Conversion of Recombinant Hirudin to the Natural Form by in Vitro Tyrosine Sulfation DIFFERENTIAL SUBSTRATE SPECIFICITIES OF LEECH AND BOVINE TYROSYLPROTEIN SULFOTRANSFERASES*

Hirudin, a tyrosine-sulfated protein secreted by the leech Hirudo medicinalis, is one of the most potent anticoagulants known. The hirudin cDNA has previously been cloned and has been expressed in yeast, but the resulting recombinant protein was found to be produced in the unsulfated form, which is known to have an at least 10 times lower affinity for thrombin than the naturally occurring tyrosine-sulfated hirudin. Here we describe the in vitro tyrosine sulfation of recombinant hirudin by leech and bovine tyrosylprotein sulfotransferase (TPST). With both enzymes, in vitro sulfation of recombinant hirudin occurred at the physiological site (Tyr-63) and rendered the protein biochemically and biologically indistinguishable from natural hirudin. However, leech TPST had an over 20-fold lower apparent Km value for recombinant hirudin than bovine TPST. Further differences in the catalytic properties of leech and bovine TPSTs were observed when synthetic peptides were tested as substrates. Moreover, a synthetic peptide corresponding to the 9 carboxyl-terminal residues of hirudin (which include Tyr-63) was sulfated by leech TPST with a similar apparent Km value as full length hirudin, indicating that structural determinants residing in the immediate vicinity of Tyr-63 are sufficient for sulfation to occur.

Hirudin, a tyrosine-sulfated protein secreted by the leech Hirudo medicinalis, is one of the most potent anticoagulants known.
The hirudin cDNA has previously been cloned and has been expressed in yeast, but the resulting recombinant protein was found to be produced in the unsulfated form, which is known to have an at least 10 times lower affinity for thrombin than the naturally occurring tyrosine-sulfated hirudin. Recombinant proteins' are increasingly being used in biology and medicine. Most of these proteins are secretory, and many of them are post-translationally modified. One posttranslational modification found in many secretory proteins is tyrosine sulfation  which occurs in the lumen of the trans Golgi and is catalyzed by an integral membrane protein, tyrosylprotein sulfotransferase (TPST)' Baeuerle and Huttner, 1987; expressed in bacteria and yeast, these recombinant proteins are not tyrosine-sulfated (Riehl-Bellon et al., 1989). This is consistent with the observations that protein tyrosine sulfation, though widespread in metazoan cells, does not occur in prokaryotes and certain lower eukaryotes Hohmann et al., 1985), presumably because of the lack of TPST.3 In several cases, tyrosine sulfation has been shown to be of major physiological importance, affecting the biological activity or half-life of specific proteins (Anastasi et al., 1966;Bodanszky et al., 1978;Nachman et al., 1986;Pauwels et al., 1987;Suiko and Liu, 1988;Hortin et al., 1989). A striking example is the anticoagulant hirudin; the desulfated form of hirudin binds to thrombin with a lo- (Stone and Hofsteenge, 1986) to 15-fold (Seemtiller et al., 1986;Dodt et al., 1987) lower affinity than the natural, tyrosine-sulfated form produced by the leech Hirudo medicinalis.
In addition, a tyrosinesulfated carboxyl-terminal dodecapeptide of hirudin was found to have a lo-fold higher anticoagulant activity than the unsulfated peptide (Maraganore et al., 1989). These differences between tyrosine-sulfated and unsulfated hirudin observed in vitro may well translate into a large increase in antithrombotic efficacy when used in vivo, in analogy to previous results with hirudin containing a single amino acid change (Degryse et al., 1989).
A hirudin cDNA has recently been cloned (Harvey et al., 1986). The recombinant protein has been expressed in yeast and is available in highly purified form (Loison et al., 1988;Riehl-Bellon et al., 1989). Since hirudin, as one of the most potent anticoagulants known, has great potential in medical therapy (Markwardt, 1970;Markwardt et al., 1984Markwardt et al., , 1988, it would be desirable to convert the recombinant, unsulfated hirudin, which can be obtained much more easily than hirudin isolated from leeches, to the natural, tyrosine-sulfated form. Here we report on this conversion by using homologous (leech) as well as heterologous (bovine) TPST in vitro. Moreover, we show that leech and bovine TPSTs have distinct catalytic properties.
resis at pH 1.9  and pH 3.5 was used to separate

TPST Preparations
All steps were performed at 4 "C. Leech TPST Preparation-Salivary glands of leeches (H. medicinalis, obtained from Ricarimpex, Audenge, France, or from a local pharmacy) were dissected and homogenized in 4 volumes of 0.3 M sucrose.
