Bacterial expression, folding, purification and characterization of soluble NTPDase5 (CD39L4) ecto-nucleotidase

Biochimica et Biophysica Acta 1747 (2005) 251 – 259 http://www.elsevier.com/locate/bba

Bacterial expression, folding, purification and characterization of soluble NTPDase5 (CD39L4) ecto-nucleotidaseB Deirdre M. Murphy-Piedmonte1, Patrick A. Crawford, Terence L. Kirley* Department of Pharmacology and Cell Biophysics, College of Medicine, University of Cincinnati, P.O. Box 670575, 231 Albert Sabin Way, Cincinnati, OH 45267-0575, United States Received 4 February 2004; received in revised form 24 August 2004; accepted 22 November 2004 Available online 8 December 2004

Abstract The ecto-nucleoside triphosphate diphosphohydrolases (eNTPDases) are a family of enzymes that control the levels of extracellular nucleotides, thereby modulating purinergically controlled physiological processes. Six of the eight known NTPDases are membrane-bound enzymes; only NTPDase 5 and 6 can be released as soluble enzymes. Here we report the first bacterial expression and refolding of soluble human NTPDase5 from inclusion bodies. The results show that NTPDase5 requires the presence of divalent cations (Mg2+ or Ca2+) for activity. Positive cooperativity with respect to hydrolysis of its preferred substrates (GDP, IDP and UDP) is observed, and this positive cooperativity is attenuated in the presence of nucleoside monophosphate products (e.g., GMP and AMP). In addition, comparing the biochemical properties of wild-type NTPDase5 and those of a mutant NTPDase5 (C15S, which lacks the single, non-conserved cysteine residue), also expressed in bacteria, suggests that Cys15 is not essential for either proper refolding or enzymatic activity (indicating this residue is not involved in a disulfide bond). Moreover, the substrate profile of bacterially expressed NTPDase5, as well as the C15S mutant, was determined to be similar to that of full-length, membrane-bound and soluble NTPDase5 expressed in mammalian COS cells. D 2004 Elsevier B.V. All rights reserved. Keywords: Nucleoside triphosphate diphosphohydrolase; NTPDase5; CD39L4; Bacterial expression; Protein refolding; Enzymatic characterization; Positive cooperativity

1. Introduction The ectonucleoside triphosphate diphosphohydrolases (eNTPDases) are a family of enzymes that hydrolyze Abbreviations: NTPDases, nucleoside triphosphate diphosphohydrolases; NTPDase5, nucleoside triphosphate diphosphohydrolase type 5; ACRs, apyrase conserved regions; B-PER, bacterial protein extraction reagent; MWCO, molecular weight cut off; Pi, inorganic phosphate; MOPS, 3-[N-morpholino]propanesulfonic acid; IPTG, isopropyl-h-d-thiogalactopyranoside; DTT, dithiothreitol; DMEM, Dulbecco’s modified Eagle medium; NTA, nitrilotriacetic acid; KLH, keyhole limpet hemocyanin; TBS, Tris-buffered saline; SEC, size exclusion chromatography B The sequence of the cDNA encoding the NTPDase5 expressed in this study has been submitted to GenBank and assigned accession no. AY430094. * Corresponding author. Tel.: +1 513 558 2353; fax: +1 513 558 1169. E-mail address: [email protected] (T.L. Kirley). 1 Current address: Amgen, One Amgen Center Drive, Thousand Oaks, CA 91320-1799, United States. 1570-9639/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.bbapap.2004.11.017

nucleotides and require divalent cations (e.g. Ca2+ or Mg2+) for activity. These enzymes share seven highly conserved regions of homology named apyrase conserved regions (ACRs) [1–4], which are essential for catalytic activity. The six well-established enzymes, NTPDase1–6, consist of four membrane-bound enzymes and two soluble, secreted enzymes [5]. NTPDase5 and NTPDase6 can be extracellularly released as soluble enzymes following cleavage of their respective N-terminal signal sequences [6–11]. Additionally, two other NTPDases have recently been described, NTPDase7 (also known as LALP1 [12]) and NTPDase8 [13]. NTPDase8 is a novel plasma membrane enzyme cloned and characterized from mouse, which is highly homologous to the previously characterized [14,15] and cloned [16] chicken ecto-ATPDase. NTPDase5 (CD39L4) was first identified as a member of the CD39/NTPDase family in 1998 [17]. Later, NTPDase5 was shown to be identical to PCPH [18], a human proto-

252

D.M. Murphy-Piedmonte et al. / Biochimica et Biophysica Acta 1747 (2005) 251–259

oncogene product that was discovered after its activation upon treatment with a chemical carcinogen [19]. To date, two studies on the characterization of bNTPDase5Q have been published [9,10]. Additionally, a study on an enzyme termed the bER-UDPaseQ, a soluble, mammalian, endoplasmic reticulum (ER) localized nucleoside diphosphatase, has been published [11]. This mouse liver ER-UDPase (GenBank Accession No. AJ238636 [11]) is approximately 88% identical to human NTPDase5 at the amino acid level. Interestingly, these previous studies have produced some seemingly contradictory information about NTPDase5, especially in relation to its hydrolytic activity towards ADP. The primary sequence of the soluble portion of NTPDase5 encodes five cysteine residues, four of which are conserved amongst the NTPDases. In NTPDase6, the other soluble NTPDase, it was shown that these four conserved cysteine residues form two disulfide linkages [7], which are very likely to be conserved in other NTPDases, including NTPDase5. This suggests that the single, non-conserved cysteine residue in the soluble NTPDase5 (C15) is not involved in a disulfide bond. Physiologically, the NTPDases are involved in many important functions including smooth muscle contraction, pain perception, and the modulation of platelet aggregation (for a review, see [20]). The NTPDases function in these processes by modulating the concentrations of nucleotide agonists in the extracellular space, thus regulating the purinergic receptors in control of each respective process. Recent studies have focused on the therapeutic potential of these enzymes. Hydrolysis of ADP (that regulates platelet aggregation and blood clotting) by the soluble portion of NTPDase1 has been shown to decrease damage caused by ischemic stroke [21]. Thus, naturally occurring, soluble nucleotidases such as the NTPDase5 enzyme studied in this work may be useful clinically, and serve as a starting point for the optimization of enzymatic properties via mutagenesis, in order to generate enzymes with modified nucleotide specificities that may be more useful as therapeutic proteins. To this end, Dai et al. [22] have recently reported that site-directed mutagenesis of an unrelated nucleotidase, the human soluble calcium activated nucleotidase (SCAN), can be used to engineer a protein much better suited as an anticoagulant protein than the naturally occurring protein. To accomplish this goal for the NTPDases, a better biochemical understanding of this class of soluble nucleotidases is necessary, and thus, this study strives to address that goal for the soluble NTPDase5/CD39L4 protein.

