Analysis of the Substrate Specificity of Tyrosylprotein Sulfotransferase Using Synthetic Peptides*

Tyrosylprotein sulfotransferase (TPST) catalyzes the sulfation of proteins at tyrosine residues. We have analyzed the substrate specificity of TPST from bovine adrenal medulla with a novel assay, using synthetic peptides as substrates. The peptides were modeled after the known, or putative, tyrosine sulfalion sites of the cholecystokinin precursor, chromogranin B (se-cretogranin I) and vitronectin, as well as the tyrosine phosphorylation sites of a-tubulin and pp6Fc. Varying the sequence of these peptides, we found that catalyzed the sulfation of a peptide corresponding to the tyrosine autophosphor- ylation site of pp60”@’ but not of a peptide corresponding to the non-autophosphorylation site of pp60+*” These results experimentally de- fine structural determinants for the substrate specificity of TPST and show that this enzyme and certain autophosphorylaGng tyrosine kinases have overlap- ping substrate specificities in vitro.

We have analyzed the substrate specificity of TPST from bovine adrenal medulla with a novel assay, using synthetic peptides as substrates.
The peptides were modeled after the known, or putative, tyrosine sulfalion sites of the cholecystokinin precursor, chromogranin B (secretogranin I) and vitronectin, as well as the tyrosine phosphorylation sites of a-tubulin and pp6Fc. Varying the sequence of these peptides, we found that (i) the apparent Km of peptides with multiple tyrosine sulfa-Gon sites decreased exponentially with the number of sites; (ii) acidic amino acids were t,he major determinant for tyrosine sulfation, acidic amino acids adjacent to the tyrosine being more important than distant ones; (iii) a carboxyl terminally located tyrosine residue may be sulfated.
Moreover, TPST catalyzed the sulfation of a peptide corresponding to the tyrosine autophosphorylation site of pp60"@' (Tyr-416) but not of a peptide corresponding to the non-autophosphorylation site of pp60+*" (Tyr-527). These results experimentally define structural determinants for the substrate specificity of TPST and show that this enzyme and certain autophosphorylaGng tyrosine kinases have overlapping substrate specificities in vitro.
Sulfation is the most abundant post-translational modification of t,yrosine residues in metazoan cells (1). As an increasing number of tyrosine-sulfated proteins have been identified, there is growing interest in the biochemistry, cell biology, and function of protein tyrosine sulfation. This posttranslational modificaCon is catalyzed by tyrosylprotein sulfotransferase (TPST)' (2), an integral membrane protein of the trans Golgi (3, 4), and occurs in soluble and membrane proteins residing in, or passing through, this compartment (1). Although TPST activity has been found in various cellfree systems (for review, see Ref. l), little information is available on the detailed biochemical properties and characteristics of this enzyme (3, 5-7). * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18  Various types of substrates have previously been used to detect TPST activity. These include: (i) physiologically tyrosine-sulfated proteins and peptides that were desulfated prior to use; (ii) physiologically tyrosine-sulfated proteins isolated in unsulfated or partially sulfated form; (iii) proteins that are not tyrosine-sulfated physiologically but serve as substrates for TPST kz vitro; (iv) synthetic, randomly condensed, tyrosine-containing polymers; and (v)  were applied to a 10 x 24.cm sheet of Whatman 3MM chromatography paper that had been prewetted with electrophoresis buffer (7.8% acetic acid, 2.2% formic acid, pH 1.9) and blotted onto Whatman 3MM paper to remove excess liquid. Samples were spotted at 2-cm distance from each other onto a line running at a 4.cm distance in parallel to one of the 24-cm edges of the sheet; this edge subsequently became the anodic side. After spotting, up to two such wet paper sheets (with up to 12 samples/sheet) were placed into an electrophoresis tank (see Fig. 1 of Miniprint for details) and subjected to electrophoresis at 4 'C. In the case of peptide CCK-1, electrophoresis was carried out until the front of the phenol red marker had migrated ~2 cm out of the origin toward the anode (~12 min at 400 V, 40 V/cm).
