Portal de Periódicos da CAPES

Identification of Cysteine Residues Responsible for Oxidative Cross-linking and Chemical Inhibition of Human Nucleoside-triphosphate Diphosphohydrolase 3

2002; Elsevier BV; Volume: 277; Issue: 8 Linguagem: Inglês

10.1074/jbc.m110105200

1083-351X

D.M. Murphy, Vasily V. Ivanenkov, Terence L. Kirley,

Pneumocystis jirovecii pneumonia detection and treatment

Cysteine-to-serine mutations were constructed to test the functional and structural significance of the three non-extracellular cysteine residues in ecto-nucleoside-triphosphate diphosphohydrolase 3 (eNTPDase3). None of these cysteines were found to be essential for enzyme activity. However, Cys10, located on the short N-terminal cytoplasmic tail, was found to be responsible for dimer formation occurring via oxidation during membrane preparation as well as for dimer cross-linking resulting from exogenously added sulfhydryl-specific cross-linking agents. The resistance to further cross-linking of these dimers into higher order oligomers by lysine-specific cross-linkers suggests that this enzyme may form its native tetrameric structure as a "dimer of dimers" with nonequivalent interactions between subunits. Cys501, located in the hydrophobic C-terminal membrane-spanning domain of eNTPDase3, was found to be the site of chemical modification by a sulfhydryl-specific reagent, p-chloromercuriphenylsulfonic acid (pCMPS), leading to inhibition of enzyme activity. The effect of pCMPS was negligible after dissociation of the enzyme into monomers by Triton X-100, suggesting that the mechanism of inhibition is dependent on the oligomeric structure. Because Cys501 is accessible for modification by the membrane-impermeant reagent pCMPS, we hypothesize that eNTPDase3 (and possibly other eNTPDases) contains a water-filled crevice allowing access of water and hydrophilic compounds to at least part of the protein's C-terminal membrane-spanning helix. Cysteine-to-serine mutations were constructed to test the functional and structural significance of the three non-extracellular cysteine residues in ecto-nucleoside-triphosphate diphosphohydrolase 3 (eNTPDase3). None of these cysteines were found to be essential for enzyme activity. However, Cys10, located on the short N-terminal cytoplasmic tail, was found to be responsible for dimer formation occurring via oxidation during membrane preparation as well as for dimer cross-linking resulting from exogenously added sulfhydryl-specific cross-linking agents. The resistance to further cross-linking of these dimers into higher order oligomers by lysine-specific cross-linkers suggests that this enzyme may form its native tetrameric structure as a "dimer of dimers" with nonequivalent interactions between subunits. Cys501, located in the hydrophobic C-terminal membrane-spanning domain of eNTPDase3, was found to be the site of chemical modification by a sulfhydryl-specific reagent, p-chloromercuriphenylsulfonic acid (pCMPS), leading to inhibition of enzyme activity. The effect of pCMPS was negligible after dissociation of the enzyme into monomers by Triton X-100, suggesting that the mechanism of inhibition is dependent on the oligomeric structure. Because Cys501 is accessible for modification by the membrane-impermeant reagent pCMPS, we hypothesize that eNTPDase3 (and possibly other eNTPDases) contains a water-filled crevice allowing access of water and hydrophilic compounds to at least part of the protein's C-terminal membrane-spanning helix. The eNTPDases 1eNTPDasesecto-nucleoside-triphosphate diphosphohydrolasespCMPSp-chloromercuriphenylsulfonic acidDSSdisuccinimidyl suberateBMOEbismaleimidoethaneNEMN-ethylmaleimideMOPS3-(N-morpholino)propanesulfonic acid 1eNTPDasesecto-nucleoside-triphosphate diphosphohydrolasespCMPSp-chloromercuriphenylsulfonic acidDSSdisuccinimidyl suberateBMOEbismaleimidoethaneNEMN-ethylmaleimideMOPS3-(N-morpholino)propanesulfonic acid (1Zimmermann H. Beaudoin A.R. Bollen M. Goding J.W. Guidotti G. Kirley T.L. Robson S.C. Sano K. Vanduffel L. Second International Workshop on Ecto-ATPases and Related Ectonucleotidases. Shaker Publishing BV, Maastricht, The Netherlands1999: 1-9Google Scholar), also known as ecto-ATPases/ecto-apyrases and E-type ATPases (2Plesner L. Int. Rev. Cytol. 1995; 158: 141-214Crossref PubMed Scopus (449) Google Scholar), rapidly hydrolyze a variety of extracellular nucleoside 5′-triphosphates and 5′-diphosphates. There are six members of the vertebrate eNTPDase family known to date (1Zimmermann H. Beaudoin A.R. Bollen M. Goding J.W. Guidotti G. Kirley T.L. Robson S.C. Sano K. Vanduffel L. Second International Workshop on Ecto-ATPases and Related Ectonucleotidases. Shaker Publishing BV, Maastricht, The Netherlands1999: 1-9Google Scholar). eNTPDases1–4 are integral membrane proteins with two membrane-spanning segments, one near each end of their respective primary structures. eNTPDases5 and 6 each have a single hydrophobic signal sequence near their N termini, which, when cleaved, results in the release of a soluble enzyme to the outside of the cell (3Hicks-Berger C.A. Chadwick B.P. Frischauf A.-M. Kirley T.L. J. Biol. Chem. 2000; 275: 34041-34045Abstract Full Text Full Text PDF PubMed Scopus (47) Google Scholar, 4Mulero J.J. Yeung G. Nelken S.T. Bright J.M. McGowan D.W. Ford J.E. Biochemistry. 2000; 39: 12924-12928Crossref PubMed Scopus (30) Google Scholar, 5Yeung G. Mulero J.J. McGowan D.W. Bajwa S.S. Ford J.E. Biochemistry. 2000; 39: 12916-12923Crossref PubMed Scopus (46) Google Scholar, 6Braun N. Fengler S. Ebeling C. Servos J. Zimmermann H. Biochem. J. 2000; 351: 639-647Crossref PubMed Scopus (59) Google Scholar). eNTPDases1–4 are all glycosylated and contain 10 conserved cysteine residues located in the extracellular region. All 10 of these conserved residues are likely to be involved in the formation of five conserved disulfide bonds. Different members of the eNTPDase family contain various numbers of nonconserved cysteine residues located either in membrane-spanning helices or in the N- or C-terminal cytoplasmic tails. These cysteine residues may exist as free sulfhydryls and are not expected to be involved in disulfide bond formation. The lone non-extracellular cysteine residue in human CD39 (eNTPDase1) is located on the N-terminal cytoplasmic tail and has been shown to be palmitoylated (7Koziak K. Kaczmarek E. Kittel A. Sevigny J. Blusztajn J.K. Schulte Am Esch J., II Imai M. Guckelberger O. Goepfert C. Qawi I. Robson S.C. J. Biol. Chem. 2000; 275: 2057-2062Abstract Full Text Full Text PDF PubMed Scopus (83) Google Scholar).The primary structure and enzymatic properties of eNTPDase3 (HB6 (8Smith T.M. Kirley T.L. Biochim. Biophys. Acta. 1998; 1386: 65-78Crossref PubMed Scopus (164) Google Scholar) and CD39L3 (9Chadwick B.P. Frischauf A.