Heparinase I from Flavobacterium heparinum. Mapping and characterization of the heparin binding domain.

In this study we have identified the primary heparin binding site of heparinase I (EC 4.2.2.7). Chemical and proteolytic digests of heparinase I were used in direct binding and competition assays, to map the regions of heparinase I that interact specifically with heparin. We find the heparin binding site contains two Cardin-Weintraub heparin binding consensus sequences and a calcium co-ordination consensus motif. We show that heparin binding to heparinase I is independent of calcium (K of 60 nM) and that calcium is able to activate heparinase I catalytically. We find that sulfhydryl selective labeling of cysteine 135 of heparinase I protects the lysines of the heparin binding sequence from proteolytic cleavage, suggesting the close proximity of the heparin binding site to the active site. Site-directed mutagenesis of H203A (contained in the heparin binding site) inactivated heparinase I; however, a H203D mutant retained marginal activity, indicating a role for this residue in catalysis. The above results taken together suggest that histidine 203 (hence the heparin binding site) is immediately adjacent to the scissile bond. We propose that the heparin binding site and active site are in close proximity to each other and that the calcium coordination motif, contained in the heparin binding site, may bridge heparin to heparinase I through calcium in a ternary complex during catalysis.

In this study we have identified the primary heparin binding site of heparinase I (EC 4.2.2.7). Chemical and proteolytic digests of heparinase I were used in direct binding and competition assays, to map the regions of heparinase I that interact specifically with heparin. We find the heparin binding site contains two Cardin-Weintraub heparin binding consensus sequences and a calcium co-ordination consensus motif. We show that heparin binding to heparinase I is independent of calcium (K d of 60 nM) and that calcium is able to activate heparinase I catalytically. We find that sulfhydryl selective labeling of cysteine 135 of heparinase I protects the lysines of the heparin binding sequence from proteolytic cleavage, suggesting the close proximity of the heparin binding site to the active site. Site-directed mutagenesis of H203A (contained in the heparin binding site) inactivated heparinase I; however, a H203D mutant retained marginal activity, indicating a role for this residue in catalysis. The above results taken together suggest that histidine 203 (hence the heparin binding site) is immediately adjacent to the scissile bond. We propose that the heparin binding site and active site are in close proximity to each other and that the calcium coordination motif, contained in the heparin binding site, may bridge heparin to heparinase I through calcium in a ternary complex during catalysis.
There have been numerous recent reports on the high degree of specificity in the binding interactions of heparin and heparan sulfate with cytokines, growth factors and other molecules of the extracellular matrix (Jackson et al., 1991;Varki, 1993;Ernst et al., 1995). In several heparin-binding proteins, the binding specificity is dependent on certain sequences of the polysaccharide as well as characteristic amino acid sequences of the protein (Jackson et al., 1991;Lindahl et al., 1994). On the protein side, Cardin and Weintraub identified two consensus sequences (XBBBXXBX or XBBXBX (B ϭ basic residues; X ϭ hydrophobic or other residues)) found in many heparin-binding proteins (Cardin and Weintraub, 1989). Subsequently, sitedirected mutagenesis and binding studies with synthetic or isolated peptides from several of these proteins have confirmed that this consensus region often is involved in binding specifically to heparin (Bae et al., 1994;Baird et al., 1988;Bober Barkalow and Schwarzbauer, 1991;Jackson et al., 1991;Smith and Knauer, 1987;Thompson et al., 1994). On the polysaccharide side, oligosaccharides with certain sequences of modifications in the disaccharide repeat have been isolated and shown to confer binding specificity for antithrombin III, bFGF, 1 and others (Lindahl et al., 1984;Maccarana et al., 1993;Maimone and Tollefsen, 1990;Parthasarathy et al., 1994). This is parallel to the substrate specificities of heparinases and suggests that the mechanism of substrate binding to heparinases may be involved in generating specificity.
Earlier studies suggested the importance of basic residues in heparinase I activity (Comfort et al., 1989;Leckband and Langer, 1991). Recently, we found that cysteine 135 of heparinase I is catalytically active . In addition, the observation of a decrease in the rate of heparinase I inactivation by chemical modification of cysteine 135 in the presence of heparin, along with other experiments, led to the hypothesis that a heparin binding domain is in close proximity to cysteine 135 . In this study, we map and characterize the heparin binding domain of heparinase I and test the hypothesis of the binding domain being in proximity to the active site.

