| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 279, Issue 19, 19955-19969, May 7, 2004
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
From the
Department of Pharmacology and Toxicology, ¶Molecular Graphics and Modeling Laboratory, the University of Kansas, Lawrence, Kansas 66045-7582, the ||Department of Biochemistry, the University of Texas Southwestern Medical Center, Dallas, Texas 75390-9050, the **Department of Chemistry and Physics, Southeastern Louisiana University, Hammond, Louisiana 70402-0878, and the Department of Physiology and Biophysics, Health Sciences Center, State University of New York, Stony Brook, New York 11794-8661
Received for publication, November 17, 2003 , and in revised form, February 2, 2004.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL PROCEDURESRESULTS AND DISCUSSIONREFERENCES |
|---|
-thio]triphosphate is a potent competitive inhibitor of AC in S49 lymphoma cell membranes. These data prompted us to study systematically the effects of 24 nucleotides on AC in S49 and Sf9 insect cell membranes, ACs 1, 2, 5, and 6, expressed in Sf9 membranes and purified catalytic subunits of membranous ACs (C1 of AC5 and C2 of AC2), sAC, sGC, EF, and ACT in the presence of MnCl2. N-Methylanthraniloyl (MANT)-GTP inhibited C1·C2 with a Ki of 4.2 nM. Phe-889 and Ile-940 of C2 mediate hydrophobic interactions with the MANT group. MANT-inosine 5'-[-thio]triphosphate potently inhibited C1·C2 and ACs 1, 5, and 6 but exhibited only low affinity for sGC, EF, ACT, and G-proteins. Inosine 5'-[-thio]triphosphate and uridine 5'-[-thio]triphosphate were mixed G-protein activators and AC inhibitors. AC5 was up to 15-fold more sensitive to inhibitors than AC2. EF and ACT exhibited unique inhibitor profiles. At sAC, 2',5'-dideoxyadenosine 3'-triphosphate was the most potent compound (IC50, 690 nM). Several MANT-adenine and MANT-guanine nucleotides inhibited sGC with Ki values in the 200–400 nM range. UTP and ATP exhibited similar affinities for sGC as GTP and were mixed sGC substrates and inhibitors. The exchange of MnCl2 against MgCl2 reduced inhibitor potencies at ACs and sGC 1.5–250-fold, depending on the nucleotide and cyclase studied. The omission of the NTP-regenerating system from cyclase reactions strongly reduced the potencies of MANT-ADP, indicative for phosphorylation to MANT-ATP by pyruvate kinase. Collectively, AC isoforms and sGC are differentially inhibited by purine and pyrimidine nucleotides. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTS AND DISCUSSIONREFERENCES |
|---|
s, for the inhibitory G-protein, Gi, and for the diterpene, forskolin. Catalytic activity of all AC isoforms depends on the presence of divalent cations (Mg2+ or Mn2+). Membranous ACs possess two Me2+-binding sites (9–11). When mixed together, purified C1 and C2 form a functional AC that is efficiently activated by forskolin and Gs-GTPS (12, 13).
AC isoforms differ from each other in their regulation (1, 2). ACs 1–9 are all activated by Gs, whereas sAC is activated by 
(14). Forskolin activates ACs 1–8 but not AC9 or sAC. Gi inhibits ACs 1, 5, and 6. G-protein 
subunits exhibit stimulatory or inhibitory effects on AC isoforms. Ca2+/calmodulin stimulates ACs 1, 3, and 8. In addition, Mg2+ and Mn2+ show differential stimulatory effects on AC isoforms (15). Moreover, AC isoforms are differentially expressed in tissues (1, 2, 16). AC1 knockout mice show impaired cerebellar long term potentiation and impaired development of the somatosensory cortex (17, 18); AC3 knockout mice exhibit anosmia (19); AC5 knockout mice show reduced sensitivity to development of heart failure (20); and AC8 knockout mice exhibit altered stress-induced anxiety responses (21). Collectively, the differential regulation, tissue distribution, and function of AC isoforms indicate that isoform-specific AC inhibitors could constitute promising drugs.
