Article No. jmbi.1999.3153 available online at http://www.idealibrary.com on
J. Mol. Biol. (1999) 293, 521±530
Crystal Structure of Human Catecholamine Sulfotransferase Lisa M. Bidwell1, Michael E. McManus1, Andrea Gaedigk2 Yoshimitsu Kakuta3, Masa Negishi3, Lars Pedersen3 and Jennifer L. Martin4* 1
Department of Physiology and Pharmacology, University of Queensland, Brisbane Queensland, 4072, Australia 2
The Children's Mercy Hospital, Department of Clinical Pharmacology and Toxicology, Kansas City, MO 64108, USA 3 Pharmacogenetics Section Laboratory of Reproductive and Developmental Toxicology National Institute of Environmental Health Sciences National Institutes of Health Research Triangle Park, NC 27709, USA
Sulfonation, like phosphorylation, can modify the activity of a variety of biological molecules. The sulfotransferase enzymes sulfonate neurotransmitters, drugs, steroid hormones, dietary carcinogens and proteins. SULT1A3 speci®cally sulfonates catecholamines such as dopamine, adrenaline and noradrenaline. The crystal structure of SULT1A3 with a sulÊ resolution. fate bound at the active site, has been determined at 2.4 A Although the core a/b fold is like that of estrogen and heparan sulfotransferases, major differences occur in and around the active site. Most notably, several regions surrounding the active site, including a section of 40 residues, are disordered in SULT1A3. Regions that are topologically equivalent to the disordered parts of SULT1A3 are involved in substrate and cofactor binding in estrogen and heparan sulfotransferase. Flexibility in these regions suggests that ligand binding elicits a disorder-order transition in and around the active site of sulfotransferases and might contribute to the broad substrate speci®city of these enzymes. # 1999 Academic Press
4
Centre for Drug Design and Development, University of Queensland, Brisbane Queensland, 4072, Australia *Corresponding author
Keywords: dopamine; SULT1A3; sulfotransferase; protein structure; catecholamine
Introduction The function of biological molecules can be regulated by the transfer and removal of phosphate groups at speci®c points in the structure, through the action of kinases and phosphatases. Similarly, the addition and removal of sulfate can modify the function of biological molecules through the action of sulfotransferase and sulfatase enzymes. Although the three-dimensional structures of several hundred kinase and phosphatase enzymes, enzyme complexes and enzyme mutants are known, the structures of only two sulfotransferase Abbreviations used: PAP, adenosine 30 ,50 -diphosphate; PAPS, 30 -phosphoadenosine 50 -phosphosulfate; E2, 17b-estradiol; PSB-loop, phosphate-sulfate binding loop. E-mail address of the corresponding author:
[email protected] 0022-2836/99/430521±10 $30.00/0
(Kakuta et al., 1997, 1999) and two sulfatase enzymes have been reported (Bond et al., 1997; Lukatela et al., 1998). We have therefore initiated structural studies on sulfotransferases to gain further insight into the structural and functional relationships of these enzymes. Sulfotransferases are a superfamily of enzymes found in species ranging from bacteria to humans, which catalyse the sulfonation reaction on a variety of endogenous and exogenous substrates. While the substrate speci®city of individual sulfotransferases differs signi®cantly, all utilize the same sulfate donor cofactor, 30 -phosphoadenosine 50 phosphosulfate (PAPS). The sulfotransferase superfamily is grouped into two major classes that share less than 20 % sequence identity and which differ in terms of their solubility, size, subcellular distribution and substrate speci®city. The membrane bound sulfotransferases (45-100 kDa and located in # 1999 Academic Press
522 the Golgi apparatus) catalyze the sulfonation of large endogenous molecules such as heparan, glycoproteins, glycosaminoglycans and tyrosyl protein residues (Hashimoto et al., 1992; Ong et al., 1998; Orellana et al., 1994; Honke et al., 1997; Fukuta et al., 1998; Niehrs & Huttner, 1990). The cytosolic sulfotransferases (30-35 kDa in size and located in the cell cytoplasm) catalyze the sulfonation of xenobiotics, dietary carcinogens, neurotransmitters and hormones (Falany, 1997). Until recently, sulfonation of these small molecules was regarded as a detoxi®cation pathway leading to the formation of water soluble metabolites that are readily excreted. However, for certain compounds including carcinogens, bioactive peptides and the antihypertensive and hair growth stimulant minoxidil, sulfonation is an activation pathway. The mammalian cytosolic sulfotransferases are classi®ed based on sequence identity{. The SULT1, SULT2 and SULT3 families share less than 45% sequence identity (Weinshilboum et al., 1997). The SULT1 family is further subdivided into SULT1A, SULT1B, SULT1C and SULT1E. Members within each subfamily share more than 60% sequence identity. Although the sequence identity of the sulfotransferases varies signi®cantly, sequence alignments reveal regions of highly conserved residues (Kakuta et al., 1998a) that have been targeted for mutational studies (Kakuta et al., 1998b; Komatsu et al., 1994; Marsolais & Varin, 1995). The recent publication of the ®rst two sulfotransferase crystal structures of the cytosolic mouse estrogen sulfotransferase (SULT1E1) and the sulfotransferase domain of the human membrane-bound N-deacetylase/N-sulfotransferase1 (HSNST-1) that sulfonates heparan shows that these conserved regions interact with the cofactor PAPS (Kakuta et al., 1997, 1999). We now report the crystal structure of a third sulfonating enzyme, the catecholamine sulfotransÊ resolution. SULT1A3 ferase SULT1A3, at 2.4 A (previously called HAST3) is the enzyme primarily responsible for sulfonation of catecholamines such as dopamine, adrenaline and noradrenaline and was originally isolated in our laboratory from human brain (Zhu et al., 1993; Veronese et al., 1994). SULT1A3 does not sulfonate heparan or estrone/estradiol, the preferred substrates of HSNST-1 and SULT1E1, respectively.