For the other peptides, the time of electrophoresis was adjusted according to the respective RF value (see Table I of Miniprint).
After electrophoresis, the paper sheet was dried at 120 "C for 10 min. For each peptide, the electrophoretic migration was examined in initial experiments by autoradiography. ["'S]CCK-1 separated by paper electrophoresis was eluted from the paper with 2 ml of HZO, lyophilized, and subjected to alkaline hydrolysis followed by two-dimensional cellulose thin layer electrophoresis as described (10).
Protein was determined using the dye binding assay (Bio-Rad) or a turbidometric assay (11).

RESULTS'
Synthetic Peptides as Substrates for TPST-The goal of the present study was to investigate the substrate specificity of TPST using a variety of synthetic peptides. In developing a suitable assay for this purpose, we encountered the problem (particularly when crude membranes were the source of TPST) that the sulfated peptides were incompletely freed from the sulfate donor [35S]PAPS and/or endogenous sulfatelabeled compounds. It seemed likely that not only PAPS but also the majority of the sulfated endogenous compounds remaining after removal of heat-precipitated proteins carry a " Portions of this paper (including part of "Results," Figs. l-3, and Table I net negative charge at pH L9 and migrate toward the anode upon electrophoresis. At this pH, peptides with 2 basic amino acid residues, even if they contained several carboxyl groups, should still be positively charged after sulfation and migrate toward the cathode; this should result in their separation from sulfated endogenous compounds as well as PAPS and its degradation products. We therefore added 2 basic residues to the various peptides and used paper electrophoresis for their separation (for details, see Miniprint).
Incubation of the peptides with the sulfate donor ["S]PAPS and a TPST preparation led to the formation of sulfated peptides that on paper electrophoresis migrated slightly slower than the unsulfated peptides. A typical electrophoretic separation is shown in Fig. 1 for the peptide CCK-1, corresponding to the carboxyl-terminal nonapeptide of rat preprocholecystokinin which is tyrosine-sulfated in vivo (12). The major radioactive compounds present on the anodic side of the paper sheet were identified as ["S]APS and ["S]PAPS by comigration with unlabeled standards. The radioactive material with faster electrophoretic mobility than PAPS was most likely inorganic "'SO.,. Both [?S]APS and free '5SOd probably are breakdown products generated by nucleotidases typically present in membrane preparations (13). One unidentified sulfate-labeled compound was present on the cathodic side after incubation for 30 min irrespective of the presence of CCK-1. This endogenous product was not observed when TPST solubilized from carbonate-treated, Golgi-enriched membranes was assayed (data not shown). ["'S]CCK-1 was clearly separated from all of these radioactive compounds.
Since CCK-1 contains serine and threonine residues, it was necessary to ascertain that its sulfation occurred only on tyrosine. When ['?S]CCK-1 was subjected to alkaline hydrolysis followed by two-dimensional thin layer electrophoresis, 65% of the starting radioactivity was recovered in the tyrosine sulfate spot (Fig. 2). This recovery was in the same range as that of authentic tyrosine ['?S]sulfate standard (typically =71%, (14)), showing that the sulfation of CCK-1 occurred only on tyrosine. Comparison of the amount of sulfate-labeled CCK-1 and sulfate-labeled endogenous proteins showed that CCK-1 was the major acceptor for sulfate on tyrosine residues (data not shown). The apparent K,,: of TPST for the cosubstrate PAPS was 2 pM (see Fig. 3A of Miniprint), which is similar to the reported value of 5 pM for TPST using EAY as sulfate acceptor (3). We therefore used 5 pM PAPS in the following experiments in which the substrate specificity of TPST was investigated.
The apparent K,,$ of TPST from adrenal medulla for SgI-1, a peptide containing a single tyrosine residue and corresponding to an identified tyrosine sulfation site of a protein from the same tissue (secretogranin I (chromogranin B) (15)), was 43 pM (Table I).
