Phosphorylation of Tyrosine Hydroxylase in Situ at Serine 8 , 19 , 31 , and 40 ”

The site specificity of tyrosine hydroxylase phosphorylation in intact PC12 cells, labeled with 32Pi, was investigated. Digestion of s2P-tyrosine hydroxylase with trypsin produced five distinct 32P-labeled peptides (termed PC-1 through PC-5). Sequencing of the peptides revealed four acceptor sites: Ser’, Ser”, Sersl, and Ser4’. The phosphorylation site in peptides PC-1 (AVSEQDAK) and PC-2 (RAVSEQDAK) was identified as Ser”. Agents which cause calcium influx increased 32P incorporation into tyrosine hydroxylase at Ser”. PC3 was identified as QAEAVTSPR, which contains the phosphorylation site Ser3’. Nerve growth factor and phorbol dibutyrate increased 32P incorporation into Ser3’. PC-4 was identified as the N-terminal amino acid sequence ((M)PTPSAPSPQPK), and the 32P incorporation occurred at Se?. Of the agents tested, only okadaic acid (a protein phosphatase inhibitor) increased the phosphorylation of Ser*. PC-5 was shown to contain Ser4’. Treatment of the PC12 cells with CAMP-acting agents increased 32P incorporation into Ser40. The present results demonstrate that some, but not all, of the phosphorylation sites demonstrated previously in vitro exist in situ. Conversely, the identification of Ser31 establishes a physiological phosphorylation site not previously reported in vitro. These four sites account for most, if not all, of the diversity in tryptic phosphopeptides reported previously for rat tyrosine hydroxylase.

Tyrosine hydroxylase (TH)' catalyzes the initial step in the biosynthesis of catecholamines. Because this step is ratelimiting, the regulation of TH activity has been the subject of numerous studies (cf. Ref. 1). Recent attention has focused on the activation of TH by protein kinase-mediated phosphorylction. Cyclic AMP-dependent protein kinase was the first shown to phosphorylate and activate tyrosine hydroxylase (2, 3). Because TH activity increased directly with phos- phate incorporation up to -0.7 mol of phosphate/m01 of TH subunit (4), a "one-site one-kinase" model was suggested for the regulation of TH activity. In situ, however, multiple-site phosphorylation of TH was demonstrated (5) and, subsequently, several other protein kinase systems were shown to phosphorylate and activate TH in vitro (6)(7)(8)(9). Campbell and co-workers (10) identified four phosphorylation sites in rat pheochromocytoma TH in uitro. Serlg was phosphorylated by CAM-PK II; Ser*' was phosphorylated by CAMP-PK, protein kinase C, and, to a small extent, CAM-PK II. Minor phosphorylation of two other sites was also demonstrated. A protein kinase activity associated with the partially purified TH phosphorylated Ser', and CAMP-PK phosphorylated Ser153 in a proteolytically degraded TH preparation.
Tryptic digestion of TH after phosphorylation in situ in rat pheochromocytoma cells (PCl2) also produced four phosphopeptides (11). However, a different separation technique was used, thus comparison of these peptides to the sites identified in vitro (10) was not possible. Subsequent studies of the multiple-site phosphorylation of TH in situ have reported anywhere from three to seven tryptic phosphopeptides (12)(13)(14)(15)(16)(17). Thus, neither the number of phosphopeptides nor a strong inference regarding which phosphorylation site(s) might be represented by each of the phosphopeptides has been possible.
In the present report, multiple-site phosphorylation of TH in intact PC12 cells has been analyzed and the properties of the peptides bearing the phosphorylation sites have been characterized in a number of separation systems, allowing comparison among the previous studies. Furthermore, the high level of TH expression in the PC12 cells provided sufficient material for determining the amino acid sequences of the phosphopeptides, allowing assignment of phosphorylation sites to the different phosphopeptides (18) elution profile of "'P-labeled peptides produced by limit tryptic digestion of TH immunoprecipitated from untreated PC12 cells after labeling with 32Pi. Five prominent peaks of radioactivity were detected and named according to their order of elution, as shown. To determine whether more than one phosphopeptide was present in any of the individual peaks, PC-l through PC-5 were concentrated and analyzed with twodimensional electrophoresis/chromatography (tryptic fingerprinting) and with isoelectric focusing. As a rule, each of the five peaks migrated as a single phosphopeptide in either of the two-dimensional separation procedures.