-M. Genomics. 1998; 50: 357-367Crossref PubMed Scopus (104) Google Scholar)) are intermediate between two other eNTPDases, eNTPDase1 (CD39 ecto-apyrase) and eNTPDase2 (ecto-ATPase) (8Smith T.M. Kirley T.L. Biochim. Biophys. Acta. 1998; 1386: 65-78Crossref PubMed Scopus (164) Google Scholar). Site-directed mutagenesis of eNTPDase3 has revealed many amino acids essential for nucleotidase activity as well as for expression and proper folding of the enzyme (10Smith T.M. Kirley T.L. Biochemistry. 1999; 38: 321-328Crossref PubMed Scopus (100) Google Scholar, 11Smith T.M. Lewis Carl S.A. Kirley T.L. Biochemistry. 1999; 38: 5849-5857Crossref PubMed Scopus (56) Google Scholar, 12Hicks-Berger C.A. Yang F. Smith T.M. Kirley T.L. Biochim. Biophys. Acta. 2001; 1547: 72-81Crossref PubMed Scopus (16) Google Scholar, 13Kirley T.L. Yang F. Ivanenkov V.V. Arch. Biochem. Biophys. 2001; 395: 94-102Crossref PubMed Scopus (29) Google Scholar, 14Yang F. Hicks-Berger C.A. Smith T.M. Kirley T.L. Biochemistry. 2001; 40: 3943-3950Crossref PubMed Scopus (49) Google Scholar). In this study, we mutated singly and in combination all three of the non-extracellular eNTPDase3 cysteine residues (not involved in disulfide bonds) to serine. These residues are localized to the N-terminal cytoplasmic tail (cysteine 10), the C-terminal membrane-spanning helix (cysteine 501), and the interface of the C-terminal membrane helix and the cytoplasm (cysteine 509), respectively. Although none of these cysteine residues were found to be essential for nucleotidase activity, cysteine 10 was identified as the residue responsible for oxidative dimer formation as well as for sulfhydryl-specific intermolecular cross-linking, and cysteine 501 was identified as the residue responsible for inhibition of the enzyme by the sulfhydryl-specific reagent pCMPS. pCMPS has been shown to inhibit several eNTPDases (15Shi X. Knowles A.F. Arch. Biochem. Biophys. 1994; 35: 177-184Crossref Scopus (22) Google Scholar, 16Caldwell C.C. Davis M.D. Knowles A.F. Arch. Biochem. Biophys. 1999; 362: 46-58Crossref PubMed Scopus (33) Google Scholar, 17Caldwell C.C. Hornyak S.C. Pendleton E. Campbell D. Knowles A.F. Arch. Biochem. Biophys. 2001; 387: 107-116Crossref PubMed Scopus (21) Google Scholar). This sulfhydryl-specific reagent is negatively charged under physiological conditions (pK a = 1.5) (18Olami Y. Rimon A. Gerchman Y. Rothman A. Padan E. J. Biol. Chem. 1997; 272: 1761-1768Abstract Full Text Full Text PDF PubMed Scopus (94) Google Scholar) and is membrane-impermeant, being incapable of partitioning into or passing through biological membranes (19Hruz P.W. Mueckler M.M. Biochemistry. 2000; 39: 9367-9372Crossref PubMed Scopus (34) Google Scholar, 20Ding P.Z. Wilson T.H. Biochemistry. 2001; 40: 5506-5510Crossref PubMed Scopus (12) Google Scholar, 21Bragg P.D. Hou C. Arch. Biochem. Biophys. 2000; 380: 141-150Crossref PubMed Scopus (10) Google Scholar, 22Yan R.T. Maloney P.C. Proc. Natl. Acad. Sci. U. S. A. 1995; 92: 5973-5976Crossref PubMed Scopus (79) Google Scholar). Therefore, its ability to react with cysteine 501, which is located in the C-terminal membrane-spanning region of the protein, was surprising. We interpret this result as indicating that the part of the C-terminal membrane-spanning helix containing residue 501 must be solvent-accessible, possibly suggesting a water-filled crevice in the membrane-spanning region involved in the formation of the quaternary structure of eNTPDase3, and perhaps in the other membranous eNTPDases as well.DISCUSSIONIn this study, we used site-directed mutagenesis to examine the role of three non-extracellular cysteine residues in eNTPDase3. None of the cysteines were essential for nucleotidase activity; however, the C10S mutation did result in an increased expression level for the enzyme in transfected COS cells, and the triple cysteine-to-serine mutant, although expressed at a higher level, was slightly less active than the wild-type enzyme when normalized for expression level.The cysteine residue present in the N-terminal cytoplasmic tail (Cys10) was found to be the residue responsible for both dimer formation occurring via oxidation of the wild-type enzyme during membrane preparation and dimer formation due to sulfhydryl-specific chemical cross-linking with BMOE (Fig. 2). There is a cysteine residue in the short N-terminal cytoplasmic tail of CD39 (eNTPDase1) that is analogous to cysteine 10 in eNTPDase 3. In eNTPDase1, this residue is palmitoylated (7Koziak K. Kaczmarek E. Kittel A. Sevigny J. Blusztajn J.K. Schulte Am Esch J., II Imai M. Guckelberger O. Goepfert C. Qawi I. Robson S.C. J. Biol. Chem. 2000; 275: 2057-2062Abstract Full Text Full Text PDF PubMed Scopus (83) Google Scholar). This is clearly not the case for Cys10of eNTPDase3, at least not for the enzyme expressed in COS cells, because this cysteine must be present as a free sulfhydryl to be cross-linked and oxidized to a disulfide. It is clear that dimers of wild-type eNTPDase3 observed on nonreducing SDS-polyacrylamide gels are due to oxidation of the Cys10 sulfhydryl to disulfides during harvesting and homogenization of cells because only a small amount of dimers was formed in eNTPDase3 COS cell membranes prepared in the presence of either alkylating or reducing agents (Fig. 3). In addition, this oxidative cross-linking of the wild-type enzyme can be increased by intentional exposure to air during freeze-thaw of the COS membrane preparations, as was done to generate the data in Fig.2 B.Experiments on sequential cross-linking of eNTPDase3 with sulfhydryl- and amino-specific reagents suggest that, in a native tetramer (25Stoscheck C.M. Anal. Biochem. 