The classic AC inhibitors are adenine nucleotides with a phosphate or polyphosphate at the 3'-O-ribosyl position (22–25). These compounds are also referred to as P-site inhibitors. P-site inhibitors are noncompetitive or uncompetitive AC inhibitors that bind to the AC-PPi conformation (26, 27). The most potent P-site inhibitors presently available are 2',5'-dd-3'-ATP (IC50 for rat brain AC, 40 nM) and 2',5'-dd-3'-A4P (IC50 for rat brain AC, 7.4 nM) (24, 25). P-site inhibitors exhibit a moderate degree of specificity for mammalian AC isoforms (28, 29), and EF is insensitive to P-site inhibitors (4). The most potent competitive AC inhibitor presently available is -L-2',3'-dd-5'-ATP (Ki for rat brain AC, 16 nM) (30). Ethyl 5-aminopyrazolo[1,5-]quinazoline-3-carboxylate is a competitive inhibitor of EF and ACT (Ki, 20 µM) (5). In contrast, the quinazoline does not inhibit ACs 1, 2, and 5. Thus, although ethyl 5-aminopyrazolo[1,5-]quinazoline-3-carboxylate is not a potent inhibitor of the AC toxins, these data nonetheless show that competitive AC inhibitors with excellent isoform specificity can be developed.
ANT- and MANT-nucleotides are environmentally sensitive fluorescence probes that show an increase in fluorescence upon interaction with a hydrophobic environment (31, 32). ANT- and MANT-nucleotides have been used successfully to study conformational changes in various nucleotide-binding proteins including G-proteins and bacterial AC toxins (32–35). 2'-d-3'-ANT-ATP inhibits EF and ACT with a Ki of 
10 µM (33). 2'-d-3'-ANT-ATP fluorescence increases upon binding of the EF and ACT activator calmodulin to the AC toxins (33, 36). The analysis of the crystal structure of the EF·Ca·calmodulin·2'-d-3'-ANT-ATP complex revealed that the ANT group interacts with Phe-586 in the switch B region of the toxin (36).
In a project that was originally aimed at developing a fluorescence assay for receptor/G-protein coupling by using fluorescent guanine nucleotides, we fortuitously identified MANT-GTPS and MANT-GMPPNP as potent competitive AC inhibitors (Ki for AC in S49 cyc- lymphoma cell membranes, 53 and 160 nM, respectively) (37). The preliminary analysis of the three-dimensional structure of AC suggested that the 2'-MANT group interacts with a hydrophobic pocket in the catalytic site (6, 37), but a precise model of the interaction of MANT-GTPS with AC remained to be established. By analogy to the studies performed with EF (33, 36), fluorescence studies with C1·C2 and MANT-nucleotides would facilitate the analysis of conformational changes in mammalian ACs associated with enzyme activation.
We were quite surprised that guanine nucleotides inhibit adenylyl cyclase with such a high potency. Most interesting, early studies with Escherichia coli AC and sGC pointed to competitive cyclase inhibition by non-cognate nucleotides including pyrimidine nucleotides, but those studies were not pursued (38–41). Moreover, GTP was reported to inhibit the soluble AC from Sf9 insect cells noncompetitively (42). These data suggest that the catalytic sites of ACs and sGC are flexible, allowing purine and pyrimidine nucleotides to bind.
The above described observations prompted us to conduct molecular modeling studies of competitive AC inhibitors with C1·C2 and a systematic analysis of the effects of purine and pyrimidine nucleotides, with particular emphasis on MANT-nucleotides, on the activity of several cyclases. These included AC in S49 and Sf9 insect cell membranes, ACs 1, 2, 5, and 6 expressed in Sf9 membranes and purified C1·C2, sAC, sGC, EF, and ACT. Here we develop a model of the interaction of C1·C2 with MANT-GTPS, and we show that purine and pyrimidine nucleotides differentially inhibit ACs and sGC, providing an excellent starting point for the development of potent and selective cyclase inhibitors.