(Occasionally, the fingerprints of PC-3 revealed a second, minor phosphopeptide (cf. Miniprint).) The correspondence of the elution pattern of the five phosphopeptide peaks in reverse-phase HPLC to their migration in the two fingerprint systems is presented in Fig. 2. Thin-layer isoelectric focusing in Servalyt 3-10 Precoat gels indicated that all of the phosphopeptides had pIs less than 6 ( Fig. 3). The p1 values of the phosphopeptides were ordered (from lowest to highest) PC-l < PC-3 < PC-2 = PC-5 < PC-4. Relative to the Coomassie staining pattern of the three different Serva isoelectric focusing standards, the p1 estimates are PC-l, -3.2; PC-2, -3.9; PC-3, -3.5; PC-4, -5.1; PC-5, -3.9.
Thus, the five 32P-labeled peaks that were separated with reverse-phase HPLC appeared to be single distinct phosphopeptides. Phosphoamino acid analysis of each of the phosphopeptides revealed phosphoserine but not phosphothreonine or phosphotyrosine (Fig. 4).

Regulation of 32P Incorporation into TH Phosphopeptides
To facilitate comparison of the phosphopeptides in the present study to those reported previously and to optimize 32P incorporation (and, by inference, stoichiometry) into each of the peptides for subsequent analyses, several test substances were evaluated for their abilities to increase 32P incorporation into TH and the phosphopeptides. Secretagogues-Elevated K' (40 mM), veratridine (100 PM), or nicotine (50 PM) increased 32P incorporation into TH. Relatively brief treatments (30-60 s) selectively increased the phosphorylation of PC-l and PC-2 (Table I)  added 60 s prior to secretagogue) abolished these effects (not shown). Longer treatments (5 min) with the same concentrations of nicotine, veratridine, and elevated K+ as in Table I also produced small (20-30%) increases in 32P incorporation into PC-5 (not shown). A23187, which causes calcium influx ionophorically as opposed to via depolarization, also increased TH phosphorylation.
In contrast to the other secretagogues, A23187 increased the phosphorylation of PC-3 in addition to PC-l and PC-2.
The selective increase in PC-l/PC-2 phosphorylation by elevated [K'], has not been observed previously. This is presumably due to the longer treatment periods (5 min to l-2 h uers'sus 30 s) and higher temperature (37 uersus 22-24 "C) in the previous studies resulting in increases in, variously, PC-3 (T3, see below) or PC-3 and PC-5 (Tl, see below) as well (11,13,14). Similarly, A23187 and ionomycin were shown previously to increase the phosphorylation of all of the peptides except what appears to be 21).
Phorbol Esters-Phorbol dibutyrate (1 pM, 15 min) increased 32P incorporation into TH, and the effect resulted predominantly from an increase (240%) in PC-3 phosphorylation. Phorbol dibutyrate also produced a substantially smaller increase (35%) in PC-5 phosphorylation (Table I); however, a similar effect was also observed with 4-a-phorbol dibutyrate (1 pM, 30 min; not shown), and the effect of phorbol dibutyrate on PC-5 phosphorylation was not observed at lower concentrations (lo-100 nM; not shown).
CAMP-acting Agents-Forskolin (10 pM, 10 min), &bromo-CAMP (1 mM, 5 min) or dibutyryl-CAMP (1 mM, 5 min) increased the phosphorylation of TH (Table I). With these agents, the increases in 32P incorporation into TH were associated entirely with an approximately 2-fold increase in PC-5 phosphorylation.
The magnitude of these effects appeared to be inversely related to the relative level of phosphorylation of PC-5 in control cells. In fact, forskolin could produce greater than 4-fold increases in 32P incorporation into PC-5 when PC-5 phosphorylation was decreased by prior treatment with theophylline. 3 Vasoactive intestinal peptide and secretin, but not substance P, neurotensin, or somatostatin (all 1 pM, 30 min), also produced selective increases in 32P incorporation into PC-5 (not shown).