2. Materials and methods 2.1. Materials The QuikChange site-directed mutagenesis kit and Escherichia coli XL-1 Blue competent cells were purchased

from Stratagene. The DNA Core Facility at the University of Cincinnati produced the synthetic oligonucleotides and performed DNA sequencing. Quality Control Biochemicals performed the peptide synthesis and conjugation to keyhole limpet hemocyanin (KLH). Rabbit antisera to the NTPDase5 C-terminal synthetic peptide ((C)ALGATFHLLQSLGISH) conjugated to KLH (via its N-terminal cysteine) were generated by Lampire Biologicals. Plasmid purification kits and Ni-NTA agarose were purchased from Qiagen. NheI, XhoI, and HindIII restriction endonucleases, T4 DNA ligase, and the mammalian expression vector pcDNA3 were obtained from Invitrogen. The bacterial expression vector pET28a and the expression host E. coli BL21 (DE3) and BL21 (DE3) Star were purchased from Novagen. Glycerol and dialysis tubing were from Fisher. Bacterial protein extraction reagent (BPER), Enhanced Chemiluminescent Reagents, and bovine serum albumin (BSA) standard were purchased from Pierce. 4–15% acrylamide Tris–Glycine gels, the S-200 size exclusion chromatography (SEC) matrix, SDS, SEC molecular weight standards, protein assay reagent, and the anion-exchange QMA cartridge were from Bio-Rad. DMEM was a generous gift from Deb Marsh at Hyclone and subsequently has been purchased from Gibco BRL/Invitrogen. Lipofectamine, Plus reagents and antibiotics/antimycotics were from Gibco BRL/Invitrogen. Kanamycin, nucleotides, IPTG, glucose, DTT, oxidized glutathione, reduced glutathione, the protease inhibitor cocktail and other reagents were from Sigma. 2.2. NTPDase5 cloning and sequencing NCBI BLAST searches of human EST databases identified an IMAGE clone (ID# 4817229) that was highly homologous to CD39L4 (NTPDase5). The clone was purchased from ATCC and completely sequenced (with T7, T3, and three custom-synthesized internal primers). This 1997 base pair IMAGE clone contained the full-length open reading frame encoding CD39L4 (NTPDase5), and was submitted to GenBank under accession number AY430094. Computer analysis of the NTPDase5 insert predicted that cleavage of the signal peptide would occur after Arg in the sequence SAVSHRNQQTWF. In order to express only the soluble portion of NTPDase5 in the bacterial expression vector, pET28a two endonuclease sites flanking the coding region of the soluble insert (which begins with amino acid residues NQQTWF. . .) were introduced with site-directed mutagenesis. To allow cloning of NTPDase5, a 5V NheI restriction site (indicated by boldface type) was introduced before the start of the soluble sequence with the following primer: 5V GTCTCCCACAGGGCTAGCAACCAGCAGACTTG3V. A 3V XhoI site (indicated in boldface) was introduced after the stop codon using the following primer: 5V GGCCACGTACTTCCTTGCTCGAGACCTGCATTTGCC3V. The mutated NTPDase5 insert was digested with NheI and XhoI (1 h at 37 8C) and ligated into the pET28a bacterial expression

D.M. Murphy-Piedmonte et al. / Biochimica et Biophysica Acta 1747 (2005) 251–259

vector (digested with the same endonucleases). This NTPDase5 cDNA construct in pET28a was used to transform non-expression E. coli XL-1 Blue bacterial cells, allowing construct confirmation by sequencing. 2.3. Expression of NTPDase5 in bacterial BL21(DE3) cells The NTPDase5 insert in the pET28a vector was transformed into BL21(DE3) or BL21(DE3)Star E. coli cells for bacterial expression. 50 AL of frozen NTPDase5 bacterial stock (in BL21(DE3) or BL21(DE3)Star cells) was used to inoculate 5 mL of LB, containing 30 Ag/AL kanamycin and 1% glucose (glucose was added to reduce the background, non-IPTG-induced, bacterial expression of NTPDase5), and grown to an OD600 of approximately 0.6. A 50-AL aliquot of this bacterial culture was used to inoculate 50 mL of LB/ kanamycin/glucose broth, and allowed to incubate overnight at 37 8C until the culture reached an OD600 of 1.0. This culture was added to 500 mL of LB/kanamycin/glucose broth and grown to an OD600 of approximately 0.6. Expression was induced with 0.5 mM (final concentration) IPTG. The culture was allowed to grow for 4 h, at 37 8C, post induction. Inclusion body preparations were performed as previously described [23], with minor modifications. Briefly, 10 mM EDTA was added to the Pierce BPER reagent during the initial solubilization of the bacterial pellet, and subsequent extractions were performed using a 1:10 dilution of BPER containing a final concentration of 10 mM EDTA. EDTA was used to limit proteolysis caused by bacterial proteases. In addition, a strain of bacterial expression cells BL21(DE3)Star, which lacks two common proteases (OmpT protease and Imp protease), was used in an attempt to further limit NTPDase5 proteolysis during refolding and purification.

