Dependence of ATP-citrate lyase kinase activity on the phosphorylation of ATP-citrate lyase by cyclic AMP-dependent protein kinase.

ATP-citrate lyase from rat liver and adipose tissue is phosphorylated by either ATP-citrate lyase kinase or catalytic subunit of cyclic AMP-dependent protein kinase to 0.5-0.6 mol/subunit. We previously demonstrated that the site phosphorylated by ATP-citrate lyase kinase (peptide B) is different from that phosphorylated by catalytic subunit of cyclic AMP-dependent protein kinase (peptide A) (Ramakrishna, S., Pucci, D. L., and Benjamin, W.B. (1981) J. Biol. Chem. 256, 10213-10216). ATP-citrate lyase phosphorylation by both protein kinases added simultaneously was increased synergistically. When ATP-citrate lyase was first phosphorylated by catalytic subunit of cyclic AMP-dependent protein kinase, the net phosphorylation of the fragments subsequently phosphorylated by lyase kinase increased about 6-fold. However, when ATP-citrate lyase was first phosphorylated by lyase kinase, there was no effect on the subsequent phosphorylation of the enzyme by cyclic AMP-dependent protein kinase. Alkaline phosphatase-dephosphorylated ATP-citrate lyase was phosphorylated by catalytic subunit of cyclic AMP-dependent protein kinase to 0.9-1.0 mol/subunit. However, dephospho-ATP-citrate lyase was not phosphorylated by lyase kinase. The addition of both protein kinases simultaneously phosphorylated ATP-citrate lyase up to 2 mol/subunit. Phosphorylation of dephospho-ATP-citrate lyase first by catalytic subunit of cyclic AMP-dependent protein kinase and ATP enabled the lyase to be phosphorylated by lyase kinase. Peptide mapping and phosphoamino acid analysis of dephospho-ATP-citrate lyase phosphorylated by catalytic subunit of cyclic AMP-dependent protein kinase and/or lyase kinase conclusively showed that phosphorylation of ATP-citrate lyase by ATP-citrate lyase kinase was completely dependent on peptide A phosphorylation by cyclic AMP-dependent protein kinase. Furthermore, increased phosphorylation when both protein kinases were added simultaneously was due to increased phosphorylation at peptide B.

ated phosphorylation is via the adenylate cyclase-cyclic AMPdependent phosphorylation system (6). Since insulin action also stimulates ATP-citrate lyase phosphorylation and experimental evidence suggests a cyclic AMP-independent pathway for this phosphorylation (Z), we proposed that an early event in insulin action is the stimulation of an insulin-sensitive protein kinase which phosphorylates ATP-citrate lyase and possibly other physiological substrates. Subsequently, we isolated from liver a cyclic AMP-independent protein kinase which phosphorylates ATP-citrate lyase in vitro to 0.5-0.6 mol/subunit ( 7 ) . In addition, rat liver and fat ATP-citrate lyase could be phosphorylated by cyclic AMP-dependent protein kinase also to 0.5-0.6 mol/subunit (with endogenous structural phosphate content of 0.5 mol/subunit) (8-lo), while rat mammary gland ATP-citrate lyase with endogenous structural phosphate content of 0.2 mol/subunit could be phosphorylated to 0.7 mol/subunit (11). However, the function of these ATP-citrate lyase phosphorylations on enzyme activity and metabolism is not known.
We have shown (8) that in vitro the ATP-citrate lyase kinase-responsive phosphorylation of lyase is different from the cyclic AMP-dependent protein kinase-responsive system since different sites in the lyase molecule are phosphorylated by the two different protein kinases. Serine is the only amino acid phosphorylated by catalytic subunit of cyclic AMP-dependent protein kinase, while lyase kinase phosphorylates both serine and threonine residues.
Recently, we have noted that incubation of ATP-citrate lyase simultaneously with both protein kinases for longer periods of time than previously reported (8) resulted in the synergistic phosphorylation of the lyase. We therefore asked whether, ( a ) ATP-citrate lyase phosphorylation by one protein kinase facilitates phosphorylation by the other protein kinase, and ( b ) ATP-citrate lyase dephosphorylation by alkaline phosphatase alters the amount of phosphorylation or the sites of phosphorylation by these protein kinases?