Catecholamine Sulfotransferase Structure
sity at the active site indicates that sulfate is bound rather than PAP (see below). SULT1A3 crystals difÊ and belong to fract to a limiting resolution of 2.4 A the hexagonal space group P3221 with unit cell Ê and c 191.1 A Ê and one dimensions a, b 56.4 A molecule in the asymmetric unit. The structure of SULT1A3 was determined by molecular replacement using as a search model a signi®cantly trimmed version of the mouse SULT1E1 crystal structure (see Materials and Methods). Initial attempts to solve the structure using either the complete mouse SULT1E1 structure or a polyalanine model were unsuccessful, presumably because of the major differences in parts of the tertiary structure. The ®nal model of SULT1A3 has an R-factor of 27.6 % and Rfree of 28.8 %. Analysis by PROCHECK showed that 91 % of residues are in the most favored regions of the Ramachandran plot with no residues in disallowed regions. Molecular structure The overall structure has approximate dimenÊ 40 A Ê 37 A Ê . The molecule consists sions of 52 A of a single a/b domain with a central fourstranded parallel b sheet (Figure 1). The ®nal re®ned structure includes 223 of the 295 residues of SULT1A3, a single sulfate ion bound in the active site and 52 water molecules. The disordered residues cluster into several regions; the N and C termini (residues 1-7 and 294-295) and regions labelled I, II and III in Figure 1 corresponding to
Results Crystallization and structure determination Crystals of human SULT1A3 grow readily from polyethylene glycol 8000 and lithium sulfate. The enzyme used for crystallization was pre-incubated with the reaction product inhibitor PAP (adenosine 30 , 50 -diphosphate) prior to crystallization, but den{ Proposed nomenclature from the Sulfotransferase Enzyme Nomenclature Workshop, Scotland, 1996.
Figure 1. Ribbon diagram of human SULT1A3. Secondary structure elements are shown in blue (b strands), green (a helices) and yellow (loops) and the PSB-loop is depicted in red. The sulfate ion at the active site is shown as a stick model. The numbering of a helices and b strands follows that of mouse SULT1E1 (Kakuta et al., 1997). Disordered regions are represented by purple broken lines and are labelled I (residues 64-77), II (residues 91-93) and III (residues 216-261).
Catecholamine Sulfotransferase Structure
523
residues 64-77, 91-93 and 216-261, respectively. A stereo view of the three dimensional structure of SULT1A3 with sulfate bound at the active site is shown in Figure 2. An interesting feature of this structure is the loop region connecting helix a3 and strand b7. This loop is positioned close to the active site, but extends away from the core of the enzyme to interact with the active site region of another monomer of SULT1A3, related by crystallographic symmetry (Figure 3). This region is poorly ordered in the SULT1A3 structure and has been modelled with half occupancy. The active site and PSB loop SULT1A3 was crystallized in the presence of the reaction product PAP (adenosine 30 ,50 -diphosphate). We initially attempted to model PAP into the active site but the re®ned occupancy was very low and there was little or no density for the ribose and purine rings. Instead, sulfate and water were modelled into density at the active site at positions corresponding to the 50 and 30 phosphates of PAP, respectively. Residues that form the active-site pocket are contributed from different regions of the SULT1A3 sequence. One of these is the PSB-loop, a region of nine amino acid residues (residues 45TYPKSGTTW-53) that connects strand b3 and helix a3 (Figure 1) and is important for binding the cofactor PAPS. It is termed the phosphate-sulfate binding loop (PSB-loop) because it interacts with phosphate, and has a classical P-loop structure that resembles the P-loop of uridylate kinase and related kinases (Kakuta et al., 1997). The PSB-loop forms numerous contacts with the bound sulfate ion in the SULT1A3 active site. Backbone amide nitrogen atoms of Lys48, Ser49, Gly50, Thr51 and Thr52 are all within hydrogen bonding distance Ê ) of the bound sulfate (Figure 4). (2.7-3.0 A The amino acids that form the PSB-loop are highly conserved with Lys48, Gly50 and Trp53 being strictly conserved amongst all mammalian
Figure 2. Stereo drawing of the Ca atom trace of human SULT1A3. Every tenth residue is labelled and denoted by a dot. The sulfate ion is represented by a CPK model.
Figure 3. Ribbon diagram of the active site region of SULT1A3 (orange) showing the symmetry-related interaction (blue) in the proposed substrate binding pocket. The bound sulfate ion of SULT1A3 is shown in balland-stick representation. The crystallographic interaction may preclude substrate binding in this crystal form of SULT1A3 because the symmetry related loop overlaps the proposed substrate binding site, indicated by estradiol (green) from the SULT1E1 crystal structure.
cytosolic sulfotransferases identi®ed to date. Mutation of Lys59 to alanine in plant ¯avonol sulfotransferase 3 and Lys614 to alanine in the membrane-bound HSNST-1 (both corresponding to Lys48 in SULT1A3) signi®cantly reduces activity, indicating an important catalytic role for this residue (Marsolais & Varin, 1995; Sueyoshi et al., 1998). Gly50 is conserved in this tight turn for stereochemical reasons (Crawford et al., 1973). The conserved tryptophan residue (Trp53) is thought to play a role in orienting the adenine ring of PAPS (Kakuta et al., 1997). It is interesting that this residue is poorly ordered in the SULT1A3 structure, in which sulfate, and not PAP or PAPS, is bound at the active site. Although a single conformation of Trp53 is modelled in SULT1A3, there is additional electron density in this region suggestive of a second conformation for the tryptophan side-chain. A similar situation occurs for the side-chain of Lys197, which is also highly conserved and located Ê of Trp53. The electron in the active site within 4 A density in this region suggests a possible second conformation (not modelled) that would remove the lysine side-chain from the active site. The average B-factors for atoms in each of these side-chains Ê 2. Presumably, these active-site sideare over 40 A chains become more ordered upon binding the cofactor. Other conserved residues are Ser138, part of helix a6, and Arg130, located on the loop region immediately preceding helix a6. In the SULT1A3 structure, a water molecule modelled in place of the 30 phosphate of PAP (Figure 4) forms a hydroÊ gen bond with the serine side-chain, and is 3.8 A from the arginine side-chain.
