin and band 3 are phosphoproteins. Based on data obtained from membrane autophosphorylation, ankyrin was found to be phosphorylated by both cyclic AMP-dependent and -independent protein kinases, whereas the phosphorylation of band 3 was due primarily to the activity of the cyclic AMP-independent enzyme (8). The effect of phosphorylation of ankyrin by either the membrane-bound
AND
TAO
cyclic AMP-independent protein kinase or a cytosolic casein kinase on its interaction with spectrin has been examined recently by our laboratory (8). We showed that the phosphorylated ankyrin exhibited a lower affinity for either phosphorylated or unphosphorylated spectrin tetramer but not spectrin dimer. In contrast, phosphorylation of spectrin had no effect on its binding to ankyrin. Bennett (9) has examined the effect of dephosphorylation of ankyrin and band 3 by phosphatases and concluded that the interaction between these two membrane proteins is not affected by their states of phosphorylation. In this report, we have examined the phosphorylation of the cytoplasmic domain of band 3 using protein kinases purified from the erythrocyte membrane and cytosol. Our results indicate that the affinity of band 3 fragment for ankyrin is not affected by phosphorylation. However, the stoichiometry of binding of band 3 fragment to ankyrin appears to decrease by the phosphorylation of ankyrin. EXPERIMENTAL
PROCEDURES
Muterids. Human blood was obtained from either the University of Illinois Hospital blood bank or the United Blood Services and used within 2 weeks of the drawing date. [y-**P]ATP and ‘l-labeled BoltonHunter reagent were purchased from Amersham Corp. cY-Chymotrypsin and aprotinin were supplied by Sigma Chemical Co. Leupeptin and pepstatin A were obtained from Fluka Chemical Corp. and Transformation Research Inc., respectively. All other reagents were of analytical grade. Preparation of the cytoplusmic domain of band 3. Human erythrocyte membranes were prepared as described earlier by Lu et al (8). The cytoplasmic domain of band 3 was obtained from treatment of acid stripped, spectrin-depleted vesicles with a-chymotrypsin according to the procedure described by Bennett (2). The fragments released from the membranes were further purified by DEAE-cellulose (DE-52) chromatography and by gel filtration on Sephacryl S-200 column. The purified preparation was analyzed by SDS’-polyacrylamide gel electrophoresis. Over 80% of the fragment was a polypeptide of Mr 43,000 and
‘Abbreviation
used: SDS, sodium
dodecyl
sulfate.
INTERACTION
BETWEEN
the remainder was that of M, 41,000. Data obtained from immunological studies indicate that the 41,000Da polypeptide is probably a subfragment of the larger polypeptide (3). No further attempt was made to separate these two fragments and the preparation will be referred to as the 43K peptide. Preparation of ankyrin and protein kinosea Ankyrin was isolated and purified from human erythrocyte membranes as previously described (8). The ankyrin preparation was judged to be homogeneous based on SDS-polyacrylamide gel electrophoresis. The human erythrocyte membrane cyclic AMP-independent protein kinase (10) and the cytosolic cyclic AMP-independent protein kinase or casein kinase A (11) were purified to homogeneity based on procedures recently developed in our laboratory (8). The kinases were assayed as described earlier; and one unit of kinase activity is defined as that amount of enzyme which catalyzes the incorporation of one nanomole of phosphate into casein per minute (8). Phosphoqlation assay. The phosphorylation of ankyrin and 43K peptide was conducted at 37°C for 1 h in a reaction mixture containing 50 mM Tris-HCl, pH 7.5, 7.5 mM MgCla, 0.2 mM [T-~‘P]ATP, 0.4 mg/ml ankyrin or 43K peptide, 3 units/ml kinase, 40 mM (for ankyrin) or 5 mM (for 43K peptide) KCI, and 10 pg/ ml each of the protease inhibitors (leupeptin, pepstatin A, and aprotinin). The incorporation of phosphate into ankyrin and 43K peptide was analyzed by SDS-polyacrylamide gel electrophoresis (10). The radioactivity incorporated was determined by excising the protein band from the dried gel and counting in a liquid scintillation spectrometer. *e51-Labeling of &%‘K peptide. The 43K peptide was radioiodinated at 0°C for 90 min in a reaction mixture containing 1 mCi of ‘?-labeled Bolton-Hunter reagent, 1 mg of 43K peptide, 0.16 M of Na-borate, pH 8.5, and 250 mM of NaCl, in a final volume of 0.4 ml. The unbound ‘=I was removed by dialysis against a buffer (binding buffer) containing 20 mM Tris-HCI, pH 7.5, 130 mM KCI, 20 mM NaC1, 0.5 mM 2-mercaptoethanol, 20 @g/ml phenylmethylsulfonyl fluoride, and 2 pgg/ml each of the protease inhibitors, followed by repeated washing with the same buffer in an Amicon microconcentrator. The final concentration of stock ‘?-labeled 43K peptide was 1 to 1.5 mg/ml with specific activity of 4-10 X 10” cpm/pg. The phosphorylation of ?-labeled 43K peptide with unlabeled ATP was conducted as described in the preceding section. The excess unreacted ATP was removed by repeated washing with the binding buffer in an Amicon microconcentrator. Binding assay. The binding of ‘ZI-labeled 43K peptide to ankyrin was determined by nondenaturing gel electrophoresis on a 2 to 4% polyacrylamide gradient gel slab according to the method described by Weaver et al. (7).