1990; 184: 111-116Crossref PubMed Scopus (136) Google Scholar,26Fiske C.H. SubbaRow Y. J. Biol. Chem. 1925; 66: 375-400Abstract Full Text PDF Google Scholar), the monomers are more efficiently cross-linked within a dimer than between two dimers. In other words, the four monomers in a tetramer are not equally likely to be cross-linked to each other. This is evident in Fig. 4, where amino group cross-linking (DSS and glutaraldehyde), subsequent to sulfhydryl-specific cross-linking via cysteine 10 (BMOE), resulted in the formation of very little additional trimers, tetramers, and higher order oligomers, leaving the dimer as the predominant cross-linked form. Considering these data and previous work, both in our laboratory (30Stout J.G. Kirley T.L. Biochemistry. 1996; 35: 8289-8298Crossref PubMed Scopus (67) Google Scholar) and by Knowles and co-workers (17Caldwell C.C. Hornyak S.C. Pendleton E. Campbell D. Knowles A.F. Arch. Biochem. Biophys. 2001; 387: 107-116Crossref PubMed Scopus (21) Google Scholar) using chicken muscle eNTPDase2 (ecto-ATPase), as well as evidence presented by Wang et al. (29Wang T.-F., Ou, Y. Guidotti G. J. Biol. Chem. 1998; 273: 24814-24821Abstract Full Text Full Text PDF PubMed Scopus (128) Google Scholar) demonstrating that CD39 (eNTPDase1) is dimeric after solubilization with the detergent sodium cholate, we propose that, although all of the membranous eNTPDases exist as native tetramers (29Wang T.-F., Ou, Y. Guidotti G. J. Biol. Chem. 1998; 273: 24814-24821Abstract Full Text Full Text PDF PubMed Scopus (128) Google Scholar, 31Hicks-Berger C.A. Kirley T.L. Int. Union of Biochem. Mol. Biol. Life. 2000; 50: 43-50Crossref Scopus (19) Google Scholar), these tetramers are not 4-fold symmetric, but instead are composed of an asymmetric "dimer of dimers," where the interaction between two monomers in each dimer is different and stronger than the interaction between the two dimers forming the tetramer.It has been previously demonstrated that some eNTPDases are inhibited by mercurial sulfhydryl group-selective reagents such as pCMPS (15Shi X. Knowles A.F. Arch. Biochem. Biophys. 1994; 35: 177-184Crossref Scopus (22) Google Scholar, 16Caldwell C.C. Davis M.D. Knowles A.F. Arch. Biochem. Biophys. 1999; 362: 46-58Crossref PubMed Scopus (33) Google Scholar, 17Caldwell C.C. Hornyak S.C. Pendleton E. Campbell D. Knowles A.F. Arch. Biochem. Biophys. 2001; 387: 107-116Crossref PubMed Scopus (21) Google Scholar). By examining pCMPS inhibition of wild-type and mutant eNTPDase3, we showed in this study that modification of cysteine 501 is primarily responsible for inhibition of nucleotidase activity by pCMPS in eNTPDase3 (Fig. 5). This result is somewhat surprising due to the location of cysteine 501 in the membrane-spanning region near the C-terminal end of the protein. Under physiological conditions and the conditions used in this study, pCMPS is a permanently charged and therefore membrane-impermeant reagent (18Olami Y. Rimon A. Gerchman Y. Rothman A. Padan E. J. Biol. Chem. 1997; 272: 1761-1768Abstract Full Text Full Text PDF PubMed Scopus (94) Google Scholar, 19Hruz P.W. Mueckler M.M. Biochemistry. 2000; 39: 9367-9372Crossref PubMed Scopus (34) Google Scholar, 20Ding P.Z. Wilson T.H. Biochemistry. 2001; 40: 5506-5510Crossref PubMed Scopus (12) Google Scholar, 21Bragg P.D. Hou C. Arch. Biochem. Biophys. 2000; 380: 141-150Crossref PubMed Scopus (10) Google Scholar, 22Yan R.T. Maloney P.C. Proc. Natl. Acad. Sci. U. S. A. 1995; 92: 5973-5976Crossref PubMed Scopus (79) Google Scholar). Therefore, it was unexpected that this reagent could react with a cysteine residue predicted to be inside the plasma membrane. We interpret this result as indicating the presence of an aqueous cavity or crevice in the vicinity of cysteine 501; and thus, the interior of the membrane-spanning region surrounding cysteine 501 must be solvent-accessible. This interpretation is consistent with other systems in which pCMPS was used to determine the solvent accessibility of cysteine residues in membrane-spanning regions of proteins containing aqueous pores or crevices (19Hruz P.W. Mueckler M.M. Biochemistry. 2000; 39: 9367-9372Crossref PubMed Scopus (34) Google Scholar, 20Ding P.Z. Wilson T.H. Biochemistry. 2001; 40: 5506-5510Crossref PubMed Scopus (12) Google Scholar, 21Bragg P.D. Hou C. Arch. Biochem. Biophys. 2000; 380: 141-150Crossref PubMed Scopus (10) Google Scholar, 32Williams D.B. Akabas M.H. Mol. Pharmacol. 2000; 58: 1129-1136Crossref PubMed Scopus (46) Google Scholar).The question remains as to how modification of cysteine 501 in the membrane-spanning domain inhibits nucleotidase activity that is catalyzed by a large extracellular domain of eNTPDase3, far removed from the membrane-spanning segments and cysteine 501. One possibility is that modification of cysteine 501 in a membrane-spanning region results in decreased or modified interactions of the membrane-spanning segments in the native tetrameric structure. This is a reasonable postulate because it has been shown that the membrane-spanning regions are important for maintaining the tetrameric structure and the enzyme activity of eNTPDase1 (CD39) (29Wang T.-F., Ou, Y. Guidotti G. J. Biol. Chem. 1998; 273: 24814-24821Abstract Full Text Full Text PDF PubMed Scopus (128) Google Scholar). To test this possibility for mechanism of inactivation, we dissociated tetrameric eNTPDase3 into monomers by solubilization with Triton X-100 (Fig. 6) and examined the effect of pCMPS on solubilized monomeric enzyme. The fact that pCMPS no longer substantially inhibited the residual nucleotidase activity after solubilization with Triton X-100 (Fig. 7) suggests that this inhibition is dependent upon the oligomeric structure of eNTPDase3. The non-additive inhibitory effects of pCMPS and Triton X-100 on the nucleotidase activities of eNTPDase3 (Fig. 8) also suggest a common mechanism of enzyme inactivation by these two reagents, mediated by a weakening or disruption of the oligomeric structure.Knowles and co-workers (17Caldwell C.C. Hornyak S.C. Pendleton E. Campbell D. Knowles A.F. Arch. Biochem. Biophys. 