253

dialysis is required to decrease the arginine concentration that interferes with the hexa-histidine tag binding to a NiNTA column. After dialysis, the approximately 350 mL sample was added to 3 mL (settled bead volume) of washed Ni-NTA agarose beads, incubated with end-overend rotation for 2 h, and poured onto a column. The column was washed with seven 4 mL aliquots of equilibration buffer (50 mM Tris–HCl, 250 mM NaCl, and 2 mM CaCl2, pH 8.0). NTPDase5 was eluted from the Ni-NTA column with 12 mL of 200 mM imidazole in equilibration buffer, and the sample was immediately diluted into 18 mL of ice-cold equilibration buffer, to reduce the imidazole concentration. 2.5. Thrombin cleavage and subsequent anion exchange chromatography of NTPDase5 After elution from the Ni-NTA column and dilution, 3 units of thrombin were added to the NTPDase5 sample, and the solution was dialyzed overnight at 4 8C against 2 L of 50 mM Tris, 150 mM NaCl, 2 mM CaCl2, pH 8.0. In preparation for anion exchange chromatography, NTPDase5 was further dialyzed at 4 8C against 4 L of 20 mM Tris– HCl, pH 8.0 for 4–6 h. After dialysis, the sample was clarified upon centrifugation in a Beckman JA20 rotor at 18,000 rpm for 30 min at 4 8C. The supernatant was loaded onto a 1 mL strong anion exchange (Bio-Rad QMA) cartridge equilibrated with approximately 40 mL of 20 mM Tris–HCl, pH 8.0. The cartridge was washed twice with 5 mL of the equilibration buffer. The sample was eluted at 1.0 mL/min with a gradient of 0–500 mM NaCl in 20 mM Tris–HCl, pH 8.0. Fractions were collected every 2 min. NaCl eluted fractions, along with the unbound and wash fractions, were assayed for Ca2+ GDPase activity. The most active fractions were pooled.

2.4. Preparative refolding and purification of NTPDase5 from inclusion bodies

2.6. Site-directed mutagenesis of the non-conserved cysteine residue in NTPDase5

Refolding of NTPDase5 was accomplished by dilution, similar to the refolding of the other soluble ecto-nucleotidase, NTPDase6 [7]. Specifically, 5 mg of NTPDase5 inclusion body preparation (at a concentration of 1 mg/mL) was solubilized in 6 M Guanidine–HCl, 5 mM EDTA, and 5 mM DTT in 100 mM Tris–HCl, and denatured by heating for 10 min at 60 8C, then allowed to cool to room temperature. Refolding was performed by dilution of the 5 mL sample into 250 mL of sterile-filtered buffer containing 150 mM Tris–HCl, 10 mM KCl, 250 mM NaCl, 2 mM CaCl2, 2 mM MgCl2, 800 mM L-arginine–HCl, 2 mM reduced glutathione (GSH), and 0.2 mM oxidized glutathione (GSSG), pH 8.3, with stirring. Refolding proceeded for approximately 3 days at 13 8C (without stirring). After refolding, the NTPDase5 solution was dialyzed three times against 4 L of 50 mM Tris–HCl, 250 mM NaCl, and 2 mM CaCl2 (pH 8.0) at 4 8C, over a period of 24 h. This extensive

Cys 15, the sole non-conserved cysteine residue in NTPDase5 (numbered from the start of the soluble sequence NQQ. . .) was replaced with serine via the QuikChange sitedirected mutagenesis kit using the following sense primer: 5V CCTGTCTTCCATGAGCCCCATCAATGTCAGC 3V. The codon responsible for the mutation is indicated in boldface. The anti-sense primer also required for mutagenesis is not shown. The presence of the desired mutation and lack of unwanted mutations were confirmed by DNA sequencing. This cysteine to serine mutation was made in both the pET28a construct used for bacterial expression and in the pcDNA3 construct used for mammalian expression of NTPDase5. 2.7. Protein assay The concentration of NTPDase5, both wild-type and C15S, was determined according to the Bio-Rad CB-250

254

D.M. Murphy-Piedmonte et al. / Biochimica et Biophysica Acta 1747 (2005) 251–259

dye binding technique following the modifications of Stoscheck [24,25] using bovine serum albumin as a standard. 2.8. Size exclusion chromatography of NTPDase5 A 46 mL, 27-cm-long Sephacryl S-200 size exclusion chromatography column, prepared as previously described [26], was equilibrated with Tris-buffered saline (TBS). 2 Ag of NTPDase5 sample in 0.5 mL TBS was chromatographed at a flow rate of 0.5 mL/min, and fractions were collected every 2 min. The apparent molecular size of the bacterially expressed, refolded, and purified NTPDase5, as determined by fractions exhibiting Ca2+-GDPase activity, was determined by comparison with a calibration plot generated using soluble proteins of known molecular mass. 2.9. Mammalian expression of NTPDase5 in COS-1 cells The NTPDase5 clone in pBluescript vector was modified to allow re-ligation into the pcDNA3 mammalian expression vector. This mammalian expression construct differs from the construct in the bacterial expression vector in that it contains the putative signal peptide, as well as the soluble portion of the NTPDase5 sequence. Including the signal peptide, the mammalian expression clone encodes a 428residue protein, with a calculated molecular weight of 47.5 kDa. For mammalian expression (in the pcDNA3 vector), a HindIII site was introduced by site-directed mutagenesis in the 5V portion of the insert (prior to the initiation Met) by the mutation of a single base pair with the following primer: 5V GCTTCTGCAACAAAAGCTTCCACCCAGCCAC 3V. The HindIII site is indicated in boldface, and the single mutated base pair is italicized. The anti-sense primer also required for mutagenesis is not shown. Following digestion with 5V HindIII site and 3V XhoI site, the NTPDase5 cDNA insert was ligated into pcDNA3 vector that had been digested with the same enzymes. As described earlier, the sole nonconserved cysteine residue (Cys15) of NTPDase5 in pcDNA3 vector was replaced by serine. Both wild-type NTPDase5 and C15S mutant pcDNA3 constructs were verified by DNA sequencing. The transient transfection of COS-1 cells with NTPDase5 was performed as previously described for NTPDase6 [8] with minor modifications. COS cells were transfected with 4 Ag of NTPDase5 cDNA in the presence of serum-free media, and incubated for 5 h at 37 8C before the addition of 13 mL of DMEM containing fetal bovine serum and antibiotic/antimycotic agents. 24 h post transfection, the cell media was changed to serum-free DMEM, and to some preparations a 1:200 dilution of Sigma’s protease inhibitor cocktail was added. The cells and media were harvested 48 h post transfection; cell media was removed and saved, while the membranes were prepared as previously described [8]. Both the media and membrane preparations were assayed for nucleotidase activity.