BAND
3 AND
ANKYRIN
511
Varying amounts of phosphorylated and unphosphorylated ?-labeled 43K peptide were incubated for 30 min on ice with 16 pg of phosphorylated or unphosphorylated ankyrin in 60 ~1 of the binding buffer according to Weaver et al. (7). Immediately prior to electrophoresis, 10 cl1 of a 40% sucrose was added to the incubation mixture and all 70 pl was loaded onto the gel. Electrophoresis was performed at 60 V for 20 h in a cold room. After the run, the gel was stained with Coomassie brilliant blue R-250, dried under vacuum, and exposed to X-ray films. The radioactive bands were also excised from the gel and counted in a gamma counter. RESULTS
Phosphorylation of .&‘K peptide. The phosphorylation of band 3 was initially reported in membrane preparations. The labeling of band 3 in membrane preparations incubated with [y-32P]ATP was found to occur via a membrane-bound cyclic AMPindependent protein kinase-catalyzed reaction (12). Subsequent studies using purified kinase preparations and dimethylmaleic anhydride-extracted membranes indicate that band 3 is a substrate of both the membrane cyclic AMP-independent protein kinase and casein kinase A isolated from the cytosolic fraction (11,12). The dimethylmaleic anhydride-extracted membranes contained mainly the integral membrane proteins, band 3, and the sialoglycoproteins, and were devoid of any kinase activity as measured by membrane autophosphorylation. In light of these observations, the possibility that the cytoplasmic domain of band 3 may serve as substrate of the membrane and cytosolic kinases was examined. Figure 1 shows the SDS-polyacrylamide gel electrophoretic profile and the phosphorylation pattern of the cytoplasmic domain of band 3. The phosphorylation was conducted using casein kinase A. As shown in the figure, radioactivity is incorporated into both the 43K and the 41K peptides. The same result was obtained with the membrane kinase (data not shown). The phosphorylation of the 43K peptide exhibited a broad pH activity profile. Maximal incorporation of phosphates into the 43K peptide was observed between pH 7
512
SOONG,
“j@g q& Y ,p2,,-mi
LU,
.p.,qr7 .pip; I,>, ,F’J1’*‘i.7, ,P?Ji3>3
;$ ? ; ,= . T $’ 9. .a -, 3’
‘,& pig I ! = :*-g:a:;,; “‘c”aY, *
A
0
+29K
FIG. 1. SDS-polyacrylamide gel electrophoresis of the 43K peptide phosphorylated by casein kinase A. The electrophoresis of 43K (6 pg), phosphorylated as described under Experimental Procedures, was conducted on a 5% polyacrylamide gel slab. The specific activities of [r-S”P]ATP used for the phosphorylation of 43K was 420 cpm/pmol. (A) stained gel; (B) radioautogram.