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTS AND DISCUSSIONREFERENCES |
|---|
S, ITPS, UTPS, and XTPS were synthesized by nucleoside diphosphokinase-catalyzed transthiophosphorylation of the respective NDPs as described, with adenosine 5'-[-thio]triphosphate and GTPS as thiophosphoryl group donors (43). NTPSs were purified by Mono Q ion-exchange chromatography as described (43), resulting in product purity of >98%. 2', 5'-dd-3'-ATP was synthesized as described (22). Nucleotides from commercial sources were of the highest purity available. MANT-ITPS, MANT-GTPS, MANT-GMPPNP, MANT-GTP, MANT-GDP, 2'-d-3'-MANT-GMPPNP, 2'-d-3'-MANT-GTP, 2'-d-3'-MANT-GDP, MANT-AMPPNP, MANT-ATP, MANT-ADP, 2'-d-3'-MANT-ATP, MANT-XMP-PNP, and MANT-XDP were obtained from Jena Bioscience (Jena, Germany). BODIPY-FL-GTPS and BODIPY-FL-GMPPNP were purchased from Molecular Probes (Eugene, OR). 2'-d-UTP was from Sigma. All other unlabeled nucleotides were from Roche Applied Science. Purified calmodulin from bovine brain (as activator of EF and ACT) was obtained from Calbiochem. sGC purified from bovine lung (44) was purchased from Alexis Biochemicals (San Diego, CA). Sodium nitroprusside (as activator of sGC) and forskolin (as AC activator) were from Sigma. [-32P]ATP (3,000 Ci/mmol), [-32P]GTP (3,000 Ci/mmol), [-32P]UTP (3,000 Ci/mmol), and [-32P]GTP (6,000 Ci/mmol) were from PerkinElmer Life Sciences. [8-3H]cGMP (8.8 Ci/mmol) was from Moravek Biochemicals (Brea, CA). All other reagents were of the highest purity commercially available. C1(AC5), C2(AC2), and Gs-GTPS were purified as described (13). Detergent-dispersed extract from rat brain and purified AC from bovine brain were prepared as described (23). Cell Culture and Membrane Preparation—Sf9 cells were cultured and infected with 1:100 dilutions of high titer virus stocks as described (45). Sf9 membranes were prepared as described (46) and stored at -80 °C until use. S49 wt and S49 cyc- cells were cultured under the conditions described (47). S49 membranes were prepared as Sf9 membranes except that S49 cells were disintegrated at 4 °C and 7000 kilopascals for 30 min with a nitrogen cavitation chamber (Parr Instruments, Moline, IL) in a buffer consisting of 50 mM KH2PO4, 100 mM NaCl, and 0.5 mM EDTA, pH 7.0.
Purine and Pyrimidine Nucleotide Cyclase Assays—S49 wt and S49 cyc- cell membranes, membranes from uninfected Sf9 cells, and membranes from Sf9 cells expressing ACs 1, 2, 5, or 6 were thawed and sedimented by a 15-min centrifugation at 4 °C and 15,000 x g to remove residual endogenous nucleotides and resuspended in 75 mM Tris·HCl, pH 7.4 (final concentration in assay was 30 mM). Assay tubes contained various membranes (15–50 µg of protein/tube), 1 mM MnCl2, 100 µM forskolin, and 10 µM GTPS (as activator of mammalian or insect cell Gs) to maximally activate AC. Assay tubes additionally contained nucleotides 1–23 at concentrations from 0.1 nM to 1 mM as appropriate to construct concentration-response curves. Assay tubes containing membranes and various additions in a total volume of 30 µl were incubated for 3 min at 37 °C before initiating reactions with 20 µl of reaction mixture containing (final) [-32P]ATP (1.0–1.5 µCi/tube), 40 µM unlabeled ATP/Mn2+, 2.7 mM mono(cyclohexyl)ammonium phosphoenolpyruvate, 0.125 IU pyruvate kinase, 1 IU myokinase, and 0.1 mM cAMP. For determination of the apparent Km values of ACs for ATP, reaction mixtures contained 10 µM to 2 mM unlabeled ATP/Mn2+ plus 1 mM MnCl2. Reactions were conducted for 20 min at 37 °C. Reactions were terminated by the addition of 20 µl of 2.2 N HCl. Denatured protein was sedimented by a 1-min centrifugation at 25 °C and 15,000 x g. Sixty five µl of the supernatant fluid were applied onto disposable columns filled with 1.3 g of neutral alumina (Sigma A-1522, super I, WN-6). [32P]cAMP was separated from [-32P]ATP by elution of [32P]cAMP with 4 ml of 0.1 M ammonium acetate, pH 7.0 (48). Recovery of [32P]cAMP was 80% as assessed with [3H]cAMP as standard. Blank values were routinely 0.01% of the total amount of [-32P]ATP added. [32P]cAMP was determined by liquid scintillation counting.