Peptides with Multiple Tyrosine Sulfation Sites-The apparent K"> of TPST for CCK-1 was 35 pM (Table II). Since this peptide contains 2 tyrosine residues, corresponding to position 111 and 113 in prepro-cholecystokinin, both of which are sulfated in uivo (l2), we tested two variants in which either tyrosine residue was replaced by a phenylalanine residue. Table II shows that replacement of either tyrosine residue resulted in a 3-to 4-fold increase in K," and in an up to 2-fold increase in V,,,zQx.
The above results prompted us to systematically investigate the effect of multiple sulfation sites in peptide substrates. To this end we compared the sulfation of SgI-1 with that of SgI-2 and SgI-3, in which the same tyrosine sulfation site was arranged in tandem twice and thrice, respectively. Strikingly, the K,,, of the latter two peptides decreased exponentially with the number of sulfation sites present in the peptide, reaching values as low as 44 nM (Table I, Fig. 3). In contrast, the Vn,,,x was largely unaffected (Table I). Both SgI-2 and SgI-3 appeared to be converted to the monosulfated form, as judged by the decrease in their electrophoretic mobility compared with the respective unsulfated form.
To investigate whether this remarkable decrease in K", was due to an increase in the number of sulfation sites or in some other parameter, we compared SgI-3 with the peptide SgI-4, in which 2 of the 3 tyrosine residues of SgI-3 are replaced by phenylalanines. The K,,, of SgI-4 was one order of magnitude higher than that of SgI-3 and two orders of magnitude lower than that of the peptide SgI-1, which like SgI-4 contains only 1 tyrosine residue. This indicated that both the tyrosine residue itself as well as the surrounding amino acid residues contributed to the decrease in K", observed for SgI-3.
The Role of Acidic Residues-Which features of the amino acid residues surrounding a tyrosine residue are important for sulfation to occur? A hallmark of the peptides SgI-1 and CCK-1 and of tyrosine sulfation sites in general appears to be the presence of acidic amino acid residues near the tyrosine residue to be sulfates (1, 16). To test the influence of flanking acidic residues on tyrosine sulfation in uitro, we investigated the sulfation of three CCK-1 variants (Fig. U). The acidic amino acid residues surrounding the tyrosines (DYEY) were replaced by the respective amides DYQY (CCK-4), NYEY (CCK-5), and NYQY (CCK-6). Fig. 4A shows that the rate of sulfate transfer to these variants decreased markedly as a function of increasing substitution of acidic amino acids. A 6and I2-fold decrease of sulfation was observed when CCK-4 and CCK-5, respectively, were compared with CCK-1. Sulfation of the doubly substituted CCK-6 was even less than that of CCK-5.
The influence of acidic amino acids distant from the tyrosine was investigated using variants of the peptide SgI-1. In these variants (SgI-5, SgI-6, and SgI-7), distant acidic amino acids were cumulatively exchanged for alanine residues (Fig. 4B). When tested at 0.1 mM, the sulfate transferred to the peptides gradually decreased 2-, 3-, and 6-fold with increasing replacement of acidic amino acids.

The Role of Turn-inducing
Amino Acids-In addition to the presence of acidic amino acid residues, CCK-1 and SgI-1 as well as tyrosine sulfation sites in general contain turn-inducing amino acids, in particular proline and glycine residues. TO test for the influence of these amino acids in sulfation sites in vitro, we investigated the sulfation of an SgI-1 variant in which the proline and glycine residues were replaced by alanines (SgZ-8) (Fig. 5). The sulfation of SgI-8 was about 2-fold less than that of SgI-I.
It has been noted that aspartate residues, which have a greater turn conformational preference than glutamate residues (17), occur more frequently adjacent to sulfated tyrosine residues than glutamates (1, 16). We therefore compared the sulfation of SgI-1 with that of the variant SgI-9, which had an Asp for Glu replacement in position -1 of the tyrosine. Fig.  5 shows that there was little difference in the sulfation of these peptides.