and 8.5. In this study, the reaction was routinely conducted at pH 7.5. Mgz+ was required for the reaction and a sharp Mgz+ optimum was observed at about 7.5 II1M. KCl, at 5 mM, was found to be slightly stimulatory; whereas at concentrations greater than 75 mM, inhibition of phosphorylation was observed. The above reaction parameters were the same for both the membrane kinase and casein kinase A. Figure 2 shows the time course of phosphorylation of the 43K peptide by the two kinases under optimal phosphorylation conditions. The results show that each kinase can catalyze the incorporation of a maximum of about 1 mol of phosphate per mole of 43K peptide. The kinetics of phosphorylation of the 43K peptide by the membrane kinase and casein kinase A have been examined using varying concentrations of ATP and 43K peptide. Figure 3 shows the kinetics of the reaction catalyzed by casein kinase A as analyzed by double reciprocal plots of the initial velocity versus the concentrations of ATP at different fixed levels of 43K peptide, and vice versa. The data suggest that the reaction mechanism of the kinase is
AND
TAO
consistent with a sequential bireactant reaction kinetics involving the formation of a ternary complex consisting of the kinase, ATP, and the 43K peptide (13). This reaction mechanism may be typical for reactions catalyzed by protein kinases (14). From replots of the slope and intercepts of the double reciprocal plots shown in Fig. 3, Km values of 2.5 PM and 80 pg/ml were obtained for ATP and the 43K peptide, respectively. Basically the same results were also obtained for the membrane kinasecatalyzed reaction. The phosphoamino acids obtained from acid hydrolysis of 32P-labeled 43K peptide were analyzed by high voltage electrophoresis at pH 1.9 and 3.5 as described earlier (15). The results indicate that a large proportion of the 32P label is attached primarily to seryl and threonyl residues. The 32P label was equally distributed between these two amino acids. Upon closer examination, however, we have also detected trace phosphorylation of tyrosyl residue (16). In order to establish the relationship, if any, between the membrane kinase and casein kinase A, we have continued to com-
::* 0.1) f R 2 0.4 P 0
40
80
.__c
120
MINUTES
FIG. 2. Time course of phosphorylation of the 43K peptide by the membrane kinase and casein kinase A. The phosphorylation of the 43K peptide was conducted in 200 pl of a reaction mixture containing 60.2 ng of the 43K peptide, 0.2 mM [T-~P]ATP (460 cpm/pmol), 50 mM Tris-HCl, pH 7.5, 7.5 mM MgCls, 5 mM KCl, and 4 units/ml of either the membrane kinase (0) or casein kinase A (0). The reaction mixture was incubated at 3’7’C; and at the time indicated, an aliquot (20 ~1) was withdrawn for determination of =P incorporation.
INTERACTION
BETWEEN
BAND
3 AND
513
ANKYRIN
A . i
e
.8 -
1/43K(mg/mlbl
I
1
0
I
.2
I
.4
1 /ATP(rM)-’
FIG. 3. Double reciprocal plots of initial velocity versus substrate concentration. The reactions were carried out in the presence of 4 units/ml casein kinase A. The incubation was conducted at 3’7°C for 4 min. (A) Double reciprocal plots of initial velocity versus 43K concentration at different fixed levels of ATP (0, 5 pM; 0,lO j.tM; A, 25 pM; and A, 50 HIM). (B) Double reciprocal plots of the initial velocity versus ATP concentration at different fixed levels of 43K peptide (0,0.04 mg/ml; 0, 0.08 mg/ml; A, 0.16 mg/ml; and A, 0.32 mg/ml).