For determination of the activity of sAC, reaction mixtures contained 30 mM NaHCO3 as activator (14), resulting in an increase in basal AC activity by 50%. The MnCl2 concentration was 5 mM. Assay tubes contained 0.5–2.0 µg of partially purified sAC. To determine the activity of C1·C2, reaction mixtures contained 3 nM C1(AC5), 15 nM C2(AC2), 50 nM Gs-GTPS, 1 mM MnCl2, and 100 µM forskolin. Because of the high activity of this system, AC assays with C1·C2 were conducted for 10–12 min at 30 °C to ensure linearity of reactions. To determine the activity of EF and ACT, reaction mixtures contained 5 mM MnCl2, 100 nM calmodulin plus 1 µM free Ca2+. Free Ca2+ concentrations were calculated with WinMaxC version 2.05 (www.stanford.edu/~cpatton/max-c.html) by using 100 µM EGTA as chelator. Assay tubes contained 10 pM EF or 10 pM ACT. EF and ACT were assayed at 30 °C for 10–12 min to ensure linearity of reactions. To search for potential GC and UC activity of recombinant ACs 2 and 5, reaction mixtures contained 10–100 µM unlabeled GTP plus 1.5–3.0 µCi of [-32P]GTP and 10–100 µM unlabeled UTP plus 1.5–3.0 µCi of [-32P]UTP, respectively. AC activity in detergent-dispersed extract from rat brain and the activity of purified bovine brain AC were determined as described (23, 30). Reaction mixtures contained 1.0–1.5 µCi of [-32P]ATP, 100 µM unlabeled ATP, 5 mM MnCl2, and 100 µM forskolin. The experiments for the determination of the potencies of 2',5'-dd-3'-ATP at various AC preparations shown in Fig. 5 were conducted in the presence of 1.0–1.5 µCi of [-32P]ATP, 100 µM unlabeled ATP, 5 mM MnCl2, and 100 µM forskolin to allow direct comparison with literature data (22, 24, 25, 28). All AC assays with purified proteins contained 0.2% (mass/volume) bovine serum albumin to prevent adsorption of proteins to reaction tubes. For initial G-protein activation studies with S49 wt and S49 cyc- cell membranes, reaction mixtures contained 5 mM MgCl2 instead of 1 mM MnCl2. Reaction mixtures for S49 cyc- cell membranes additionally contained 100 µM forskolin.
|
-32P]GTP (1.0–1.5 µCi/tube), 20 µM unlabeled GTP/Mn2+, 2.7 mM mono(cyclohexyl)ammonium phosphoenolpyruvate, 0.125 IU pyruvate kinase, 1 IU myokinase, and 0.1 mM cGMP. For determination of the apparent Km value of sGC for GTP, reaction mixtures contained 1–100 µM unlabeled GTP/Mn2+ plus 5 mM MnCl2. Reactions were conducted for 20 min at 30 °C. Subsequent processing of tubes was performed as for the AC assay. Recovery of [32P]cGMP from alumina columns was 80% as assessed with [3H]cGMP as standard. Blank values were routinely 0.01% of the total amount of [-32P]GTP added. To determine the AC and UC activities of sGC, the amount of sGC was increased to 6.6 ng of protein/tube, and reaction mixtures contained 2–200 µM unlabeled ATP plus 1.5–3.0 µCi of [-32P]ATP and 3–300 µM unlabeled UTP plus 1.5–3.0 µCi of [-32P]UTP, respectively. Separation of [-32P]UTP and bona fide [32P]cUMP ([3H]cUMP as standard for chromatography was not available to us) was performed as for the corresponding adenine and guanine nucleotides. Given the fact that the elution efficiency of our chromatography system for cAMP and cGMP was very similar (48), it is reasonable to assume that cUMP behaves similarly as the cyclic purine nucleotides. In the AC experiments shown in Table IV, we omitted the NTP-regenerating system consisting of mono(cyclohexyl)ammonium phosphoenolpyruvate, pyruvate kinase, and myokinase (49) to study the impact of MANT-NDP phosphorylation on the potencies of MANT-NDPs at inhibiting AC. In the AC experiments shown in Table V, ATP/Mn2+ was replaced by ATP/Mg2+, and 1 mM MnCl2 was replaced by 5 mM MgCl2.
|
|
2AR-Gs-short fusion protein or 10 µM N-formyl-L-methionyl-L-leucyl-L-phenylalanine to fully activate the formyl peptide receptor-Gi2 fusion protein. Reaction mixtures (80 µl) were incubated for 3 min at 25 °C before the addition of 20 µl of [-32P]GTP (0.2 µCi/tube). Reactions were conducted for 20 min at 25 °C. Reactions were terminated by the addition of 900 µl of slurry consisting of 5% (mass/volume) activated charcoal and 50 mM NaH2PO4, pH 2.0. Charcoal-quenched reaction mixtures were centrifuged for 15 min at room temperature at 15,000 x g. Seven hundred µl of the supernatant fluid of reaction mixtures were removed, and 32Pi was determined by liquid scintillation counting. Non-enzymatic [-32P]GTP degradation was determined in the presence of 1 mM unlabeled GTP and was <1% of the total amount of radioactivity added.