Here we present evidence that the ability of lyase kinase to phosphorylate ATP-citrate lyase at a unique peptide sequence (peptide B) is absolutely dependent on the lyase's prior phosphorylation at the site phosphorylated by cyclic AMP-dependent protein kinase (peptide A). With increasing lyase phosphorylation by cyclic AMP-dependent protein kinase, lyase kinase increasingly is able to phosphorylate ATP-citrate lyase at peptide B. Therefore, the phosphorylation of the enzyme by both protein kinases added together is more than the sum of the phosphorylation of native ATP-citrate lyase when it is incubated with each protein kinase separately. protein) was prepared from rat adipose tissue' and liver by modification (12) of a published method (13). Iodoacetate, guanidine hydrochloride, ATP, SDS, Tris, Escherichia coli alkaline phosphatase, and other biochemicals were purchased from Sigma. TPCK-trypsin' was from Worthington. [Y-~*P]ATP (3000 Ci/mmol) was purchased from Amersham Corp. ATP-citrate lyase kinase was purified from rat liver as described previously (7) with some modification.' Catalytic subunit was prepared from rabbit skeletal muscle by method I1 of Bechtel et al. (14) with an additional hydroxylapatite chromatogaphy step. TLC cellulose plates were from E. Merck. X-ray films XAR-2 were from Kodak.
HPLC of ATP-citrate Lyase-ATP-citrate lyase phosphorylated by either protein kinase or both kinases was reduced, carboxymethylated, and digested with TPCK-trypsin (8). Two-dimensional separation of the peptides and phosphoamino acid analysis have been described (8). Trypsin (trypsinsubstrate, 1:50)-digested ATP-citrate lyase was chromatographed on a Waters reversed-phase pBondapak Cle column (0.39 X 30 cm). Sample (40 p1) was injected onto the column equilibrated with 5% acetonitrile and 0.1% triiluoroacetic acid. The column was washed for 5 rnin with 5% acetonitrile, and peptides were eluted with a linear gradient of 5 4 0 % acetonitrile in 0.1% trifluoroacetic acid for 70 min followed by 40% acetonitrile in 0.1% trifluoroacetic acid for 5 min. The flow rate was 1 ml/min and peptides were monitored by absorbance at 210 nm. Radiolabeled phosphopeptides were located by collecting 1-min fractions and measuring their Cerenkov radiation. Chromatography of a completely digested sample of catalytic subunit phosphorylated lyase gave a single radiolabeled phosphopeptide with RE = 26 min (peptide A, Fig. 1B). With incomplete digestion, another phosphopeptide with a retention time of 36 this study confirmed that further digestion of phosphopeptide A with min (peptide A ) was found. Pierce et al. (16) have shown and we in trypsin yielded phosphopeptide A on rechromatography. Peptides A and A on acid hydrolysis contained only 3'P-labeled phosphoserine. Chromatography of lyase kinase-phosphorylated ATP-citrate lyase, however, gave a major radiolabeled phosphopeptide B (R, = 72 rnin) and few minor radiolabeled phosphopeptides with Rr = 47, 50, 55, 64, and 67 min ( Fig. l A , 0). All minor peptides and peptide B contained both phosphoserine and phosphothreonine labeled with 32P. Fractions of each phosphopeptide peak were pooled, desalted with Sep-Pak Cla (Waters), and lyophilized. Phosphopeptides were dissolved in 40% acetonitrile containing 0.2 M phosphate buffer and chromatographed on a calibrated column of (0.38 X 60 cm) Spherogel TSK-2000 SW (Beckman) at a flow rate of 0.2 ml/min. Preparation of Dephospho-and Phosphorylated ATP-citrate Lyase-ATP-citrate lyase was dephosphorylated by incubating 1-1.5 m g / d of homogenous liver ATP-citrate lyase with 0.4 mg/ml of E. coli alkaline phosphatase for 3-6 h at 30 "C in a reaction mixture containing 20 m Tris-HC1 (pH 8.5), 0.2 mg/ml of bovine serum albumin, and 2 I I M~ MgS04. The reaction mixture was passed through a Sephadex G-150 superfine column (0.8 X 16 cm) equilibrated with 50 mM MES buffer (pH 6.7) containing 50 mM NaCl and 5 mM 2mercaptoethanol. Dephospho-ATP-citrate lyase free of the phosphatase was recovered in the void volume. ATP-citrate lyase incubated in the above reaction mixture without added alkaline phosphatase and processed similarly served as the control in the comparative ' S. Ramakrishna   experiments. ATP-citrate lyase was phosphorylated with unlabeled ATP using catalytic subunit or lyase kinase. The reaction mixture (200 pl) was passed through a Sephadex G-150 superfine column (0.7 x 3.0 cm), and ATP-citrate lyase free of ATP and the added protein kinase was collected in the void volume.