524
Catecholamine Sulfotransferase Structure
Figure 4. Stereo view of the 2Fo ÿ Fc electron density in the region of the PSB-loop of human SULT1A3 showing the modelled sulfate ion and water molecule. The map is contoured at 1.25 s. Atoms are colored as follows: green, carbon; blue, nitrogen; red, oxygen; yellow, sulfur. Hydrogen bonds are represented by broken lines.
The substrate binding pocket Residues lining the proposed substrate binding pocket of SULT1A3 are predominantly hydrophobic (Ile21, Tyr23, Phe24, Phe81, Val84, Phe142, Glu146, His149, Tyr169) and are contributed by helices a1 and a6, strand b2 and other loop regions (Figure 5). Most of these residues are not generally conserved amongst sulfotransferases, consistent with the varying substrate speci®city of these enzymes. However, an exception is Phe142 that is highly conserved and which may interact with the aromatic/phenoxide ring of the substrate. Of these residues, Glu146 has been directly implicated in determining the catecholamine substrate speci®city of SULT1A3 (Figure 5). We and others have recently shown that mutation of Glu146 to alanine modi®es the substrate speci®city of SULT1A3 so that the enzyme behaves more like SULT1A1, a sulfotransferase which utilises p-nitrophenol as its preferred substrate (Dajani et al., 1998; Brix et al., 1999a). Similarly, the reverse mutation in SULT1A1 changes its substrate speci®city so that it behaves more like SULT1A3 (Brix
et al., 1999a,b). The hypothesis is that the acidic side-chain of Glu146 forms a charge:charge interaction with the amine group of the catecholamine substrate to produce this substrate speci®city (Brix et al., 1999a). In the SULT1A3 structure the Glu146 side-chain is positioned in the active site close to where the substrate is expected to bind (Figure 5). The substrate binding pocket in this structure of SULT1A3 has two unusual features. Firstly, many of the residues that would line the pocket are disordered, including a region of over 40 residues (region III, residues 216-261). Secondly, part of the proposed substrate binding pocket is occupied by residues 86-90 from a symmetry-related molecule (Figure 3). This interaction is most probably a crystallization artefact but could help explain why substrate co-crystallization and soaking experiments have so far proved unsuccessful with this SULT1A3 crystal form. The disorder observed for residues and regions in and around the active site of SULT1A3 probably results from the absence of stabilizing interactions with bound substrate and/ or cofactor.
Figure 5. Ribbon diagram comparing the PAPS and substrate binding sites of SULT1A3 (green) and SULT1E1 (pink). The PSB-loops of the two enzymes are highlighted in light green. The bound ligands of SULT1E1 are shown in yellow, with PAP in the upper right overlapping with the active site sulfate of SULT1A3, and E2 in the lower left. An arrow points to a region of the active site that differs signi®cantly between the two enzyme structures. In SULT1E1, this region folds into a b sheet (b5-b6) and a short helix (a4) but in SULT1A3 it forms a loop which is translated upwards and to the left, away from the substrate binding site, compared with SULT1E1. The region of SULT1E1 that forms part of the PAPS and substrate binding sites and corresponds to the disordered region III of SULT1A3 would be positioned above the active site in this orientation (see also Figure 6).
Catecholamine Sulfotransferase Structure
Comparison with other sulfotransferase structures At this time, three structures of sulfotransferases have been solved; the SULT1A3:sulfate complex reported here, mouse SULT1E1 complexed with PAP and PAP/E2 (Kakuta et al., 1997), and the sulfotransferase domain of human N-deacetylase/Nsulfotransferase1 (hHSNST-1) complexed with PAP (Kakuta et al., 1999). Human SULT1A3 and mouse SULT1E1 are both cytosolic enzymes comprising 295 residues and share 48 % sequence identity. In contrast, HSNST-1 is a membrane-bound Golgi enzyme that is responsible for the ®rst sulfonation event that triggers the formation of heparan and heparan sulfates. The sulfotransferase domain of this enzyme is 324 residues. We have compared the structures of SULT1A3, SULT1E1 and HSNST1 (Figure 6). As might be pre-
525 dicted given their sequence identity, the two cytosolic enzymes superimpose well giving an r.m.s.d. Ê for 212 of the 223 modelled Ca atoms of of 1.24 A SULT1A3. Residues implicated in PAPS binding, 45-53 (PSB-loop), Arg130, Arg257, and Ser138 are conserved between human SULT1A3 and mouse SULT1E1 sequences (Figure 6). Likewise, residues that have been implicated in catalysis (Lys48, Lys106, and His108) are also conserved (Kakuta et al., 1997). Given the difference in substrate speci®city it is not unexpected that residues in the substrate binding site are not as well conserved as the cofactor binding site. However, Phe24, Phe142, Tyr139, and Tyr169 are conserved and may be important for structural integrity or maintaining the hydrophobicity of the substrate binding pocket. The sequence identity between the two cytosolic enzymes and the membrane-bound Golgi sulfotransferases is very low (<20 %). However, the
Figure 6. Comparison of three sulfotransferases (upper panel) Ca traces of crystal structures of human SULT1A3 (left), mouse SULT1E1 (Kakuta et al., 1997) (middle) and human HSNST-1 (Kakuta et al., 1999) (right). Regions of the structures that can be superimposed are shown in light blue. Regions that are not structurally aligned are shown in red. Residues at either end of the disordered region III of SULT1A3 are labelled in the SULT1A3 and SULT1E1 and the conserved PSB-loop is highlighted in green. Ligands are shown at the active site (sulfate for SULT1A3, PAP and E2 for SULT1E1 and PAP for HSNST-1). Lower panel, Structure-based sequence alignment of the three sulfotransferases. The same color scheme is used, with the addition that residues not modelled in the crystal structures are shown in grey. Secondary structure elements are shown for the mouse SULT1E1.