Molecular Modeling—Evaluation of competitive AC inhibitors was performed via combined molecular docking and molecular dynamics simulations based on the crystal structure of AC with ATPS (Rp-diastereoisomer) bound to the substrate-binding site formed at the junction of the C1 and C2 subunits and populated by a Mg2+ ion and a Mn2+ ion (11). Initial validation was performed by using AutoDock (50) to redock ATPS into the substrate-binding site. Proton positions, unresolved within the 3.0-Å resolution crystal structure, were added according to the default protonation scheme in SYBYL (51). Residue charges were set such that glutamate and aspartate monomers were modeled in their anionic form, and lysines were described as cations. All other amino acid residues were left as neutral species. By default, N-
on histidines was protonated, whereas N-
was left in unprotonated sp2-hybridized form. AC metal centers (Mg2+ and Mn2+) were assigned to their II oxidation state. ATPS, GTPS, and 2'-MANT-GTPS were described as tetra-anionic species with the negative charges assigned to the phosphate chain as depicted in Fig. 1. Partial atomic charges were computed for both AC and ligands via the Gasteiger Marsili formalism (52). Where available, parameters for van der Waals interactions were left at their default values as defined within AutoDock (50). However, nonbonding parameters had to be devised for the Mg2+ and Mn2+ atoms present in AC-binding sites. These values were obtained by taking the elemental Lennard-Jones parameters ((Mg2+) = -0.045 kcal/mol, 
(Mg2+) = 1.439 Å, (Mn2+) = -0.700 kcal/mol, (Mn2+) = 1.600 Å) and applying a sixth-power mixing formalism (Equation 1),
 | (Eq. 1) |
S was set to coincide with its original crystallographic location. Initial randomizations of position, orientation, and torsional degrees of freedom were employed according to the AutoDock defaults. GTPS and 2'-MANT-GTPS starting structures were chosen to maximize alignment of the triphosphate, ribosyl, and base structures of ATPS. One hundred genetic algorithm runs were performed for each ligand-AC pair in order to obtain a high quality sampling of the docking space. All other docking parameters were left at their default values.
|
To achieve a balanced assessment of AC/ligand interaction, a subset of high scoring ligand·AC complexes was isolated for more rigorous evaluation. Based on AutoDock clustering analysis (50), we isolated the top 10 scoring unique structures for each ligand·AC combination and computed an array of different scoring schemes, including the standard AutoDock score, the FlexX scoring function (F-Score) (53), the D score (54), G score (55), ChemScore (56), and PMF score (57). A recent assessment of docking methodologies advocates integrating the results from multiple scoring schemes to maximize correlation with existing affinity data of a given compound for a specific receptor (58). Accordingly, we ranked compounds according to a balanced consensus score using the following expression (Equation 2):
 | (Eq. 2) |
Miscellaneous—Protein concentrations were determined with the Bio-Rad DC protein assay kit. Substrate saturation experiments and competition experiments were analyzed by non-linear regression with the Prism 3.02 software (GraphPad, San Diego).
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTS AND DISCUSSIONREFERENCES |
|---|
S, GTPS, 2'-MANT-GTPS, and 3'-MANT-GTPS with C1·C2—It was quite surprising that MANT-GTPS, a guanine nucleotide, turned out to be a potent competitive AC inhibitor (37). The preliminary analysis of the three-dimensional structure of C1·C2 suggested that the MANT group attached to the 2'-O-ribosyl position interacts with a hydrophobic pocket in the catalytic site of AC (6, 11, 37). In contrast, the MANT group attached to the 3'-O-ribosyl position should face toward a more hydrophilic environment, decreasing inhibitor affinity (6, 11, 37). In order to test this hypothesis, we conducted molecular modeling studies using the C1·C2·ATPS·Mg2+ + Mn2+ crystal (11) as starting point. Like MANT-GTPS and MANT-GMP-PNP (37), ATPS (Rp-diastereoisomer) is a competitive AC inhibitor (IC50 for C1·C2, 1 µM) (11, 59).
