Protein kinases of the thylakoid membrane.

The claim of Racker and co-workers (Lin, Z. F., Lucero, H. A., and Racker, E. (1982) J. Biol. Chem. 257, 12153-12156 and Lucero, H. A., Lin, Z. F., and Racker, E. (1982) J. Biol. Chem. 257, 12157-12160) that two protein kinases, designated CPK1 (25 kDa) and CPK2 (38 kDa), are present in spinach thylakoid membranes was investigated in light of results from this laboratory (Coughlan, S. J., and Hind, G. (1986) J. Biol. Chem. 261, 11378-11385) showing that 75-80% of the measurable protein kinase activity of isolated thylakoids is attributable to a protein kinase of 64 kDa apparent molecular mass. Extraction of thylakoid membranes with octyl glucoside/cholate according to the procedure of Lin et al. (Lin, Z. F., Lucero, H. A., and Racker, E. (1982) J. Biol. Chem. 257, 12153-12156) released proteins assignable to CPK1 and CPK2 on the basis of photoaffinity labeling with 8-azido-[32P]ATP. The 64-kDa protein kinase was present in this extract and accounted for greater than 80% of the total phosphotransferase activity toward lysine-rich histone as substrate; it was not labeled by the photoaffinity reagent. The three presumptive kinases were purified by ammonium sulfate precipitation, sucrose density gradient centrifugation, hydroxylapatite chromatography, and affinity chromatography. CPK1 was specifically eluted from Cibacron blue-Sepharose by 10 mM ATP; it electrophoresed on denaturing polyacrylamide gels as a single band with apparent molecular mass of 25 kDa. Its specific activity toward lysine-rich histone as substrate was approximately 250 pmol of phosphate transferred (mg protein)-1 min-1. The 64-kDa protein kinase was eluted from the affinity column by 1% (w/v) lithium dodecyl sulfate or from a histone IIIs-Sepharose affinity column by 0.25 M NaCl. Its specific activity towards lysine-rich histone was 100-200 times greater than that of CPK1. CPK2 eluted from the Cibacron blue affinity column in 10 mM NADP+; it had an apparent molecular mass of 38 kDa, possessed NADPH-dependent diaphorase activity (specific activity: 225 nmol of ferricyanide reduced (mg protein)-1 min-1), and cross-reacted with immunoglobulin raised against purified ferredoxin:NADP+ oxidoreductase, with which it was thus identified. Kinase activity was not detectable in CPK2 or in reductase isolated by conventional procedures.

3 To whom correspondence should be addressed. tyrosine residues in proteins (1,2). These enzymes are intimately involved in the control of metabolism in animal cells (3). Relatively little is known about the properties and functions of protein kinases in plants (4, 5), with one notable exception: the well-documented effectuation of State transitions in the thylakoid by reversible phosphorylation of LHC.' The responsible protein kinase is associated with the thylakoid membrane (5) and is regulated by the redox status of the plastoquinone pool (6).
Lin et al. (7) and Lucero et al. (8) reported the isolation and partial purification of two protein kinases from spinach thylakoid membranes. One had an apparent molecular mass of 25 kDa (CPK,) and the other 38 kDa (CPK,), as determined by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate. Since neither enzyme was able to utilize LHC as a substrate, the discovery of a third kinase was anticipated.
This enzyme has now been isolated and purified (9, 10). It has an apparent molecular mass of 64 kDa and phosphorylates isolated LHC at a modest rate (9, 10). Evidence for the existence of CPK, and CPK, was not obtained in these studies, even though the isolation protocols were similar (detergent solubilization of the thylakoid membrane by octyl glucoside/cholate). In order to reconcile the two bodies of data, we examined thylakoid membranes, octyl glucoside/ cholate extracts, and partially or completely purified subfractions with special attention to the reliability of 8-azid0- [~~P] ATP in detecting each of the three putative kinases. C P K , and CPK, were also purified to homogeneity for the first time, by applying affinity column chromatography to the final products from the procedure of Lin et al. (7).

