Phosphorylation of High-mobility-group Proteins by the Calcium-Phospholipid-dependent Protein Kinase and the Cyclic AMP-dependent Protein Kinase*

have investigated as poten- tial substrates for the Ca2+-phospholipid-dependent protein kinase and the CAMP-dependent protein ki- nase. HMG proteins 1, 2, and are phosphorylated by the Ca2+-phospholipid-dependent protein kinase; the reactions are totally Ca2+ and lipid dependent and are not inhibited by the inhibitor protein of the CAMP- dependent protein kinase. HMG 17 is phosphorylated predominantly in a single seryl residue, Ser 24 in the sequence Gln-Arg-Arg-Ser 24-Ala-Arg-Leu-Ser 28-Ala-Lys, with the second seryl moiety, Ser 28, modified to a markedly lesser degree. HMGs 1 and 2 are also phosphorylated in only seryl residues but with each there are multiple phosphorylation sites. HMG 17, but not HMG 1 or 2, is also phosphorylated by the CAMP-dependent protein kinase with the site phosphorylated being the minor of the two phosphorylated by the Ca2+-phospholipid-dependent protein kinase; the K,,, for phosphorylation by the CAMP-dependent enzyme is 50-fold higher than by the Ca2+-phospholipid-depend- ent enzyme. is an equally effective substrate for the Ca2+-phospholipid-dependent

HPLC, high-performance liquid chromatography; EGTA, ethylene glycol bis(@-aminoethyl ether)-N,N,N',N'-tetraacetic acid. acids which are located in separate domains (21, and this amphoteric nature leads to the potential for electrostatic interactions with both DNA and other chromosomal basic proteins. While the function(s) of the HMG proteins remain to be fully elucidated, various key roles in chromosomal structure and the regulation of its function are currently under active consideration (3-15). The HMG proteins can undergo a variety of post-translational modifications including acetylation (16,17), poly(ADP-ribosylation) (18, 191, glycosylation (20), methylation (21), and phosphorylation (22)(23)(24)(25)(26)(27). Concerning phosphorylation, one or several protein kinases may be of importance. In in vitro studies, the CAMP-dependent protein kinase (28,29), the cGMP-dependent protein kinase (30), casein kinase I1 (24), and endogenous nuclear protein kinases (31-33) have all been shown to phosphorylate HMG proteins with HMG 14 being the predominant substrate and the other HMG proteins being phosphorylated less or not at all. With intact cells there remains ambiguity as to which HMG proteins might get phosphorylated. Clearly HMG 14 is phosphorylated in response to a variety of stimuli and in several cell types (22)(23)(24)(25)(26)(27) but, whereas some data has implicated HMG 17 phosphorylation to occur in intact cells (22, 23), other studies suggest that the phosphoprotein identified as HMG 17 might not be HMG 17 but a protein of some similar physiochemical characteristics (24, 34). Most current evidence suggests that HMG 1 and 2 are not phosphorylated in uiuo, but clearly the physiological implications of the phosphorylation of HMG proteins need further resolution. Recently, a Ca2+-phospholipid-dependent protein kinase has been shown to be present in a wide range of tissues, both in the soluble and particulate fractions (35,36). In this study we have examined the ability of purified HMG proteins to serve as substrates for this enzyme.

EXPERIMENTAL PROCEDURES
Purification of HMG Proteins-HMG proteins 1, 2, and 17 were purified to homogeneity from fresh lamb thymus glands according to the procedure of Goodwin et al. (37). Briefly, lamb thymus was homogenized in 5% (v/v) perchloric acid, and the HMG proteins were precipitated with acidified acetone. The crude HMG proteins were further fractionated with acidified ethanol and the individual proteins 1, 2, and 17 separated by chromatography on CM-Sephadex C-25, eluted with a linear gradient of 0.1 to 1.2 M NaCl in 7.5 mM sodium borate (pH 8.8). The purified proteins were precipitated with 6 volumes of acidified acetone, washed once with acetone, and dried under vacuum. The purified HMG proteins 1, 2, and 17 showed a single Coomassie Blue staining band when subjected to SDS/PAGE. For lamb thymus HMG 17, the amino acid analysis of the entire protein and the sequence of at least the first 38 residues were identical to those previously reported for calf thymus HMG 17 (2). HMG 14 was purified from lamb thymus by a modification of the procedure of Goodwin et al. (37) in which the fraction obtained by ethanol precip-Ca2+-PhmphOlipid-dependent Phosphorylation of HMG Proteins itation was chromatographed on CM-Sephadex C-25 equilibrated in 8 mM borate buffer, pH 8.8, and eluted with a 0.2 to 0.6 M NaCl gradient. The trailing edge of the HMG 14, which also contained HMG 17, was discarded. The final product was at least 75% homogeneous by SDS-PAGE, and only the band corresponding to HMG 14 was phosphorylated by either the CaZ+-phospholipid-dependent and CAMP-dependent protein kinases. Lamb thymus HMG proteins have similar mobilities to the respective calf thymus proteins on SDS-PAGE.