Chemicals and Materials
Heparin, from porcine intestinal mucosa with an average molecular mass of 12 kDa and activity of 157 USP units/mg was from Hepar (Franklin, OH). Heparin was radioiodinated to a specific activity of 5.86 ϫ 10 8 cpm/nmol (6 kDa, average molecular mass), and was purified as described (Lee and Lander, 1991). Low melting point agarose (Sea-Plaque) and GelBond were from FMC (Rockland, ME). Urea, dithiothreitol, iodoacetamide, trifluoroacetic acid, PCMB, and acetonitrile were from Allied Chemicals (Deerfield, IL). Chondroitin sulfate was from Sigma. Other chemicals were from Mallinckrodt. Molecular weight standards were obtained from Life Technologies, Inc. Escherichia coli BL21(DE3) host was from Novagen (Madison, WI). Molecular biology reagents and their sources are listed in the appropriate sections below.

Heparinase I, Protein Analyses, and Heparin-binding Peptides (HBP-I and Maxadilan)
Heparinase I was purified as described (Yang et al., 1985;Sasisekharan et al., 1993). The heparinase I used for activity measurements was extensively desalted using a Centricon P-30 (Amicon, Beverly, MA). Protein concentration was determined using Micro-BCA reagent (Pierce) relative to a bovine serum albumin standard. The heparinbinding peptides HBP-I and Maxadilan were synthesized using t-butoxycarbonyl chemistry (Stewart and Young, 1984) on an Applied Biosystems model 430A solid phase peptide synthesizer (Biopolymers Laboratory, MIT, Cambridge, MA). The peptides were purified by RPHPLC on a Vydac C 18 reverse-phase column using a gradient of 0 -80% acetonitrile for 40 min. HBP-I is an 18-amino acid peptide representing residues 196 -213, 2 with the sequence IFKKNIAHD-KVEKKDKDG. Maxadilan was used as the control peptide. This peptide contains a basic cluster with a sequence VFKESMKQKKKKEF-SSEK (Lerner and Shoemaker, 1992).

Heparinase I Activity Assays
UV 232-nm Assay-For studying the effect of Ca 2ϩ , 0.5 g of heparinase I was added to 25 mg/ml heparin in 100 mM (MOPS) buffer at pH 7.0 with varying amounts of calcium acetate, in a total volume of 900 l. The reaction mixture was incubated at 37°C. At various time intervals (up to 6 min), aliquots of 50 -70 l were withdrawn in duplicate and quenched in 1.5 ml of 0.03 N HCl, and the absorbance at 232 nm was measured essentially as described by Bernstein et al. (1988). For measuring r-heparinase I activity during purification, the assay was done essentially as described by Sasisekharan et al. (1993).
HPLC of Heparin Oligosaccharides-Heparin (0.2 or 2 mg/ml) was incubated with heparinase I, r-heparinase I, or mutant enzymes in 5 mM calcium acetate, 100 mM MOPS buffer, pH 7.0, for 18 h. The reaction was then subjected to anion-exchange HPLC to separate the oligosaccharide products, as described by Sasisekharan et al. (1993).

Heparin Affinity Chromatography
Heparin affinity chromatography was carried out using a heparin POROS (4.6 mm ϫ 100 mm) column (PerSeptive BioSystems, Cambridge, MA). A nanomole (ϳ40 g) of heparinase I in 100 mM MOPS buffer, pH 7.0 (with and without 5 mM calcium acetate), was loaded onto the heparin POROS column, connected to a BIOCAD workstation (Per-Septive BioSystems, Cambridge, MA). The protein was eluted in a salt gradient of 0 -1 M NaCl (in 5 min.), in 10 mM Tris-HCl, pH 7.0, and monitored at 210 and 277 nm.

Affinity Co-electrophoresis
Affinity co-electrophoresis (ACE) on heparin binding to heparinase I was performed with 0 (buffer 1) or 5 mM (buffer 2) calcium acetate essentially as described (Lee and Lander, 1991). The heparinase I concentrations were from 0.05 to 25 g/ml, either untreated, or pretreated with IAA as reported previously . The binding constant was determined using Scatchard analysis, using weighted, nonlinear least-squares curve fitting (Lee and Lander, 1991).

Cyanogen Bromide Digest and Heparin Hybridization
One nmol (ϳ40 g) of heparinase I was digested with CNBr as described by Matsudaira (1993). The digest was diluted 10-fold with water and then lyophilized under vacuum. SDS-PAGE of the CNBrdigested heparinase I fragments was carried out on a 15% polyacrylamide gel using a Mini PROTEAN II electrophoresis apparatus (120 V for 90 min) (Bio-Rad). Proteins were visualized with a 0.1% Coomassie Brilliant Blue R-250 solution, followed by destaining with a 40% methanol and 10% acetic acid (v/v) solution.