Protein Kinases
of the Thylakoid Membrane 14063 the bottom of the tube. The pellet was redissolved in 0.5 ml of 0.1% Triton X-100, 10 mM sodium Tricine (pH 8.0). Samples were frozen and stored as above.
Photoaffinity Lubeling of Membranes with ~-Az~~o-[~'P]ATP-Samples were incubated with 4 pCi of 8-a~ido-[~~P]ATP, radiolabeled in the a or y position, for 5 min on ice in the dark, and then were irradiated for 10 min in the light from a General Electric germicidal lamp G15 T8 filtered through plate glass (4 J m-2 s-' at the sample surface).
Amlytical Techniques-Polyacrylamide gel electrophoresis was routinely performed in sodium dodecyl sulfate as in Ref. 11. Gels were fixed and stained with Coomassie Brilliant Blue (12) or Silver (13). Protein was determined according to Bradford (14) or, when detergent would interfere, Bensadoun and Weinstein (15). Protein kinase activities (7) and ferredoxin:NADP+ oxidoreductase activity (16) were measured by standard procedures. Chlorophyll concentration was determined by the method of Arnon (17).

Materials-8-A~ido-[y-~~P]ATP
, specific activity 22 mCi/pmol, 300 pCi in 300 pl of MeOH, and 8-a~ido-[a-~~P]ATP, specific activity 7.3 mCi/pmol, 120 pCi in 370 pl of MeOH were obtained from ICN Tracerlab. Cytochrome c oligomers used as molecular mass markers were from U. S. Biochemical, other markers were from Bio-Rad. Octyl glucoside was obtained from Behring Diagnostics and other reagents from Sigma.