Phosphorylntion of HMG Proteins-The reaction mixture contained 20 mM Tris chloride (pH 7.35), 10 mM MgCl,, 45 mM 2mercaptoethanol, 50 p~ [y-32P]ATP (300-600 dpm/pmol), and the purified HMG proteins. When present, as indicated in the figure legends, the concentration of CaCI, was 50 p~, EGTA was 0.5 mM, phosphatidylserine was 16 pg/ml, and dioleine was 1.2 pg/ml. The amounts of Ca2+-phospholipid-dependent protein kinase and CAMPdependent protein kinase were 1.25 X and 4.0 X lo-' units/ml, respectively. The reaction temperature was 30 "C. The phosphorylation reaction was initiated by the addition of [y-32P]ATP, and at the varying time points indicated, 15-p1 aliquots were removed and the reaction terminated by adding 30 p1 of SDS sample buffer containing 120 mM Tris chloride (pH 6.8), 15% (v/v) glycerol, 3.75% (v/v) 2mercaptoethanol, 3% (w/v) SDS, and 0.0075% (w/v) bromphenol blue. The samples were heated at 100 "C for 5 min and then applied to a 1.5-mm thick 17.5% (w/v) polyacrylamide gel made according to Shreier et al. (38). Electrophoresis was carried out until the dye front just ran off the gel. The gels were stained with 0.03% (w/v) Coomassie Brilliant Blue in 45% (v/v) methanol and 10% (v/v) acetic acid and then destained for 1 h in 25% (v/v) methanol plus 10% (v/v) acetic acid and subsequently for 4-6 h in 10% (v/v) acetic acid plus 1% (v/ v) ethanol. The gels were vacuum dried onto Whatman 3MM filter paper and exposed to Kodax X-Omat AR x-ray film with intensifying screens at room temperature. For quantitation, the bands corresponding to the HMG proteins were cut out and counted in a toluene scintillation fluor. Blank controls were for protein samples in the absence of added enzyme.
Purification of Protein Kinases-The catalytic subunit of the CAMP-dependent protein kinase was purified to homogeneity from bovine heart by a modification (39) of the procedure of Beavo et al. (40). The specific activity of the enzyme, assayed with histone IIA (Sigma) as previously reported (39, 41), was 1.6 units/mg. The Ca2+phospholipid-dependent protein kinase (35,36) was partially purified from bovine heart by ammonium sulfate fractionation of the DE-52 eluate containing the Ca2+-phospholipid-dependent protein kinase, followed by twice-repeated Sephacryl S-200 gel filtration. The details of this preparation will be presented elsewhere. The specific activity of the final preparation, assayed with histone H1 with reaction conditions identical to those used for HMG proteins, was 0.15 X lo-* pmol of 32P incorporated/min/mg. A unit of activity for either enzyme is that which catalyzes the incorporation of 1 pmol of phosphate per min under the conditions described above.
Preparation of Mononucleosomes from HeLa Cell Nuclei and Reconstitution with HMG 17-The isolation, fractionation, and characterization of mononucleosomes from HeLa cells and the reconstitution of mononucleosomes with HMG 17 were performed as we have described earlier (7, 42). Briefly, mononucleosomes, isolated by sucrose density gradient, were passed through a Sephadex G-25 column equilibrated with 10 mM Tris chloride (pH 7.4). Purified HMG 17 was added at a ratio of 2 molecules/nucleosome and incubated for 5 min at 0 "C. The HMG 17-reconstituted nucleosomes were passed through Sephadex G-25 equilibrated with 10 mM Tris chloride (pH 7.4). The AZm-absorbing materials were pooled and used in the phosphorylation reactions. Under these conditions, as described before, all of the added HMG 17 is bound to the nucleosome (7).