526 structural elements of SULT1A3, SULT1E1 and HSNST-1 are surprisingly similar with the cytosolic and membrane-bound enzymes utilizing the same a/b fold to bind PAPS (Figure 6). Superposition of Ê SULT1A3 and HSNST-1gives an r.m.s.d. of 1.85 A a for 123 equivalent C atoms. The sequence identity based only on these 123 structurally similar residues is 23 %. However, for the 49 residues that make up the PAPS binding site (residues 40-61 and 121-147 of SULT1A3 corresponding to 606-627 and 695-721of HSNST-1) the sequence identity is signi®cantly increased (39 %). There are, of course, major differences in the cytosolic and membrane-bound sulfotransferase structures. The hydrophilic heparan substrates of membrane bound HSNST-1 are much larger than the small aromatic substrates (catecholamine/estradiol) of the cytosolic SULT1A3 and SULT1E1 enzymes. This is apparent in the different substrate binding sites of the enzymes, with that of HSNST1 forming a large open cleft to accommodate the polysaccharide chain (Figure 6) and the two cytosolic enzymes forming a buried hydrophobic pocket. A striking difference between the structure of SULT1A3 and those of SULT1E1 and HSNST-1 is that the disordered region of the SULT1A3 structure (residues 216-261) is ordered in the other two structures (Figure 6). In mouse SULT1E1, these residues form a signi®cant part of the hydrophobic substrate binding pocket and contribute several speci®c hydrogen bond interactions to PAP, including a number from the highly conserved 257-RKGxxGxxK-265 motif (Marsolais & Varin, 1995). A major difference between the structures of Ê movement in the two cytosolic enzymes is a 4 A the region incorporating strands b5 and b6 of SULT1E1 (residues 78-90) corresponding to the loop region linking disordered sections I and II in SULT1A3. This region is also likely to contribute to substrate binding in SULT1A3, since in SULT1E1 Asn86 forms a hydrogen bond to the 17b-hydroxyl group of E2 and Ile90 lines the hydrophobic binding pocket. The fact that many of the residues involved in cofactor and substrate binding are disordered or ¯exible in the SULT1A3 structure is further evidence that this enzyme incorporates mobile motifs that close down on the active site upon ligand binding. Catalytic residues The sulfonation reaction has been proposed to involve an SN2 in-line displacement mechanism through nucleophilic attack on the PAPS sulfate by the 3a-phenoxide of the substrate (Kakuta et al., 1997). Three residues (Lys48, Lys106 and His108) have been shown to be important for catalysis, and are proposed to stabilize the transition state (Kakuta et al., 1998b). His108 may also function as a catalytic base by deprotonating the hydroxyl group of the substrate (Kakuta et al., 1998b). This residue is conserved in all cytosolic sulfotrans-
Catecholamine Sulfotransferase Structure
ferases and mutation of this residue in mouse SULT1E1 and in ¯avonol sulfotransferase abolishes almost all activity (Kakuta et al., 1998b; Marsolais & Varin, 1997). The position of this residue is similar in the SULT1A3 and SULT1E1 structures. In the SULT1A3 structure, His108 Nd1 forms hydrogen bonds with the main-chain oxygen atom of Thr45 and the side-chain oxygen atom of Thr51. The sidechain of Thr51 forms a hydrogen bond with the side-chain of Thr45 which in turn interacts with the main-chain oxygen atom of Tyr46. A similar network surrounding His108 is observed in the structure of SULT1E1. This hydrogen bond network is consistent with the proposed catalytic role of His108 as a proton acceptor. Site-directed mutagenesis studies have shown that Lys48 and Lys106 are essential for sulfotransferase activity (Kakuta et al., 1998b; Marsolais & Varin, 1997). While the conformation of Lys48 is conserved between SULT1A3 and SULT1E1, Lys106 adopts different conformations in these two structures (Figure 5). In SULT1E1, Lys106 Nz forms a hydrogen bond with the sulfonate acceptor Ê from hydroxyl group of the substrate and is 4 A e2 the N atom of His108. In the SULT1A3 structure, the Lys106 side-chain is oriented away from His108 so that the distance between these two Ê . This is accompanied by a change in atoms is 7.3 A position of the nearby Phe81 side-chain of SULT1A3 relative to the equivalent Tyr81 sidechain of SULT1E1 so that it is oriented between His108 and Lys106 side-chains in SULT1A3. The essential role of Lys106 in catalysis suggests that if this residue does stabilize the transition state it must undergo a conformational change upon ligand binding in SULT1A3. Of the three catalytic residues discussed above (Lys48, Lys106 and His108), only Lys48 is strictly conserved amongst cytosolic and membranebound sulfotransferases. HSNST-1 has no residue equivalent to His108; however, the crystal structure of this enzyme suggests that an additional residue (Lys833 which is not conserved in SULT1A3) interacts with the 50 phosphate of PAP and may play a role in catalysis (Kakuta et al., 1999).