Following gel electrophoresis CNBr digests of heparinase I were blotted onto nitrocellulose sheets (Schleicher & Schü ll) according to the method of Matsudaira (1993). Transfer was carried out in the following amperage sequence: 30 min at 50 mA, 30 min at 100 mA, and 60 min at 275 mA. The nitrocellulose membrane was stained for 1 min with 0.1% Ponceau S solution (Fluka, Buchs, Switzerland) in 0.1% acetic acid and then washed with 1% acetic acid for 1 min (Tempst et al., 1990).
For heparin hybridization, the membrane was further processed by washing it with 10 mM Tris⅐HCl, 10 mM NaCl and 0.5 mM EDTA, pH 7.0 (TNE-10) twice, and once with the probing buffer. The heparin binding domain was probed using 125 I-labeled heparin at a concentration 10 6 cpm/ml in the probing buffer. The membrane was hybridized at 37°C for 18 h followed by washing of the membrane to remove excess radiolabel. The membrane was cut into bands corresponding to the stained peptide fragments, and each band was counted for 125 I incorporation on a scintillation counter (model LS 3801, Beckman, Fullerton, CA). The bands that were positive for 125 I incorporation were washed with phosphate-buffered saline, pH 7.0, and sequenced as described below. Alternatively, the stained membrane was cut into bands before hybridization and the bands were then washed in a microcentrifuge tube, prehybridized, and probed individually. CNBr peptides were extracted from the nitrocellulose membrane using essentially the procedure described by Matsudaira (1993) for sequencing (see below).

Tryptic Digest, Heparin Competition, Dot Blot Assay, and Peptide Sequencing
The tryptic digestion was carried out essentially as described by Sasisekharan et al., 1993. In some experiments, heparin was added to heparinase I before trypsin digestion. The reaction was terminated by heating the sample at 65°C for 2 min. In the PCMB protection experiment, heparinase I was first labeled with PCMB as described , under nonreduced conditions, and then digested with trypsin as described above.
A total of 2.5 nmol of heparinase I was digested and then divided into five batches (500 pmol each; ϳ20 g) inclusive of one control digest. For the competition experiments, heparin was added in varying concentrations (5, 50, 100, and 200 g) to the digests, which were subsequently separated using RPHPLC, with a 20-min isocratic run, followed by 0 -80% acetonitrile (in 0.1% trifluoroacetic acid) for 120 min, and monitored at 210 and 277 nm. Chondroitin sulfate was used as control for nonspecific competition. In dot blot assays, 2.5 nmol of heparinase I digested with trypsin were separated by RPHPLC as described above. The peptide fragments were collected in microcentrifuge tubes and blotted onto nitrocellulose paper. The blots were washed with TNE-10 twice and then probed with 125 I-labeled heparin for 2 h, at a concentration of 10 ng/ml, 1 mg/ml chondroitin sulfate, and 0.01% Tween 20 in 10 mM MOPS containing 5 mM calcium acetate, pH 7.0 (probing buffer). The blots were washed twice with 10 mM MOPS containing 5 mM calcium acetate, pH 7.0, and then counted for 125 I-heparin binding. Binding data was analyzed using weighted, nonlinear least-squares curve fitting (McPherson, 1985). Small amounts of heparin (0.05% of total counts) bound to nitrocellulose filters dotted either with buffer or with 1 mM bovine serum albumin.
Peptides were sequenced using an Applied Biosystems model 477 sequencer, with an on-line model 120 PTH amino acid analyzer (Biopolymers Laboratory, MIT).

Mutagenesis, Expression, and Purification of r-Heparinase I
The recombinant and mutant heparinases I were expressed without the putative F. heparinum leader sequence, i.e. as a construct ((ϪL)rheparinase I) that reads Met-Glu 22 -Glu 23 -. . . (Sasisekharan et al., 1993). To facilitate purification, the heparinase I gene was expressed using the pET15b system with a histidine tag and a thrombin cleavage site in a 21-amino acid N-terminal leader sequence .
Expression and Purification-The constructs were transformed in BL21(DE3) (Novagen), and the protein purified as described previously . SDS-PAGE (Laemmli, 1970) was carried out using precast 12% gels and a Mini PROTEAN II apparatus and stained with the Silver Stain Plus kit .