For validation, ATPS was re-docked into the AC receptor as derived directly from the crystal structure (11). The conformer with the top AutoDock score differed by an r.m.s.d. score of only 1.66 Å relative to the originally co-crystallized ligand. Seventy of 100 docked structures exhibited an r.m.s.d. of less than 2.0 Å relative to the original structure, thus providing solid validation for the technique chosen. The top scoring structures for GTPS, 2'-MANT-GTPS, and 3'-MANT-GTPS docking to AC(Mg2+ + Mn2+) differed from their corresponding start structures (aligned to the crystallized ATPS) by r.m.s.d. scores of 2.09, 3.70, and 4.25 Å, respectively. These data indicate qualitative similarity in GTPS binding conformation relative to that of ATPS and modest differences upon addition of the MANT group.
Table I summarizes the binding free energies and consensus scores for those conformers of ATPS, GTPS, 2'-MANT-GTPS, and 3'-MANT-GTPS with the top consensus scores for docking into the catalytic site of C1·C2. The binding free energies were all strongly negative for Mn2+ + Mg2+ and Mn2+ + Mn2+ models and did not vary greatly by changing ligand or receptor model. These data indicate that all four ligands readily bind to AC and that the exchange of adenine against guanine does not grossly interfere with nucleotide binding to AC. Thus, the initial modeling studies supported our hypothesis that base specificity of AC is much less stringent than is generally assumed (37). The base-specificity of AC has not yet been studied extensively, but GTP inhibits soluble AC of Sf9 cells non-competitively (42), and UTP inhibits the particulate AC of E. coli competitively (38). Those experimental data fit to our modeling data.
|
S, GTPS, 2'-MANT-GTPS, and 3'-MANT-GTPS with AC, we integrated five other scoring functions as described under "Experimental Procedures." These results are reported in parentheses in Table I. In fact, some differences in the consensus score were large enough to permit identification of trends. Specifically, the consensus score pointed to stabilization of the 2'-MANT-GTPS·AC·(Mn2+ + Mn2+) complex relative to 2'-MANT-GTPS·AC·(Mg2+ + Mn2+) complex, whereas ATPS was destabilized by going from Mg2+ + Mn2+ to Mn2+ + Mn2+. These data fit the experimental data showing that the exchange of Mg2+ against Mn2+ increases the affinity of AC for 2'(3)'-MANT-GTPS (37). GTPS interactions with the AC receptor were also enhanced by going to the Mn2+-rich system but not by such a great extent. In accordance with the consensus score, GTPS at concentrations of up to 100 µM was devoid of AC-inhibitory effects in S49 cyc- cell membranes in the presence of Mn2+ (37). Only at concentration 
300 µM, a direct inhibitory effect of GTPS on AC in S49 wt and S49 cyc- membranes became apparent (data not shown). Most important, the consensus score of 3'-MANT-GTPS was much lower than the score of 2'-MANT-GTPS, supporting the hypothesis that the hydrophobic pocket in the catalytic site faces to the 2'-O-position of the ribosyl group (37).
Two- and Three-dimensional Models of the Interactions of ATPS and 2'-MANT-GTPS with C1·C2—The similarities in interactions of ATPS (Rp-diastereoisomer) and 2'-MANT-GTPS with AC were also evident in the analysis of specific H-bonding interactions. Most of the AC residues that interact with ATPS (Fig. 1A) also play a role in 2'-MANT-GTPS binding (Fig. 1B). The only residues for which H-bonding interactions with ATPS/AC(Mg2+ + Mn2+) were completely lost in the 2'-MANT-GTPS/AC(Mn2+ + Mn2+) case are Asp-1018 and Ile-1019 (whose H-bond accepting carbonyl oxygens are well positioned to interact with the adenine NH2 group but not with that of the guanine) and Val-1024 (the hydroxyl proton for which it had been an H-bond donor is replaced by the MANT group). The interaction of 2'-MANT-GTPS with Phe-400 is impaired due to phosphate rearrangement in the AC(Mn2+ + Mn2+) case to compensate for the larger metal ion. Loss of this interaction is compensated by bidentate interactions of Arg-484 with the -phosphate oxygens and Ser-1028 with the MANT group and adjoining ribosyl-O that were not observed for ATPS.
