Site-selected Mutagenesis of a Conserved Nucleotide Binding HXGH Motif Located in the ATP Sulfurylase Domain of Human Bifunctional 3′-Phosphoadenosine 5′-Phosphosulfate Synthase*

3′-Phosphoadenosine-5′-phosphosulfate (PAPS) synthase is a bifunctional protein consisting of an NH2-terminal APS kinase and a COOH-terminal ATP sulfurylase. Both catalytic activities require ATP; the APS kinase domain involves cleavage of the β-γ phosphodiester bond of ATP, whereas the ATP sulfurylase domain involves cleavage of the α-β phosphodiester bond of ATP. Previous mutational studies have suggested that β-γ phosphodiesterase activity involves a highly conserved NTP-binding P-loop motif located in the adenosine-5′-phosphosulfate kinase domain of PAPS synthases. Sequence alignment analysis of PAPS synthases and the superfamily of TagD-related nucleotidylyltransferases revealed the presence of a highly conserved HXGH motif in the ATP sulfurylase domain of PAPS synthases, a motif implicated in the α-β phosphodiesterase activity of cytidylyltransferases. Thus, site-selected mutagenesis of the HXGH motif in the ATP sulfurylase domain of human PAPS synthase (amino acids 425–428) was performed to examine this possibility. Either H425A or H428A mutation produced an inactive enzyme. In contrast, a N426K mutation resulted in increased enzymatic activity. A G427A single mutant resulted in only a modest 30% reduction in catalytic activity, whereas a G427A/H428A double mutant produced an inactive enzyme. These results suggest an important role for the HXGH histidines in the ATP sulfurylase activity of bifunctional PAPS synthase and support the hypothesis that the highly conserved HXGH motif found in the ATP sulfurylase domain of PAPS synthases is involved in ATP binding and α-β phosphodiesterase activity.

the activated sulfate molecule is 3Ј-phosphoadenosine 5Ј-phosphosulfate (PAPS). 1 The activation of inorganic sulfate to form PAPS results from the concerted action of two enzymes (1). The first step is catalyzed by ATP sulfurylase (ATP:sulfate adenylyltransferase, EC 2.7.7.4) and involves the reaction of inorganic sulfate with ATP to form adenosine 5Ј-phosphosulfate (APS) and inorganic pyrophosphate. This reaction results in the formation of a high energy phosphoric-sulfuric acid anhydride bond that is the chemical basis for sulfate activation (2). The second step involves APS reacting with another molecule of ATP to form PAPS and ADP and is catalyzed by APS kinase (ATP:adenylylsulfate 3Ј-phosphotransferase, EC 2.7.1.25). ATP sulfurylase and APS kinase cloned from bacteria, Fungi, yeast, and plants are found on separate polypeptide chains; however, in mammalian species, the marine worm, and the fruit fly, gene fusion has occurred, and the two enzymes are integral to bifunctional PAPS synthase. It has been clearly established that the NH 2 -terminal region of human (h) PAPS synthase constitutes the APS kinase domain, whereas the ATP sulfurylase domain is located in the COOH-terminal portion of this bifunctional protein (3). That is, the NH 2 -terminal 268 amino acids of hPAPS synthase (623 amino acids) when overexpressed and purified functions independently as an APS kinase (3), whereas the overexpressed and purified COOHterminal 405 amino acids of hPAPS synthase functions independently as an ATP sulfurylase. 2 A highly conserved P-loop NTP-binding motif (GXXGXGK(S/ T)) is present within the NH 2 -terminal APS kinase domain of human (3), guinea pig, 2 mouse (4), marine worm (5), and Drosophila (6) PAPS synthase. The P-loop motif is highly conserved among NTP-binding proteins where it is involved in binding the phosphate moiety and in the cleavage of the ␤-␥ phosphodiester bond of NTP (7,8). It was recently reported that site-selected mutagenesis of the P-loop motif in mouse PAPS synthase markedly impaired APS kinase activity (9). The latter finding would be in keeping with the hypothesis that the P-loop motif located near the NH 2 terminus of PAPS synthase is involved in cleavage of the ␤-␥ phosphodiester bond of ATP and transfer of the terminal phosphoryl group of ATP to the 3Ј-hydroxyl position of adenosine 5Ј-phosphosulfate.