Discussion Sulfotransferases are a superfamily of enzymes that catalyze the sulfonation of a wide variety of structurally and chemically diverse compounds. Until now, structural information has been available for only two of these enzymes, mouse SULT1E1 and human HSNST-1 which differ signi®cantly from SULT1A3 in both their primary sequence and substrate speci®cities. Our structural comparison of SULT1A3 with SULT1E1 and HSNST-1 reveals that the enzymes incorporate a common structural fold consistent with a common evolutionary ancestor. In addition, many of the residues involved in cofactor binding are conserved between these three enzymes. However, the
527
Catecholamine Sulfotransferase Structure
residues which line the substrate binding pocket are not well conserved, re¯ecting the different size and chemistry of the preferred substrates for each enzyme (SULT1A3:dopamine, SULT1E1:estradiol, HSNST-1:heparan). A major ®nding from this work, that was not apparent previously, is that several regions and residues in and around the active site that are ordered in the ligand-bound sulfotransferase structures are disordered in the unliganded SULT1A3 complex. This suggests that these regions and residues may undergo a disorder-order transition upon ligand binding. Such conformational changes have been reported for a number of other proteins including ribulose-1,5bisphosphate carboxylase/oxygenase, triosephosphate isomerase and streptavidin (Schreuder et al., 1993; Mande et al., 1994; Fritag et al., 1997). The ¯exibility of speci®c regions around the active site of the sulfotransferases may be of importance in their ability to utilize a variety of related substrates in the sulfonation reaction. A similar effect has been described for the cyclin-dependent kinase (Cdk) inhibitor p21Waf1/Cip1/Sdi1 whose N-terminal region lacks stable secondary or tertiary structure in the free state but adopts a stable ordered structure when bound to Cdk2 (Kriwacki et al., 1996). The ability of this protein to adopt multiple conformations has been associated with its recognition of a range of biological targets (Kriwacki et al., 1996). SULT1A3 is the predominant sulfotransferase involved in the sulfonation of catecholamines such as dopamine, adrenaline and noradrenaline. In humans, 90-95 % of circulating dopamine and approximately 70 % of plasma adrenaline and noradrenaline are found in their sulfonated form (Johnson et al., 1980). Both dopamine 3-O-sulfate and dopamine 4-O-sulfate have been shown to be active in their sulfonated form through interactions with central GABA receptors (Buu et al., 1984). While the precise function of SULT1A3 is not well understood, this enzyme may play an important role in regulating the levels of free dopamine in the brain and/or other tissues. Altered levels of sulfotransferase activity have been found in diseases such as Alzheimer's, dietary migraine, obsessive-compulsive disorder, mania and unipolar depression and essential hypertension (Mao & Barger, 1998; Littlewood et al., 1982; Marazziti et al., 1992, 1996; Yoshizumi et al., 1996). It has been suggested that a de®ciency of SULT1A3 may be responsible for the condition known as pseudopheochromocytoma, where patients display the symptoms of pheochromocytoma (adrenal tumours that synthesise and secrete catecholamines) but have decreased levels of conjugated catecholamines rather than a tumour (Goldstein et al., 1996; Kuchel et al., 1981). The structure of SULT1A3 reported here and the proposed disorder-order transition upon ligand binding will facilitate characterization of this enzyme by mutational and binding studies. Ultimately, this will provide invaluable information for under-
standing the catalytic mechanism and physiological role of sulfotransferase enzymes in general.
Materials and Methods Expression, purification and crystallization Recombinant SULT1A3 was expressed as described by Gaedigk et al. (1998) and bacterial cytosol was prepared by the method described by Gillam et al. (1993). SULT1A3 protein was puri®ed using DEAE Sepharose CL-6B chromatography as described by Falany et al. (1989). Partially puri®ed protein was dialyzed against buffer A (5 mM NaPO4 (pH 6.8), 0.25 M sucrose, 10 % (v/v) glycerol) and applied to a Macro-Prep Ceramic Hydroxyapatite type I column (BioRad Laboratories, Hercules, CA, USA) which had been pre-equilibrated with buffer A. The eluate, containing the SULT1A3 protein, was collected and further puri®ed using PAP agarose af®nity chromatography as described by Falany et al. (1990). Puri®ed SULT1A3 protein was dialyzed against 5 mM sodium phosphate (pH 6.8), 1 mM DTT then concentrated to 15-20 mg/ml. Purity was estimated to be above 95 % by SDS-PAGE and mass spectrometry. For crystallization, the protein was pre-incubated with excess PAP at 4 C for one to two hours. Crystals were grown at 20 C by hanging drop vapour diffusion after mixing 1 ml of the enzyme with 1ml of the reservoir solution (0.5 M lithium sulfate and 5-7 % (w/v) polyethylene glycol 8000). This was equilibrated over the reservoir solution and crystals of average size 0.3 mm 0.2 mm 0.1 mm appeared in the drop after two to three days. Crystals of SULT1A3 grown under these conditions are trigonal, with space group P3221, Ê , c 191.1 A Ê , a, unit cell dimensions a, b 56.4 A Ê , g 120 A Ê and one molecule per asymmetric b 90 A unit. Diffraction data measurement X-ray diffraction data were measured from a single cryocooled (100 K) crystal on a MAR image plate system at Beamline 7.1 of the Stanford Synchrotron Radiation Laboratories