2.10. Nucleotidase assay Nucleotidase activity was determined by measuring the amount of inorganic phosphate released from nucleotide substrates at 37 8C using a modification of the technique of Fiske and Subbarow [27] as previously described [28]. Nucleotide-hydrolyzing units are reported in micromoles of Pi liberated per milligram of protein per hour. Activity was measured using a nucleotide concentration of 2.5 mM, in 20 mM MOPS, pH=7.1, in the presence of either 5 mM CaCl2 or 5 mM MgCl2, as indicated. Assays for determining K m were performed using the more sensitive phosphate assay utilizing a malachite green reagent [29], as previously described [7]. K m and V max values were determined by fitting data using the Origin 7 computer program, to either the standard Michaelis–Menten hyperbolic equation (V=(V max[S])/K m+[S]), or in the case of the sigmoidal curves arising from positive cooperativity, with the Hill equation (V=(V max[S]n )/K m+[S]n ), where n is the Hill coefficient. Means and standard errors for K m and V max values were obtained from multiple, independent experiments and fits. 2.11. Electrophoresis and Western blot analysis SDS-PAGE was performed following the method of Laemmli [30], using 4–15% pre-cast gradient gels. The NTPDase5 samples were boiled for 5 min in SDS sample buffer with or without 30 mM dithiothreitol before SDSPAGE. Native gel electrophoresis was performed as previously described [23], using 4–15% pre-cast gradient Laemmli gels, without SDS in any component. Prior to electrophoresis, the samples were diluted with a solution of 125 mM Tris–HCl, pH 6.8, buffer saturated with sucrose, and, without heating, were loaded onto the gel and electrophoresed until after the Bromphenol Blue tracking dye reached the bottom of the gel. Western blot analyses were performed as previously described [8], using an affinity-purified antibody raised against the Cterminal NTPDase5 peptide.

3. Results 3.1. Cloning, bacterial expression, refolding and purification of NTPDase5 The cDNA sequence encoding the soluble part of NTPDase5 was cloned into the pET28a expression vector and expressed in bacteria. The resultant bacterial inclusion bodies were purified as previously described for NTPDase6 [7]. Several parameters were evaluated to determine their effect on the efficiency of refolding of NTPDase5. An alkaline pH, the presence of a high concentration of l-arginine, and the redox pair of oxidized and reduced glutathione were each essential

D.M. Murphy-Piedmonte et al. / Biochimica et Biophysica Acta 1747 (2005) 251–259

for the successful refolding of NTPDase5 into an active nucleotidase. Previously, two soluble human nucleotidases have been successfully refolded from inclusion bodies in our laboratory (NTPDase6 [7] and human soluble calcium activated nucleotidase-1 (SCAN) [23]). The refolding and purification of NTPDase5 presented three problems that had not been encountered with the aforementioned nucleotidases. The first was the susceptibility of refolded NTPDase5 to proteolysis. Proteolysis was apparent from SDS-PAGE analysis of the refolded and purified NTPDase5 in early experiments (data not shown), and was eliminated by adding EDTA during the inclusion body preparation, suggesting the proteolysis was due to metalloproteases originating from the expressing bacteria. Second, major losses (80–90%) of both protein and activity occurred during centrifugal concentration of purified, bacterially expressed wild-type and C15S NTPDase5. Loss upon concentration was also observed for the small amount of NTPDase5 secreted into the media by COS cells after mammalian cell transfection. No such losses were observed for either bacterially expressed NTPDase6 [7] or SCAN [23] nucleotidases. Although the reason for this protein loss upon concentration is unclear, the purification of refolded NTPDase5 was designed to eliminate all centrifugal concentration steps. Third, the bacterially expressed, refolded, and purified NTPDase5 described in this work lost enzyme activity upon storage faster than the corresponding bacterially expressed, refolded and purified NTPDase6 [7]. Starting from 5 mg of EDTA-treated inclusion bodies, typically 0.4–0.5 mg of pure wild-type or C15S NTPDase5 was obtained with a Ca2+-GDPase activity of about 40,000– 50,000 Amol/mg/h (yields are similar to those obtained with bacterially expressed, refolded, and purified NTPDase6 [7]). After refolding, the active NTPDase5 protein was purified by metal affinity chromatography, the hexa-histidine tag was removed by thrombin digestion, and the protein was further purified by ion exchange chromatography, as was described previously for NTPDase6 [7]. The resultant pure protein was analyzed by reducing and nonreducing SDS-PAGE (lanes 2 and 3 of Fig. 1A), with the apparent molecular weight as judged by SDSPAGE (42 kDa) being consistent with the 45.5 kDa theoretical molecular weight of the NTPDase5 construct in pET28a (after thrombin cleavage). Western blot analysis of bacterially expressed NTPDase5 with an affinity-purified antibody raised against its C-terminal peptide [8] showed an immunoreactive band corresponding to the same molecular weight as seen in the Coomassie-stained gel in Fig. 1A (lane 4). Native gel electrophoresis of purified NTPDase5 is seen in lane 2 of Fig. 1B, and indicates the conformational homogeneity of the preparation. The more highly negatively charged, and therefore faster migrating, bovine serum albumin protein