Phosphoamino Acid Analysis and Fingerprint Analysis of Tryptic Digests of HMG Proteins-HMG proteins were phosphorylated under the conditions described above. The reaction was terminated by the addition of acetic acid to a final concentration of 30% (v/v), [y-"P] ATP was removed by adsorption on a column (1 ml) of Dowex AG 1-X8 (43), and the eluted HMG proteins were diluted with Hz0 and lyophilized. For phosphoamino acid analysis, the lyophilized powder was dissolved in 6 N HC1, vacuum sealed, and then hydrolyzed at 110 "C for 4 h. Phosphoamino acids were separated by thin-layer electrophoresis using 0.1-mm Polygram cell 300 thin-layer cellulose plates (Brinkmann Instruments) with a solvent system of formic acidacetic acidwater, 4:1:45, and with electrophoresis for 3 h at 480 V and 6 "C. Phosphoserine and phosphothreonine were identified using standards detected with 0.3% (w/v) ninhydrin in col1idin:acetic acidethanol, 3:1087. For tryptic fingerprints and HPLC analyses, the lyophilized HMG proteins were dissolved in 100 mM N-ethylmorpholine acetate (pH 8.01, the solution heated at 100 "C for 10 min, and then the protein hydrolyzed by digestion at 23 "C for 24 h with tosylphenylalanyl chloromethyl ketone-treated trypsin at a trypsin concentration of 27 pg/ml and a trypsin to protein ratio of approximately 1:2.5. Tryptic maps were developed using Polygram cellulose thin-layer plates with separation in the first dimension by electrophoresis at 500 V for 3 h with a solvent system of acetic acidpyridine:water; 5:1:44, and in the second dimension by chromatography using pyridine:l-butano1:acetic acidwater, 1015:3:12. The plates were dried for 16 h before autoradiography. The conditions for HPLC analysis are given in Fig. 9. Sequence Determination of Phosphorylated HMG 17"HMG 17 (300 pg) was phosphorylated to a stoichiometry of 0.82 and 0.31 mol/ mol with the Ca2+-phospholipid-dependent protein kinase (5 X lo-' units/ml) and CAMP-dependent protein kinase (4 X lo-' units/ml), respectively, using the reaction conditions described above with a 90min incubation. The phosphorylated HMG 17 was separated from [y-32P]ATP using Dowex AG 1-X8 and purified by reverse-phase HPLC using a Waters C,, column. The protein was absorbed in 0.05% trifluoroacetic acid and the column developed with a linear gradient of acetonitrile (0-60%). HMG 17 was eluted at 40% acetonitrile and lyophilized. The protein was sequenced using a Beckman 89OC, using the program 031281, utilizing 2 mg of Polybrene carrier (44) and double cleavage for proline residues. The amino acid residues were identified as their phenylthiohydantoin derivatives separated by HPLC according to Bhowan et al. (45), and in addition, either by gasliquid chromatography (46) or thin-layer chromatography (47). At each cycle an aliquot of the chlorobutane wash was counted for 32P.
Other Methods-Amino acid analyses were performed for 24-and 48-h hydrolysates with a Durrum D500 analyzer. Protein concentration of the HMG proteins and the protein kinases was determined by the method of Lowry et al. (48) with bovine serum albumin as standard.
Materials-Phosphatidylserine and dioleine were from Sigma; 160 pg of phosphatidylserine plus 12 pg of dioleine, dissolved in chloroform/methanol, were dried with N, and the lipid dissolved in 2 ml of 20 mM Tris chloride (pH 7.4) by sonicating for 5 min at 0 "C. The source of all other materials is as reported previously (7,39,41,42).