2001; 387: 107-116Crossref PubMed Scopus (21) Google Scholar) have reported that pCMPS inhibits chicken muscle ecto-ATPase (eNTPDase2). These authors hypothesized that a cysteine residue located at the interface between the N-terminal membrane-spanning region and the large extracellular loop (cysteine 23) is the most likely target for modification by pCMPS, leading to inhibition of nucleotidase activity by interfering with that enzyme's oligomerization. The data presented in our study, using site-directed mutagenesis of eNTPDase3, are consistent with their hypothesis in that the pCMPS-induced inhibition of eNTPDase3 is mediated by interfering with monomer-monomer interactions in the native tetrameric quaternary structure. However, the proposed location of the cysteine residue responsible for inhibition of nucleotidase activity by pCMPS in chicken eNTPDase2 (the extracellular face of the N-terminal membrane-spanning segment) is different from the location of the pCMPS-reactive cysteine reside found in this work because cysteine 501 of eNTPDase3 is located deep within the C-terminal membrane-spanning region, slightly closer to the cytoplasmic face than to the extracellular face of the cell membrane.A diagram summarizing the results and conclusions from this study is presented in Fig. 9. Cysteine 10, located in the N-terminal cytoplasmic tail, is shown to be the site of sulfhydryl cross-linking. Cysteine 501, located in the C-terminal membrane-spanning helix, is depicted to be accessible to pCMPS and water. Cysteine 509, located at the interface of the membrane and the C-terminal cytoplasmic tail, is depicted at the mouth of the hypothesized aqueous crevice. DSS and glutaraldehyde cross-linking is depicted as occurring on the extracellular loops because most of the lysine residues and most of the protein mass are contained in the extracellular domain.Because the transmembrane domains greatly modulate the nucleotidase activity of eNTPDases (17Caldwell C.C. Hornyak S.C. Pendleton E. Campbell D. Knowles A.F. Arch. Biochem. Biophys. 2001; 387: 107-116Crossref PubMed Scopus (21) Google Scholar, 29Wang T.-F., Ou, Y. Guidotti G. J. Biol. Chem. 1998; 273: 24814-24821Abstract Full Text Full Text PDF PubMed Scopus (128) Google Scholar, 30Stout J.G. Kirley T.L. Biochemistry. 1996; 35: 8289-8298Crossref PubMed Scopus (67) Google Scholar), future understanding of the interactions of the membrane-spanning helices, both within monomers and between monomers in the native tetramer, will be dependent on the delineation of the details of the aqueous crevice that has been postulated in this work. Further investigation of the extent and functional significance of this aqueous crevice in the transmembrane regions of eNTPDase3 and the existence of such an aqueous crevice in other membranous eNTPDases should lead to a better understanding of the structure and function of these enzymes. The eNTPDases 1eNTPDasesecto-nucleoside-triphosphate diphosphohydrolasespCMPSp-chloromercuriphenylsulfonic acidDSSdisuccinimidyl suberateBMOEbismaleimidoethaneNEMN-ethylmaleimideMOPS3-(N-morpholino)propanesulfonic acid 1eNTPDasesecto-nucleoside-triphosphate diphosphohydrolasespCMPSp-chloromercuriphenylsulfonic acidDSSdisuccinimidyl suberateBMOEbismaleimidoethaneNEMN-ethylmaleimideMOPS3-(N-morpholino)propanesulfonic acid (1Zimmermann H. Beaudoin A.R. Bollen M. Goding J.W. Guidotti G. Kirley T.L. Robson S.C. Sano K. Vanduffel L. Second International Workshop on Ecto-ATPases and Related Ectonucleotidases. Shaker Publishing BV, Maastricht, The Netherlands1999: 1-9Google Scholar), also known as ecto-ATPases/ecto-apyrases and E-type ATPases (2Plesner L. Int. Rev. Cytol. 1995; 158: 141-214Crossref PubMed Scopus (449) Google Scholar), rapidly hydrolyze a variety of extracellular nucleoside 5′-triphosphates and 5′-diphosphates. There are six members of the vertebrate eNTPDase family known to date (1Zimmermann H. Beaudoin A.R. Bollen M. Goding J.W. Guidotti G. Kirley T.L. Robson S.C. Sano K. Vanduffel L. Second International Workshop on Ecto-ATPases and Related Ectonucleotidases. Shaker Publishing BV, Maastricht, The Netherlands1999: 1-9Google Scholar). eNTPDases1–4 are integral membrane proteins with two membrane-spanning segments, one near each end of their respective primary structures. eNTPDases5 and 6 each have a single hydrophobic signal sequence near their N termini, which, when cleaved, results in the release of a soluble enzyme to the outside of the cell (3Hicks-Berger C.A. Chadwick B.P. Frischauf A.-M. Kirley T.L. J. Biol. Chem. 2000; 275: 34041-34045Abstract Full Text Full Text PDF PubMed Scopus (47) Google Scholar, 4Mulero J.J. Yeung G. Nelken S.T. Bright J.M. McGowan D.W. Ford J.E. Biochemistry. 2000; 39: 12924-12928Crossref PubMed Scopus (30) Google Scholar, 5Yeung G. Mulero J.J. McGowan D.W. Bajwa S.S. Ford J.E. Biochemistry. 2000; 39: 12916-12923Crossref PubMed Scopus (46) Google Scholar, 6Braun N. Fengler S. Ebeling C. Servos J. Zimmermann H. Biochem. J. 2000; 351: 639-647Crossref PubMed Scopus (59) Google Scholar). eNTPDases1–4 are all glycosylated and contain 10 conserved cysteine residues located in the extracellular region. All 10 of these conserved residues are likely to be involved in the formation of five conserved disulfide bonds. Different members of the eNTPDase family contain various numbers of nonconserved cysteine residues located either in membrane-spanning helices or in the N- or C-terminal cytoplasmic tails. These cysteine residues may exist as free sulfhydryls and are not expected to be involved in disulfide bond formation. The lone non-extracellular cysteine residue in human CD39 (eNTPDase1) is located on the N-terminal cytoplasmic tail and has been shown to be palmitoylated (7Koziak K. Kaczmarek E. Kittel A. Sevigny J. Blusztajn J.K. Schulte Am Esch J., II Imai M. Guckelberger O. Goepfert C. Qawi I. Robson S.C. J. Biol. Chem. 2000; 275: 2057-2062Abstract Full Text Full Text PDF PubMed Scopus (83) Google Scholar). ecto-nucleoside-triphosphate diphosphohydrolases p-chloromercuriphenylsulfonic acid disuccinimidyl suberate bismaleimidoethane N-ethylmaleimide 3-(N-morpholino)propanesulfonic acid ecto-nucleoside-triphosphate diphosphohydrolases p-chloromercuriphenylsulfonic acid disuccinimidyl suberate bismaleimidoethane N-ethylmaleimide 3-(N-morpholino)propanesulfonic acid The primary structure and enzymatic properties of eNTPDase3 (HB6 (8Smith T.M. Kirley T.L. Biochim. Biophys. Acta. 1998; 1386: 65-78Crossref PubMed Scopus (164) Google Scholar) and CD39L3 (9Chadwick B.P. Frischauf A.-M. Genomics. 1998; 50: 357-367Crossref PubMed Scopus (104) Google Scholar)) are intermediate between two other eNTPDases, eNTPDase1 (CD39 ecto-apyrase) and eNTPDase2 (ecto-ATPase) (8Smith T.M. Kirley T.L. Biochim. Biophys. Acta. 1998; 1386: 65-78Crossref PubMed Scopus (164) Google Scholar). Site-directed mutagenesis of eNTPDase3 has revealed many amino acids essential for nucleotidase activity as well as for expression and proper folding of the enzyme (10Smith T.M. Kirley T.L. Biochemistry. 