255

Fig. 1. SDS-PAGE and native gel analysis of refolded, purified wild-type NTPDase5. Panel A—SDS-PAGE. The samples were denatured in Laemmli [30] sample buffer (either with or without the reductant DTT) prior to loading on a 4–15% gradient gel. For lanes 1–3, the gel was stained with Coomassie blue, while lane 4 shows a Western blot of the bacterially expressed, reduced protein, probing with an affinity-purified polyclonal antibody against the C-terminus of NTPDase5. Molecular mass standards are in lane 1. Lanes 2 and 3 show 1 Ag of reduced and nonreduced NTPDase5, respectively. Panel B—Native gel electrophoresis. The samples were diluted with 125 mM Tris–HCl, pH 6.8 buffer saturated with sucrose, and, without heating, were loaded onto a 4–15% acrylamide Laemmli gradient gel in the absence of SDS. The gel was stained with Coomassie blue. BSA (10 Ag in lane 1) was used for comparative purposes (note the darker stained monomer and the lighter stained dimer of BSA present). Wild-type NTPDase5 (2 Ag) is shown in lane 2.

(Fig. 1B, lane 1) was included for comparative purposes. Size exclusion chromatography of bacterially expressed wild-type and C15S mutant NTPDase5 indicated an apparent molecular weight of 42 kDa (data not shown), consistent with a monomeric protein, and in agreement with the theoretical molecular weight (45.5 kDa) of the NTPDase5 construct in pET28a, as well as with the SDSPAGE results shown in Fig. 1A. 3.2. Substrate preference and kinetics experiments Fig. 2 shows the nucleotidase activities, relative to GDPase, of the bacterially expressed and purified NTPDase5. Using 5 mM calcium as the divalent cation and a nucleotide concentration of 2.5 mM, the rank order of hydrolysis is GDPcIDPNUDPNNITPNGTPNUTPNADPc CDPNCTPNATPc0. This rank order agrees for the most part with the mammalian COS cell membranes expressing soluble wild-type enzyme (GDP(100%)cIDP(91%)NUDP (80%) NNITP(7.4%)cGTP(4.8%)cUTP(8.4%)). The data are normalized to the GDPase activity in crude COS cell media of 116 Amol/mg/h (100%). Other nucleotidase activities (e.g., ATPase, ADPase, CTPase, etc.) are too low to measure accurately in the COS cell expressed cell media. Thus, the preferred substrates (GDP, IDP, and UDP) are the same for both COS cell and bacterial cell expressed NTPDase5, with the bacterially expressed NTPDase5

256

D.M. Murphy-Piedmonte et al. / Biochimica et Biophysica Acta 1747 (2005) 251–259

Fig. 2. Substrate preference of bacterially expressed and refolded wild-type NTPDase5. The nucleotidase profile of NTPDase5 was generated by measuring activity in the presence of various nucleotide substrates, all at an initial concentration of 2.5 mM, in 20 mM MOPS, pH 7.1, in the presence of 5 mM CaCl2. Three experiments were performed (each in duplicate), and the nucleotidase activities are reported as the meansFstandard deviations (S.D.), relative to the maximal activity measured (GDPase). The maximum activity (100% in the figure, for GDPase activity) was 46,979F1578 Amol/ mg/h for this preparation of NTPDase5.

ADPase activity being only about 3.5% of the GDPase activity under these conditions. Data obtained by varying the GDP concentration and measuring the bacterially expressed NTPDase5 GDPase activity were not fit adequately by the Michaelis–Menten hyperbolic equation normally used to calculate K m. Instead, the Ca2+-GDPase data were best fit by a modified kinetic model including a Hill coefficient to account for the observed positive cooperativity (Fig. 3, solid curve). Similar

Fig. 3. Rate of GDP hydrolysis as a function of substrate concentration for wild-type NTPDase5 and NTPDase6. Ca2+-GDPase enzyme activity assays were carried out in 5 mM CaCl2, 2.5 mM GDP (initial concentration), and 20 mM MOPS (pH 7.1). For ease of visual comparison, the data for NTPDase6 (CD39L2) was normalized by dividing all data points by 1.6, to make the maximum of both curves similar. Note the positive cooperativity for NTPDase5 that is not present for NTPDase6 assayed in parallel under the same conditions.

positive cooperativity was also observed with other preferred substrates for NTPDase5 (IDP and UDP), but not with ADP, a relatively poor substrate for NTPDase5 (data not shown). These experiments were performed in parallel with experiments utilizing bacterially expressed and refolded NTPDase6 (CD39L2) [7], as a methodological control and a negative control for cooperativity. The NTPDase6 protein used was prepared and assayed in precisely the same manner as the NTPDase5 protein, i.e., it was bacterially synthesized, isolated from inclusion bodies, refolded using exactly the same protocol, purified, and assayed for enzyme activity using all the same techniques as used for NTPDase5. This negative control (NTPDase6) data was well fit by a simple Michaelis– Menten hyperbolic equation, with no sign of positive cooperativity (Fig. 3, dashed curve), virtually eliminating the possibility of the NTPDase5 positive cooperativity being due to some impurity or other artifact of preparation or purification. In an effort to find an effector/activator of this positive cooperativity, GMP was chosen, since it is a product of the GDP hydrolysis reaction and GMP was not hydrolyzed by NTPDase5 in control experiments (data not shown). Inclusion of GMP resulted in a concentration-dependent increase of NTPDase5 Ca2+-GDPase activity at low GDP concentrations (Fig. 4, inset). GMP inclusion over the full range of GDP concentrations used attenuated the degree of positive cooperativity (Fig. 4, filled circles), and yields data that is better fit by the Michaelis–Menten hyperbolic equation (Fig. 4, dashed upper curve). A similar attenuation of the positive cooperativity was also observed with AMP (data not shown). From the data obtained for Ca2+-GDPase activity

Fig. 4. Rate of GDP hydrolysis as a function of substrate concentration for wild-type NTPDase5 in the presence and absence of 2 mM GMP. The inset shows the dependence of NTPDase5 Ca2+-GDPase activity (at a GDP concentration of 0.05 mM) on the concentration of GMP added. Enzyme activity assays were carried out in 5 mM CaCl2 and 20 mM MOPS (pH 7.1). In control experiments, no hydrolysis of GMP by NTPDase5 was detectable. Note that the presence of the reaction product, GMP, decreases the extent of positive cooperativity and allows the data to be better fit by the standard Michaelis–Menten hyperbolic equation (dashed line).