RESULTS AND DISCUSSION
When thylakoid membranes were photoreacted with 8azido-[y-"P]ATP and analyzed by denaturing gel electrophoresis and autoradiography, two strongly labeled peptides were revealed ( Fig. 1). With reference to a mixture of cytochrome c oligomers as molecular mass markers, their relative masses were -38 and 23 kDa. The control lane showing thylakoid peptides, however, suggests a higher mass for the latter, close to that of the 25-kDa LHC peptide. Labeling of the 23-kDa peptide was enhanced by Mg'. Both peptides failed to incorporate 3ZP if 1 mM ATP or 1 mM 8-azido-ATP was present during the photoreaction; other nucleotides (ADP, AMP) were without effect. Dark controls showed no labeled proteins. Washing of membranes in 2 M NaBr to remove coupling factor had no effect on the labeling pattern; however, pretreatment of membranes with the detergent CHAPS, to remove ferredoxin:NADP+ oxidoreductase, abolished the labeling a t 38 kDa (Fig. 1).
Thylakoid membranes were treated with octyl glucoside/ cholate and the extract was fractionated as previously described (10) up to the Bio-Gel P300 step. Photolabeling of this kinase-enriched material with 8-azido-[y-"P]ATP resulted in major incorporation of radiolabel into peptides of apparent mass 38, 23, 19, and 16 kDa (Fig. 2, right). This labeling was light-dependent and was abolished by the presence of either ATP or 8-azido-ATP, as in membranes (data not shown).
The photoaffinity-labeled Bio-Gel P300 eluate was subjected to density gradient centrifugation followed by analysis of the fractions for radioactively labeled peptides, using gel electrophoresis, silver staining, and autoradiography. The cytochrome b/f complex, peaking between lanes 4 and 8 (Fig. 3), was not labeled (Fig. 2). The major labeled peptides evident in Fig. 2 are at 38 kDa (lanes 10-15), 23 kDa (lanes 11-15), 19 and 16 kDa (lanes 12-16). Less strongly labeled peptides were observed at 42 kDa (lanes 8-10), 34 kDa (lanes 2-6), and 32 kDa (lanes 11-15). In the converse experiment, shown in Fig. 4, P300 eluate was fractionated at the outset by density gradient centrifugation, then individual fractions were photoreacted with 8-azido-[a-"'P]ATP and analyzed as before. The principal labeling in this case was of the 23-kDa peptide (lanes [10][11][12][13][14][15] and the 38-kDa peptide (lanes 10-13). Again, this labeling required light and was prevented by prior addition of 1 mM ATP or of 1 mM 8-azido-ATP (not shown). The 38-kDa peptide was identified as ferredoxin:NADP+ oxidoreductase on the basis of its copurification with diaphorase activity and immune blotting with rabbit immunoglobulin raised to the purified enzyme from spinach (data not shown). Peptides of 42, 34, and 32 kDa showed minor labeling; however, there was no evident incorporation of label into 19-or 16-kDa components. No labeling of the cytochrome b/f complex was found. Comparable data were obtained in experiments corresponding to Figs. 2 and 4, in which the position of the radiolabel was interchanged or the temperature during incubation varied between 0 and -196 "C. None of the labeled peptides corresponded to the peak of histone kinase activity (not shown) centering upon fraction 10; furthermore, there was no measurable incorporation of radiolabel from either 8azido-[a-"PIATP or 8-azido-[y-"P]ATP into a 64-kDa peptide previously identified as an active protein kinase (10).
A direct attempt was made to isolate and purify the two  Table I) that no measurable protein kinase activity remained in the detergent-insoluble membrane fragments pelleted by ultracentrifugation. Solubilized kinase activity was precipitated by 55% (saturation) ammonium sulfate ( Table I). Separation of kinase activity from the coupling factor complex was effected by sucrose density gradient centrifugation (not shown), as originally reported (7). Fractions containing the peak of kinase activity and coincident diaphorase activity were pooled and examined by denaturing gel electrophoresis (Fig. 6, lane A ) , which revealed numerous peptide components. This material was adsorbed on hydroxylapatite and eluted with a linear 0-0.5 M phosphate gradient (7). All the protein was initially bound to the column. Almost half of the protein eluted at 50 mM phosphate (Figs. 5 and 6, lane 5), together with most of the kinase activity, the 32-kDa peptide as the major band staining with Coomassie Blue, and numerous other peptides present in minor amounts (Fig. 6). The 38and 16-kDa peptides eluted at 150 and 300 mM phosphate, respectively, the former being associated with a peak in diaphorase activity. Procedures"; 1 nmol of ATP was added, followed by trichloroacetic acid (25% (w/v) final concentration). After 30 min on ice, the samples were centrifuged (12,000 x g, 10 s); the pellets were rinsed with 250 pl of 5% (w/v) trichloroacetic acid and prepared for electrophoresis by dissolving in 40 pl of the medium given in   Thus far, no obvious candidate(s) for the kinase(s) had been established. However, introduction of an affinity chromatography step resolved this issue, as in the previous report (10). Fractions 5-7 from the hydroxylapatite column, containing the majority of histone kinase activity, were pooled, made to 0.5% Triton X-100, and loaded on a Cibacron blue-Sepharose column. Of the total applied protein, 75% was recovered in the throughput and initial washings (Fig. 7), including all of the 32-kDa peptide. All protein kinase and diaphorase activity was bound to the column. Elution with 10 mM NADP+ Fractions from the hydroxlapatite column, containing the peak in kinase activity, were combined, desalted, and brought to 0.5% Triton X-100, then were loaded (-1 mg protein in 1 ml) on a Cibacron blue-Sepharose column (0.5 X 3 cm) previously equlibrated with 0.1% Triton X-100, 0.1 mM PMSF, 10 mM sodium Tricine (pH 8.0). After 10 min (4 "C) the throughput (fraction I) was collected. The column was rinsed sequentially with this buffer and additives as follows: fractions 2 and 3, 1 ml each of buffer; fraction 4, 1 ml of buffer containing 50 mM NaCI; fraction 5, same as fraction 4, supplemented with 5 mM NADP+; fraction 6, repeat of fraction 4; fraction 7, 1 ml of buffer containing 5 mM ATP; fraction 8, repeat of fraction 4; fractions 9-13, 1 ml each of buffer containing 0.1,0.25,0.5, 1, and 2 M NaCI, respectively; fraction 14, 2 ml of buffer; fraction 15, 1 ml of 30 mM octyl glucoside/0.5% cholate, 10 mM sodium Tricine (pH 8.0); fraction 16, 1 ml of 1% (w/v) lithium dodecyl sulfate. Fractions 4-13 were dialyzed overnight against 0.1% Triton buffer. The open histogram shows protein content (pg/fraction), the solid area shows histone Vs kinase activity, and the hatched area shows reductase activity (units as for Fig. 5). Loadings (recoveries) were: protein, 1.8 (1.6) mg; histone kinase activity, 500 (100) units; reductase activity, 35 (32) units.  Fig. 7. Aliquots (50 pl) of gradient fractions were subjected to denaturing electrophoresis and the gel was silver stained (13). Lune A , material loaded on the affinity column (combined fractions 5-7 from hydroxylapatite chromatography (Fig. 6)); lanes 1-16, respective affinity column fractions from Fig. 7 (Fig. 7, fractions 5 and 6). The calculated specific diaphorase activity was 246 pmol of 2,6-dichlorophenolindophenol reduced (mg protein)" min" at pH 8.0. Electrophoretic analysis showed a major band lying at an apparent molecular mass of 38 kDa (Fig. 8), identifiable by Western blotting as ferredoxin:NADP+ oxidoreducatase. No measurable protein kinase activity was associated with these column fractions.