In vitro, protein kinase C phosphorylates TH on the same peptide as that phosphorylated by CAMP-PK (7,35). In contrast, phorbol esters have been reported to increase 32P incorporation into TH in PC12 cells predominantly, if not entirely, in association with a peptide other than that influenced by CAMP-acting agents (11,13,21).
In that this peptide appears to be PC-3, there is, with one exception (17), agreement between the present and previous studies in intact PC12 cells. However, the disparity between in uitro and in situ results suggests that the involvement of protein kinase C in the phorbol ester-induced phosphorylation of TH in situ is indirect.
The increase in PC-5 phosphorylation by forskolin and CAMP analogues agrees with previous reports. With comparable treatment conditions, a similar selective increase was reported (14, 21), whereas with longer treatments, increases in PC-3 phosphorylation were also observed (11). A similar effect of vasoactive intestinal peptide has been reported in other systems (cf. Ref. 22) and secretin is predicted to have an effect similar to vasoactive intestinal peptide on pharma-Growth Factors-NGF produced effects similar to those produced by high concentrations of phorbol dibutyrate, although the magnitude of the NGF effect on PC-3 was larger (Table I). EGF, on the other hand, was without effect on TH phosphorylation in the present studies. The use of different treatment durations (l-120 min), EGF concentrations (lo-1000 rig/ml), levels of confluence (cells passed at %o to l/2), and PC12 cells (two different sources) all failed to produce an EGF-stimulated increase in 32P incorporation into TH (not shown).
Several laboratories have shown that NGF increases TH phosphorylation in PC12 cells (11,13,(23)(24)(25). In the two studies which analyzed tryptic phosphopeptides, both also reported increases in PC-3 (T3) and PC-5 (Tl) with the effects on PC-3 being similarly larger than those on  reported larger overall effects with a longer (l-2 h) treatment, and Cahill et al. (13) reported smaller overall effects with a shorter (5 min) treatment. phorylation on PC-4 (T4) as well as PC-3. One hypothesis for the failure to observe any effect of EGF in the present studies is that the transduction system(s) influenced by EGF is already activated in control cells. At least some variation of this hypothesis seems likely, because, in contrast to the present data, '?'P incorporation into T4 was barely discernible without EGF treatment in the previous report (11) Table I and as will be described elsewhere in greater detail, okadaic acid (1 PM, 30 min) increased 32P incorporation into all of the peptides. As shown in Table I, though, the increase in TH phosphorylation was greater than the increase in "P incorporation into total cellular protein, indicating that the overall phosphate content of TH is turned over more rapidly than that of the general population of cellular proteins. In terms of the individual phosphopeptides, this was true for all of the peptides except PC-4, which showed a 2-fold increase in phosphorylation, comparable with the increase in ,"P-labeled total protein.

Analysis of Phosphorylation Sites
The phosphopeptides were subjected to manual Edman degradation, and subtractive analyses were performed to identify the location of phosphoserine residues and to test for the presence of multiple phosphorylation sites within a given peptide. As illustrated in Fig. 5, a single phosphorylation site was present in PC-l and in PC-5 at the third amino acid from the N terminus. A single phosphorylation site was present in PC-2 at the fourth amino acid from the N terminus (not pH shown). No phosphorylation sites were revealed in either PC-3 or PC-4 through six degradation cycles. In the first sequencing attempt, cells were treated with veratridine, phorbol dibutyrate, and forskolin to increase "'P incorporation into (and presumably the total phosphate content of) PC-l, -2, -3 and -5. Sequences were obtained for PC-1, -2, and -5. In the second sequencing experiment, 50% more cells were used, the cells were treated with NGF and okadaic acid, and a higher threshold for the collection of the radioactive peaks was selected. With this approach, sequence data for both PC-3 and PC-4 were obtained.
The PC-l sample produced a sequence of ALZV1XElfiQ9L-Di9AliKR, wherein the numbers in subscript indicate the picomole yield. From the subtractive Edman analysis (Fig. 5), the loss of a positive charge was observed at both pH 1.9 and 8.9, presumably reflecting modification of the Lys side chain. At pH 8.9, the additional increase in mobility presumably reflects deamidation of the Gln to Glu. Also, the release of 'Y2P1 at position 3 allows the assignment of phosphoserine for the X. Thus, PC-l was assigned the sequence AVS(P)EQDAK containing Serlg as the phosphorylation site.