In contrast to the NH 2 -terminal APS kinase domain of PAPS synthase, the COOH-terminal ATP sulfurylase domain does not and would not necessarily be expected to contain a classical P-loop motif for ATP binding. The reason being that although the COOH-terminal ATP sulfurylase domain of PAPS synthase binds ATP as does the NH 2 -terminal domain, the ultimate fate of the bound ATP is not the same. That is, APS kinase catalyzes the removal of the terminal ␥-phosphate for transfer to the acceptor molecule, APS, whereas ATP sulfurylase catalyzes the removal of the ␤-␥ diphosphate (inorganic pyrophosphate) of ATP and condensation of the formed AMP with inorganic sulfate to form APS. It might, therefore, be expected that the latter function would require a different type of nucleotidebinding site, one that would function as an ␣-␤ phosphodiesterase rather than a ␤-␥ phosphodiesterase. Inspection of the * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  data base search (10) output for the COOH-terminal domain of hPAPS synthase revealed a HXGH motif that is conserved throughout the PAPS synthase family. This motif has been previously characterized as the signature of a large family of (predicted) nucleotidylyltransferases, prototyped by the Bacillus subtilis cytidylyltransferase TagD; all functionally characterized proteins of this family cleave the ␣-␤ phosphate bond of an NTP (11).
In this paper, we report that site-selected mutagenesis of the HXGH motif implicates this motif in the ATP sulfurylase reaction of bifunctional hPAPS synthase, a finding consistent with the concept that the HXGH sequence is involved in ATP binding and ␣-␤ phosphodiesterase activity.

EXPERIMENTAL PROCEDURES
Materials-Radionucleotides [␣-35 S]ATP for DNA sequencing, inorganic [ 35 S]SO 4 (1300 Ci/mmol), and [ 35 S]PAPS for enzyme assays were purchased from NEN Life Science Products. Oligonucleotides were obtained from Life Technologies, Inc. and Gene Probe Technology (Gaithersburg, MD). Site-directed mutagenesis kit was obtained from Stratagene (La Jolla, CA). Version-2 sequencing kit was obtained from U. S. Biochemical Corp. Agarose was purchased from FMC BioProducts (Rockland, ME) and polyethyleneimine cellulose (PEI)-TLC plates were purchased from Merck.
Overexpression of Human PAPS Synthase and Mutant Constructs-Wild type full-length hPAPS synthase (GenBank TM accession number U53447) and mutant hPAPS synthase constructs were amplified by PCR using primers designed to contain BamHI restriction sites, cloned into BamHI digested pET-19b vectors containing a proprietary 122base pair NcoI-NdeI cassette (Veritas, Potomac, MD) encoding the calmodulin-binding site of calcineurin followed by a histidine tag and an enterokinase cleavage site, and used for transformation of Escherichia coli essentially as described previously (3). Plasmids were used for transformation of DH-5␣ competent E. coli cells by the CaCl 2 method. Transformants were isolated and miniprepped, and plasmids were sequenced for correct orientation of the initiator codon with respect to the T7 promoter sequence. The pET-19b vectors containing the correct inserts were isolated and used for transformation of expression host cells from Stratagene (BL21-DE3 plyz).