Ca2+-Phospholipid-dependent Phosphorylation of
HMG Proteins-The ability of purified lamb thymus HMG proteins 1, 2, and 17 to serve as substrates for the partially purified beef heart Ca2+-phospholipid-dependent protein kinase is depicted in Fig. l. With HMG 17 (right panels), the autoradiogram of one-dimensional SDS-polyacrylamide gels showed a single radioactive band that was coincident with the Coomassie-stained protein band (indicated by arrow). The phosphorylation of HMG 17 was totally Ca2+-and phospholipiddependent (Fig. 1, c and f ) and achieved a stoichiometry of 1 mol of phosphate per mol of HMG 17 (see Fig. I showing longer periods of incubation). Two-dimensional gel electrophoresis, using Triton acid-urea gels in the first dimension and SDS/PAGE in the second dimension, confirmed HMG 17 was the protein phosphorylated ( Fig. 1, right lower panels). The results obtained with HMG 1 (Fig. 1, left panels) and HMG 2 ( Fig. 1, center panels) were similar to those observed with HMG 17, although there were some specific differences. Both HMG 1 and 2 were phosphorylated by the Ca2+-phospholipid-dependent protein kinase, the reactions with each were totally Ca2+ and lipid dependent, and at least 1 mol of phosphate was incorporated into each. (Even at the longest reaction times, however, the reactions do not appear to be complete, Fig. 1, d and e, insets.) With HMG 1 or 2 as substrates, two to three higher-molecular-weight proteins were also phosphorylated, and with HMG 2 one of these was quite prominent. Based upon Coomassie staining, the HMG preparations used appear homogeneous. If the higher-molecular-weight proteins were derived from the protein kinase  or HMG 17 (c) for reaction conditions using (as indicated) either EGTA plus lipid mixture (i.e. phosphatidylserine plus diolene), Ca2+ plus lipid, or Ca2+ alone, for the time periods indicated. Concentrations of EGTA, lipid mixture, and Ca" used are given under "Experimental Procedures." The final concentrations of HMG proteins 1.2, and 17 in the assay were 75, 110, and 75 pg/ml, respectively. The concentration of protein kinase was 125 microunits/ml. The arrows indicate the specific HMG proteins detected by Coomassie staining. In d-f are indicated the time courses of phosphorylation of the respective HMG proteins for data from the polyacrylamide gels, obtained by counting the 3zP in the bands corresponding to the HMG proteins. Data is for either the complete reaction mixture (i.e. Caz+ plus lipid), u, in the presence of EGTA plus lipid, A-& or with Ca2+ alone, A-A.

Ca2+-Phospholipid-dependent Phosphorylation of HMG Proteins
In the insets of d-f the phosphorylation reactions were performed for extended periods of time under essentially identical conditions. In g-i are presented autoradiograms from two-dimensional gels for proteins phosphorylated for 120 min. The first dimension of the gel system used Triton-acid urea gels (49), and the second dimension used SDS/PAGE with the conditions as given under "Experimental Procedures." In j-1 is the Coomassie protein staining of the twodimensional gels. For all experiments, the left panels are for HMG 1; center panels, HMG 2; and right panels, HMG 17.
preparation, it is curious that they were not detected when HMG 17 was used as substrate (Fig. IC). They were also not detected when no exogenous substrate was added (data not shown). It is possible that the HMG protein preparations contain low levels of other nonhistone proteins that are not readily detected with Coomassie Blue but are highly phosphorylated or that the HMG proteins differentially affect the phosphorylation of proteins present in the protein kinase preparation. Confirmation that HMG 1 and 2 are indeed phosphorylated by the Ca2+-phospholipid-dependent protein kinase was provided by analysis with two-dimensional gels. As indicated (Fig. 1, lower panels), a good coincidence was observed between the major spot of '*P labeling and the Coomassie staining.
HMG 17 is a slightly better substrate for the Ca2+-phospholipid-dependent protein kinase than either HMG 1 or 2 and is phosphorylated a t about 2-to 4-fold more rapidly than either of the other two under similar conditions (Fig. 1, d-f). All three of the HMG proteins have Ca2+ and lipid concentration dependencies (Fig. 2) similar to those that have been observed for other protein substrates of this enzyme (35, 36). The phosphorylation of HMG 1, 2, or 17 by the Ca2+-phospholipid-dependent protein kinase was not affected by the inhibitor protein of the CAMP-dependent protein kinase (39) (data not presented).

Comparison of HMG 17 Phosphorylation by the CAMPdependent Protein Kinase and by the Ca2+-Phospholipid-dependent Protein Kinme-Lamb
thymus HMG 17 is also a substrate for the CAMP-dependent protein kinase (Fig. 3a), although,as Walton et al. (30) and Taylor (28) have shown, it is not a major substrate for this enzyme. As indicated (compare Figs. 1 and 3), the amount of CAMP-dependent protein kinase required to demonstrate phosphorylation was considerably more than is needed with the Ca2+-phospholipid- dependent enzyme. K,,, value determinations confirmed that HMG 17 was a better substrate for the Ca2+-phospholipiddependent enzyme than for the CAMP-dependent protein kinase (Table I). For the Ca2+-phospholipid-dependent protein kinase, the K,,, value for HMG 17 is well within a range that would be reasonable for it being a physiological substrate. In contrast, the K,,, value of the CAMP-dependent protein kinase for HMG 17 is substantially higher than has been observed with other, apparently more physiological substrates (50). Neither HMG 1 nor HMG 2 was a substrate for the CAMP-dependent protein kinase (Fig. 3a).