The PC-2 produced the sequence RAVXEQDAK at a level of 6-12 pmol. Consistent with this sequence, subtractive Edman analysis at pH 1.9 revealed a loss of two positive charges after the first cycle and a release of "'PI at the fourth cycle (not shown). Thus, the sequence RAVS(P)EQDAK, with the phosphorylation site being Ser'", was assigned to PC-2.
The first analysis of PC-3 produced readings in the 6-20 pmol range for one to two amino acids/cycle. Two possible TH sequences could be matched with the data: VSDDVR and XAEAVTXPR, corresponding respectively to nonphosphorylated Ser'"" and potentially phosphorylated Ser.". Reversephase HPLC chromatography of the radioactivity remaining on the filter (30) indicated that the phosphoserine was cleaved within the eight cycles run. The second analysis, with modified experimental conditions, produced the sequence Q A E A V T S PrrR R4 141 120 118 115 142 Id .a.> 2" with no indication of the presence of VSDDVR.
Based on sequencing yields of serine from known amounts of phospho versus dephospho-LRRASVA, the 13 pmol of serine in position 7 is consistent with the original residue at this position being entirely from phosphoserine.
(Also, the phosphopeptides subjected to sequencing analysis would be expected to elute prior to their cognate dephosphopeptides (31).) From the subtractive Edman analysis at pH 1.9 (not presented), cycle one decreased the mobility of a portion of the molecules, and the shift was consistent with a loss of one positive charge from +2 to +l. The mobility of the shifted portion of peptide remained unchanged with subsequent cycles, which was taken to reflect conversion of Gln to pyro-Glu in a portion of the molecules during the Edman procedure. At pH 8.9, a decrease in mobility was observed after cycle 3, consistent with the loss of Glu. Together, these data identified PC-3 as QAEAVTS(P)PR, wherein Ser"' is the phosphorylation site.
In the first analysis of PC-4, all of the readings subsequent to the first cycle were below 10 pmol, and the analysis was terminated after six cycles. Consistent with the manual Edman analyses, HPLC analysis of the filter as above indicated that most of the 32P was still associated with peptide after six cycles. Given the low yield of Pro in sequencing PC-3 and the possibility that PC-4 was the Pro-rich peptide containing Ser' (lo), Pro-l cycles were utilized for cycles 1, 3, 6, and 8 during the second attempt at sequencing PC-4. This analysis resulted in the sequence P. T -P S A,rP. 31 3, 3s 2s 2, LO XPi6 for the eight cycles performed.
Because PC-4 contained more than 1 Ser, the portion of each cycle not injected into the HPLC (collected into vials by the ABI 477A) was analyzed for radioactivity. Cycles l-6 produced 30-40 cpm, whereas cycle 7 produced 110 cpm. This and the relatively high yield of Ser at position 4 indicated that Ser* was the phosphorylated serine in PC-4.
In the subtractive Edman analysis (not presented), a loss of one positive charge was apparent after the first cycle and maintained through six cycles, consistent with modification of the side chain of Lys presumed to be at the C-terminal of the phosphopeptide. From these data, PC-4 was assigned the sequence PTPSAPS(P)PQPK.
The PC-5 sample resulted in a sequence of R40Q&8Ls3177E59-D55As2R52. From the subtractive Edman analysis (Fig. 5), the loss of a positive charge at cycle 1 was indicated by a decrease in mobility at pH 1.9 and an increase in mobility at pH 8.9. From this, the release of 32Pi, at cycle 3, and the presence of a single serine having a low yield characteristic of phosphoserine, PC-5 was assigned the sequence RQS(P)LIEDAR containing Ser4' as the phosphorylation site.