Heparin Blotting of CNBr Digests of Heparinase I-CNBr
digested heparinase I separated by SDS-PAGE, transferred onto nitrocellulose, resulted in 10 peptide fragments (Fig. 1). Heparinase I contains five internal methionine residues (CNBr sites), two of which are adjacent to each other, so for complete digestion, only five fragments should be expected. However, only four of the 10 peptides could be sequenced and two others were contaminating peptides ( Table I). The remaining four fragments, based on molecular weights and sequencing, probably represented partial digests from the N terminus, which was shown previously to be blocked (Sasisekharan et al., 1993). The CNBr-generated heparinase I fragments on nitrocellulose were hybridized with 125 I-labeled heparin, and counted for 125 I incorporation. The binding of 125 I-heparin to CNBr-8 was significantly higher compared to the other bands and to controls (Table I). Similar results were obtained by an alternative method, where the peptide bands on nitrocellulose were cut out and hybridized individually. CNBr-8 spans amino acids 195 to approximately 290 of the heparinase I primary sequence. It has a lysine-rich N-terminal region, containing two potential Cardin-Weintraub heparin binding consensus sequences (Cardin and Weintraub, 1989) and a calcium binding loop of the "EFhand" structural domain (Kretsinger, 1975(Kretsinger, , 1980. It is interesting to note that CNBr-9 also contains a putative Cardin-Weintraub heparin binding consensus sequence; however, this fragment does not seem to bind heparin significantly. Since CNBr-7 (residues 272-360) did not bind heparin and inasmuch as residues 272-290 are common to CNBr-7 and CNBr-8, this region is excluded from being a part of the heparin binding domain. Therefore, the region 195-270 contains the primary heparin binding site of heparinase I.
It is important to point out that an unavoidable limitation of the technique described here is that when ligand-protein binding is probed on nitrocellulose (with the protein being denatured), only the primary (non-folded) structures that bind to the ligand can be detected. In order to further investigate the heparin binding region of heparinase I, we carried out proteolytic digests of heparinase I as described below.

Competition Experiments and Dot Blot with Tryptic Digests
of Heparinase I-Tryptic digestion was used to map the heparin binding site of bFGF, by showing that heparin protects this site from trypsin cleavage (Coltrini et al., 1993). We tested the ability of heparin in protecting the heparin binding domain of heparinase I from trypsin cleavage. Under the conditions tested, heparin was ineffective in protecting the heparin binding domain; however, it was able to compete specifically with the binding of some heparinase I tryptic peptides to the reverse-phase column. Peaks that shifted significantly in their elution time, or disappeared (presumably eluting in the void volume), represent tryptic peptides that bind to heparin. Chondroitin sulfate was used as a control to account for nonspecific ionic effects. Compared to a control tryptic map ( Fig. 2A), in the presence of increasing concentrations of heparin, the following peaks were altered reproducibly: peak at 34 min (td4), peak at 43 min (td9), peaks eluting between 52 and 58 min, peak at 71 min (td45), peak at 77 min (td39), and peak at 84 min (td50) (Fig. 2B). As the peptides td9, those eluting in the region between 52 and 58 min, and td50 were all altered by both heparin and chondroitin sulfate (data not shown), it can be inferred that only td4, td39, and td45 bind heparin specifically. In a dot blot assay using 125 I-heparin and in the presence of a 100-fold excess of cold chondroitin sulfate, only td45 showed 125 I signal (data not shown). It should be pointed out that td45 is a partial digest of td39, with the common region being residues 215-221. From the competition and dot blot experiments, it can be concluded that td45 (residues 215-221) and td4 (residues 132-141) are the only peptides that bind specifically to heparin. While this is consistent with td45 being a part of CNBr-8 peptide, it is interesting to note that td4 contains the active site cysteine 135 (see "Discussion") .