D.M. Murphy-Piedmonte et al. / Biochimica et Biophysica Acta 1747 (2005) 251–259

Fig. 5. The pH dependence of the hydrolysis of GDP by wild-type NTPDase5 in Ca2+ and Mg2+ MOPS buffers. 2.5 mM GDP was used as the substrate for NTPDase5 in buffers containing 50 mM MOPS or 50 mM Tris and either 5 mM Ca2+ (open circles) or 5 mM Mg2+ (filled circles) at the pH values indicated. The nucleotidase activity is reported as the mean of triplicate measurementsFstandard error (S.E.).

using wild-type NTPDase5 in the absence of GMP (see Fig. 4, solid lower curve) the calculated enzymatic parameters are V max=34,000F300 Amol/mg/h, K m=0.14F0.01 mM GDP, and a Hill coefficient of 1.8F0.1. While the pH profile of NTPDase5 activity has never been reported, there is data that suggests CD39L4 (NTPDase5) has substantially more activity in the presence of Mg2+ versus Ca2+ [10]. Additionally, the ER-UDPase was shown to be substantially more active in the presence of Mg2+ versus Ca2+ when UDP is used as the substrate [11]. Therefore, the GDPase activity of the bacterially expressed NTPDase5 was determined in both Ca2+ and Mg2+ buffers as a function of pH (Fig. 5). As seen in Fig. 5, NTPDase5 has approximately 1.6fold more activity in the Mg2+-containing buffers than in Ca2+-containing buffers below pH 8. The activity in both divalent cation-containing buffers decreases fairly steeply with increasing pH above pH 8.0. For Mg2+-GDPase, maximal activity for NTPDase5 was observed at a pH of 7.25.

4. Discussion Through BLAST searches of human EST databases, an IMAGE clone that had been isolated from the hippocampus of the human brain and that was highly homologous to NTPDase5 (CD39L4 [17]) was identified. The IMAGE clone cDNA was sequenced and submitted to GenBank (accession number AY430094). Minor differences exist between this clone and the NTPDase5 (CD39L4) clone reported by Chadwick [17]; however, none of these differences occur in the protein coding region (see GenBank accession # AY430094). Bacterial expression of a pET28a construct containing the soluble sequence of NTPDase5 (the

257

sequence following the predicted signal peptide cleavage site) resulted in synthesis of inactive NTPDase5 contained in inclusion bodies. Refolding conditions were screened, and the conditions needed to generate active NTPDase5 nucleotidase were determined. The refolded protein was purified on an immobilized metal affinity column. After removal of the hexa-histidine tag by thrombin cleavage, the N-terminal sequence of NTPDase5 is GSHMASNQQTWF. Including the six extra amino acids (indicated in boldface type) N-terminal to the start of the predicted soluble NTPDase5 sequence (underlined), the pET28a NTPDase5 construct expressed in bacteria produces a 410-residue, 45.5-kDa protein, which is consistent with the experimentally determined size (Fig. 1A). Comparison of the wild type and C15S mutant refolding and purification data suggests that elimination of the single non-conserved cysteine (C15S) did not adversely affect folding or enzymatic activity. It should be noted that the other four cysteines in NTPDase5, which are conserved in all NTPDases, are likely involved in the same two conserved disulfide bonds as were determined for NTPDase6 [7]. Fig. 1 (panel A) demonstrates the purity of a typical wildtype NTPDase5 preparation after refolding and purification, and the presence of one band during native gel electrophoresis (Fig. 1, panel B) suggests that bacterially expressed wild-type NTPDase5 is also conformationally homogeneous. Size exclusion chromatography data suggests that bacterially expressed NTPDase5 is monomeric, which is consistent with the mammalian NTPDase5 expression results of Mulero et al. [9,10], and Trombetta and Helenius [11]. NTPDase5 preferentially hydrolyzes diphosphates over triphosphates, as shown in Fig. 2, with GDP, UDP, and IDP being the preferred substrates. A similar rank order of nucleotidase activities was observed for both bacterially expressed and mammalian COS cell-expressed NTPDase5. Since bacteria are unable to perform certain posttranslational modifications, including glycosylation, the high nucleotidase activity of bacterially expressed NTPDase5 indicates that glycosylation is not required for activity. This result is consistent with the findings of Mulero et al. [9] who reported that enzymatic de-glycosylation of mammalian expressed NTPDase5 did not result in decreased nucleotidase activity. Additionally, Trombetta and Helenius [11] found that the ER-UDPase (an enzyme very similar to the one studied in this work) exhibited a similar substrate preference (GDP, UDP, and IDP being the preferred substrates). However, the substrate specificity profile obtained for NTPDase5 in this work does not agree with the data published by Mulero et al. [10], who reported a nucleotidase profile of UDPNGDPNCDPNADPNGTPNCTPNUTPNATP, with an ADPase activity of about 30% of the GDPase activity, using 1 mM nucleotide concentrations (see Table II in [10]). In contrast to Mulero et al. [10], a much lower ADPase activity, approximately 3.5% relative to GDPase activity (Fig. 2), was measured in this study for NTPDase5