Approximately equal amounts of these phosphorylated ATP-citrate lyase samples were again phosphorylated by incubating them with [y3*P]ATP and the appropriate protein kinase as described under "Phosphorylation Assay". Radiolabeled ATP-citrate lyase was isolated as the 115,000dalton subunit by SDS-gel electrophoresis. Enzyme activities of dephospho-and phosphorylated and native liver ATP-citrate lyase were comparable when assayed either at pH 8.7 or 7.5, suggesting that during these procedures the enzyme did not undergo extensive denaturation at least in the region of the complex active site@). Stained SDS gels did not show any increase in ATP-citrate lyase fragments in phosphorylated and dephosphorylated samples compared to native enzyme, suggesting proteolysis was negligible during alkaline phosphatase treatment and phosphorylation experiments.

RESULTS
The rate of native ATP-citrate lyase phosphorylation by lyase kinase, catalytic subunit, or combinations of both kinases is shown in Fig. 2. When lyase was phosphorylated by lyase kinase, phosphate incorporation was rapid and linear for 30 min, reaching about 0.25 mol of P/subunit at 3 h. When lyase was phosphorylated by catalytic subunit, phosphate incorporation was linear and rapid for the first 30 m i n , with 0.62 mol of P/subunit incorporated at 3 h (Fig. 2). The calculated phosphate incorporation into ATP-citrate lyase, if the protein kinases were added together using the previous data determined when each protein kinase was added separately, should be 0.87 mol of P/subunit. However, when ATP-citrate lyase was incubated with lyase kinase and catalytic subunit together, the phosphate incorporation was linear for about 2 h and reached 1.55 mol of P/subunit at 3 h. After the simdta-

FIG. 2.
Phosphorylation of rat liver ATP-citrate lyase. Phosphorylation assays were performed for 15 min to 3 h as described under "Experimental Procedures." Rat liver ATP-citrate lyase (300 pg/ml) was phosphorylated with ATP-citrate lyase kinase (W), catalytic subunit (X-X), and combination of the two protein kinases (W). The phosphates incorporated in ATP-citrate lyase by lyase kinase and catalytic subunit were added (---) to calculate the expected phosphorylation by the addition of both protein kinases to the assay. neous addition of both protein kinases, phosphate incorporation was approximately 20% higher than expected during the fiist 30 min of incubation, 70% higher at 1 h, and 80% higher at 3 h.
To demonstrate whether phosphorylation first by one protein kinase increases phosphorylation by the other or vice versa. ATP-citrate lyase was phosphorylated first with unlabeled ATP for various times using catalytic subunit or lyase kinase as the protein kinase (Table I). The phospho-ATPcitrate lyase was isolated and then phosphorylated for 0.5, 1.5, and 3 h with [y3'P]ATP and the other appropriate protein kinase. Lyase kinase phosphorylated control ATP-citrate lyase to 0.13-0.21 mol of P/subunit. However, phospho-ATPcitrate lyase generated by prior phosphorylation with catalytic subunit for 1 h was further phosphorylated by lyase kinase to 0.58 mol of P/subunit at 0.5 h and to 0.96 mol of P/subunit at 3 h. When phospho-ATP-citrate lyase prepared by 4 h of incubation with catalytic subunit was incubated for an additional 3 h with lyase kinase, 1.26 mol of P/subunit were incorporated. This represents approximately a 6-fold increase in phosphorylation by lyase kinase after the substrate was first phosphorylated by catalytic subunit. Control and phospho-ATP-citrate lyase (phosphorylated by lyase kinase) incubated with catalytic subunit phosphorylated up to 0.17 mol of P/subunit at 0.5 h and 0.42 mol of P/subunit at 3 h. No significant differences were observed between control and phospho-ATP-citrate lyase first phosphorylated by lyase kinase when these samples were used as substrates for further phosphorylation by catalytic subunit. These results indicate that ATP-citrate lyase phosphorylation by catalytic subunit enhances the ability of lyase kinase to phosphorylate ATPcitrate lyase, while ATP-citrate lyase phosphorylation by lyase kinase had no effect on the ability of catalytic subunit to phosphorylate the lyase.