Site-selected Mutagenesis-The conserved purported NTP-binding motif (HXGH) located in the COOH-terminal region of human PAPS synthase was subjected to mutational analysis. Thus, amino acid substitutions were carried out in the HNGH sequence (amino acids 425-428). Site-selected mutagenesis was performed according to Stratagene's quik change mutagenesis kit. For example, oligonucleotides (54 nucleotides) containing the respective base substitutions were synthesized. Employing the wild type hPAPS synthase expression vector plasmid pET-19b, mutations were performed by PCR using the substituted primer and Pfu DNA polymerase. Thermal cycle phases consisted of either 12 cycles (single mutation) or 16 cycles (double mutation) of denaturation at 94°C for 30 s, annealing at 55°C for 1 min, and extension at 68°C for 12 min. After PCR, the methylated parent template plasmid was digested with the DpnI restriction enzyme, and the circular dsDNA was used to transform XL1-Blue supercompetent cells. Colonies were isolated, miniprepped, and sequenced. Verified mutant plasmids were used to transform pLyz BL21-DE3 overexpression bacteria that are deficient in proteases. Plasmids were isolated and sequenced to verify the presence of the proper mutation.
Preparation of Bacterial Cell Extracts-Colonies were grown in LB broth containing ampicillin to an A 595 nm of ϳ0.5, and IPTG was added to a final concentration of 1 mM to induce expression. Induction was carried out for 3 h. Cells were collected by centrifugation, the pellets were resuspended in 150 l of lysis buffer (20 mM Tris-HCl, pH 7.5, containing 50 mM KCl, 1 mM dithiothreitol, 10% glycerol, and 1.2 mg/ml lysozyme), and the cell suspension was transferred to a microcentrifuge tube. The original tube was washed with 100 l of lysis buffer, the wash was added to the cell suspension, cell lysis was carried out by incubation at 25°C for 7 min, and lysates were centrifuged for 15 min at 10,000 ϫ g at 4°C.
Purification of Histidine Fusion Proteins-Wild type hPAPS synthase and mutant proteins were purified using nickel columns. Bacteria were grown in LB broth to an A 595 nm of ϳ0.5, at which time 1 mM IPTG was added to induce expression as noted above. Cells were collected by centrifugation and lysed using lysis buffer as before, and the clear extract was mixed with 1 ml of slurry of nitrilotriacetic acid agarose (Qiagen, Chatsworth, CA). The contents were gently rotated on a wheel at 4°C for 1.5 h. The contents were then poured into small columns and washed with buffer A (20 mM Tris-HCl, pH 8.0, 100 mM KCl, 10% glycerol, 1 mM dithiothreitol) until the A 280 nm was less than 0.01. The columns were subsequently washed with six column volumes of buffer A containing 10 mM imidazole and eluted with 1 ml of buffer A containing 100 mM imidazole.
PAPS Synthase Assay-Enzyme activity was determined in a total volume of 10 l consisting of 3 l of sample, 3 l of reaction buffer (150 mM Tris-HCl, pH 8.0, 50 mM KCl, 15 mM MgCl 2 , 3 mM EDTA, and 45 mM dithiothreitol), 1 l of 50 mM ATP, and 3 l of inorganic [ 35 S]SO 4 (ϳ3.4 Ci). Reactions were carried out for 30 min at 37°C and stopped by placing the reaction tubes in boiling water for 5 min. Aliquots (1 l) were transferred to PEI-TLC plates and developed using 0.9 M LiCl as the solvent system. Following chromatography, the PEI-TLC plates were dried and exposed overnight to x-ray film (Eastman Kodak Co.). The respective spots for PAPS, APS, and SO 4 were excised, and the radioactivity was determined by liquid scintillation.