The homogenate was centrifuged at 800 X g for 10 min and the resulting supernatant at 12,000 x g for 40 min. The membrane pellet was resuspended in 4 volumes of 0.3 M sucrose, layered on top of a 1.3 M sucrose cushion and centrifuged at 36,000 X g for 30 min. The membranes at the interface were collected and subjected to carbonate treatment which has been found to increase the specific activity of TPST ) and resulted in TPST assays being linear for several hours (data not shown). For this, 700 ~1 of interface membranes were incubated for 15 min in 10 ml of 0.1 M NaC03/NaHC03, pH 11.0, 1 M KCl, 0.025% (w/v) saponin, 2 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, and 1 mM 5-aminocaproic acid and centrifuged for 30 min at 130,000 x g. The membranes were resuspended in 1 ml of 10 mM MES-NaOH, pH 6.5, and 2 mM EDTA, centrifuged for 30 min at 130,000 X g, resuspended in 0.5 ml of 10 mM MES-NaOH, pH 6.5, 2 mM EDTA, and 0.3 M sucrose, and stored at -25 "C.  ) and used as the source of bovine TPST.

AND DISCUSSION
When purified rHV2 was incubated with the sulfate donor [35S]3'-phosphoadenosine 5'-phosphosulfate (PAPS) and a membrane preparation from leech salivary glands enriched in TPST, i.e. the enzyme that physiologically sulfates hirudin, a sulfated product was formed which on HPLC eluted slightly before the peak of unsulfated rHV2 (Fig. lA). No such product was found when a soluble fraction of leech salivary glands was used as potential source of TPST or when rHV2 was omitted from the sulfation reaction (not shown). The product of the in vitro sulfation corn&rated with natural, tyrosinesulfated hirudin purified from leeches (Fig. lA), showing that sulfation alone is sufficient to convert the recombinant protein to a form indistinguishable from natural hirudin. As an alternative to leech TPST, we have tested a heterologous TPST preparation, TPST from bovine adrenal medulla, which has been previously characterized  and is available in larger amounts than the leech enzyme. The rationale behind using this latter enzyme was the previous observation that protein substrate recognition by TPST has been sufficiently conserved during evolution to allow stoichiometric sulfation of an insect protein by mammalian TPST (Friederich et al., 1988). Incubation of purified rHV2 with [35S]PAPS and a membrane preparation enriched in bovine TPST resulted in the formation of sulfated rHV2 which, like the product of the reaction using leech TPST, comigrated with natural hirudin on HPLC (Fig. 1B).
To investigate whether the [35S]S04 in rHV2 was linked to tyrosine, we subjected the in vitro sulfated rHV2 to alkaline hydrolysis, a condition in which tyrosine sulfate is released from proteins (Huttner, 1984). Thin-layer electrophoresis of the hydrolysate showed that, indeed, the radioactivity was recovered as tyrosine sulfate (Fig. 2). Serine sulfate and threonine sulfate were not detected.  Hirudin contains 2 tyrosine residues, one at position 3 and one at position 63, 3 residues from the carboxyl terminus (Bagdy et al., 1976). Only tyrosine 63 is sulfated in the leech in uiuo (Bagdy et al., 1976). To investigate whether the in vitro sulfation of rHV2 by the leech and the bovine TPST preparation occurred specifically at tyrosine 63, purified [%I rHV2 was digested with carboxypeptidase Y at pH 5.5, a condition known to selectively release the carboxyl-terminal amino acids (Chang, 1983). Thin-layer electrophoresis of the digests showed that free tyrosine [35S]sulfate was released from [%S]rHV2 sulfated by either TPST preparation (Fig. 3). Serine sulfate and threonine sulfate were not detected (data not shown). To demonstrate that the release of tyrosine sulfate was caused by carboxypeptidase Y itself rather than by a contaminating endoprotease, a control digestion was performed at pH 7.4, a pH at which carboxypeptidase Y is inactive (Hayashi, 1977). No significant quantities of tyrosine [?S]sulfate were released from [%]rHV2 under these conditions (Fig. 3). Thus, in vitro sulfation of recombinant hirudin by either leech or bovine TPST occurred at tyrosine 63.
It has previously been shown that the presence of a sulfate group on tyrosine 63 increases the affinity of hirudin toward thrombin (Stone and Hofsteenge, 1986;Seemuller et al., 1986;Dodt et al., 1987). Since in vitro sulfation of rHV2 occurred specifically at this site, we did not attempt to confirm these observations using the minute amounts of tyrosine-sulfated recombinant hirudin produced under the present in vitro conditions. It was, however, important to ascertain that the conditions of in uitro incubation did not unspecifically impair the biological activity of hirudin. The biological activity of hirudin can be demonstrated qualitatively by its binding to immobilized thrombin (Walsmann, 1981). To determine whether recombinant hirudin retained this property after in vitro incubation, purified [35S]rHV2 was subjected to affinity chromatography on a thrombin-Sepharose column ( Table I). All of the applied radioactive rHV2 bound to the column, and 70% could be specifically eluted with the thrombin inhibitor 4-aminobenzamidine.