pare the substrate specificity of these enzymes. Our earlier study with ankyrin indicates that the two kinases may recognize the same phosphorylation sites on this protein (8). In this study, we likewise examined the activity of both kinases toward the 43K peptide. First, we found that the degree of phosphorylation of the 43K peptide in the presence of both kinases was the same as in the presence of either kinase alone (Fig. 2). Since the phosphorylation is not additive, it suggests that the kinases catalyze the phosphorylation of the same
sites on the 43K peptide. The data obtained from the phosphopeptide maps of 43K peptide phosphorylated by the membrane kinase and casein kinase A appear to support this possibility. We found no discernible differences between the phosphopeptide maps of the 43K peptide phosphorylated by the two kinases. E3ect.s of phosphorglation on the interaction between the .@K peptide and ankgriti The binding of ankyrin to both band 3 and spectrin has been widely believed to represent the major mechanism by which the
514
SOONG,
LU,
cytoskeletal network is attached to the membrane. The possibility that these interactions may be modified by phosphorylation has been investigated in light of the common knowledge that band 3, ankyrin, and spectrin are all phosphoproteins. Previous studies indicate that phosphorylation of ankyrin reduces significantly its affinity for spectrin tetramer (8). Therefore, it is of interest to determine whether phosphorylation also affects the interaction between band 3 and ankyrin. The association between the 43K peptide of band 3 with ankyrin has been investigated by the method of Weaver et al. (7) using nondenaturing gradient gel electrophoresis. In order to monitor the formation of complexes, the 43K band 3 fragment was labeled with ‘%I. As shown in Fig. 4, the mobilities of the 43K peptide and ankyrin on the 2-4% gradient gel are distinctly different and that the two polypeptides are clearly resolved on the gel. It is of interest to note that the phosphorylated ankyrin (Fig. 4, lane 2) appears to migrate further to the anode on the gel than the unphosphorylated form (lane 1). This increase in mobility is not due to proteolysis of ankyrin during incubation. Phosphorylated and unphosphorylated ankyrin exhibit the same mobility on SDS-polyacrylamide gel electrophoresis. Since ankyrin was phosphorylated to the extent of about 7 mol of phosphate per mole of ankyrin, these added negative charges could conceivably account for the observed increase in mobility on the gradient gel. Somewhat surprisingly we found that incubation of the 43K peptide with ankyrin resulted in the formation of two different complexes, Cl and C2, which were resolved on the gel (Fig. 4, lanes 5-8). One of these complexes, C2, has about the same mobility as the corresponding phosphorylated or unphosphorylated ankyrin. It should be noted that in the absence of band 3, phosphorylated and unphosphorylated ankyrin migrated as a single protein band on the gel as shown in Fig. 4. The relative distribution of proteins in Cl and C2 has been determined by scanning the Coomassie blue stained gel and estimated to be about
AND
TAO
12
345676
'*, +@ 12
3-
4
5
6
7
6
WC. 4. Nondenaturing gradient gel electrophoretic analysis of the interaction between ‘?-labeled 43K peptide and ankyrin. The labeling of the 43K peptide with ‘%I, the phosphorylation of the ‘zI-labeled 43K peptide and ankyrin, and the binding assay were as described under Experimental Procedures. Lanes 1 and 2 contain only ankyrin (15.6 pg) and phosphoankyrin (15.8 pg), respectively. Lanes 3 and 4 contain 15 rg of unphosphorylated and phosphorylated ‘=I-labeled 43K peptide, respectively. Lane 5 contains an incubation mixture of ankyrin (15.6 pg) and ‘?-labeled 43K peptide (4.2 pg); whereas lane 6 contains ankyrin (15.6 ng) and phosphorylated ‘=I-labeled 43K peptide (4.2 gg). Lane 7 contains an incubation mixture of pbosphoankyrin (15.8 rg) and ‘l-labeled 43K peptide (4.2 ng); whereas lane 8 contains phosphoankyrin and phosphorylated ‘l-labeled 43K peptide. (A) Coomassie blue stained gel; (B) radioautogram. The Cl and C2 complexes correspond to the top and lower bands of the complexes between the 43K peptide and ankyrin or their phosphorylated forms.
1:5. As shown in Figure 4B, both Cl and C2 contained ‘l-labeled 43K peptide. The formation of the Cl and C2 complexes was investigated in the presence of varying amounts of the 43K peptide. The radioactivities in the Cl and C2 complexes were analyzed by excising the protein bands and counted in a gamma counter. Figure 5 shows that ankyrin binds the 43K peptide in a saturable manner. Although
INTERACTION
BETWEEN
BAND
3 AND
ANKYRIN
515
no significant effect on its binding to phosphorylated and unphosphorylated ankyrin to form the Cl and C2 complexes; and the binding isotherm was in general agreement with that shown in Fig. 5. DISCUSSION
40 43K
’
60
(rrg/ml)
FIG. 5. Binding of 1251-labeled 43K peptide to unphosphorylated and phosphorylated ankyrin. Ankyrin was phosphorylated with ATP in a reaction catalyzed by casein kinase A as described under Experimental Procedures. The associations of the 43K peptide with unphosphorylated (A, A) and phosphorylated (0, 0) ankyrin in the Cl and C2 complexes were determined by excising and counting the radioactive bands after separation of the complexes by gel electrophoresis as described in Fig. 4.