Western Blot Analysis-Expressed purified proteins (30 l from 1 ml peak fractions) were resolved by SDS-polyacrylamide gel electrophoresis using 12% polyacrylamide gels obtained from Novex, Inc. (San Diego, CA) and electroblotted onto Immobilon-P membranes from Millipore (Bedford, MA). Membranes were soaked in a solution of 0.1% bovine serum albumin in phosphate-buffered saline containing 0.1% Tween 20 (blocking buffer) for 45 min by gentle shaking, after which they were exposed to primary antibody (1:1000), which had been preadsorbed with E. coli extract (0.5 mg/ml), for 1 h. The primary antibody was generated against a COOH-terminal peptide fragment of human PAPS synthase as previously reported (3). Blots were washed three times with wash buffer (1ϫ phosphate-buffered saline with 0.01% Tween 20) and incubated with the secondary antibody in blocking buffer (1:5000) for an additional 1 h. Finally, the membranes were subjected to three washes in wash buffer and two additional washes in reaction buffer (0.1 M Tris-HCl, pH 8.0). The blots were developed by color reaction according to the manufacturer's instructions (Kirkegaard and Perry Laboratories, Inc., Gaithersburg, MD).
Amino Acid Sequence Alignment Analyses-Searches of the protein sequence data base (the nonredundant data base at the NCBI) were performed using the gapped BLAST program, the position-specific iterative BLAST (PSI-BLAST), and the pattern-hit initiated BLAST (PHI-BLAST) programs (10,12). Multiple alignments were constructed using either the Clustal W program (13) or the MACAW program (14,15).

Amino Acid Sequence Analyses
Initially, the presence of the HX(G/A)H motif that is strictly conserved in the COOH-terminal, ATP sulfurylase domain of the PAPS synthases as well as ATP sulfurylases was noticed by visual inspection of the data base search outputs and multiple alignments of this domain (data not shown). Data base searches started with the PAPS synthase sequences because the queries failed to detect significant similarity to any members of the TagD superfamily, although limited similarity to some of them was detectable when the search was performed using the PHI-BLAST program that combined BLAST analysis with the detection of the HX(G/A)H pattern (12). By contrast, in a search initiated with the sequence of the nucleotidylyltransferase domain of the NadR protein from Mycobacterium tuberculosis using the PSI-BLAST program (10), members of the PAPS synthase superfamily were retrieved at a statistically significant level (p Ͻ 0.01) within 8 -10 search iterations (10,16). Although quite subtle (11% identity on a 189-amino acid overlap), the alignment between NadR and PAPS synthases encompassed almost the whole nucleotidylyltransferase domain (data not shown). Multiple alignment analysis using the MACAW program (14,15) showed that not only the HX(G/A)H but also the surrounding hydrophobic positions that form the basis of a predicted ␤-loop-␣ structural unit were conserved in the PAPS synthases and the TagD superfamily (Fig. 1).

Expression of Wild Type and Mutant hPAPS Synthase
Western Blotting-Expression of partially purified wild type and mutant protein preparations was clearly demonstrated (Fig. 2). The level of expression of the mutant construct N426A was consistently higher than the other constructs in three independent experiments, whereas the level of expression for all other constructs was nearly equal. In a mock control, i.e. vector without a hPAPS synthase insert, no protein expression was observed (Fig. 2).
Catalytic Activity-Vectors (wild type and mutants) not under the control of promoter induction by IPTG, as well as mock controls, exhibited no demonstrable catalytic activity.