Thus, the in vitro sulfated recombinant hirudin was biologically active. The results described so far show that a recombinant secretory protein can be converted to the physiological form by performing the appropriate post-translational modification, tyrosine sulfation, in vitro. Although this modification occurs late in the secretory pathway (truns Golgi), this was not necessarily to be expected since a protein purified from the extracellular medium may not apriori have the same structure as in the truns Golgi. Differences in structure between the trans Golgi and the secreted form of a protein result, for example, from post-translational modifications occurring later in the secretory pathway than tyrosine sulfation, such as proteolytic processing (for review, see Steiner et al., 1984) and oligomerization uia disultide bonds (e.g. von Willebrand factor; for review, see Verweij, 1988). Thus, it could not be excluded that the tram Golgi form of a protein is specifically competent to undergo tyrosine sulfation. However, the present results show that TPST can not only sulfate proteins endogenously present in subcellular Golgi-containing fractions, as shown previously (for review see Huttner and Baeuerle, 1988), but also full length proteins purified after secretion, implying that complete passage through the eukar- [%]rHV2 sulfated by the leech TPST preparation and purified through HPLC (gradient I, Fig. 1) was subjected to tyrosine sulfate analysis. An autoradiogram of the cellulose thin-layer sheet is shown. The dashed line indicates the position of the tyrosine sulfate (7'yrG)) standard detected by ninhydrin staining. [""S]rHV2 obtained from in vitro sulfation reactions with either the leech or the bovine TPST preparation was purified by paper electrophoresis, and aliquots containing 2205 cpm (leech TPST) or 3283 cpm (bovine TPST) were digested with carboxypeptidase Y at pH 5.5 or 7.4. Released tyrosine [%]sulfate was separated from [""S]rHVZ by electrophoresis and is expressed as percent of total (sum of ""S radioactivity present in the tyrosine [%]sulfate plus [""S]rHV2 spots). yotic secretory pathway is not incompatible with subsequent tyrosine sulfation.
The nine carboxyl-terminal amino acid residues of hirudin, which include the tyrosine sulfation site, are of particular relevance for the inhibitory action of hirudin on thrombin, probably by interacting with a noncatalytic domain of thrombin that binds to fibrinogen (Fenton et al., 1988;Noe et al., 1988). Deletion of these residues increases the apparent Ki value of hirudin lO,OOO-fold (Degryse et al., 1989). Conversely, the lo-12 carboxyl-terminal amino acid residues (in relatively high amounts) are alone sufficient for inhibition of thrombinmediated clotting (Krstenansky et al., 1987;Mao et al., 1988;Maraganore et al., 1989). It was of interest to investigate whether the structural requirements for the enzymatic tyrosine sulfation of hirudin were also contained in this part of [%]rHV2 sulfated by the leech TPST preparation and purified through HPLC (gradient I, Fig. 1) was chromatographed on a thrombin-Sepharose column. The flow-through was collected, the column was washed, and [""S]rHV2 was eluted with 4-aminobenzamidine. The radioactivity recovered in the various fractions is given after subtraction of background. For this comparison, we determined the kinetic parameters of leech as well as bovine TPST for these two substrates (Table II). Sulfation of Hir-(57-65) occurred on tyrosine (data not shown). The apparent K,,, values of leech and bovine TPST for Hir-(57-65) were in the same range. In contrast, only leech TPST had an apparent K,,, for full length hirudin that was similar to that for Hir-(57-65), whereas bovine TPST had a 23-fold higher apparent K,,, for full length hirudin than for Hir-(57-65) (Table II). These data suggest that the structural information required for the recognition of hirudin by TPST is contained within the nine carboxylterminal residues of hirudin, and that the tertiary structure of full length hirudin does not promote this recognition. Rather, the tertiary structure of full length hirudin can impose steric hindrance on this recognition, unless the TPST is evolutionarily adapted for this substrate, as appears to be the case for leech, but not bovine, TPST. The different apparent K,,, values of leech and bovine TPST for full length hirudin ver.suS Hir-(57-65) suggested that leech and bovine TPST have differential substrate specificities. Differences in catalytic properties between leech and bovine TPST were also observed with respect to V,,,,,, which was in the same range for full length hirudin and Hir-(57-65) in the case of leech TPST but differed 20-fold in the case of bovine TPST (Table II). Moreover, further differences between leech and bovine TPST became apparent when we assayed both enzymes with a second synthetic peptide, CCK-(107-115), corresponding to the carboxyl-terminal sulfation sites of preprocholecystokinin (Adrian et al., 1986;Eng et al., 1986). Leech TPST exhibited a 24-fold lower apparent K,,, value for CCK-(107-115) than for Hir-(57-65), with little change in V max, whereas bovine TPST showed a 16-fold higher V,,,,, for CCK-(107-115) than for Hir-(57-65), with little change in apparent Km (Table II). Thus, leech and bovine TPST are distinct in their catalytic properties toward two synthetic peptides, although both peptides conform to the previously suggested consensus features for tyrosine sulfation (Huttner