1999; 38: 321-328Crossref PubMed Scopus (100) Google Scholar, 11Smith T.M. Lewis Carl S.A. Kirley T.L. Biochemistry. 1999; 38: 5849-5857Crossref PubMed Scopus (56) Google Scholar, 12Hicks-Berger C.A. Yang F. Smith T.M. Kirley T.L. Biochim. Biophys. Acta. 2001; 1547: 72-81Crossref PubMed Scopus (16) Google Scholar, 13Kirley T.L. Yang F. Ivanenkov V.V. Arch. Biochem. Biophys. 2001; 395: 94-102Crossref PubMed Scopus (29) Google Scholar, 14Yang F. Hicks-Berger C.A. Smith T.M. Kirley T.L. Biochemistry. 2001; 40: 3943-3950Crossref PubMed Scopus (49) Google Scholar). In this study, we mutated singly and in combination all three of the non-extracellular eNTPDase3 cysteine residues (not involved in disulfide bonds) to serine. These residues are localized to the N-terminal cytoplasmic tail (cysteine 10), the C-terminal membrane-spanning helix (cysteine 501), and the interface of the C-terminal membrane helix and the cytoplasm (cysteine 509), respectively. Although none of these cysteine residues were found to be essential for nucleotidase activity, cysteine 10 was identified as the residue responsible for oxidative dimer formation as well as for sulfhydryl-specific intermolecular cross-linking, and cysteine 501 was identified as the residue responsible for inhibition of the enzyme by the sulfhydryl-specific reagent pCMPS. pCMPS has been shown to inhibit several eNTPDases (15Shi X. Knowles A.F. Arch. Biochem. Biophys. 1994; 35: 177-184Crossref Scopus (22) Google Scholar, 16Caldwell C.C. Davis M.D. Knowles A.F. Arch. Biochem. Biophys. 1999; 362: 46-58Crossref PubMed Scopus (33) Google Scholar, 17Caldwell C.C. Hornyak S.C. Pendleton E. Campbell D. Knowles A.F. Arch. Biochem. Biophys. 2001; 387: 107-116Crossref PubMed Scopus (21) Google Scholar). This sulfhydryl-specific reagent is negatively charged under physiological conditions (pK a = 1.5) (18Olami Y. Rimon A. Gerchman Y. Rothman A. Padan E. J. Biol. Chem. 1997; 272: 1761-1768Abstract Full Text Full Text PDF PubMed Scopus (94) Google Scholar) and is membrane-impermeant, being incapable of partitioning into or passing through biological membranes (19Hruz P.W. Mueckler M.M. Biochemistry. 2000; 39: 9367-9372Crossref PubMed Scopus (34) Google Scholar, 20Ding P.Z. Wilson T.H. Biochemistry. 2001; 40: 5506-5510Crossref PubMed Scopus (12) Google Scholar, 21Bragg P.D. Hou C. Arch. Biochem. Biophys. 2000; 380: 141-150Crossref PubMed Scopus (10) Google Scholar, 22Yan R.T. Maloney P.C. Proc. Natl. Acad. Sci. U. S. A. 1995; 92: 5973-5976Crossref PubMed Scopus (79) Google Scholar). Therefore, its ability to react with cysteine 501, which is located in the C-terminal membrane-spanning region of the protein, was surprising. We interpret this result as indicating that the part of the C-terminal membrane-spanning helix containing residue 501 must be solvent-accessible, possibly suggesting a water-filled crevice in the membrane-spanning region involved in the formation of the quaternary structure of eNTPDase3, and perhaps in the other membranous eNTPDases as well. DISCUSSIONIn this study, we used site-directed mutagenesis to examine the role of three non-extracellular cysteine residues in eNTPDase3. None of the cysteines were essential for nucleotidase activity; however, the C10S mutation did result in an increased expression level for the enzyme in transfected COS cells, and the triple cysteine-to-serine mutant, although expressed at a higher level, was slightly less active than the wild-type enzyme when normalized for expression level.The cysteine residue present in the N-terminal cytoplasmic tail (Cys10) was found to be the residue responsible for both dimer formation occurring via oxidation of the wild-type enzyme during membrane preparation and dimer formation due to sulfhydryl-specific chemical cross-linking with BMOE (Fig. 2). There is a cysteine residue in the short N-terminal cytoplasmic tail of CD39 (eNTPDase1) that is analogous to cysteine 10 in eNTPDase 3. In eNTPDase1, this residue is palmitoylated (7Koziak K. Kaczmarek E. Kittel A. Sevigny J. Blusztajn J.K. Schulte Am Esch J., II Imai M. Guckelberger O. Goepfert C. Qawi I. Robson S.C. J. Biol. Chem. 2000; 275: 2057-2062Abstract Full Text Full Text PDF PubMed Scopus (83) Google Scholar). This is clearly not the case for Cys10of eNTPDase3, at least not for the enzyme expressed in COS cells, because this cysteine must be present as a free sulfhydryl to be cross-linked and oxidized to a disulfide. It is clear that dimers of wild-type eNTPDase3 observed on nonreducing SDS-polyacrylamide gels are due to oxidation of the Cys10 sulfhydryl to disulfides during harvesting and homogenization of cells because only a small amount of dimers was formed in eNTPDase3 COS cell membranes prepared in the presence of either alkylating or reducing agents (Fig. 3). In addition, this oxidative cross-linking of the wild-type enzyme can be increased by intentional exposure to air during freeze-thaw of the COS membrane preparations, as was done to generate the data in Fig.2 B.Experiments on sequential cross-linking of eNTPDase3 with sulfhydryl- and amino-specific reagents suggest that, in a native tetramer (25Stoscheck C.M. Anal. Biochem. 1990; 184: 111-116Crossref PubMed Scopus (136) Google Scholar,26Fiske C.H. SubbaRow Y. J. Biol. Chem. 1925; 66: 375-400Abstract Full Text PDF Google Scholar), the monomers are more efficiently cross-linked within a dimer than between two dimers. In other words, the four monomers in a tetramer are not equally likely to be cross-linked to each other. This is evident in Fig. 4, where amino group cross-linking (DSS and glutaraldehyde), subsequent to sulfhydryl-specific cross-linking via cysteine 10 (BMOE), resulted in the formation of very little additional trimers, tetramers, and higher order oligomers, leaving the dimer as the predominant cross-linked form. Considering these data and previous work, both in our laboratory (30Stout J.G. Kirley T.L. Biochemistry. 1996; 35: 8289-8298Crossref PubMed Scopus (67) Google Scholar) and by Knowles and co-workers (17Caldwell C.C. Hornyak S.C. Pendleton E. Campbell D. Knowles A.F. Arch. Biochem. Biophys. 2001; 387: 107-116Crossref PubMed Scopus (21) Google Scholar) using chicken muscle eNTPDase2 (ecto-ATPase), as well as evidence presented by Wang et al. (29Wang T.