in San Francisco, USA. The crystal was immersed in cryoprotectant (0.5 M lithium sulfate, 10 % polyethyleneglycol 8000 and 20 % (v/v) glycerol) immediately prior to ¯ash cooling in the nitrogen gas stream. For data measurement, an oscillation range of 2 , exposure time of 90 seconds and crystal to detector distance of 300 mm were used. Diffraction data extended Ê and were integrated, merged and scaled to 2.4 A using the DENZO/SCALEPACK program package (Otwinowski, 1997). Statistics for crystallographic data processing are shown in Table 1. Structure determination The structure was solved by molecular replacement using AMoRe (Navaza, 1994). Initial attempts to phase the structure using mouse SULT1E1 coordinates (PDB code 1AQU) (Kakuta et al., 1997) as the search model were unsuccessful whether the complete structure or a polyalanine model was used. After many trials, a solution was found using a search model based on monomer A of the SULT1E1 structure but with residues other than 26-64, 73-148 and 168-260 removed and with nonconserved residues replaced by alanine. This gave a 4.6 s peak in the rotation function and a 8.2 s peak in the
528
Catecholamine Sulfotransferase Structure
Table 1. Crystallographic data collection statistics
Table 2. Crystallographic re®nement statistics
Unit cell Ê) a (A Ê) b (A Ê) c (A a (deg.) b (deg.) g (deg.) Space group Solvent content (%) Molecules in asymmetric unit Observations (I > 1s(l)) Unique reflections (I > 1s(l)) Rasym(%) Rasym (outer shell)b (%) I/s(I) I/s(I) (outer shell)b Completeness (%) Completeness (outer shell)b (%)
Ê) Resolution range (A Number of reflections (F > 0sF) R-factora Rbfree Number of water molecules Ê 2) Average B-factor (A r.m.s.d. from ideal Ê) Bond length (A Bond angle (deg.) Dihedral angle (deg.) Improper angle (deg.) Ramachandran statistics Residues in most favored regions (%) Residues in additionally allowed regions (%) Residues in disallowed regions (%)
56.4 56.4 191.1 90 90 120 P3221 58 1 63,776 15,080 7.0 33.0 14.3 2.7 96.8 86.8
a
Rsym jI ÿ hIij/hIi. b Ê. The outer shell is 2.35-2.43 A
Ê . After rigid translation function, using data from 8-3 A body re®nement in AMoRe this solution gave a correlation coef®cient of 57.9 % and an R-factor of 52.4 %. A rotated and translated model of the SULT1A3 structure incorporating residues 7-64, 75-242, 244-295 of SULT1E1 was subjected to rigid body re®nement in X-PLOR (BruÈnger, 1992) giving an R-factor of 47.9 % and Rfree of Ê with F > 2 sF. The elec48.7 % using data between 6-3 A tron density map generated from this model was subjected to solvent ¯attening and histogram matching in DM (Cowtan, 1994). A new model of SULT1A3 was built into this map using the program O (Jones et al., 1991). This model included residues 36-62, 76-84, 101211, 217-230, 247-292 and after 100 cycles of positional re®nement in X-PLOR this gave an R-factor of 34.2 % and Rfree of 41.8 %. Several rounds of model building and re®nement, including bulk solvent correction, were performed to reduce the R-factor and Rfree to 29.8 % and Ê and 33.1 %, respectively, for all data between 2.4 A Ê . Water molecules were included in the model 100 A where density was present in both the 2Fo ÿ Fc (1.0 s) and Fo ÿ Fc (2.8 s) maps and a sulfate ion was added in the ®nal stages of re®nement. The ®nal re®ned model of SULT1A3 includes residues 8-63, 78-90, 94-215 and 262-293, plus 52 water molecules and a single sulfate ion (Figures 1 and 2). Residues 85-90 were modelled with half occupancy due to the disorder in this region. Poor density in the region of residues 216224 was modelled with water molecules. Residues with poorly de®ned side-chain density were modelled as alanine. These residues are Ser8, Glu13, Lys22, Tyr23, Leu28, Leu31, Gln32, Gln56, Gln63, Arg78, Asn85, Asp86, Pro90, Glu94, Leu96, Lys97, Asp98, Asp120, Lys201, Arg292 and Ser293. Two alternate conformations were modelled for His186. The stereochemical quality of the ®nal model was assessed by WHATIF (Vriend & Sander, 1993) and PROCHECK (Laskowski et al., 1993). Crystallographic statistics for the ®nal SULT1A3 model are shown in Table 2. Structural alignments with SULT1E1 and HSNST-1 were performed using the lsq options in O (Jones et al., 1991). Figures were generated using programs MOLSCRIPT (Kraulis, 1991), Raster3D (Merrit & Murphy, 1994) and SETOR (Evans, 1993).
100-2.4 14,265 27.6 28.8 52 30.0 0.005 1.1 23 1.1 91.0 9.0 0.0
a
R-factor jFo ÿ Fcj/Fo. b Rfree as de®ned by BruÈnger (1992) calculated from 10 % of the data.
Accession numbers The coordinates and structure factors for SULT1A3 have been deposited at the Brookhaven Protein Data Bank with accession number 1CJM.
Acknowledgments We thank Alun Jones for help with mass spectrometric data measurement and analysis, Dr Luke Guddat for help with X-ray data measurement and Joel Tyndall for help with Figures. We also acknowledge Dr Elizabeth Gillam for advice on the puri®cation of SULT1A3. The SULT1A3 research is supported by grants from the National Health and Medical Research Council of Australia and the University of Queensland (Mayne Bequest Fund). This work is based on research conducted at the Stanford Synchrotron Radiation Laboratory (SSRL), which is funded by the Department of Energy, Of®ce of Basic Energy Sciences. The SSRL Biotechnology Program is supported by the National Institutes of Health, National Centre for Research Resources, Biomedical Technology Program and the Department of Energy, Of®ce of Biological and Environmental Research. J.L.M. is supported by an Australian Research Council Fellowship.