The potential additivity of phosphorylation of HMG 17 by ' K , values were determined using the standard assay procedure from double-reciprocal plots drawn by weighted linear regression analysis using at least seven concentrations of substrates, each in triplicate. Rates were determined from a minimum of two time points. All reaction conditions were as described under "Experimental Procedures." protein kinase protein kinase the Cazf-phospholipid-dependent and CAMP-dependent protein kinases is presented in Fig. 3b. For HMG 17 phosphorylated maximally by the Ca2+-phospholipid-dependent enzyme to a level of near 1 mol/mol, subsequent addition of the CAMP-dependent protein kinase resulted in no further phosphorylation. Even with long periods of reaction and with high levels of enzyme, phosphorylation of HMG 17 by the CAMPdependent protein kinase resulted in an incorporation of only 0.4 to 0.5 mol/mol of protein. Subsequent addition of the Ca2+-phospholipid-dependent protein kinase resulted in phosphorylation to a level equivalent to that observed with the latter enzyme alone. The phosphorylation of HMG 17 by the two protein kinases was also examined with tryptic peptide mapping. Both the Ca2+-phospholipid-dependent protein kinase and the CAMPdependent protein kinase phosphorylate only seryl residues in HMG 17 (Fig. 4a). Tryptic fingerprint analysis showed that only two peptide sites were phosphorylated by the Ca2+phospholipid-dependent protein kinase, one of which was in marked abundance over the other. The CAMP-dependent protein kinase phosphorylated a single peptide that was identical to the minor of the two peptides phosphorylated by the Ca2+-phospholipid-dependent enzyme (Fig. 4, b and c ) .
The sites of HMG 17 phosphorylated by the two protein kinases were determined by sequence analyses of the intact  (30), the site phosphorylated by the CAMP-dependent protein kinase was shown to be Ser 28 (Fig. 5a). From the data given in Fig. 4 (6   13499 and c) Ser 28 is most likely also the minor site phosphorylated by the Ca2+-phospholipid-dependent protein kinase; this was not directly identifiable from the sequence analysis because of the typical spillover of 32P into the subsequent cycles.
In contrast to the results obtained with HMG 17, both HMG 1 and 2 showed considerably less specificity in the sites phosphorylated by the Ca2+-phospholipid-dependent protein kinase. Although in each, only seryl residues were phosphorylated (Fig. 4a), with both, tryptic peptide maps indicated that there were a minimum of at least six phosphorylation sites (Fig. 4, d and e). Presumably, none of these sites achieves stoichiometric incorporation even with long-term phosphorylation (see Fig. 1).

Phosphorylation of HMG 17 in Nucleosomes and Nucleosomes Reconstituted with Stoichiometric Lmek of HMG 17-
Isolated nucleosomes contain only low amounts of HMG 17 apparently associated with regions of DNA derived from actively transcribing genes (8). Incubation of mononucleosomes with HMG 17 results in the stoichiometric reconstitution to two discrete sites located at the DNA ends of the core particle (4-7, 51). Isolated nucleosomes from HeLa cells and these nucleosomes reconstituted with 2 mol of HMG 17 were tested as substrates for the Ca2'-phospholipid-dependent protein kinase and the CAMP-dependent protein kinase. With the directly isolated nucleosomes, the Ca2+-phospholipid-dependent protein kinase catalyzed the phosphorylation of the residual HMG 17 (Fig. 6a) with the low level of HMG 17 rapidly being maximally phosphorylated (Fig. 6c). The Ca2+phospholipid-dependent protein kinase also catalyzed some phosphorylation of the core histones H2B, H3, and H4 (Fig.   6a). With the HMG 17-replenished nucleosomes, HMG 17 was clearly the major protein phosphorylated (Fig. 66). As indicated, it was phosphorylated to a level of at least 0.6 mol/ mol of HMG 17, and the rate of phosphorylation (Fig. 6c) was, within error, identical to that which is obtained with the pure protein at an equivalent concentration. Consistent with what had been observed with the pure protein, nucleosomebound HMG 17 was a poor substrate for the CAMP-dependent protein kinase. Some phosphorylation of it was observed with the reconstituted nucleosomes but only by using a high concentration of enzyme. As reported by Taylor (28), nucleosomebound H3 is an effective substrate for the CAMP-dependent protein kinase (Fig. 6).