-F., Ou, Y. Guidotti G. J. Biol. Chem. 1998; 273: 24814-24821Abstract Full Text Full Text PDF PubMed Scopus (128) Google Scholar) demonstrating that CD39 (eNTPDase1) is dimeric after solubilization with the detergent sodium cholate, we propose that, although all of the membranous eNTPDases exist as native tetramers (29Wang T.-F., Ou, Y. Guidotti G. J. Biol. Chem. 1998; 273: 24814-24821Abstract Full Text Full Text PDF PubMed Scopus (128) Google Scholar, 31Hicks-Berger C.A. Kirley T.L. Int. Union of Biochem. Mol. Biol. Life. 2000; 50: 43-50Crossref Scopus (19) Google Scholar), these tetramers are not 4-fold symmetric, but instead are composed of an asymmetric "dimer of dimers," where the interaction between two monomers in each dimer is different and stronger than the interaction between the two dimers forming the tetramer.It has been previously demonstrated that some eNTPDases are inhibited by mercurial sulfhydryl group-selective reagents such as pCMPS (15Shi X. Knowles A.F. Arch. Biochem. Biophys. 1994; 35: 177-184Crossref Scopus (22) Google Scholar, 16Caldwell C.C. Davis M.D. Knowles A.F. Arch. Biochem. Biophys. 1999; 362: 46-58Crossref PubMed Scopus (33) Google Scholar, 17Caldwell C.C. Hornyak S.C. Pendleton E. Campbell D. Knowles A.F. Arch. Biochem. Biophys. 2001; 387: 107-116Crossref PubMed Scopus (21) Google Scholar). By examining pCMPS inhibition of wild-type and mutant eNTPDase3, we showed in this study that modification of cysteine 501 is primarily responsible for inhibition of nucleotidase activity by pCMPS in eNTPDase3 (Fig. 5). This result is somewhat surprising due to the location of cysteine 501 in the membrane-spanning region near the C-terminal end of the protein. Under physiological conditions and the conditions used in this study, pCMPS is a permanently charged and therefore membrane-impermeant reagent (18Olami Y. Rimon A. Gerchman Y. Rothman A. Padan E. J. Biol. Chem. 1997; 272: 1761-1768Abstract Full Text Full Text PDF PubMed Scopus (94) Google Scholar, 19Hruz P.W. Mueckler M.M. Biochemistry. 2000; 39: 9367-9372Crossref PubMed Scopus (34) Google Scholar, 20Ding P.Z. Wilson T.H. Biochemistry. 2001; 40: 5506-5510Crossref PubMed Scopus (12) Google Scholar, 21Bragg P.D. Hou C. Arch. Biochem. Biophys. 2000; 380: 141-150Crossref PubMed Scopus (10) Google Scholar, 22Yan R.T. Maloney P.C. Proc. Natl. Acad. Sci. U. S. A. 1995; 92: 5973-5976Crossref PubMed Scopus (79) Google Scholar). Therefore, it was unexpected that this reagent could react with a cysteine residue predicted to be inside the plasma membrane. We interpret this result as indicating the presence of an aqueous cavity or crevice in the vicinity of cysteine 501; and thus, the interior of the membrane-spanning region surrounding cysteine 501 must be solvent-accessible. This interpretation is consistent with other systems in which pCMPS was used to determine the solvent accessibility of cysteine residues in membrane-spanning regions of proteins containing aqueous pores or crevices (19Hruz P.W. Mueckler M.M. Biochemistry. 2000; 39: 9367-9372Crossref PubMed Scopus (34) Google Scholar, 20Ding P.Z. Wilson T.H. Biochemistry. 2001; 40: 5506-5510Crossref PubMed Scopus (12) Google Scholar, 21Bragg P.D. Hou C. Arch. Biochem. Biophys. 2000; 380: 141-150Crossref PubMed Scopus (10) Google Scholar, 32Williams D.B. Akabas M.H. Mol. Pharmacol. 2000; 58: 1129-1136Crossref PubMed Scopus (46) Google Scholar).The question remains as to how modification of cysteine 501 in the membrane-spanning domain inhibits nucleotidase activity that is catalyzed by a large extracellular domain of eNTPDase3, far removed from the membrane-spanning segments and cysteine 501. One possibility is that modification of cysteine 501 in a membrane-spanning region results in decreased or modified interactions of the membrane-spanning segments in the native tetrameric structure. This is a reasonable postulate because it has been shown that the membrane-spanning regions are important for maintaining the tetrameric structure and the enzyme activity of eNTPDase1 (CD39) (29Wang T.-F., Ou, Y. Guidotti G. J. Biol. Chem. 1998; 273: 24814-24821Abstract Full Text Full Text PDF PubMed Scopus (128) Google Scholar). To test this possibility for mechanism of inactivation, we dissociated tetrameric eNTPDase3 into monomers by solubilization with Triton X-100 (Fig. 6) and examined the effect of pCMPS on solubilized monomeric enzyme. The fact that pCMPS no longer substantially inhibited the residual nucleotidase activity after solubilization with Triton X-100 (Fig. 7) suggests that this inhibition is dependent upon the oligomeric structure of eNTPDase3. The non-additive inhibitory effects of pCMPS and Triton X-100 on the nucleotidase activities of eNTPDase3 (Fig. 8) also suggest a common mechanism of enzyme inactivation by these two reagents, mediated by a weakening or disruption of the oligomeric structure.Knowles and co-workers (17Caldwell C.C. Hornyak S.C. Pendleton E. Campbell D. Knowles A.F. Arch. Biochem. Biophys. 2001; 387: 107-116Crossref PubMed Scopus (21) Google Scholar) have reported that pCMPS inhibits chicken muscle ecto-ATPase (eNTPDase2). These authors hypothesized that a cysteine residue located at the interface between the N-terminal membrane-spanning region and the large extracellular loop (cysteine 23) is the most likely target for modification by pCMPS, leading to inhibition of nucleotidase activity by interfering with that enzyme's oligomerization. The data presented in our study, using site-directed mutagenesis of eNTPDase3, are consistent with their hypothesis in that the pCMPS-induced inhibition of eNTPDase3 is mediated by interfering with monomer-monomer interactions in the native tetrameric quaternary structure. However, the proposed location of the cysteine residue responsible for inhibition of nucleotidase activity by pCMPS in chicken eNTPDase2 (the extracellular face of the N-terminal membrane-spanning segment) is different from the location of the pCMPS-reactive cysteine reside found in this work because cysteine 501 of eNTPDase3 is located deep within the C-terminal membrane-spanning region, slightly closer to the cytoplasmic face than to the extracellular face of the cell membrane.A diagram summarizing the results and conclusions from this study is presented in Fig. 9. Cysteine 10, located in the N-terminal cytoplasmic tail, is shown to be the site of sulfhydryl cross-linking. Cysteine 501, located in the C-terminal membrane-spanning helix, is depicted to be accessible to pCMPS and water. Cysteine 509, located at the interface of the membrane and the C-terminal cytoplasmic tail, is depicted at the mouth of the hypothesized aqueous crevice. DSS and glutaraldehyde cross-linking is depicted as occurring on the extracellular loops because most of the lysine residues and most of the protein mass are contained in the extracellular domain.Because the transmembrane domains greatly modulate the nucleotidase activity of eNTPDases (17Caldwell C.C. Hornyak S.C. Pendleton E. Campbell D. Knowles A.F. Arch. Biochem. Biophys. 