NHZ-and COOH-terminal
Tyrosine Residues-Since the existence of tyrosine sulfation sites in which the tyrosine is located directly at either the NHt-or the COOH-terminus of a protein has not been reported, it was of interest to investi- Vtn-1, Vtn-2, Vtn-3, and Vtn-4 were assayed in triplicate at 0.1 mM concentration. gate whether such tyrosine residues could be sulfated by TPST. We tested the peptides Vtn-1 and Vtn-3, which corresponded to the two putative tyrosine sulfation sites of vitronectin (18), in which tyrosine 56 is preceded, and tyrosine 59 is succeeded, by 3 acidic amino acid residues, respectively.
In an analogous manner, we tested peptides containing (Vtn-3) or lacking (Vtn-4) 2 amino acid residues at the NHzterminal side of a tyrosine residue (Fig, 6). Vtn-3 was a very poor substrate for TPST, its sulfation being =180-fold less than that of SgI-1. In the absence of the NHz-terminally preceding threonine and valine residue, with the tyrosine being located directly at the NH* terminus (Vtn-4), sulfation was reduced at least 2-fold, being at the limits of detection.
Peptide Chain L,ength and Hydrophobicity-To test for the influence of the substrate chain length, we compared the sulfation of SgI-1 to a peptide containing 3 additional alanine residues at the COOH terminus (Sg1-11) (Fig. 7). A 3-fold decrease in the sulfate transferred to the alanine-extended substrate was observed. Replacement of the alanine residues by the more hydrophobic leucine (Sgl-12) resulted in a 2-fold decrease in sulfation compared with SgI-1 (Fig. 7). and phosphorylation sites. To be able to evaluate the significance of the observed acidic amino acid frequencies, the frequency distribution of basic amino acid residues in these sites was also determined (Fig. 8, bottom). This comparison indicates a strong prevalence of acidic amino acids in both tyrosine sulfation and phosphorylation sites. Moreover, in both classes of sites, the probability for an acidic amino acid to be located NH* terminally of the modified tyrosine was found to be greater than that to be located COOH terminally.
To investigate whether tyrosine kinases and TPST have overlapping substrate specificities, we compared the sulfation of CCK-1 with that of peptides whose sequences corresponded FIG. 9. TPST sulfates tyrosine kinase substrates.
The sequences of the synthetic peptides are given in the snzgle-lerrer code. The tyrosine kinase substrate-related peptides Tub-l, srcd,e, and srcslr and, for comparison, CCK-1 were assayed at three time points at I mM concentration.
to (i) the putative tyrosine phosphorylation site of a-tubulin (Tub-l) (19), (ii) the tyrosine autophosphorylation site of pp60v-src (srcdls, (20)), and (iii) the non-autophosphorylation site of pp60ce'rc (srceZ7, (21)). As shown in Fig. 9, Tub-1 and CCK-1 were sulfated to a similar extent, consistent with the apparent Km of TPST for these two peptides being of the same order of magnitude (140 and 35 pM, Fig. 3B of Miniprint and Table II, respectively).
The srcdlG peptide, which has been shown to be tyrosine-phosphorylated by the epidermal growth factor receptor kinase in uitro (22), was also sulfated by TPST, but with a 6-fold lower transfer rate compared with CCK-1. No sulfation was detectable for the peptide srcs2, in which the acidic amino acids are not adjacent to the tyrosine residue.

DISCUSSION
The purification of TPST and the characterization of its substrate specificity requires a rapid and versatile TPST assay. We here describe such an assay using synthetic peptides as substrates. Although it cannot be excluded that the presence of 2 basic residues at the amino terminus of the peptides, facilitating their separation from other sulfate-labeled compounds, might be a potential limitation, the present assay has been found to be very versatile. It allowed the use of peptides with widely varying sequences and, hence, the characterization of the substrate specificity of TPST.