Because of the possibility that the conditions of phosphorylation may have caused dissociation of HMG 17 from the nucleosome, the phosphorylation of nucleosomal proteins was further examined using nondenaturing particle gels. Consistent with past observations (4, 5), reconstitution of mononucleosomes with HMG 17 caused a decrease in the electrophoretic mobility with the addition of 2 mol of HMG 17 (Fig. 7a). For either the directly isolated nucleosomes or HMG 17replenished nucleosomes, phosphorylated with either the Ca"-phospholipid-dependent or CAMP-dependent protein kinases, the "P incorporated co-migrated with the intact nucleosome (Fig. 76). The phosphorylated nucleosomes were examined further on a second dimensional SDS/PAGE (Fig.  7, c and d ) to show in each case that phosphorylated HMG 17 had remained associated with the nucleosome. These findings show clearly that HMG 17 is bound to the nucleosome both before and after phosphorylation and, in consequence, that nucleosome-bound HMG 17 is an effective substrate for the Ca"-phospholipid-dependent protein kinase.
Phosphorylation of HMG 14"Although only limited amounts have been available so as not to permit as full a study as reported above for HMG 17, lamb thymus HMG 14 has been tested as a substrate for the Ca2+-phospholipiddependent protein kinase. Previous studies by others (28-30) have shown that HMG 14 (calf thymus) is an effective substrate for the CAMP-dependent protein kinase. As indicated in Fig. 8, with either the Ca2+-phospholipid-dependent protein kinase or the CAMP-dependent enzyme, lamb thymus HMG 14 is phosphorylated to a level of close to 1 mol/mol. The rates of reaction observed are similar to those seen with HMG 17, and the phosphorylation of HMG 14 by the Ca2+-phospholipid-dependent enzyme is totally Ca2+ and phospholipid dependent (not shown). The K,,, values of HMG 14 and 17 for the Ca2+-phospholipid-dependent enzyme are closely similar ( Table I) whereas, by that criteria, HMG 14 appears to be a better substrate than HMG 17 for the CAMP-dependent protein kinase. As is also depicted in Fig. 8, additional phosphorylation of HMG 14 is observed if either the Ca2+-phospholipid enzyme or the CAMP-dependent protein kinase is added after the phosphorylation by the other enzyme had reached a plateau value.
The sites of HMG 14 and 17 have been compared by HPLC analysis of tryptic peptides. To do so, the tryptic peptides were first chromatographed on a C-18 reverse-phase column developed with an acetonitrile gradient, and then the phosphopeptides obtained were rechromatographed on a pBondapak-NH, column eluted with a triethylamine gradient. The phosphotryptic peptide obtained from either HMG 14 or HMG 17, as phosphorylated by the Ca2+-phospholipid-dependent protein kinase, behaved identically. With each a single major peptide eluted in the unretained fraction in the first column (Fig. 9, c and a, respectively, solid symbols) and then at 155 mM triethylamine from the second column (Fig.  9, d and b, respectively, solid symbols). It has been shown that calf thymus HMG 14 contains the sequence Pro-Lys-Arg-Arg-Ser-Ala-Arg-Leu (2) which is an identical sequence to the site phosphorylated in HMG 17 by the Ca2+-phospholipiddependent protein kinase, except that the Ser in HMG 17 is residue 24 in the protein whereas in HMG 14, it is residue 20. Assuming a similarity between calf and lamb thymus HMG 14s, it would appear that Ser 20 (or its lamb thymus equivalent) is the site in HMG 14 phosphorylated by the Ca2+phospholipid-dependent protein kinase.  7. Analysis of nucleosome protein phosphorylation by particle gel electrophoresis. Native nucleosomes or nucleosomes reconstituted with HMG 17 were phosphorylated for 90 min under the conditions described for Fig. 6. At the termination of the reaction a 2 0 4 aliquot of the reaction mixture was mixed with 1 pl of 0.005% bromphenol blue and immediately applied onto a 5% polyacrylamide gel (acry1amide:methylene bisacrylamide; 29:l) using 89 mM Tris borate (pH 8.3) containing 2.5 mM EDTA as gel and electrode buffers and with the gels pre-electrophoresed for 1 h prior to sample loading. In a the location of HMG 17-reconstituted nucleosomes ( L a n e I ) or native nucleosomes ( L a n e 2) was detected by ethidium bromide staining for the nucleosomes prior to phosphorylation. In b is shown the autoradiogram for reconstituted nucleosomes (Lanes I and 2) or native nucleosomes (Lanes 3 and 4 ) phosphorylated with either the Ca2+-phospholipid-dependent protein kinase (Lanes 1 and 3) or the CAMP-dependent protein kinase (Lanes 2 and 4). In c and d samples of reconstituted nucleosomes, phosphorylated with the Caz+-phospholipid-dependent protein kinase @anel c) or CAMP-dependent protein kinase (panel d), were initially separated in the first dimension by particle gel electrophoresis and then in the second dimension with the SDS/PAGE buffer system. Location of the nucleosome in the first dimension is denoted by the symbol 1-1.