References Bond, C. S., Clements, P. R., Ashby, S. J., Collyer, C. A., Harrop, S. J., Hopwood, J. J. & Guss, J. M. (1997). Structure of a human lysosomal sulfatase. Structure, 5, 277-289. Brix, L. A., Barnett, A., Duggleby, R. G. & McManus, M. E. (1999a). Analysis of the substrate speci®city of human sulfotransferases SULT1A1 and SULT1A3: site-directed mutagenesis and kinetic studies. Biochemistry, 38, 10474-10479. Brix, L. A., Duggleby, R. G., Gaedigk, A. & McManus, M. E. (1999b). Structural characterization of human aryl sulphotransferases. Biochem. J. 337, 337-343.
Catecholamine Sulfotransferase Structure BruÈnger, A. T. (1992). X-PLOR Version 3.1 Manual, Yale University Press, New Haven, CT. Buu, N. T., Duhaime, J. & Kuchel, O. (1984). The bicuculline-like properties of dopamine sulfate in rat brain. Life Sci. 35, 1083-1090. Cowtan, K. (1994). Joint CCP4 and ESF-EACBM Newsletter on Protein Crystallography, vol. 31, pp. 34-38. Crawford, J. L., Lipscomb, W. N. & Schellman, C. G. (1973). The reverse turn as a polypeptide conformation in globular proteins. Proc. Natl Acad. Sci. USA, 70, 538-542. Dajani, R., Hood, A. M. & Coughtrie, M. W. H. (1998). A single amino acid, Glu146, governs the substrate speci®city of a human dopamine sulfotransferase, SULT1A3. Mol. Pharmacol. 54, 942-948. Evans, S. V. (1993). SETOR: hardware lighted threedimensional solid model representations of macromolecules. J. Mol. Graph. 11, 134-138. Falany, C. N. (1997). Enzymology of human cytosolic sulfotransferases. FASEB J. 11, 206-216. Falany, C. N., Vazquez, M. E. & Kalb, J. M. (1989). Puri®cation and characterization of human liver dehydroepiandrosterone sulphotransferase. Biochem. J. 260, 641-646. Falany, C. N., Vazquez, M. E., Heroux, J. A. & Roth, J. A. (1990). Puri®cation and characterization of human liver phenol-sulfating phenol sulfotransferase. Arch. Biochem. Biophys. 278, 312-318. Fritag, S., Le, Trong I., Klumb, L., Stayton, P. S. & Stenkamp, R. E. (1997). Structural studies on the streptavidin binding loop. Protein Sci. 6, 1157-1166. Fukuta, M., Kobayashi, Y., Uchimura, K., Kimata, K. & Habuchi, H. (1998). Molecular cloning and expression of human chondroitin 6-sulfotransferase. Biochim. Biophys. Acta, 1399, 57-61. Gaedigk, A., Lekas, P., Berchuk, M. & Grant, D. M. (1998). Novel sulfotransferases cloned by RT-PCR: real proteins or PCR artifacts? Chem. Biol. Interact. 109, 43-52. Gillam, E. M. J., Baba, T., Kim, B.-R., Ohmori, S. & Guengerich, F. P. (1993). Expression of modi®ed human cytochrome P450 3A4 in Escherichia coli and puri®cation and reconstitution of the enzyme. Arch. Biochem. Biophys. 305, 123-131. Goldstein, D. S., Lenders, J. W. M., Kaler, S. G. & Eisenhofer, G. (1996). Catecholamine phenotyping: clues to the diagnosis, treatment, and pathophysiology of neurogenetic disorders. J. Neurochem. 67, 1781-1790. Hashimoto, Y., Orellana, A., Gil, G. & Hirschberg, C. B. (1992). Molecular cloning and expression of rat liver N-heparan sulfate sulfotransferase. J. Biol. Chem. 267, 15744-15750. Honke, K., Tsuda, M., Hirahara, Y., Ishii, A., Makita, A. & Wada, Y. (1997). Molecular cloning and expression of cDNA encoding human 30 -phosphoadenylylsulfate:galactosylceramide 30 -sulfotransferase. J. Biol. Chem. 272, 4864-4868. Johnson, G. A., Baker, C. A. & Smith, R. T. (1980). Radioenzymatic assay of sulfate conjugates of catecholamines and DOPA in plasma. Life Sci. 26, 1591-1598. Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjeldgaard, M. (1991). Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallog. sect. A, 47, 110-119. Kakuta, Y., Pedersen, L. G., Carter, C. W., Negishi, M. & Pedersen, L. C. (1997). Crystal structure of
529 estrogen sulphotransferase. Nature Struct. Biol. 11, 904-908. Kakuta, Y., Pedersen, L. G., Pedersen, L. C. & Negishi, M. (1998a). Conserved structural motifs in the sulfotransferase family. Trends Biochem. Sci. 23, 129-130. Kakuta, Y., Petrotchenko, E. V., Pedersen, L. C. & Negishi, M. (1998b). The sulfuryl transfer mechanism; crystal structure of a vanadate complex of estrogen sulfotransferase and mutational analysis. J. Biol. Chem. 273, 27325-27330. Kakuta, Y., Sueyoshi, T., Negishi, M. & Pedersen, L. C. (1999). Crystal structure of the sulfotransferase domain of human heparan sulfate N-deacetylase/Nsulfotransferase-1. J. Biol. Chem. 274, 10673-10676. Komatsu, K., Driscoll, W. J., Koh, Y. C. & Strott, C. A. (1994). A P-loop related motif (GxxGxxK) highly conserved in sulfotransferases is required for binding the activated sulfate donor. Biochem. Biophys. Res. Commun. 204, 1178-1185. Kraulis, P. J. (1991). MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallog. 