Single Versus Multiple Sulfation Sites-The apparent Km for peptides containing only 1 tyrosine (SgI-1, CCK-2,  was in the range of 10m4 to lo-' M, with the notable exception of peptide SgI-4 (apparent K,,, = 10e7 M) which is discussed below. The former values are in good agreement with the observed apparent K,,, of lob4 M for the peptide tert-butoxycarbonyl-CCK-8 sulfated by non-solubilized TPST from rat brain (5). It remains to be investigated whether the moderate affinity of TPST observed in zdro toward small synthetic peptides containing only 1 tyrosine sulfation site reflects the affinity of this enzyme in uivo toward the larger physiological protein substrates in the lumen of the trans-Golgi.
In the case of the 65-residue, tyrosine-sulfated protein hirudin, the kinetic constants of leech TPST toward the full-length polypeptide were found to be similar to those toward a nonapeptide corresponding to the single tyrosine sulfation site of hirudin (23).
However, in the case of larger proteins and peptide precursors, which may contain multiple recognition sites for TPST, the apparent K,,, may well turn out to be lower than lO-,5 M: we found remarkably low apparent Km values for peptides with multiple sulfatable tyrosines. The apparent Km values of the CCK-1 variants increased 3-4-fold as a result of the presence of 2 sulfatable tyrosine residues and decreased ex-8530 Tymylprotein Sulfotransferase ponentially in SgI-1 variants with increasing number of sulfation sites, reaching 44 nM for the peptide SgI-3 which contains three tandemly repeated tyrosine sulfation sites. One possible explanation for this increase in affinity may be the presence of binding subsites in TPST, in analogy to other enzymes (24). Since the peptide SgI-4, which like SgI-1 contains one tyrosine sulfation site but three times the number of acidic amino acids, had a >lOO-fold higher affinity for TPST than SgI-1, it is likely that binding subsites in TPST recognize acidic amino acids in substrate proteins. This, however, does not fully explain the increase in affinity for the peptide SgI-3, which differed from SgI-4 by the presence of 2 additional tyrosines (as opposed to 2 phenylalanines) and by a further =lO-fold increase in affinity. Previously, TPST from bovine adrenal medulla (3) and rat brain (25) was found to have a very low apparent K,,, (0.3 pM) toward the random synthetic polymer EAY, which may contain multiple tyrosine sulfation sites per single polymer molecule. Our findings might explain this low apparent Km in light of the above observations.
The increased affinity of TPST for substrates with multiple sulfation sites may be of physiological importance since a number of proteins containing such sites are known, e.g. prepro-cholecystokinin (12, 26), cionin (27), complement C4 (28), and heparin cofactor II (29). The presence of multiple sulfation sites might promote stoichiometric sulfation.
Acidic Amino Acids Are a Major Structural Requirement for Tyrosine Sulfation-Our data demonstrate that the previously noted presence of acidic amino acid residues in the vicinity of sulfated tyrosine residues is not merely a correlation and useful for the prediction of potential tyrosine sulfation sites (1,16), but is indeed a structural requirement for the sulfation reaction per se. Both adjacent and distant acidic amino acids contributed to the sulfation of the peptides, adjacent acidic amino acids being more important. It will be interesting to determine whether acidic amino acid residues are only necessary for substrate recognition or whether they are actively involved in catalysis, e.g. by forming an acyl-enzyme intermediate.
Although turn-inducing amino acids are frequently found in tyrosine sulfation sites of proteins (1, 16), we observed little effects of such residues on the sulfation of synthetic peptides. The replacement of two turn-inducing amino acids by alanine residues resulted in a 2-fold reduction in the rate of sulfation in one case and in a 2-fold increase in another case. The substitution of a glutamate by an aspartate, which has a higher turn-conformational preference than the former, did not affect the sulfation of the peptide markedIy. One interpretation of these results is that the high frequency of turn-inducing amino acids in tyrosine sulfation sites of proteins is coincidental, reflecting the fact that both tyrosine sulfation sites and loops (which are rich in turn-inducing amino acids, Ref. 30) are preferentially located on the protein surface. Alternatively, the presence of turn-inducing amino acids may promote tyrosine sulfation of proteins, but not of peptides since the latter are less restrained in adopting an optimal conformation for recognition by TPST.