Relative PAPS synthase specific activities in cellular extracts for wild type hPAPS synthase, the mutant constructs, and the mock control are presented in Fig. 3. The mutant constructs H425A, H428A, and G427A/H428A exhibited only background activity similar to that of the mock control. On the other hand, the mutant construct N426K formed 185.8 pmol PAPS/min⅐mg, which was about 2-fold higher than the activity of 96.8 pmol PAPS/min⅐mg demonstrated by wild type hPAPS synthase. The G427A mutant formed 72.2 pmol PAPS/min⅐mg, a value slightly less than wild type. DISCUSSION Sequence analysis has shown the relatedness of the nucleotidylyltransferase superfamily and TagD and revealed a connection between this category of enzymes and the class I aminoacyl-tRNA synthetases whose structure is known (11). Based on crystal structure for aminoacyl-tRNA synthetases, structural predictions for the nucleotidylyltransferases mapped a conserved HXGH motif to a position analogous to the location of the ATP-binding HIGH motif in the tRNA synthetases (11). Thus, a structural and functional link was proposed for the TagD superfamily and the class I aminoacyl-tRNA synthetases with the HXGH motif in the cytidylyltransferase superfamily functioning as a NTP-binding site analogous to the HIGH motif serving as an ATP-binding site for the tRNA synthetases. Interestingly, the HXGH motif is completely conserved in cloned bifunctional PAPS synthase proteins, i.e. the marine worm (5), mouse (4,17), human (3), guinea pig, 2 and fruit fly (6) as well as the ATP sulfurylases from fungi (18,19), plants (20,21), and yeast (22). Importantly, the ATP sulfurylase domain of bifunctional PAPS synthase is located in the COOH-terminal region of the protein (3), and it is in this segment of PAPS synthase that the HXGH motif is located.
A unifying functional feature of the TagD nucleotidylyltransferase superfamily and the aminoacyl-tRNA synthetases is their ␣-␤ phosphodiesterase activity. It has been proposed that the conserved HXGH motif in this superfamily is involved in NTP binding (23), an hypothesis strengthened by recent mutational analyses (24). Our finding that mutating either H425A or H428A of the HXGH motif in hPAPS synthase produced an inactive enzyme is comparable with the results of similar mutations carried out with CTP:phosphocholine cytidylyltransferase (24). The G427A mutation of hPAPS synthase resulted in approximately a 30% reduction in overall enzymatic activity, a finding also comparable with a similar mutation carried out with CTP:phosphocholine cytidylyltransferase (24). Thus, as with the TagD superfamily, mutational studies with hPAPS synthase support the proposition that the highly conserved HXGH motif in the ATP sulfurylase domain of PAPS synthases is involved in NTP binding.
Crystal structures of aminoacyl-tRNA synthetases indicate that the HIGH sequence is located in the first loop of a ␤-␣-␤ unit, which allows the imidazole nitrogens to form hydrogenbinding contacts with the phosphate oxygens of ATP (11). Mutational studies of the HXGH motif in CTP:phosphocholine cytidylyltransferase also implicate the involvement of similar contacts with CTP (24). It follows that the results of the mutational studies involving the HXGH motif in hPAPS synthase should be analogously interpreted. Whereas the H425A and H428A mutants of hPAPS synthase produced inactive en- zymes, only a modest reduction (ϳ30%) in PAPS synthase activity was noted with the G427A mutant compared with the wild type enzyme. On the other hand, the N426K mutant yielded an increase in PAPS synthase activity, suggesting that the ⑀-NH 3 group of lysine creates a more efficient enzyme than the amide NH 2 of asparagine, perhaps by creating enhanced hydrogen contact with ATP.
During PAPS formation by bifunctional PAPS synthase (ATP sulfurylase/APS kinase) two molecules of ATP are required, one for each reaction. Site-selected mutagenesis of a P-loop motif (GXXGXGK(T/S)) located in the NH 2 -terminal, APS kinase domain of mouse PAPS synthase markedly impaired APS kinase activity (9), a finding in keeping with the purported role of the NTP-binding P-loop motif in the cleavage of the ␤-␥ phosphodiester bond of ATP and transfer of the terminal phosphoryl group to the 3Ј-hydroxyl position of APS. In contrast to the NH 2 -terminal, APS kinase domain, the COOH-terminal, ATP sulfurylase domain of PAPS synthase, acting as an ␣-␤ phosphodiesterase during the formation of APS from ATP and inorganic sulfate, requires a different type of NTP-binding site. Based on the results of site-selected mutagenesis of hPAPS synthase, we propose that the domain homologous to the TagD superfamily nucleotidylyltransferases and in particular, the completely conserved HXGH motif located in the ATP sulfurylase domain of all bifunctional PAPS synthases fulfills such a function.