24, 946-950. Kriwacki, R. W., Hengst, L., Tennant, L., Reed, S. I. & Wright, P. E. (1996). Structural studies of p21Waf1/Cip1/Sdi1 in the free and Cdk2-bound state: Conformational disorder mediates binding diversity. Proc. Natl Acad. Sci. USA, 93, 11504-11509. Kuchel, O., Buu, N. T., Hamet, P., Larochelle, P., Bourque, M. & Genest, J. (1981). Essential hypertension with low conjugated catecholamines imitates pheochromocytoma. Hypertension, 3, 347-355. Laskowski, R. A., MacArthur, M. W., Moss, D. S. & Thornton, J. M. (1993). PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Crystallog. 26, 283-291. Littlewood, J., Glover, V., Sandler, M., Petty, R., Peat®eld, R. & Rose, F. C. (1982). Platelet phenolsulphotransferase de®ciency in dietary migraine. Lancet, 1, 983-986. Lukatela, G., Krauss, N., Theis, K., Selmer, T., Gieselmann, V., von Figura, K. & Saenger, W. (1998). Crystal structure of human arylsulfatase A: the aldehyde function and the metal ion at the active site suggest a novel mechanism for sulfate ester hydrolysis. Biochemistry, 37, 3654-3664. Mande, S. C., Mainfroid, V., Kaklk, K. H., Goraj, K., Martial, J. A. & Hol, W. G. J. (1994). Crystal structure of recombinant human triosephosphate isomerÊ resolution. Triosephosphate isomerasease at 2.8 A related human genetic disorders and comparison with the trypanosomal enzyme. Protein Sci. 3, 810821. Mao, X. & Barger, S. W. (1998). Neuroprotection by dehydropiandrosterone sulfate: role of an NFKBlike factor. Neuropharmacology, 9, 759-763. Marazziti, D., Hollander, E., Lensi, P., Ravagli, S. & Cassano, G. B. (1992). Peripheral markers of serotonin and dopamine function in obsessive-compulsive disorder. Psych. Res. 42, 41-51. Marazziti, D., Palego, L., Dell'Osso, L., Batistini, A., Cassano, G. B. & Akiskal, H. S. (1996). Platelet sulfotransferase in different psychiatric disorders. Psych. Res. 65, 73-78. Marsolais, F. & Varin, L. (1995). Identi®cation of amino acid residues critical for catalysis and cosubstrate binding in the ¯avonol 3-sulfotransferase. J. Biol. Chem. 270, 30458-30463.
530
Catecholamine Sulfotransferase Structure
Marsolais, F. & Varin, L. (1997). Mutational analysis of domain II of ¯avonol 3-sulfotransferase. Eur. J. Biochem. 247, 1056-1062. Merrit, E. A. & Murphy, M. E. P. (1994). Raster3D version 2.0: a program for photorealistic molecular graphics. Acta Crystallog. sect. D, 50, 869-873. Navaza, J. (1994). AMoRe: an automated package for molecular replacement. Acta Crystallog. sect. A, 50, 157-163. Niehrs, C. & Huttner, W. B. (1990). Puri®cation and characterization of tyrosylprotein sulfotransferase. EMBO J. 9, 35-42. Ong, E., Yeh, J. C., Ding, Y., Hindsgaul, O. & Fukuda, M. (1998). Expression cloning of a human sulfotransferase that directs the synthesis of the HNK-1 glycan on the neural cell adhesion molecule and glycolipids. J. Biol. Chem. 273, 5190-5195. Orellana, A., Hirschberg, C. B., Wei, Z., Swiedler, S. J. & Ishihara, M. (1994). Molecular cloning and expression of a glycosaminoglycan N-acetylglucosaminyl N-deacetylase/N-sulfotransferase from a heparin-producing cell line. J. Biol. Chem. 269, 22702276. Otwinowski, Z. (1997). Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307-326. Schreuder, H. A., Knight, S., Curmi, P. M., Andersson, I., Casio, D., BraÈndeÂn, C.-I. & Eisenberg, D. (1993). Formation of the active site of ribulose-1,5-bisphosphate carboxylase/oxygenase by a disorder-order
transition from the unactivated state to the activated form. Proc. Natl Acad. Sci. USA, 90, 9968-9972. Sueyoshi, T., Kakuta, Y., Pedersen, L. C., Wall, F. E., Pedersen, L. G. & Negishi, M. (1998). A role of Lys614 in the sulfotransferase activity of human heparan sulfate N-deacetylase/N-sulfotransferase. FEBS Letters, 433, 211-214. Veronese, M. E., Burgess, W., Zhu, X. & McManus, M. E. (1994). Functional characterisation of two human sulphotransferase cDNAs that encode monoamineand phenol-sulphating forms of phenol sulphotransferase: substrate kinetics, thermal-stability and inhibitor-sensitivity studies. Biochem. J. 302, 497-502. Vriend, G. & Sander, C. (1993). Quality control of protein models: directional atomic contact analysis. J. Appl. Crystallog. 26, 47-60. Weinshilboum, R. M., Otterness, D. M., Aksoy, I. A., Wood, T. C., Her, C. & Raftogianis, R. B. (1997). Sulfotransferase molecular biology: cDNAs and genes. FASEB J. 11, 3-14. Yoshizumi, M., Kitagawa, T., Hori, T., Katoh, I., Houchi, H., Ohuchi, T. & Oka, M. (1996). Physiological signi®cance of plasma sulfoconjugated dopamine in patients with hypertension - clinical and experimental studies. Life Sci. 59, 323-330. Zhu, X., Veronese, M. E., Bernard, C. C. A., Sansom, L. N. & McManus, M. E. (1993). Identi®cation of two human brain aryl sulfotransferase cDNAs. Biochem. Biophys. Res. Commun. 195, 120-127.
Edited by R. Huber (Received 6 May 1999; received in revised form 16 August 1999; accepted 18 August 1999)