We found that the COOH-terminal tyrosine residue in the peptide Vtn-2 could be sulfated by TPST, and, surprisingly, with a 5-fold increase in sulfate transfer when compared with the parent peptide which contained two additional COOHterminal amino acids. This increase may be interpreted as a positive effect exerted by the carboxyl group of the tyrosine, which provides a close-by negative charge and thus might mimic an adjacent acidic amino acid residue. In contrast, a peptide containing an NH*-terminal tyrosine residue could not be sulfated significantly by TPST.
Variants of the peptide SgI-1 which were extended by three hydrophobic amino acids were worse substrates for TPST than the parent peptide. This decrease in sulfate transfer might be due to the stabilization of a conformer unfavorable for recognition by TPST or to an increased hydrophobicity.

TPST and Autophosphorylating Protein Tyrosine Kinases
Have Similur Substrate Specificity-Acidic amino acids have also been implicated as a major structural feature in substrates for certain protein tyrosine kinases (31). Indeed, tyrosine sulfation and phosphorylation share a variety of biochemical properties (1). In particular, it has been shown that the tyrosine sulfation site of gastrin can be tyrosine-phosphorylated in vitro (32) and that a-tubulin, which is an in vitro substrate for tyrosine phosphorylation by the insulin receptor kinase (19), is also efficiently tyrosine-sulfated in vitro by TPST (3). Furthermore, the random polymer EAY can be modified by both, certain tyrosylprotein kinases (33), and TPST (3). It is interesting that Tub-l and srcd16, which are targets for the autophosphorylating tyrosine kinase pp60v~src, and which contain acidic residues adjacent to the tyrosine, are sulfated by TPST, whereas src5z7, which is a very poor target for this kinase (21)4 and whose acidic residues are further distant, is not. These data support the notion that the substrate specificity of TPST overlaps with that of certain autophosphorylating tyrosine kinases. This homologous substrate specificity raises the possibility that TPST and protein tyrosine kinases may have evolved from a common ancestor. 7lte electmphoresis tank uxd (shown in cmss-section) was made-of plexiglass and was 60 cm Iong, 6 cm wide and 8 ctn high (iier dimensions) to wzcomodate two 24 cm long paper sheets with I2 samples each. 'Ilx tank contained a a cm long, 3 cm high and 0.8 cm tick pkastic bar rDivi&r") to separate the attodic and cathcdic buffer cbambets which contained 300 mI of pH I .9 buffer each. Tltx ti further comained 2.S I of Es80 Vats01 14%?Cnl which served as a co&ant and prevented evapotatiort of the buffer from the pre-wetted paper. Tote papet sheet is indicated by the cttrwd Ii= (urcw 1). the 10 an edge of dte sheet bcbtg shown in the cmss-se&at, me otigitt of sample qplicatiott on the papet (4 cm from the ancde) is indicated (arrow 2). me tank was kept at 40 C.

T.&n= I
The Rf value of the peptides after paper e1ectmphoresi.t at pH 1.9 is ttomuJized to that of CCK-1 which was arbitrarily set to 1.0. llte proteins from which dte peptide sequences we= tktived are given on dte left md the numbers in brackets refer to the position of ti tyrositx residue in dte pmteins. Using CCK-1 as substrate, we found tie sulfadon reaction to be liiat wh tie of incubation (Fig. 2  A) and pmtein cottwum.tion (Fig. 2 B). The standard deviation of dtree independent e.ssws was found to h cl0 % (not show). The recovery of 13-%lCCK-1 upon re.elec&photesis wai 74% indicadng that dte low pH electmphomsis condition employed malted in little hydrolysis of sulfate ester and dmt most of [3%lCCK-I was contained in tie atw of dte pe,Fer excised for qmmitaticn. As IittIe as 0.5 fmol of stdfated peptide were detectable when 135SlPAPS of higb specific activity (37 TBqhnmol) was used. Tote signal-to-noise-ratios of dx TPST asays tanged fmm 2 to SO. The sppmwtt Km vaIues for PAPS and Tub1 were 2 fl attd 140 fl, reqectively (Fig. 3).