The combined heparin binding experiments, with CNBr and tryptic digests of heparinase I, point to the residues 195-221 as being primarily involved in heparin binding. As mentioned earlier, this region contains two potential Cardin-Weintraub heparin binding consensus sequences (residues 197-204 and 207-212) and a calcium binding loop of the EF-hand structural domain (residues 206 -220). Importantly, as this region contains multiple lysines it is likely to be degraded to very short peptides (di-and tripeptides) by trypsin. Thus, it was not expected to show up on the tryptic digest chromatogram, and its potential heparin binding properties might therefore have  Table I for details. gone undetected in these experiments. To investigate the heparin-binding properties of these residues, a peptide corresponding to this region was synthesized and characterized as described below. Synthetic Peptide Corresponding to the Heparin Binding Site of Heparinase I-A synthetic peptide (HBP-I) corresponding to the residues 196 -213 has a ϳ4 M binding affinity for heparin dodecasaccharides. Interestingly, HBP-I affected the product profile of heparinase I degradation of heparin. As mentioned earlier, heparinase I depolymerization of heparin results in two disaccharides, three tetrasaccharides (1-3), and a hexasaccharide (Fig. 3A) (Rice and Linhardt, 1989). In a concentration-dependent manner, the addition of HBP-I to the reaction mixture caused the peak corresponding to tetrasaccharide 3 (⌬U 2S H NS,6S I 2S H NS,6S ) to disappear (Fig. 3, B and C). When tetrasaccharide 3 was isolated and degraded with heparinase I in the presence of HBP-I, a marked increase in the amount of disaccharide was observed (data not shown). A control peptide (Maxadilan) with similar charge properties (and at the concentration ranges tested above) had no effect on the enzyme activity or on the oligosaccharide product profile (Fig. 3D). This demonstrates that HBP-I affects the selectivity of heparin degradation by heparinase I; tetrasaccharide 3, but not tetrasaccharides 1 or 2, is degraded in the presence of HBP-I. It must be pointed out that HBP-I does not degrade heparin or heparin oligosaccharides.
Role of Calcium-Region 206 -220 was found to be homologous to the calcium binding loop of the EF-hand structural domain (Table II) (Kretsinger, 1975(Kretsinger, , 1980. Four of the five amino acids that are involved in coordinating calcium in the loop of the EF-hand structural domain are conserved in heparinase I. This led us to investigate the role of calcium in hepa-rinase I activity. There are conflicting reports on the effect of calcium on heparinase I activity (Linker and Hovingh, 1965;Dietrich et al., 1973;Linhardt et al., 1986;Lohse and Linhardt, 1992). The relative heparinase I activity increases with calcium concentrations up to around 5 mM as shown in Fig. 4. In order to further investigate the calcium-based heparinase I activation, the role of calcium in heparin binding to heparinase I was studied using heparin affinity chromatography. These experiments indicate that heparin binding to heparinase I is independent of calcium (heparinase I elutes at a salt concentration of 200 mM) and the presence of calcium leads to a loss in binding to heparin column (as calcium activates heparinase I catalytically) (data not shown). The above results suggest a possible co-factor role for calcium in heparinase I activity.
Heparin Binding Constant of Heparinase I: Affinity Co-electrophoresis-The ability of heparinase I to bind heparin, in the absence of calcium, led us to investigate the heparin binding affinity of heparinase I. ACE was used to determine a heparin binding constant for heparinase I (Lee and Lander, 1991). ACE was carried out in the presence or absence of IAA to determine the importance of the active site cysteine 135 in the binding of heparin to heparinase I , and in the absence of calcium to prevent heparin degradation. The K d for heparinase-heparin binding was found to be 60 nM by this technique. Fig. 5A shows the ACE gel of native heparinase I (nondenatured), and Fig. 5B shows the ACE gel of heparinase I modified with IAA. There is no difference in the retardation of heparin for the IAA-modified heparinase I when compared to the unmodified heparinase I. This result indicates that blocking the active site cysteine does not alter heparin binding. An ACE gel of heparin-heparinase I carried out in the presence of calcium (data not shown) showed extensive smearing of the FIG. 2. Tryptic digest of heparinase I and heparin competition. Heparinase I was digested with trypsin essentially as described by Sasisekharan et al. (1993). Following trypsin digestion, the tryptic peptides were separated by RPHPLC with a 10-min isocratic run, followed by 0 to 80% acetonitrile (in 0.1% trifluoroacetic acid) for 120 min. Tryptic peptides were monitored at 210 and 277 nm (in mAU). A, control digest; B, digest with heparin. Note peak at 34 min (td4), peak at 43 min (td9), peaks eluting between 52 and 58 min (not numbered, shown by arrows) at 77 min (td39), peak at 71 min (td45), and peak at 84 min (td50) are all marked in chromatogram A. Arrows in B show the changes in the peak profiles when compared to the control. Refer to Table I in Sasisekharan et al. (1993) for peptide sequences (tdLx in the table corresponds to td50). heparin band.