SILVER STAIN
Elution of the Cibacron blue-Sepharose column with 10 mM ATP (Fig. 8, fractions 7 and 8 ) yielded 10-20% of the total protein kinase activity a t a specific activity of -0.8 nmol of phosphate transferred (mg protein)" min" (-100 pmol of phosphate incorporated min" total activity associated with 120 pg of protein). Electrophoresis revealed a major peptide of apparent molecular mass 23 kDa in these fractions. Further washing of the column with high salt or ionic detergents removed no more protein kinase activity. Lithium dodecyl sulfate (1% w/v) removed several peptides from the column including a 64-kDa component previously identified as the major thylakoid protein kinase (10).
If the hydroxylapatite column fractions containing the peak of protein kinase activity were pooled and fractionated on a histone 111s-Sepharose affinity column, as previously described (lo), 95% of the total protein and 10-20% of the total protein kinase activity were recovered in the throughput and initial washings ( Table I). Eluting the column with 0.25-0.5 M NaCl yielded 80-90% of the total protein kinase activity and about 1% of the total protein (Table I). Analysis of these fractions by denaturing gel electrophoresis revealed the presence of a 23-kDa peptide in the throughput and initial washings (not shown) and a 64-kDa peptide (Fig. 8) in the 0.25 M NaCl eluate. The specific kinase activity of the latter, with histone 111s as substrate, was 20-30 nmol of phosphate transferred (mg protein)" min", in agreement with a previously reported value for this enzyme (IO). Thus, all protein kinase activity in the hydroxylapatite column fractions can be accounted for by the 23-kDa component (providing 10-20% of the total activity at a specific activity of 0.5-1.0 nmol of phosphate transferred (mg protein)" min" and the 64-kDa component (providing the remainder, a t a specific activity of 20-30 nmol (mg protein)" min-I).

CONCLUSIONS
Most of the protein kinase activity in these extracts, as in the thylakoid itself, is attributable to the 64-kDa enzyme (10).  (10). Kinase activity with histone 111s as substrate was assayed a t 25 "C over a linear, 30-min time course. Assay components were: 25-250 p~ ATP, 0-1 mM 8-azido-ATP, 25-50 ng of kinase (specific activity 50 nmol of phosphate transferred (mg protein)" min") and 1 pCi of [y"'P]ATP. Radioactive phosphoprotein was estimated by adsorption to paper ( 7 ) . The line of best fit of the data points was obtained by linear regression analysis.
We show here that use of 8-azido-ATP photoaffinity labeling for identifying kinase(s) is inappropriate in that ferredoxin:NADP+ oxidoreductase, and other polypeptides devoid of phosphotransferase activity are heavily labeled whereas the 64-kDa protein kinase is not, either in extracts or in the membrane. An explanation for this is presented in Fig. 9, which is a Dixon plot showing the effect of 8-azido-ATP on the rate of histone phosphorylation by the purified 64-kDa kinase, in the presence of different ATP concentrations. The photoaffinity reagent proved to be only a weak competitive inhibitor of the enzyme with a KI of -200 p~, compared to a 10-fold lower KM for ATP. The same values were found for the membrane-bound kinase, activated either by light or chemical reductant (data not shown). This low affinity, together with the low abundance of 64-kDa kinase in thylakoid membranes (-0.02% of the total membrane protein) and the low efficiency of cross-linking (between 0.01-O.l%), presumably weighs against the formation and detection of a crosslinked product between the kinase and 8-azido-ATP.

Protein Kinases of the Thylakoid Membrane
Our conclusions agree with those of Lin et al. (7) who proposed that a kinase other than CPKl or CPKz was responsible for phosphorylating LHC: we have shown that the 64-kDa enzyme could have this role (10). We also confirm here, the existence of CPK, as a peptide of apparent molecular mass 23 or 25 kDa (depending on the reference adopted) possessing weak kinase activity toward lysine-rich histone. This enzyme has been purified to homogeneity, using their protocol (7) followed by affinity chromatography on Cibacron blue-Sepharose.
We cannot confirm the existence of CPK2, however. The 38-kDa peptide designated CPKz on the basis of photoaffinity labeling is devoid of kinase activity and can be identified as ferredoxin:NADP' oxidoreductase. We have isolated this enzyme by several conventional procedures and confirmed its lack of phosphotransferase activity. It seems reasonable to conclude that the activity denoted CPKz in the preparation of Lin et al. (7) was due to contamination with 64-kDa protein kinase. Some, though not all, of the phosphotransferase activity assigned to CPKl could have this same origin, for Lucero et al. (8) showed that autophosphorylation of their crude CPK, fraction led to the labeling of a -60-kDa component.