If catalytic subunit-mediated lyase phosphorylation poten-tiates ATP-citrate lyase's further phosphorylation by lyase kinase but the analogous reverse experiment is without effect, then ATP-citrate lyase dephosphorylation may affect its phosphorylation by each protein kinase differently. In a separate set of experiments, alkaline phosphatase treatment for 1 h, as described under "Experimental Procedures," removed all the radiolabel from 32P-labeled ATP-citrate lyase that had been prepared using catalytic subunit and/or lyase kinase followed by gel filtration. SDS-gel electrophoresis of these dephosphorylated samples followed by autoradiography demonstrated that no new radiolabeled fragments were generated during the incubation with alkaline phosphatase. These results were consistent with the stain pattern of the dephosphorylated lyase samples on SDS gels, which also did not show an increase in ATP-citrate lyase fragments, suggesting that there was no detectable proteolysis of the lyase during these incubation procedures.
To study the effect of the dephosphorylation of ATP-citrate lyase on its subsequent phosphorylation by lyase kinase and catalytic subunit, ATP-citrate lyase was dephosphorylated and the experiments outlined in Table I1 were performed.
Native liver ATP-citrate lyase was phosphorylated by lyase kinase to 0.5 mol of P/subunit, by catalytic subunit to 0.65 mol of P/subunit, and by the combination of the two protein kinases to 1.75 mol of P/subunit (Table 11). Native fat ATPcitrate lyase was phosphorylated to 0.64,0.59, and 1.83 mol of P/subunit by lyase kinase, catalytic subunit, and the combination of the two protein kinases, respectively. When liver ATP-citrate lyase was fiist phosphorylated for 1 h by lyase kinase and then by catalytic subunit for an additional 1 h, the phosphate incorporated was 1.45 mol of P/subunit. If the order of protein kinase addition was reversed, 1.77 mol of P/ subunit were incorporated. Liver or fat dephospho-ATP-citrate lyase was phosphorylated by catalytic subunit to 1.0 mol of P/subunit, consistent with recent reports (10,16,17). However, the dephospho-ATP-citrate lyase either from liver or adipose tissue was not phosphorylated at all by lyase kinase.
Incubation of dephospho-ATP-citrate lyase by simultaneous addition of lyase kinase and catalytic subunit phosphorylated the lyase to 2.18 mol of P/subunit for liver dephosphoenzyme and to 1.84 mol of P/subunit for fat dephosphoenzyme. Sequential addition of lyase kinase and then catalytic subunit PK-C, catalytic subunit of cyclic AMP-dependent protein kinase; PK-A, ATP-citrate lyase kinase.

TABLE 11
Phosphorylation of native and dephospho-ATP-citrate lyase by ATP-citrate lyase kinase and catalytic subunit Rat liver ATP-citrate lyase (1 mg/ml) and fat ATP-citrate lyase (0.06 mg/ml) were dephosphorylated by incubation with alkalime phosphatase for 3 or 5.5 h. The dephospho-ATP-citrate lyase was isolated as described under "Experimental Procedures." The native and dephospho-ATP-citrate lyases from liver (0.4 mg/ml) and fat (0.015 mg/ml) were phosphorylated for 2 h with lyase kinase (70 pg/ml) or catalytic subunit (20 pg/ml) or for 1 h with one protein kinase followed by an additional 1 h with the addition of the second protein kinase. The values are the average of three to five experimental determinations f S.D.

Phosphorylation by
PK-A + pK-c PK-A (first), PK-PK-C ( h t ) , PKmol P/mol subunit  Phosphorylation of dephospho-ATP-citrate lyase by ATP-citrate lyase kinase or by catalytic subunit after the removal of the first protein kinase Dephospho-ATP-citrate lyase was prepared by incubation with alkaline phosphatase for 4 h. The dephosphoenzyme was then phosphorylated for 30 min with cold ATP by lyase kinase or catalytic subunit and the phosphorylated ATP-citrate lyase was isolated by gel filtration and further phosphorylated with [Y-~'P]ATP by the other protein kinase for 0.5 and 2 h. PK-C, catalytic subunit of cyclic AMP-dependent protein kinase; PK-A, ATP-citrate lyase kinase. incorporated 1.19-1.40 mol of P/subunit into the dephosphoenzyme, consistent with phosphorylation being mostly due to the action of catalytic subunit. If the order of addition was reversed, 1.98-2.20 mol of P/subunit were incorporated, again demonstrating the potentiating effect of catalytic subunit phosphorylation on lyase kinase activity. No significant differences were noted in the dephospho-ATP-citrate lyase produced as substrate for the phosphorylation reaction by varying incubation times with alkaline phosphatase (3 or 5.5 h).