Role of Heparin Binding in Catalysis-Heparinase I derivatization by sulfhydryl-specific reagent PCMB inactivated the enzyme due to selective modification of cysteine 135, and the inactivation kinetics were altered in the presence of heparin, suggesting a heparin binding site in close proximity to cysteine 135 . To test this hypothesis, tryptic digestion of PCMB-modified heparinase I was carried out to determine if PCMB (being a bulky negatively charged molecule) could impede trypsin access to the basic residues contained in the heparin binding domain. Such a protection of the basic residues from proteolytic digestion could argue for the close proximity of these residues to cysteine 135. The PCMBheparinase I tryptic map (data not shown) has the appearance of a new peak (about 31 min) with a sequence representing residues 199 -209. The results suggest that PCMB-labeled cysteine 135 protects the lysine-rich peptide (contained in the heparin binding sequence) from trypsin cleavage, when compared to a control digest where this peptide is not observed. This result supports the notion that a heparin binding site is in close proximity to the active site cysteine 135 .
To further investigate a functional role of the heparin binding domain, site-directed mutagenesis was carried out. Histi- FIG. 3. Effects of HBP-I on heparinase I activity. Heparin (0.2 mg/ml) was incubated with heparinase I in 5 mM calcium acetate, 100 mM MOPS buffer, pH 7.0, for 18 h as described under "Materials and Methods." The reaction was then subjected to anion-exchange HPLC using a POROS Q/M (4.6 mm ϫ 100 mm) column (PerSeptive BioSystems) with a salt gradient of 0 -1 M NaCl in 5 min and monitored at 232 nm. A shows the product profile of heparin degradation by heparinase I in the absence of HBP-I; B and C show the product profile of heparin degradation in the presence of HBP-I (1:2 and 1:10 heparinase I to HBP-I). Note the decrease of tetrasaccharide 3 (retention time of about 5 min; shown by arrow in B and C). However, when the digest is carried out in the presence of a control peptide, there is no change in the product profile or tetrasaccharide 3 (shown by arrow in D). In chromatogram A, I represents the disaccharide, II represents the three tetrasaccharides (1-3), and III represents the hexasaccharide.  (Kretsinger, 1975(Kretsinger, , 1980. "c" indicates the Ca 2ϩ coordinating amino acids (D, N, S, T, E, or Q). "h" indicates amino acids with hydrophobic side groups (I, L, or V). "n" indicates nonpolar amino acids (I, L, V, M, F, Y, and W). "x" indicates any amino acid.
dine 203, as a basic residue, was targeted for site-directed mutagenesis, because it is known to play a role in both heparin binding as well as an acid/base catalyst in some proteins (Fersht, 1985;Talpas and Lee, 1993;Fan et al., 1994). Furthermore, it has been suggested that histidine may act as a nucleophilic residue in the elimination mechanism of lyases (Greiling et al., 1975). H203A and H203D mutations were introduced, and mutant r-heparinases I were expressed in E. coli. While the H203A mutant had no detectable enzymatic activity, the H203D mutant showed a 30-fold decrease in activity. The kinetic parameters for (ϪL)r-heparinase I and for the mutants are shown in Table III. These results strongly indicate that histidine 203 is an essential amino acid for enzyme activity. In addition, the results point to the fact that the heparin binding region around residue 203 is in close proximity to the scissile bond during catalysis.

DISCUSSION
For enzymatic reactions the interplay between binding and catalysis determines both specificity and rate of reaction. This interplay is especially interesting for eliminases/lyases that degrade complex acidic polysaccharides like pectin and heparin, which are highly heterogeneous polymers. In the case of the pectate lyases, while the three-dimensional structures of the enzymes are available, there is limited information and understanding of the residues and the domains of the enzyme involved in catalysis (Yoder et al., 1993a). Little is known of the role of binding in catalysis for the eliminases in general, and the current study addresses this issue.