the overall radioactivity found in the C2 complex was significantly greater than that in Cl, the Cl complex contained more radioactivity per microgram of protein than the C2 complex. Based on our preliminary estimates, the Cl complex was found to contain approximately 1.7 mol of 43K peptide per mole of ankyrin; whereas the C2 complex was found to contain about 0.43 mol of 43K peptide per mole of ankyrin. These values are based on rough estimates of the amount of protein in each band by densitometric scanning of the gel. The error limits between duplicate experiments were 3-10s .Interestingly, phosphorylation of ankyrin significantly reduced its ability to bind band 3 and the amounts of 43K peptide bound to the phosphorylated ankyrin in the Cl and C2 complexes were estimated to be 0.8 and 0.28 mol per mole, respectively. The effect of phosphorylation of the 43K peptide on its binding to phosphorylated and unphosphorylated ankyrin was also investigated. As shown in Fig. 6, phosphorylation of the 43K peptide exhibited
Due to the difficulty in working with the integral membrane band 3 protein in solution in the absence of detergent and the propensity of the protein to form irreversible molecular aggregates (1’7), studies on the cytoskeletal function of band 3 have focused on the cytoplasmic domain of the protein. The band 3 cytoplasmic domain can be obtained by digestion with cu-chymotrypsin and has been isolated in highly purified form (1, 2). The 43K proteolytic fragment appears to retain the binding properties of band 3 (3,4,7). In this report, we have shown that the 43K peptide can also serve as substrate of the membranebound and cytosolic casein kinases. Although preliminary estimates indicate that approximately 1 mol of phosphate is incorporated per mole of 43K peptide, this
FIG. 6. Binding of phosphorylated ‘?-labeled 43K peptide to unphosphorylated and phosphorylated ankyrin. The experimental details are essentially as described under Fig. 5, except that the 43K peptide is phosphorylated with casein kinase A. A, A, binding to unphosphorylated ankyrin (15.6 pg/ml); 0, l , binding to phosphorylated ankyrin (15.8 pg/ml).
516
SOONG.
LU,
value may represent the sum of phosphates incorporated into several phosphorylation sites. That there may be more than one phosphorylation site on the band 3 fragment has been deduced from phosphoamino acid analysis. Phosphoserine, phosphothreonine, and phosphotyrosine have been detected in the acid hydrolysates of band 3 fragment phosphorylated by casein kinase A. Approximately equal amounts of 3ZP-labeled phosphoserine and phosphothreonine were recovered from the acid hydrolysate. The presence of phosphotyrosine was detected only when the 32P-labeled peptide was subjected to partial acid hydrolysis (16). Thus it can be concluded from these observations that there are at least three different phosphorylation sites on the 43K fragment of band 3. Whether all of these phosphorylation sites are localized within the lO,OOO-Da region of the NH2 terminal remains to be determined (18). That band 3 may contain multiple phosphorylation sites has been suggested earlier by Drickamer (19). One of these sites has been identified as tyrosine 8 located in the NH2 terminal cytoplasmic domain. This residue is found to be phosphorylated by a membrane-bound tyrosine kinase (l&20). It is of interest to note that the solubilized chymotryptic fragment representing the cytoplasmic domain can also serve as substrate for the tyrosine kinase (20). In an effort to gain insight into the role of protein phosphorylation in the regulation of membrane cytoskeletal function and dynamics, we also examined the effect of phosphorylation on the interaction between the 43K fragment and ankyrin. The binding studies employing nondenaturing gel electrophoresis have yielded an unexpected and interesting result. Our data indicate that two different complexes are formed between ankyrin and the 43K peptide. The results suggest that there may be two different species of ankyrin exhibiting a difference in their binding stoichiometry for the 43K peptide. This interpretation is based on the observation (Fig. 5) that the formation of the complexes is found to reach saturation in the presence of varying