2001; 387: 107-116Crossref PubMed Scopus (21) Google Scholar, 29Wang T.-F., Ou, Y. Guidotti G. J. Biol. Chem. 1998; 273: 24814-24821Abstract Full Text Full Text PDF PubMed Scopus (128) Google Scholar, 30Stout J.G. Kirley T.L. Biochemistry. 1996; 35: 8289-8298Crossref PubMed Scopus (67) Google Scholar), future understanding of the interactions of the membrane-spanning helices, both within monomers and between monomers in the native tetramer, will be dependent on the delineation of the details of the aqueous crevice that has been postulated in this work. Further investigation of the extent and functional significance of this aqueous crevice in the transmembrane regions of eNTPDase3 and the existence of such an aqueous crevice in other membranous eNTPDases should lead to a better understanding of the structure and function of these enzymes. In this study, we used site-directed mutagenesis to examine the role of three non-extracellular cysteine residues in eNTPDase3. None of the cysteines were essential for nucleotidase activity; however, the C10S mutation did result in an increased expression level for the enzyme in transfected COS cells, and the triple cysteine-to-serine mutant, although expressed at a higher level, was slightly less active than the wild-type enzyme when normalized for expression level. The cysteine residue present in the N-terminal cytoplasmic tail (Cys10) was found to be the residue responsible for both dimer formation occurring via oxidation of the wild-type enzyme during membrane preparation and dimer formation due to sulfhydryl-specific chemical cross-linking with BMOE (Fig. 2). There is a cysteine residue in the short N-terminal cytoplasmic tail of CD39 (eNTPDase1) that is analogous to cysteine 10 in eNTPDase 3. In eNTPDase1, this residue is palmitoylated (7Koziak K. Kaczmarek E. Kittel A. Sevigny J. Blusztajn J.K. Schulte Am Esch J., II Imai M. Guckelberger O. Goepfert C. Qawi I. Robson S.C. J. Biol. Chem. 2000; 275: 2057-2062Abstract Full Text Full Text PDF PubMed Scopus (83) Google Scholar). This is clearly not the case for Cys10of eNTPDase3, at least not for the enzyme expressed in COS cells, because this cysteine must be present as a free sulfhydryl to be cross-linked and oxidized to a disulfide. It is clear that dimers of wild-type eNTPDase3 observed on nonreducing SDS-polyacrylamide gels are due to oxidation of the Cys10 sulfhydryl to disulfides during harvesting and homogenization of cells because only a small amount of dimers was formed in eNTPDase3 COS cell membranes prepared in the presence of either alkylating or reducing agents (Fig. 3). In addition, this oxidative cross-linking of the wild-type enzyme can be increased by intentional exposure to air during freeze-thaw of the COS membrane preparations, as was done to generate the data in Fig.2 B. Experiments on sequential cross-linking of eNTPDase3 with sulfhydryl- and amino-specific reagents suggest that, in a native tetramer (25Stoscheck C.M. Anal. Biochem. 1990; 184: 111-116Crossref PubMed Scopus (136) Google Scholar,26Fiske C.H. SubbaRow Y. J. Biol. Chem. 1925; 66: 375-400Abstract Full Text PDF Google Scholar), the monomers are more efficiently cross-linked within a dimer than between two dimers. In other words, the four monomers in a tetramer are not equally likely to be cross-linked to each other. This is evident in Fig. 4, where amino group cross-linking (DSS and glutaraldehyde), subsequent to sulfhydryl-specific cross-linking via cysteine 10 (BMOE), resulted in the formation of very little additional trimers, tetramers, and higher order oligomers, leaving the dimer as the predominant cross-linked form. Considering these data and previous work, both in our laboratory (30Stout J.G. Kirley T.L. Biochemistry. 1996; 35: 8289-8298Crossref PubMed Scopus (67) Google Scholar) and by Knowles and co-workers (17Caldwell C.C. Hornyak S.C. Pendleton E. Campbell D. Knowles A.F. Arch. Biochem. Biophys. 2001; 387: 107-116Crossref PubMed Scopus (21) Google Scholar) using chicken muscle eNTPDase2 (ecto-ATPase), as well as evidence presented by Wang et al. (29Wang T.-F., Ou, Y. Guidotti G. J. Biol. Chem. 1998; 273: 24814-24821Abstract Full Text Full Text PDF PubMed Scopus (128) Google Scholar) demonstrating that CD39 (eNTPDase1) is dimeric after solubilization with the detergent sodium cholate, we propose that, although all of the membranous eNTPDases exist as native tetramers (29Wang T.-F., Ou, Y. Guidotti G. J. Biol. Chem. 1998; 273: 24814-24821Abstract Full Text Full Text PDF PubMed Scopus (128) Google Scholar, 31Hicks-Berger C.A. Kirley T.L. Int. Union of Biochem. Mol. Biol. Life. 2000; 50: 43-50Crossref Scopus (19) Google Scholar), these tetramers are not 4-fold symmetric, but instead are composed of an asymmetric "dimer of dimers," where the interaction between two monomers in each dimer is different and stronger than the interaction between the two dimers forming the tetramer. It has been previously demonstrated that some eNTPDases are inhibited by mercurial sulfhydryl group-selective reagents such as pCMPS (15Shi X. Knowles A.F. Arch. Biochem. Biophys. 1994; 35: 177-184Crossref Scopus (22) Google Scholar, 16Caldwell C.C. Davis M.D. Knowles A.F. Arch. Biochem. Biophys. 1999; 362: 46-58Crossref PubMed Scopus (33) Google Scholar, 17Caldwell C.C. Hornyak S.C. Pendleton E. Campbell D. Knowles A.F. Arch. Biochem. Biophys. 