258

D.M. Murphy-Piedmonte et al. / Biochimica et Biophysica Acta 1747 (2005) 251–259

expressed in bacteria, and similar results were obtained in the present study with COS cell expressed NTPDase5. A similar situation exists for CDPase activity. Interestingly, Trombetta and Helenius [11] published a substrate profile for the ER-UDPase, similar to the present findings, including extremely poor hydrolysis of ADP, amounting to b1% of the GDPase activity. The poor hydrolysis of ADP relative to other nucleoside di- and triphosphates by NTPDase5 suggests that hydrolysis of ADP (and therefore the physiological processes controlled by ADP) may not be the primary function of NTPDase5 (CD39L4). Thus, the potential role for NTPDase5 in control of hemostasis and platelet aggregation, as suggested by Mulero et al. [9] seems limited. However, NTPDase5 may be involved in the prevention of UDP accumulation in the ER lumen, as suggested by Trombetta and Helenius [11]. This proposed function is supported by the high UDPase activity reported in this work, as well as by that reported by Mulero et al. [10] (UDPase activity is 122% of the GDPase activity). Here, we demonstrate that bacterially expressed NTPDase5 is an allosteric enzyme, and displays positive cooperativity for the hydrolysis of its preferred substrates (GDP, IDP, and UDP). The most common explanation for positive cooperativity is the change in binding/hydrolysis at one site when a ligand or substrate is bound to another site in a multi-subunit protein complex, as exemplified by the binding of O2 by hemoglobin. Since the bacterially expressed NTPDase5 studied in this work is monomeric (as shown by size exclusion chromatography), in agreement with earlier findings on mammalian expressed, soluble NTPDase5 (CD39L4 [9]) and ER-UDPase [11], this possible explanation for the observed positive cooperativity is eliminated. However, it is not unusual for simple, monomeric enzymes to exhibit positive cooperativity, which can be attenuated by a downstream or upstream molecule in the metabolic pathway of the enzyme. There are many examples of this, and one recently published example of positive cooperativity displayed by a monomeric, nucleotide-handling enzyme is the allosteric regulation of NAD kinase by quinolinic acid, a central metabolite in NAD(P) biosynthesis [31]. Accordingly, when a product of the GDPase reaction (GMP, which is not hydrolyzed by NTPDase5) was added to the GDPase assay, a concentration-dependent increase in GDPase activity at a relatively low concentration of GDP was observed (0.05 mM GDP, Fig. 4 inset), resulting in a reduction in the degree of positive cooperativity observed for the hydrolysis of GDP (as well as for the other preferred substrates, IDP and UDP). A similar allosteric effector/activator effect, resulting in an attenuation of the positive cooperativity, was observed for the hydrolysis of GDP in the presence of AMP (which, like GMP, is also not hydrolyzed by NTPDase5, data not shown). The significance of the positive cooperativity and its attenuation by the hydrolytic products GMP and AMP are unknown, but one may speculate that it suggests a

regulatory mechanism for NTPDase5 that is not present for NTPDase6. This positive cooperativity was not observed by Trombetta and Helenius [11] with ER-UDPase; however, a different assay was used to measure the amount of inorganic phosphate released by ER-UDPase. Additionally, sufficiently low concentrations of substrate may not have been used to effectively detect the positive cooperativity in that previous work. In summary, NTPDase5 has been expressed in bacteria and refolded into an active nucleotidase from inclusion bodies, indicating that glycosylation of NTPDase5 is not essential for its activity. A C15S NTPDase5 mutant was also expressed in bacteria, and displayed nucleotidase activity comparable to wild type, suggesting the Cys15 residue is not essential for either refolding or activity of NTPDase5. Based on homology between the NTPDases, it is likely that the remaining four cysteine residues of soluble NTPDase5 exist in the same disulfide bond arrangement as was determined for NTPDase6 [7]. Similar to the mammalian COS cell expressed NTPDase5, bacterially expressed NTPDase5 was shown to preferentially hydrolyze the nucleoside diphosphates GDP, UDP, and IDP. The bacterially expressed enzyme also showed positive cooperativity with respect to the hydrolysis of these preferred substrates that is attenuated by nucleoside monophosphate reaction products.

Acknowledgments This work was supported by NIH grants HL59915 and HL72382 (to TLK).

References [1] M. Handa, G. Guidotti, Purification and cloning of a soluble ATPdiphosphohydrolase (apyrase) from potato tubers (Solanum tuberosum), Biochem. Biophys. Res. Commun. 218 (1996) 916 – 923. [2] B. Kegel, N. Braun, P. Heine, C.R. Maliszewski, H. Zimmermann, An Ecto-ATPase and an Ecto-ATP diphosphohydrolase are expressed in rat brain, Neuropharmacology 36 (1997) 1189 – 1200. [3] E.G. Vasconcelos, S.T. Ferreira, T.M.U. De Carvalho, W. De Souza, A.M. Kettlun, M. Mancilla, M.A. Valenzuela, S. Verjovski-Almeida, Partial purification and immunohistochemical localization of ATP diphosphohydrolase from schistosoma mansoni-immunological crossreactivities with potato apyrase and Toxoplasma gondii nucleoside triphosphate hydrolase, J. Biol. Chem. 271 (1996) 22139 – 22145. [4] T.L. Kirley, F. Yang, V.V. Ivanenkov, Site-directed mutagenesis of human nucleoside triphosphate diphosphohydrolase 3: the importance of conserved glycine residues and the identification of additional conserved protein motifs in eNTPDases, Arch. Biochem. Biophys. 395 (2001) 94 – 102. [5] H. Zimmermann, A.R. Beaudoin, M. Bollen, J.W. Goding, G. Guidotti, T.L. Kirley, S.C. Robson, K. Sano, in: L. Vanduffel (Ed.), Second International Workshop on Ecto-ATPases and Related Ectonucleotidases, Shaker Publishing, Maastricht, The Netherlands, 1999, pp. 1 – 9. [6] G. Yeung, J.J. Mulero, D.W. McGowan, S.S. Bajwa, J.E. Ford, CD39L2, a gene encoding a human nucleoside diphosphatase,