Location of standard HMG 17 and other histones is denoted for the second dimension. All other conditions were as presented under "Experimental Procedures." TIME ( m i n )

FIG. 8. Additivity of phosphorylation of HMG 14 by the
Ca*+-phospholipid-dependent and CAMP-dependent protein kinases. The reaction conditions are as described under "Experimental Procedures" except that the concentration of ATP was 100 p~, phosphatidylserine, 25 pg/ml; diolein, 2.5 pglrnl; and HMG 14, 98 pg/ml. In panel a: W, HMG 14 was initially phosphorylated by the CaZ+-phospholipid-dependent protein kinase (176 microunits/ ml), and at 60 min, 0.02 unit/ml of CAMP-dependent protein kinase was added to the reactions denoted by W. In p a n e l b A-A, HMG 14 was initially phosphorylated by the CAMP-dependent protein kinase (0.02 unit/ml), and at 60 min, 176 microunits/ml of Caz+phospholipid-dependent protein kinase were added to the reaction denoted by A-A.
Two tryptic phosphopeptides were produced from the phosphorylation of HMG 14 by the CAMP-dependent protein kinase. One of these behaved identically to the single tryptic phosphopeptide obtained from the CAMP-dependent protein kinase-catalyzed phosphorylation of HMG 17. It eluted a t 25% acetonitrile in the first column (Fig. 9, a and c ) and 85 m M triethylamine in the second column (Fig. 9, b and d).
Most likely, this peptide contains Ser 24 of HMG 14 (or its equivalent in lamb thymus HMG 14) in the sequence Ser 20-Ala-Arg-Leu-Ser 24-Ala; this is an identical sequence to that containing Ser 28 in HMG 17. The second tryptic phospho-peptide obtained from the CAMP-dependent phosphorylation of HMG 14 eluted in the unretained fraction in both columns. This peptide might possibly contain Ser 6 which is the site identified by Walton et al. (30) as the predominant site phosphorylated by the cGMP-dependent protein kinase and for which there is no analogous sequence in HMG 17.

DISCUSSION
There is increasing recognition that HMG 14 and 17 may play an important role in transcriptional control. Data supporting such a postulate includes evidence for their selective association with actively transcribing genes (8, lo), the enrichment of acetylated histones in polynucleosomes containing HMG 14 and 17 (10, 13), the inhibition of transcription by microinjection of anti-HMG 17 or the Fab fragment (12), and the preferential association of anti-HMG 14 and anti-HMG 17 with the puffed regions of Drosophila chromosomes (15). The identification of HMG 14 and 17 as substrates for various protein kinases (Refs. 28-33 and this report) and the observation that particularly HMG 14 is phosphorylated in intact cells (22)(23)(24)(25)(26)(27) lead to the suggestion that transcription may be specifically regulated by the phosphorylation of these two proteins. Of particular note, Cooper et al. (26,27) have shown that thyroid-stimulating hormone promotes rapid phosphorylation of thyroid HMG 14 associated with transcriptionally active chromatin. The evidence presented here has shown that both HMG 14 and 17 may be highly active substrates for the Ca2+-phospholipid-dependent protein kinase. There appears to be a high selectivity for the phosphorylation of each in a specific site, and the K , value for each is in a concentration range that is similar to the nuclear concentration of each. Although a specific nuclear localization has yet to be shown for the Caz+-phospholipid-dependent protein kinase, it has been shown to be localized both membrane bound and in the cytosol and to be translocated upon stimulation of the cell (52). Of note, the Ca2+-phospholipiddependent protein kinase appears to be a primary target for the tumorogenic agents, the phorbol esters, for which modi-

FRACTION NUMBER
FIG. 9. Comparison by HPLC of tryptic phosphopeptides from HMG 14 and 17. Tryptic peptides from HMG 17 (panels a and b) and HMG 14 (panels c and d) were prepared as described under "Experimental Procedures." Indicated in panels a and c are their initial separation using a Waters C-18 reverse-phase HPLC that was developed with an acetonitrile gradient containing 0.05% trifluoroacetic acid. The fractions of the 32Pcontaining peaks from that separation were lyophilized, dissolved in 25 m M triethylamine acetate (pH 4.4), and rechromatographed using a Waters pBondapak-NHz column developed with a linear triethylamine gradient (panels b and d). Conditions of phosphorylation were as described under "Experimental Procedures" and respectively. Panel d, rechromatography of the HMG 14 phosphotryptic peptides obtained from U , Ca2+-phospholipid-dependent protein kinase; A-A, the CAMP-dependent protein kinase, peak eluting at 25% in the acetonitrile gradient (panel c); W, the CAMP-dependent protein kinase, peak eluting in unretained fractions (panel c). The heauv arrows in a and c indicated the point of elution of undigested phospho-HMG 17 and phospho--HMG 14, respectively. fication of transcriptional control must be a key element in their action (53, 54).