To examine critically whether phosphorylation by catalytic subunit potentiates phosphorylation by lyase kinase, we performed the following experiment. Dephospho-ATP-citrate lyase was isolated as described. Dephospho-ATP-citrate lyase was phosphorylated with unlabeled ATP plus one protein kinase. The phosphorylated lyase was isolated free of the protein kinase and ATP by chromatography and again phosphorylated using the other protein kinase and [y3*P]ATP. The amount of phosphate incorporated during the second stage phosphorylation is shown in Table 111. As predicted, dephospho-ATP-citrate lyase wasphosphorylated by catalytic subunit to the same extent whether or not it was first incubated with lyase kinase (0.97 and 0.87 mol of P/subunit, respectively). As expected, the dephosphoenzyme was not phosphorylated by lyase kinase. Dephospho-ATP-citrate lyase f i s t phosphorylated by catalytic subunit was phosphorylated to 0.55 mol of P/subunit by lyase kinase.
To determine if ATP-citrate lyase first phosphorylated by [32P]ATP-citrate lyase (160 pg) phosphorylated by lyase kinase plus catalytic subunit as described in the legend to Table V was trypsin-digested (1:50), and the sample was processed by HPLC as described under "Experimental Procedures." Retention times ofpeaks A, A', and B are 26,36, and 72 mi n, respectively. Peak V is the radioactivity not retained on the column. This peak after acid hydrolysis contained little radioactive phosphoamino acids. catalytic subunit is a better substrate for lyase kinase phosphorylation by increasing phosphate incorporation at the lyase kinase-specific site (peptide B) or at a new site, ATPcitrate lyase phosphorylated in vitro with both protein kinases was digested with trypsin and the radioactive phosphopeptides were analyzed by HPLC (Fig. 3). Phosphopeptides A and A , generated by catalytic subunit phosphorylation alone, and phosphopeptide B and the minor peptides (labeled O), generated by lyase kinase phosphorylation alone (Fig. l), were present in the lyase phosphorylated by the combination of the two protein kinases (Fig. 3). No new phosphopeptides were detected on the chromatogram.
To determine whether the dephosphorylation of ATP-citrate lyase opened up new phosphorylation sites for either protein kinase, the trypsin-digested samples of phosphorylated native and dephospho-ATP-citrate lyase samples were analyzed by two-dimensional peptide mapping. The results suggest that the same site was phosphorylated by catalytic subunit in native and dephospho-ATP-citrate lyase.3 Similarly, lyase kinase phosphorylated the same site in native and dephosphoenzyme. Two-dimensional tryptic maps of native and dephosphoenzyme phosphorylated by the addition of the two protein kinases together were identical. Irrespective of the order of addition, sequential addition of the two protein kinases gave qualitatively similar peptide maps. However, when lyase kinase addition was preceded by catalytic subunit, lyase kinase-specific peptide was phosphorylated to a greater extent than if the order of addition of the protein kinases was reversed. Phosphoamino acid analysis of some of these samples is given in Table IV. When native or dephospho-ATPcitrate lyase was phosphorylated by catalytic subunit, only serine was phosphorylated. Lyase kinase phosphorylated both serine and threonine almost equally in native ATP-citrate lyase. Absolutely no phosphorylation was detected in the dephosphoenzyme incubated with lyase kinase. When the two protein kinases were added together, serine and threonine were phosphorylated to 73 and 27%, respectively, in the native ATP-citrate lyase and 70 and 30%, respectively, using dephospho-ATP-citrate lyase. Since lyase kinase phosphorylated serine and threonine residues of peptide B and minor fragments (Fig. lA, labeled O), it is possible that the two phosphorylated amino acids reside within a small sequence or reflect two quite distant phosphorylation sites within a much larger fragment. To study this, lyase kinase-phosphorylated ATP-citrate lyase was treated with increasing amounts of trypsin (Fig. 4). At a ratio of trypsin to substrate of 12, peptide B was quantitatively converted to fragments with the same retention time as the minor fragments produced by less exhaustive trypsin treatment. Note that as the radioactivity associated with peak 3 decreased, radioactivity associated with the radiolabeled fragments increased (Fig. 4a). Indeed, the per cent of the total radioactivity associated with the defined lyase kinase-phosphorylated peptides were similar (Fig. 4, a-c). The molecular weights of peptides A, A', B, and B were determined by gel permeation chromatography to be 1000,1800,8500, and 2100, respectively. The molecular weight of phosphopeptide (R, = 53 min) was 5400, suggesting it is a precursor of peptide B'. However, we do not have an estimate of the molecular weight of phosphopeptide (R, = 48 min). In another series of experiments using a new batch of trypsin, when ATP-citrate lyase phosphorylated with lyase kinase was treated with trypsin (1:2), two additional phosphopeptides were found on HPLC analysis which eluted with R, less than peptide B' with M, = 1600 and 1200. The molecular weight of peptide A in this report is simiiar to the molecular weight of the phosphopeptide generated by trypsin treatment of ATPcitrate lyase from rat liver cells (16) and 3T3-Ll cells (17). In