Comparison to Previously Reported Tryptic Phosphopeptides from TH
The results of the sequence analyses are summarized in Table II (top). Also presented in Table II (bottom) is a collation of previously described tryptic phosphopeptides from 32Plabeled TH labeled in intact PC12 cells. The assignments were made on the bases of the correlative phosphopeptide separations in Fig. 2 and the pharmacological profile in Table I. DISCUSSION Little is yet known about the secondary or tertiary structure of TH. In the absence of such data and on the basis of the primary structure of rat TH inferred from cDNA clones, many of the 42 serine residues, distributed throughout the molecule, are potential phosphorylation sites. Campbell et al. (10) demonstrated that at least four of these (serines 8, 19, 40, and 153) could be phosphorylated in vitro. The present studies, however, show that all of the sites of phosphorylation of TH in intact PC12 cells occur within 40 amino acids of the N terminus (serines 8, 19, 31, and 40). Such data are consonant with the hypothesis that the N-terminal region of the enzyme constitutes a regulatory domain which, in the dephosphorylated state, inhibits the catalytic center(s) located further toward the carboxyl portion of the molecule. Phosphorylation of the N-terminal region then relieves the inhibitory influence.
Phosphorylution Sites Previously Identified in Vitro-Ser" and Ser4' are within canonical substrate sequences for CAM-PK II and CAMP-PK, respectively (cf. Ref. 32) and have been considered to be strong candidates for the phosphorylation sites in respectively (33). Depolarizing secretagogues selectively increased PC-l/PC-2 phosphorylation while CAMP-acting agents selectively increased PC-5 phosphorylation (Table I). In oitro, CAM-PK II can be promiscuous with respect to the sites on TH that it phosphorylates. At levels of phosphorylation up to -0.25 mol of phosphate/ no1 of subunit, CAM-PK II phosphorylates predominantly Ser" (34). At intermediate stoichiometry (-1 mol/mol subunit), Ser" and Ser4' phosphorylation are roughly equal (35). And, at "optimal" stoichiometry (3.9 mol/mol (subunit)), CAM-PK II produced 32P incorporation into five tryptic phosphopeptides observed after labeling in situ (17). Such observations have prompted the suggestion that CAM-PK mediates the phosphorylation of virtually all of the phosphorylation sites observed in situ. However, from the present data, calcium influx via voltage-sensitive channels elicits a selective increase in Ser" phosphorylation. The effects of secretagogues on the phosphorylation of other sites seen with, e.g. longer treatments with elevated [K'],, seems more likely to involve the recruitment of other protein kinase systems via some form of cascade than to result from the promiscuity observed for CAM-PK II in uitro.
A protein kinase capable of phosphorylating Se? (10) has recently been characterized as a novel proline-directed protein kinase which phosphorylates -Xaa-Ser/Thr-Pro-Xaa-sequences (36). Tissue distribution studies (37) indicate that the proline-directed protein kinase is extremely low in brain, adrenal medulla, and other non-mitotic tissues. In agreement, Ser' phosphorylation is high in PC12 cells (Fig. 2, Table I) but exceedingly low in corpus striatal synaptosomes (15), corpus striatum in vivo (38), bovine adrenal chromaffin cells (33), and perfused rat adrenal (39). The low abundance of the proline-directed protein kinase in neural tissues, the low 32P incorporation into the Ser'-containing phosphopeptide in these tissues, and the failure of most treatments (including neuronal activation ( 12))4 to increase the phosphorylation of Ser' all suggest that Sera phosphorylation may only play a role in the regulation of TH in pathology and/or development.
Although Ser'53 can be a substrate for CAMP-PK in uitro (lo), no evidence for Ser153 phosphorylation in situ was obtained in the present studies. Presumably the higher order structure of TH in its native state is not favorable.