As indicated above, HMG 14 and 17 appear to be preferentially associated with transcribing genes. Based upon studies with reconstituted nucleosomes, there appear to be specific binding sites for HMG 14 and 17 at the entrance and exit regions of the nucleosome (4-7). Whether or not the localization of HMG 14 and 17 is the same in actively transcribing genes as in the reconstituted nucleosomes is not established, but it is of note that there was rapid phosphorylation of HMG 17 in either the initially isolated nucleosomes or in nucleosomes reconstituted with stoichiometric levels of HMG 17 (Fig. 6). Of particular interest, the regions of residues 12 to 40 in HMG 14 and 16 to 44 for HMG 17 are highly conserved, and these conserved basic regions are the primary binding segments of these proteins to DNA (55)(56)(57). Within these sequences are the primary phosphorylation sites for the Ca2+phospholipid-dependent protein kinase. As indicated in Figs. 6 and 7, HMG 17 is phosphorylated with equal efficacy when bound to nucleosomes, and despite this phosphorylation occurring in the region of DNA binding, the phosphorylated HMG 17 remained bound. Thus, the phosphorylation of HMG 17 might be expected to have a significant effect on the DNA structure in this region with possible important consequences on the processes of transcription.
The data presented here provides, to the best of our knowledge, the first identification of the peptide sequence for a substrate with high efficacy for the Cat+-phospholipid-dependent protein kinase and thus may provide an avenue to probe what are the critical residues that dictate substrate specificity. Of interest, a peptide derived from histone H2B, of sequence Arg-Lys-Arg-Ser-Arg-Lys-Glu, has also been shown to be phosphorylated by the Ca''-phospholipid-dependent protein kinase, albeit that both it and H2B are poor substrates (58). Possibly the basic amino acids in close proximity to the phosphorylated serine, as with the CAMP-dependent protein kinase (59, 60), play an important role in dictating substrate specificity for the Ca2+-phospholipid-dependent protein kinase. In the case of the CAMP-dependent protein kinase, however, a peptide with two arginines immediately next to the serine is only a very poor substrate (61). As has now been observed with several other proteins (62-64), the phosphorylation of one site on HMG 17 modified the efficacy for phosphorylation at a second site. Thus, as indicated in Fig. 3, panel b, when HMG 17 was phosphorylated by the CAMP-dependent protein kinase to a level of 0.3 mol/ mol in Ser 28, the maximum level of phosphorylation achieved with the subsequent addition of the Ca2+-phospholipid-dependent protein kinase was only 1 mol/mol. Similarly, although the Ca2+-phospholipid-dependent protein kinase catalyzed the phosphorylation of two sites, one major, one minor (Fig. 4, panel b), the maximum level of phosphorylation was only 1 mol/mol (Fig. 1, panel f, inset). Presumably in both cases, those molecules of HMG 17 already phosphorylated in Ser 28 could not be subsequently phosphorylated in Ser 24. The significance of this in the case of HMG 17 is questionable since it. is quite doubtful that HMG 17 is phosphorylated in the cell by the CAMP-dependent protein kinase (28, 30), but, nevertheless, this phenomena of intersite effects on phosphorylation is becoming increasingly recognized in the interpretation of both in vivo and in vitro studies.