Heparin Binding Domain of Heparinase I-Chemical and proteolytic digests of heparinase I in direct binding and in competition assays, were used to identify peptides that interact specifically with heparin. CNBr digests of heparinase I repro-ducibly gave eight fragments, six of which were partial digests from the blocked N terminus. Hybridization of the CNBr-generated fragments to radiolabeled heparin showed that only one fragment (CNBr-8) bound significant amounts of heparin (Table I). CNBr-8 spanned residues 195-290 and is also a partial digest, as it contained a part of CNBr-7 (residues 273-290). Since CNBr-7 did not bind heparin, the binding site is most likely located in the former half of . It is noteworthy that undigested heparinase I and partial Nterminal CNBr-generated fragments (CNBr-1, -2, -3, and -5) that contained the heparin binding sequence bound relatively less heparin than the CNBr-8 fragment. It is possible that greater heparin binding to CNBr-8 was due to better accessibility of the heparin binding site in CNBr-8. Tryptic peptides td4, td39, and td45 were selectively affected by heparin, and not chondroitin sulfate, thus reinforcing the specificity of binding of these peptides to heparin. Peptides td39 and td45 are contained in CNBr-8, while td4 contains the active site cysteine 135 (see below). The results of heparin binding and competition experiments, taken together, point to the residues 195-221 as being directly involved in heparin binding. Furthermore, a synthetic 18-mer peptide (HBP-I; corresponding to residues 196 -213 in the region) not only bound heparin specifically in a dot blot assay, but also altered enzyme activity. It can thus be concluded that residues 195-221 constitute the primary heparin binding site in heparinase I. This site belongs to a very basic region of heparinase I, with two Cardin-Weintraub heparin binding consensus sequences and a calcium coordination motif. However, heparin competition experiments indicate heparin binding to td4. It is possible that in the native heparinase I, protein folding brings the lysine and/or arginine residues contained in td4 close to the heparin binding site, and all these residues together constitute a heparin binding domain in heparinase I. The presence of such a domain is supported by the PCMB labeling results, where PCMB-labeled heparinase I protects the lysine residues in the heparin binding site from tryptic cleavage.
Heparin binding to heparinase I (K d 60 nM) was demonstrated using ACE. This is about 1 order of magnitude higher than the dissociation constants determined by the same technique for binding of heparin to bFGF (2.2 nM) and AT III (11 nM) (Lee and Lander, 1991). This also is consistent with heparinase I eluting from a heparin affinity column at 200 mM NaCl, while bFGF and other heparin binding growth factors typically elute around 1-1.5 M NaCl (Folkman and Klagsbrun, 1987). When ACE was performed with modified heparinase I (where the cysteines were blocked with IAA), there was no alteration on the heparin retardation, indicating that cysteine 135 does not influence heparin binding, but affects only enzyme catalysis.
Role of Calcium-As a region of the heparin binding site of heparinase I (residues 206 -220) is homologous to the calcium binding loop of EF-hand domains, we set out to investigate the role of calcium in heparinase I activity. We find that heparinase I activity increases with calcium concentrations up to around 5 mM. It is possible that purification of heparinase I FIG. 4. Effect of calcium on heparinase I activity. Plot of the relative reaction rate as a function of calcium concentration. Heparin concentration was fixed at 25 mg/ml, and the calcium concentration was varied in the assay. Heparinase I (0.5 g) was desalted and added to the reaction mixture.
FIG. 5. Affinity co-electrophoresis of heparinase I. Autoradiographs of gels in which radiolabeled heparin was electrophoresed through zones containing heparinase I in increasing concentrations (in g) from left to right. A (left) represents the ACE gel of native heparinase I in the absence of calcium. B (right) represents the ACE gel of modified heparinase I (the cysteines blocked by IAA) also in the absence of calcium. with a hydroxyl apatite-binding step in earlier reports (Dietrich et al., 1973;Langer et al., 1982;Linhardt et al., 1986;Yoshida, 1991) obscured calcium-based activation of heparinase I. From equilibrium dialysis, it was found that calcium binds to heparin at a stoichiometry of about 1 calcium ion/tetrasaccharide (Boyd et al., 1980;Hunter et al., 1988), and fiber diffraction studies show that calcium causes a significant change in the helical structure of heparin, heparan sulfate, and other glycosaminoglycans (Nieduszynski, 1985). This suggests that calcium, by binding to specific regions of heparin, alters heparin conformation and hence activates heparinase I. Furthermore, pectate lyase, an enzyme from the same family of acidic polysaccharide eliminases as heparinase I, is also activated by calcium (Yoder et al., 1993b). The recently solved crystal structures of three pectate lyases showed that the substrate binding cleft was embedded with one or more calcium ions (at the calcium binding loop of EF-hand domains), indicating a direct interaction between calcium and the enzyme (Pickersgill et al., 1994;Yoder et al., 1993aYoder et al., , 1993b. Interestingly, the putative calcium binding sequence in heparinase I is contained in the heparin binding site, which appears to be close to the active site cysteine 135 in the correctly folded active protein. Therefore, the effect of calcium on heparinase I and other acidic polysaccharide lyases, is possibly mediated through a conformational change of the substrate upon calcium binding, which is further stabilized by direct interaction between enzyme and calcium at the active site. Role of Heparin Binding in Catalysis-Specific interactions between the heparin binding domain of heparinase I and unique heparin sequences might determine the catalytic specificity of heparinases. The three heparinases from F. heparinum are specific for different linkages in the heparin chain (Desai et al., 1993), and heparinase I is much less active in cleaving tetrasaccharides compared to full-length heparin (Rice and Linhardt, 1989). This argues that heparin binding to heparinase I involves at least a pentasaccharide or longer region of heparin, similar to the heparin binding regions of bFGF and antithrombin III, which have rather large (20 Å or more) binding regions to accommodate close contacts (Rapraeger, 1993;Margalit, 1993;Schreuder et al., 1994). It is interesting to note that in the presence of HBP-I, tetrasaccharide 3 (⌬U 2S H NS,6S ) (Rice and Linhardt, 1989) was cleaved by heparinase I. It is possible that HBP-I "presents" this saccharide to heparinase I and facilitates cleavage. Alternately, tetrasaccharide 3 containing sequence in heparin perhaps is a "hot spot" region to which heparinase I binds during its random endolytic cleavage of heparin (Cohen and Linhardt, 1990).