TABLE IV Distribution of 32P-Zabeled serine and threonine residues in phosphorylated native and dephospho-ATP-citrate lyase
Native and dephospho-ATP-citrate lyases (treated with alkaline phosphatase for 5.5 h) were phosphorylated with either lyase kinase or catalytic subunit or both. Reaction was terminated by adding 10 pg of bovine serum albumin and trichloroacetic acid to 15%. The precipitated protein was collected by centrifugation and washed with ether:alcohol (1:l) followed by ether. The protein was dissolved in 25 rnM Tris-HC1 (pH 9). Concentrated HCl was added to find concentration of 6 N and digested for 3 h at 110 "C. Phosphoamino acids were analyzed as described (8). Correction was not made for losses during hydrolysis.
" PK-C, catalytic subunit of cyclic AMP-dependent protein kinase;  Table V. ATP-citrate lyase (500 pg) was added as citrate lyase (100 pg) was phosphorylated by lyase kinase as described carrier, and the sample was divided into three equal parts and treated with TPCK-trypsin for 24 h at 37 "C at a trypsin to lyase ratio of 1 :IOO ( a ) , 1:8 ( b ) , and 1:2 (c). Sample preparation and HPLC chromatography were as described in the text with the following exceptions. Aqueous phase was 0.2 M phosphate buffer (pH 3.5) and organic phase was acetonitrile. Retention time ofpeuks B' and B were 43 and 73 min, respectively, slightly different from the retention times given in Figs To study whether a given peptide phosphorylation is potentiated by prior phosphorylation of the other peptide, we normalized the results by setting the radioactivity incorporated by each protein kinase at its unique peptide as 100% (Table V). ATP-citrate lyase incubated with both protein kinases incorporated phosphate into both peptides. Incorporation of [32P]phosphate into peptide B increased about 267% compared to that incorporated by lyase kinase alone, while there was no change in the incorporation of phosphate at the catalytic subunit site (94%).
To evaluate the possible significance of ATP-citrate lyase phosphorylation on enzyme activity, its activity was determined after phosphorylation or dephosphorylation. Under normal assay conditions (13) at both pH 8.7 and 7.5, enzyme activity was unchanged by its prior phosphorylation by either protein kinase or by the combination of both. Enzyme activity also remained unaltered by the dephosphorylation of the lyase by alkaline phosphatase treatment or rephosphorylation of the dephosphoenzyme by catalytic subunit or by both protein kinases added together.

ATP-citrate Lyase
Phosphorylation by Lyase Kinase 4955 TABLE V Dependence of the phosphorylation of the lyase kinase-specific site on phosphorylation of the catalytic subunit-specific site Rat liver ATP-citrate lyase (660 pg) was phosphorylated with ATPcitrate lyase kinase (10 pg), catalytic subunit (10 pg), or combination of the two kinases (same amounts) by incubation for 1 h. The phosphate incorporation into ATP-citrate lyase was 0.08 mol of P/ subunit by lyase kinase, 0.24 mol of P/subunit by catalytic subunit, and 0.48 mol of P/subunit by lyase kinase and catalytic subunit together. Note that only one-third of the lyase kinase required to phosphorylate ATP-citrate lyase equal to the phosphorylation by catalytic subunit was used. Phosphorylated lyase was digested with trypsin as described (8), lyophilized, dissolved in 50 pl of 0.1% trifluoroacetic acid, and 40 pl was analyzed by HPLC. Radioactivity not retained on the column (peak V) was not considered for calculation of the radioactivity in the catalytic subunit tryptic fragment (peaks A + A') and lyase kinase tryptic fragments (Fig. 3,  PK-A, ATP-citrate lyase kinase; PK-C, catalytic subunit of cyclic AMP-dependent protein kinase. DISCUSSION Following the discovery that rabbit muscle glycogen phosphorylase exists in a relatively inactive and active form, determined by its phosphorylation state, it has become apparent that there are many enzyme systems whose net enzyme activity depend on continuingly controlled cyclic processes consisting of, in the most simple case, rates of phosphorylation and dephosphorylation. ATP-citrate lyase, which catalyzes the formation of acetyl-coA in the cytosol for anabolic purposes, undergoes both rapid reversible phosphorylation-dephosphorylation in vivo and enzyme induction under control of the hormonal and metabolic state of the animal. ATP-citrate lyase, like acetyl-coA carboxylase, should be a likely candidate to be controlled by interconvertible enzyme cascades (18). However, as yet no convincing evidence has been presented demonstrating that enzyme activity as measured in vitro is affected by its phosphorylation state, whether phosphorylated at the site phosphorylated by catalytic subunit or at the unique site phosphorylated by lyase kinase. Recalling the difficulties in deciphering the regulatory role played by the multisite phosphorylation of glycogen synthase and the paramount importance of allosteric effectors (19) and enzyme assay conditions (20), we thought it important to completely characterize the phosphorylation sites of ATP-citrate lyase as phosphorylated in vitro by lyase kinase and catalytic subunit.