AND
TAO
amounts of 43K peptide and a constant amount of ankyrin. If all ankyrin species were alike, one would suspect that complex C2 would be converted to Cl at increasing concentrations of 43K peptide. As a result, there would be a decrease in the C2 complex and an increase in the Cl complex as the concentration of the 43K peptide was increased; and the result obtained would be quite different from that shown in Fig. 5. The possibility of microheterogeneity in the ankyrin preparation and the presence of multiple forms of ankyrin is presently under investigation in our laboratory. The complex Cl, which represents the minor component, contains about 2 mol of 43K peptide per mole of ankyrin. On the other hand, complex C2 exhibits a binding stoicbiometry of about 0.5 mol of 43K peptide per mole of ankyrin. The phosphorylation of the 43K peptide did not affect its interaction with ankyrin and the same binding stoichiometric values were obtained for the Cl and C2 complexes. In contrast, phosphorylation of ankyrin appears to decrease significantly the ratios of 43K peptide to ankyrin in the Cl and C2 complexes. The reason for this change in the binding property of ankyrin is not known but preliminary data indicate that this does not appear to be due to a change in the affinity constants of ankyrin for the 43K peptide (C-J. Soong, unpublished work). These data, together with our previous observations (8, 21), strongly suggest that pbosphorylation of ankyrin and band 4.1 may play an important role in the regulation of membrane cytoskeletal structure and dynamics. REFERENCES 1. APELL, K. C., AND Low, P. S. (1981) J. Biol Chem. 256,11104-11111. 2. BENNETT, V. (1983) in Methods in Enzymology (Fleischer, S., and Fleischer, B., Eds.), Vol. 96, pp. 313-342, Academic Press, Orlando, FL. 3. BENNETT, V., AND STENBUCK, D. J. (1980) J. Biol Chem 255,6424+X32. 4. BENNETP, V. (1985) Annu Rev. Biochem. 54,273304. 5. WALDER, J. A., CHATl’ERJEE, R., STECK, T. L., LOW, P. S., Musso, G. F., KAISER, E. T., ROGERS, P. H.,
INTERACTION
AND ARNONE, 10238-10246. 6. HARGREAVES,
A. (1984)
W. R., GIEDD,
AND BRANTON, 11965-11972.
J. Biol
BETWEEN
Chem
K. N., VERKLEIJ,
BAND
259, A.,
14.
D. (1980) J. Biol. Chem 255,
7. WEAVER, D. C., PASTERNACK, G. R., AND MARCHESI, V. T. (1984) J. BioL Chem. 259,61’70-61’75. 8. Lu, P.-W., SOONG, C. J., AND TAO, M. (1985) J. Biol. Chem 260,14958-14964. 9. BENNE~, V. (1982) J. Cell. B&hem l&49-65. 10. TAO, M., CONWAY, R., AND CHETA, S. (1980) J. BioL Chem. 255,2563-2568. 11. SIMKOWSKI, K. W., AND TAO, M. (1980) J. BioL Chem 255,6456-6461. 12. HOSEY, M. M., AND TAO, M. (1977) J. Sup-raw& Stmct. 6, 61-75. 13. CLELAND, W. W. (1970) in The Enzymes (Boyer,
15. 16. 17. 18. 19. 20. 21.
3 AND
ANKYRIN
517
P. D., Ed.), Vol. 2, pp. l-65, Academic Press, New York. MATSUO, M., CHANG, L., HUANG, C.-H., AND VILLAR-PALASI, C. (19%) $%Bs htt. 87,77-79. YAN, T.-F., AND TAO, M. (1982) J. BioL C&m. 257, 7044-7049. Lu, P.-W., AND TAO, M. (1986) Biochem Biophys. Res. Commun 139,855-860. Yu, J., AND STECK, T. L. (1975) J. BioL Chem. 250, 9170-9175. DEKOWSKI, S. A., RYBICKI, A., AND DRICKAMER, K. (1983) J. BioL Chem. 253,2750-2753. DRICKAMER, L. K. (1976) J BioL Chem. 251,55155123. MOHAMED, A. H., AND STECK, T. L. (1986) J. BioL Chem 261,2804-2809. EDER, P., SOONG, C.-J., AND TAO, M. (1986) Biochemistry 25,1764-1770.