Histidine residues are involved in several enzymatic reactions via different mechanisms, which include acid/base catalysis, electrophilic catalysis, and substrate binding through electrostatic interactions (Munier et al., 1992). To differentiate between a substrate binding and a catalytic role for histidine 203 in heparinase I, this residue was replaced with alanine (which can perform neither of the above functions) and with aspartic acid (which can play a role in acid/base catalysis but is not known to participate in heparin binding). The H203A mutation abolished enzyme activity, while the H203D mutant retained residual activity, strongly suggesting a catalytic role for this residue. 3 It is considered unlikely that a change in the binding site far away from the scissile bond could produce such a dramatic change in enzyme activity. It is possible that histidine acts as a nucleophilic residue in the elimination mecha-nism, similar to that suggested for hyaluronate lyase (Greiling et al., 1975). Alternately, the scissile bond cleavage might occur through a proton relay mechanism involving histidine 203 and another nucleophile in close proximity to this residue during elimination reaction (Gacesa, 1987). The site-directed mutation experiments taken together with the biochemical studies provide compelling evidence for histidine 203 being in close proximity to the scissile HI linkage during catalysis.
In a previous report, an active site residue (cysteine 135) was identified and characterized by chemical modification and sitedirected mutagenesis . It was found that this active site has a net positive environment and may be close to the heparin binding region. From the results described in this work and from the earlier study , a molecular mechanism for the interplay between binding and catalysis is emerging; cysteine 135 is catalytically active, but is not a determinant for heparin binding. The heparin binding site (residues 195-221) drives enzymatic selectivity in terms of substrate size, and the histidine 203 from this region also assists the catalytic mechanism, possibly by acting as a secondary nucleophile or as an essential amino acid in a possible "proton relay system." This site also contains the calcium co-ordination site, which bridges heparin to heparinase through calcium and may orient the substrate to the active site region. In conclusion, we propose that the heparin binding site (residues 195-221) and the basic residues (lysine 132, arginine 141) close to cysteine 135 together constitute a heparin binding domain in heparinase I. The heparin binding domain provides the necessary charge complementarity for very specific heparin binding on the one hand, while on the other it creates a basic environment for biasing the active site reactivity (Fig. 6). FIG. 6. Schematic representation of the active site of heparinase I. The figure shows a heparin tetrasaccharide (about 0.165 nm in length; Nieduszynski (1985)) bound to the 18-amino acid heparin binding site. It is interesting to note that several secondary structure prediction algorithms suggest that the 18-amino acid heparin binding site contained the main ␣-helix forming potential in heparinase I, which is otherwise dominated by ␤-sheet forming potential (Sasisekharan, 1991;Yoder et al., 1993a). It is hypothesized that the core fold of heparinase I is dominated by ␤-sheets (as seen in pectate lyase; Yoder et al., 1993a) and that the region around 195-221 forms a large loop region with perhaps a partial ␣-helical secondary structure. Cysteine 135 and histidine 203 are in close proximity to each other and to the hydrogen (of the iduronate) marked by an arrow. Flanking these residues (cysteine 135 and histidine 203) are the heparin binding consensus sequences (lysine 198 to aspartic acid 204 and glutamic acid 207 to aspartic acid 212), the calcium co-ordinating motif (valine 206 to glycine 213), and lysine 132 and arginine 141 which together form the heparin binding domain of heparinase I. also thank Prof. Phil Robbins, Center for Cancer Research, MIT for help, discussion, and suggestions.