Serine 31 Phosphorylation-The identification of Ser31 as the phosphorylation site in PC-3 is both exciting and perplexing. Phorbol esters and NGF increase Ser31 phosphorylation; yet, the amino acid sequence of PC-3 (QAEAVTSPR) does not contain determinants that would make it an obvious substrate for a particular protein kinase. Ser31 does reside within the -X-S/T-P-X-sequence suggested to confer specificity for the recently described proline-directed protein kinase; however, additional determinants clearly must exist, because this protein kinase phosphorylates TH exclusively on Ser' (10,36).5 Alternatively, basic residues on either side of a serine are important in promoting phosphorylation by protein kinase C (40, 41), raising the possibility that Arg33 might confer reactivity with protein kinase C. Cremins et al. (42) contend that protein kinase C mediates the effects of phorbol esters and NGF on peptide T3 (Ser31) phosphorylation in situ. In contrast, the response of T3 phosphorylation to phorbol ester, but not NGF, is lost in PC12 cells pretreated with phorbol ester (13). Such procedures do not, however, necessarily down-regulate all forms of protein kinase C (43-455) or address whether protein kinase C is the direct effector of Ser31 phosphorylation. In fact, Ser4' is the preferred (17) or only (7,35) TH substrate for protein kinase C in vitro. Thus, Ser31 does not appear to be a substrate for protein kinase C, and the effects of phorbol esters on TH phosphorylation in situ appear to be mediated only indirectly by protein kinase C. Studies to identify the protein kinase(s) directly responsible for Ser31 phosphorylation are currently underway.
Although not observed previously in vitro, Ser31/PC-3 phosphorylation has been previously observed as peptide T3 in pH 8.9 fingerprints and as peptide 2 in HPLC analyses (14, 21) of TH phosphorylated in intact PC12 cells. (In the reported analysis of pH 1.9 fingerprints (17), PC-l, PC-3, and PC-4 were not resolved (cf. Fig. 2).) All of the laboratories are in agreement that NGF and phorbol esters increase Ser31 phosphorylation in PC12 cells; however, in contrast to the present data, the previous studies reported that elevated [K'10 and, in some cases, CAMP-acting agents also increased T3 phosphorylation. One possible explanation for these discrepancies is that shorter treatment periods were used in the present study (30 s uersm 5 min to l-2 h for elevated [K'],, 5-10 min uersus l-2 h for CAMP-acting agents).

Functional
Consequences of Site-specific Phosphorylation- in both Serlg and Ser4' phosphorylation) could provide for more efficient catecholamine synthesis under conditions wherein diversion of precursors (e.g. GTP) from tetrahydrobiopterin synthesis is anticipated. This form of activation may be less tightly coupled to neuronal activity in the sense of being temporally dampened. An influence of Ser' phosphorylation on TH activity has not yet been observed, but the stoichiometries achieved have been below 25% (36). In terms of neuronal or chromaffin cell function, however, the issue may be moot given the low levels of the proline-directed protein kinase in brain and adrenal (37).
In terms of Ser3r phosphorylation, treatment of PC12 cells with NGF, phorbol esters, and diacylglycerols increase TH activity in association with the increase in TH phosphorylation (11,21,47), and, in one report (21), the activation was associated with a selective increase in Ser3* phosphorylation. Thus, Ser3* phosphorylation appears to influence TH activity.
In that low frequency stimulation of the splanchnic nerve in perfused rat adrenal increases Ser31 phosphorylation,4 a physiological role for Ser31 phosphorylation in regulating catecholamine synthesis seems likely.
The phosphorylation site at Ser31 in PC-3 (QAEAVTSPR) is also present in bovine and human (HTH-1) TH as the tryptic peptide QAEAIMSPR. If, by analogy, this site is also phosphorylated in HTH-1, a new consequence of alternative splicing arises. Human TH mRNA undergoes alternative splicing which results in the insertion of amino acids between Met3' and Ser3'. In HTH-2, four amino acids are inserted resulting in the amino acid sequence QAEAIMVRGQSPR and creating the Arg-X-Y-Ser consensus phosphorylation site for CAM-PK II (cf. Ref. 32). Whereas it was previously suggested that alternative splicing creates a phosphorylation site (48), the present data indicate that alternative splicing produces a chunge in substrate specificity for an existing phosphorylation site. That is, the site influenced previously by phorbol dibutyrate and NGF would now respond to calcium influx similarly to Ser". A shift in expression of HTH-1 uersus HTH-2 could present functionally different enzymes to the cell; HTH-1 would be capable of responding to the protein kinase(s) stimulated by phorbol esters and NGF. Alternatively, HTH-2 could have an enhanced response to calcium influx but lose it responsiveness to the phorbol/NGFstimulated protein kinase(s).