During studies of the time course of ATP-citrate lyase phosphorylation, we noted that when both catalytic subunit and lyase kinase were added simultaneously to the reaction mixture, lyase phosphorylation was unexpectedly more than additive. Furthermore, when lyase was phosphorylated f i s t with lyase kinase and the kinase removed and then phosphorylated with catalytic subunit, the phosphorylation was unaffected (Table I). However, when ATP-citrate lyase was phosphorylated first with catalytic subunit and the protein kinase removed and then phosphorylated with Iyase kinase, there was a marked increase in the phosphorylation of the enzyme. We interpreted these data to indicate that catalytic subunit phosphorylation at its site (peptide A) increased the phosphorylation of lyase by lyase kinase at its site (peptide B) .
Was peptide A phosphorylation (the site phosphorylated by catalytic subunit) absolutely necessary for peptide B phosphorylation (the site phosphorylated by lyase kinase) or was its role only to increase peptide B phosphorylation? To answer this question directly, we prepared alkaline phosphatasetreated liver and fat ATP-citrate lyase and phosphorylated them by either protein kinase, both protein kinases added simultaneously, each protein kinase added sequentially, and in both orders. The dephosphoenzyme was not a substrate for phosphorylation by lyase kinase. When the protein kinases were added together or sequentially, phosphorylation increased when catalytic subunit was added first. The data suggested that catalytic subunit phosphorylation was an absolute requirement for lyase kinase phosphorylation. Since the fist protein kinase added was not removed before the addition of the second, the results of these experiments did not rigorously prove the hypothesis since the remaining protein kinase might have had an unforeseen effect. In Table I11 are presented data showing that when both alkaline phosphatase and the fist added kinase were removed before the second addition, the phosphorylation by catalytic subunit (peptide A) is an absolute requirement for phosphorylation by lyase kinase at peptide B. Could this all or nothing result be due to an artifact produced by the denaturation or cleavage of ATP-citrate lyase by alkaline phosphatase treatment? This is most unlikely because ATP-citrate lyase after dephosphorylation was isolated by gel filtration. In addition, no increase in proteolytic cleavage of the lyase was observed after alkaline phosphatase treatment. Furthermore, all phosphorylation data were calculated from the radioactivity associated with the intact 115,000-dalton subunit. Therefore, it is not likely that a denatured or cleaved substrate had been rendered unresponsive to lyase kinase in a manner different from that of the dephosphoholoenzyme since there was no evidence that a substantial amount of the enzyme was cleaved or denatured by our dephosphorylation procedure. Furthermore, when phosphorylated ATP-citrate lyase was treated with trypsin and the percentage of radioactivity associated with each site determined, it was found that all of the increase in radioactivity associated with both protein kinases added simultaneously could be accounted for by an increase in radioactivity associated with peptide B (the lyase kinase-specific site) ( Table V).
Finally, the dephosphoenzyme when assayed at either pH 7.5 or 8.7 was enzymatically equivalent to the native enzyme, suggesting that there was no extensive denaturation of the enzyme at least in the region of the catalytic site.