D.M. Murphy-Piedmonte et al. / Biochimica et Biophysica Acta 1747 (2005) 251–259

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

predominantly expressed in the heart, Biochemistry 39 (2000) 12916 – 12923. V.V. Ivanenkov, D.M. Murphy-Piedmonte, T.L. Kirley, Bacterial expression, characterization, and disulfide bond determination of soluble human NTPDase6 (CD39L2) nucleotidase: implications for structure and function, Biochemistry 42 (2003) 11726 – 11735. C.A. Hicks-Berger, B.P. Chadwick, A.-M. Frischauf, T.L. Kirley, Expression and characterization of soluble and membrane-bound human nucleoside triphosphate diphosphohydrolase 6 (CD39L2), J. Biol. Chem. 275 (2000) 34041 – 34045. J.J. Mulero, G. Yeung, S.T. Nelken, J.M. Bright, D.W. McGowan, J.E. Ford, Biochemical characterization of CD39L4, Biochemistry 39 (2000) 12924 – 12928. J.J. Mulero, G. Yeung, S.T. Nelken, J.E. Ford, CD39-L4 is a secreted human apyrase, specific for the hydrolysis of nucleoside diphosphates, J. Biol. Chem. 274 (1999) 20064 – 20067. S. Trombetta, A. Helenius, Glycoprotein reglucosylation and nucleotide sugar utilization in the secretory pathway: identification of a nucleoside diphosphatase in the endoplasmic reticulum, EMBO J. 18 (1999) 3282 – 3292. J.D. Shi, T. Kukar, C.Y. Wang, Q.Z. Li, P.E. Cruz, A. DavoodiSemiromi, P. Yang, Y. Gu, W. Lian, D.H. Wu, J.X. She, Molecular cloning and characterization of a novel mammalian endo-apyrase (LALP1), J. Biol. Chem. 276 (2001) 17474 – 17478. F. Bigonnesse, S.A. Levesque, F. Kukulski, J. Lecka, S.C. Robson, M.J. Fernandes, J. Sevigny, Cloning and characterization of mouse nucleoside triphosphate diphosphohydrolase-8, Biochemistry 43 (2004) 5511 – 5519. R.S. Strobel, A.K. Nagy, A.F. Knowles, J. Buegel, M.D. Rosenberg, Chicken oviductal ecto-ATP-diphosphohydrolase: purification and characterization, J. Biol. Chem. 271 (1996) 16323 – 16331. J.G. Stout, R.S. Strobel, T.L. Kirley, Identification and immunolocalization of ecto-ATPDase in chicken stomach, Biochem. Mol. Biol. Int. 36 (1995) 529 – 535. A.F. Knowles, A.K. Nagy, R.S. Strobel, M. Wu-Weis, Purification, characterization, cloning, and expression of the chicken liver ectoATP-diphosphohydrolase, Eur. J. Biochem. 269 (2002) 2373 – 2382. B.P. Chadwick, A.M. Frischauf, The CD39-like gene family: identification of three new human members (CD39L2, CD39L3, and CD39L4), their murine homologues, and a member of the gene family from Drosophila melanogaster, Genomics 50 (1998) 357 – 367. J.G. Paez, J.A. Recio, A. Rouzaut, V. Notario, Identity between the PCPH proto-oncogene and the CD39L4 (ENTPD5) ectonucleoside triphosphate diphosphohydrolase gene, Int. J. Oncol. 19 (2001) 1249 – 1254.

259

[19] J.A. Recio, N. Zambrano, L. de La Pena, C. Powers, D. Siwarski, K. Huppi, V. Notario, cDNA isolation, expression, and chromosomal localization of the mouse pcph proto-oncogene, Mol. Carcinog. 26 (1999) 130 – 136. [20] L. Plesner, T.L. Kirley, A.F. Knowles (Eds.), Ecto-ATPases: Recent Progress on Structure and Function, Plenum Press, New York, 1997. [21] D.J. Pinsky, M.J. Broekman, J.J. Peschon, K.L. Stocking, T. Fujita, R. Ramasamy, E.S. Connolly Jr., J. Huang, S. Kiss, Y. Zhang, T.F. Choudhri, R.A. McTaggart, H. Liao, J.H. Drosopoulos, V.L. Price, A.J. Marcus, C.R. Maliszewski, Elucidation of the thromboregulatory role of CD39/ectoapyrase in the ischemic brain, J. Clin. Invest. 109 (2002) 1031 – 1040. [22] J. Dai, J. Liu, Y. Deng, T.M. Smith, M. Lu, Structure and protein design of a human platelet function inhibitor, Cell 116 (2004) 649 – 659. [23] D.M. Murphy, V.V. Ivanenkov, T.L. Kirley, Bacterial expression and characterization of a novel, soluble, calcium binding and calcium activated human nucleotidase, Biochemistry 42 (2003) 2412 – 2421. [24] C.M. Stoscheck, Protein assay sensitive at nanogram levels, Anal. Biochem. 160 (1987) 301 – 305. [25] C.M. Stoscheck, Increased uniformity in the response of the Coomassie blue G protein assay to different proteins, Anal. Biochem. 184 (1990) 111 – 116. [26] T.M. Smith, C.A. Hicks-Berger, S. Kim, T.L. Kirley, Cloning, expression, and characterization of a soluble calcium-activated nucleotidase, a human enzyme belonging to a new family of extracellular nucleotidases, Arch. Biochem. Biophys. 406 (2002) 105 – 115. [27] C.H. Fiske, Y. Subbarow, The colorometric determination of phosphorous, J. Biol. Chem. 66 (1925) 375 – 400. [28] T.M. Smith, T.L. Kirley, Site-directed mutagenesis of a human brain Ecto-Apyrase: evidence that the E-type ATPases are related to the actin/heat shock 70/sugar kinase superfamily, Biochemistry 38 (1999) 321 – 328. [29] A.A. Baykov, O.A. Evtushenko, S.M. Avaeva, A malachite green procedure for orthophosphate determination and its use in alkaline phosphatase-based enzyme immunoassay, Anal. Biochem. 171 (1988) 266 – 270. [30] U.K. Laemmli, Cleavage of structural proteins during the assembly of the head of bacteriophage T4, Nature (Lond.) 227 (1970) 680 – 685. [31] S. Garavaglia, A. Galizzi, M. Rizzi, Allosteric regulation of bacillus subtilis NAD kinase by quinolinic acid, J. Bacteriol. 185 (2003) 4844 – 4850.