2001; 387: 107-116Crossref PubMed Scopus (21) Google Scholar). By examining pCMPS inhibition of wild-type and mutant eNTPDase3, we showed in this study that modification of cysteine 501 is primarily responsible for inhibition of nucleotidase activity by pCMPS in eNTPDase3 (Fig. 5). This result is somewhat surprising due to the location of cysteine 501 in the membrane-spanning region near the C-terminal end of the protein. Under physiological conditions and the conditions used in this study, pCMPS is a permanently charged and therefore membrane-impermeant reagent (18Olami Y. Rimon A. Gerchman Y. Rothman A. Padan E. J. Biol. Chem. 1997; 272: 1761-1768Abstract Full Text Full Text PDF PubMed Scopus (94) Google Scholar, 19Hruz P.W. Mueckler M.M. Biochemistry. 2000; 39: 9367-9372Crossref PubMed Scopus (34) Google Scholar, 20Ding P.Z. Wilson T.H. Biochemistry. 2001; 40: 5506-5510Crossref PubMed Scopus (12) Google Scholar, 21Bragg P.D. Hou C. Arch. Biochem. Biophys. 2000; 380: 141-150Crossref PubMed Scopus (10) Google Scholar, 22Yan R.T. Maloney P.C. Proc. Natl. Acad. Sci. U. S. A. 1995; 92: 5973-5976Crossref PubMed Scopus (79) Google Scholar). Therefore, it was unexpected that this reagent could react with a cysteine residue predicted to be inside the plasma membrane. We interpret this result as indicating the presence of an aqueous cavity or crevice in the vicinity of cysteine 501; and thus, the interior of the membrane-spanning region surrounding cysteine 501 must be solvent-accessible. This interpretation is consistent with other systems in which pCMPS was used to determine the solvent accessibility of cysteine residues in membrane-spanning regions of proteins containing aqueous pores or crevices (19Hruz P.W. Mueckler M.M. Biochemistry. 2000; 39: 9367-9372Crossref PubMed Scopus (34) Google Scholar, 20Ding P.Z. Wilson T.H. Biochemistry. 2001; 40: 5506-5510Crossref PubMed Scopus (12) Google Scholar, 21Bragg P.D. Hou C. Arch. Biochem. Biophys. 2000; 380: 141-150Crossref PubMed Scopus (10) Google Scholar, 32Williams D.B. Akabas M.H. Mol. Pharmacol. 2000; 58: 1129-1136Crossref PubMed Scopus (46) Google Scholar). The question remains as to how modification of cysteine 501 in the membrane-spanning domain inhibits nucleotidase activity that is catalyzed by a large extracellular domain of eNTPDase3, far removed from the membrane-spanning segments and cysteine 501. One possibility is that modification of cysteine 501 in a membrane-spanning region results in decreased or modified interactions of the membrane-spanning segments in the native tetrameric structure. This is a reasonable postulate because it has been shown that the membrane-spanning regions are important for maintaining the tetrameric structure and the enzyme activity of eNTPDase1 (CD39) (29Wang T.-F., Ou, Y. Guidotti G. J. Biol. Chem. 1998; 273: 24814-24821Abstract Full Text Full Text PDF PubMed Scopus (128) Google Scholar). To test this possibility for mechanism of inactivation, we dissociated tetrameric eNTPDase3 into monomers by solubilization with Triton X-100 (Fig. 6) and examined the effect of pCMPS on solubilized monomeric enzyme. The fact that pCMPS no longer substantially inhibited the residual nucleotidase activity after solubilization with Triton X-100 (Fig. 7) suggests that this inhibition is dependent upon the oligomeric structure of eNTPDase3. The non-additive inhibitory effects of pCMPS and Triton X-100 on the nucleotidase activities of eNTPDase3 (Fig. 8) also suggest a common mechanism of enzyme inactivation by these two reagents, mediated by a weakening or disruption of the oligomeric structure. Knowles and co-workers (17Caldwell C.C. Hornyak S.C. Pendleton E. Campbell D. Knowles A.F. Arch. Biochem. Biophys. 2001; 387: 107-116Crossref PubMed Scopus (21) Google Scholar) have reported that pCMPS inhibits chicken muscle ecto-ATPase (eNTPDase2). These authors hypothesized that a cysteine residue located at the interface between the N-terminal membrane-spanning region and the large extracellular loop (cysteine 23) is the most likely target for modification by pCMPS, leading to inhibition of nucleotidase activity by interfering with that enzyme's oligomerization. The data presented in our study, using site-directed mutagenesis of eNTPDase3, are consistent with their hypothesis in that the pCMPS-induced inhibition of eNTPDase3 is mediated by interfering with monomer-monomer interactions in the native tetrameric quaternary structure. However, the proposed location of the cysteine residue responsible for inhibition of nucleotidase activity by pCMPS in chicken eNTPDase2 (the extracellular face of the N-terminal membrane-spanning segment) is different from the location of the pCMPS-reactive cysteine reside found in this work because cysteine 501 of eNTPDase3 is located deep within the C-terminal membrane-spanning region, slightly closer to the cytoplasmic face than to the extracellular face of the cell membrane. A diagram summarizing the results and conclusions from this study is presented in Fig. 9. Cysteine 10, located in the N-terminal cytoplasmic tail, is shown to be the site of sulfhydryl cross-linking. Cysteine 501, located in the C-terminal membrane-spanning helix, is depicted to be accessible to pCMPS and water. Cysteine 509, located at the interface of the membrane and the C-terminal cytoplasmic tail, is depicted at the mouth of the hypothesized aqueous crevice. DSS and glutaraldehyde cross-linking is depicted as occurring on the extracellular loops because most of the lysine residues and most of the protein mass are contained in the extracellular domain. Because the transmembrane domains greatly modulate the nucleotidase activity of eNTPDases (17Caldwell C.C. Hornyak S.C. Pendleton E. Campbell D. Knowles A.F. Arch. Biochem. Biophys. 2001; 387: 107-116Crossref PubMed Scopus (21) Google Scholar, 29Wang T.-F., Ou, Y. Guidotti G. J. Biol. Chem. 1998; 273: 24814-24821Abstract Full Text Full Text PDF PubMed Scopus (128) Google Scholar, 30Stout J.G. Kirley T.L. Biochemistry. 1996; 35: 8289-8298Crossref PubMed Scopus (67) Google Scholar), future understanding of the interactions of the membrane-spanning helices, both within monomers and between monomers in the native tetramer, will be dependent on the delineation of the details of the aqueous crevice that has been postulated in this work. Further investigation of the extent and functional significance of this aqueous crevice in the transmembrane regions of eNTPDase3 and the existence of such an aqueous crevice in other membranous eNTPDases should lead to a better understanding of the structure and function of these enzymes.