It was important td determine whether the specific amino acids phosphorylated after such treatment were consistent with our previous findings of serine and threonine phosphorylations (8). As shown in Table IV, native ATP-citrate lyase phosphorylated with lyase kinase phosphorylated serine and threonine almost equally. When catalytic subunit was added, the per cent serine phosphorylated increased appropriately. When the dephosphoenzyme was used as a substrate, the ratios of phosphoserine to phosphothreonine changed concordant with the results presented in Tables I11 and V.
Our data indicate that the lyase kinase-phosphorylated sites are subject to second site regulation. The possibility that the phosphorylation of two amino acids near each other or widely separated or even of several amino acid residues is controlled by a singre phosphorylation at a cycIic-AMP-regulated site must be considered. This has been addressed directly by treating lyase kinase-phosphorylated ATP-citrate lyase with increasing concentrations of trypsin. Higher concentrations of trypsin produced a pattern of radiolabeled phosphopeptides with R1 values similar to those of the minor fragments seen in chromatography after normal (1:50) trypsin-treatment. Furthermore, when peptide B was purified by HPLC, treated with 1:2 trypsin, and rechromatographed, only radiolabeled fragments with similar R, values to those shown in Fig. lA (0) were found: These data indicate that peptide bonds unapproached by low trypsin treatment were hydrolyzed by higher concentrations of trypsin, producing smaller (less hydrophobic) fragments. Since the molecular weights of phosphopeptides B and B' were determined to be 8500 and 2100, respectively, and the radioactivity content of phosphothreonine and phosphoserine was conserved, we interpret the data to indicate that the lyase kinase-phosphorylated amino acids are all contained within the peptides B and B' sequences.
Note that the peptide segment may contain more than the minimal number of two radiolabeled amino acids.
Are there precedents in the literature similar to the results reported in this paper? A similar, but opposite phosphorylation effect was found in studies on the glycogen synthase system. Glycogen synthase is phosphorylated by cyclic AMPdependent protein kinase, phosphorylase kinase, and cyclic AMP-independent protein kinases, converting glycogen synthase I to the D-form (21). Phosphorylation of glycogen synthase by the simultaneous incubation with phosvitin kinase and cyclic AMP-dependent protein kinase or first phosphorylation with phosvitin kinase followed by cyclic AMP-dependent protein kinase resulted in a more complete conversion of the I to the D-form (22). However, if synthase I is phosphorylated with cyclic AMP-dependent protein kinase, there is only a partial conversion of the I to the D-form. Unexpectedly, this phosphorylated glycogen synthase was not further phosphorylated by phosvitin kinase (22). Thus, the phosphorylation of glycogen synthase by cyclic AMP-dependent protein kinase altered the synthase molecule rendering it refractory to the action of phosvitin kinase and preventing its more complete conversion from the I to the D-form.
In this report, we have described an in vitro system whereby the phosphorylation at one site absolutely determines the phosphorylations at the other. There is as yet no evidence that such a control system is operative in vivo. Indeed, Swergold et al. (17) recently reported that insulin enhanced phosphorylation in the same ATP-citrate lyase tryptic fragment in 3T3-Ll cells as that increased by isoproterenol treatment. Pierce et al. (23) demonstrated that the same serine residue of rat liver ATP-citrate lyase is phosphorylated in response to both insulin and glucagon. These puzzling findings were contrary to expectations, since these hormones have opposite actions and have been found to phosphorylate different sites on the S-6 protein (24, 25).
If the findings presented in this paper are relevant to physiology, then the phosphorylated state of the enzyme prior to hormone action would determine the site(s) phosphorylated by the hormone action. In recent experiments studying in vivo ATP-citrate lyase phosphorylation by incubating rat fat pads with 32Pi and purifying the radiolabeled enzyme, we found that after trypsin-treatment and HPLC analysis, peptides with Rt values identical with peptides phosphorylated by ATP-citrate lyase kinase action in vitro (peptides B and B') and phosphopeptides with Rt values identical with peptides D. L. Pucci, S. Ramakrishna, and W. B. Benjamin, unpublished observations. phosphorylated by cyclic AMP-dependent protein kinase (peptides A and A') were significantly radiolabeled in vivo. 5 This evidence suggests that the observations of the dependence of "site B" phosphorylation on "site A" phosphorylation may have physiological significance and be involved in the metabolism of ATP-citrate lyase. One additional caveat should be mentioned. ATP-citrate lyase is extraordinarily protease sensitive (13). In vivo experimental results might be difficult to evaluate since selective loss of phosphorylated fragments is almost to be expected.