The most significant differences between ATPS and 2'-MANT-GTPS binding concern hydrophobic interactions. In both cases, the base inserts into a slot-shaped pocket with relatively flat hydrophobic walls parallel to the base plane and a combined acid (Asp-1018)/base (Lys-938) end well suited to accommodating purine bases. This is illustrated by the lowest energy structures for ATPS and 2'-MANT-GTPS bound to AC (Fig. 2). Most interesting, the hydrophobic surface extends beyond the region that accommodates the purine base. Specifically, Ile-940 (immediately to the left of the purine pocket) and Phe-889 (above the plane of the base; position indicated in parentheses) provide an adjacent region that can be readily exploited by hydrophobic substituents attached to the 2'-O-position of the ribosyl group. In fact, the lowest energy structure shows that the MANT group occupies this pocket. The strong hydrophobic interactions of 2'-MANT-GTPS with Ile-940 and Phe-889 are an explanation for its higher consensus score relative to ATPS and GTPS (Table I) and the high potency of MANT-GTPS at inhibiting AC (37). In contrast to the 2'-O-ribosyl group, the 3'-O-ribosyl group does not face a hydrophobic but rather a polar environment (Fig. 2). Thus, the hydrophobic MANT/AC interactions are not possible with 3'-MANT-GTPS without significantly disrupting other interactions, providing a rationale for its low consensus score (Table I). All these results fit to our previous data showing that 2'-d-3'-MANT-GMPPNP was less potent than 2'(3')-MANT-GMPPNP at inhibiting AC (37). Conversely, 3'-d-2'-MANT-GMPPNP should be a more potent AC inhibitor than 2'(3')-MANT-GMP-PNP. Unfortunately, however, 3'-d-2'-MANT-GMPPNP was not available to us. Phe-889 and Ile-940 are highly conserved among mammalian ACs and sGC (2). Thus, a MANT group at the 2'-O-ribosyl position should generally increase the affinity of nucleotides for ACs 1–9 and sGC compared with non-modified nucleotides. Leu-438 is another conserved amino acid in the catalytic site of mammalian ACs and sGC (2) that is localized between the MANT group and the purine (Fig. 2). Leu-438 exhibits van der Waals interactions with the protons on both of the above ring systems, stabilizing nucleotide/AC interactions as well. However, those interactions are less important than the interactions of the MANT group with Phe-889 and Ile-940.
|
S affinity was obtained in going from the AC·(Mg2+ + Mn2+) receptor to AC·(Mn2+ + Mn2+) (consensus score in Table I). A smaller increase was noted for GTPS, whereas no increase was obvious for ATPS. This difference is partially attributed to steric interactions between the larger Mn2+ and the thiophosphate tail. In ATPS, the S on the tail (inherently bulkier than an O) comes in close conjunction with the large Mn2+ center, hindering an interaction between the S and a crystallographic water (a cyancolored lump shown in Fig. 2 immediately beside the metal ions). No such disruption is observed with GTPS and 2'-MANT-GTPS, affording them some advantage in the Mn2+-rich receptor. Fig. 2B also shows that weak electrostatic interactions are possible between lone electron pairs on MANT's carbonyl O and amino N and the uppermost Mn2+ center on the receptor. This is not observed in the case of 2'-MANT-GTPS docking to AC·(Mg2+ + Mn2+) because the smaller Mg2+ permits it to bind more tightly to an anionic pocket formed by residues Asp-396 and Asp-440, making the ion less accessible to the large MANT group than is the case for a second Mn2+. This could be a primary reason for the greater advantage derived by MANT-GTPS (relative to GTPS) in the AC· (Mn2+ + Mn2+) receptor.
Enzymatic Analysis of ACs and sGC
Identification of ITPS and UTPS as Mixed G-protein Activators and AC Inhibitors—The modeling data discussed showed that the catalytic site of AC is flexible and spacious and can readily accommodate guanine nucleotides. Although MANT-GTPS is one of the most potent AC inhibitors known, it is not optimal for studying the effects of G-proteins on conformational change in AC by fluorescence spectroscopy because MANT-GTPS also binds to Gs- and Gi-proteins (34, 35, 37). Such interaction would compromise the proper interpretation of fluorescence studies in which C1·C2 and G-proteins are combined. Therefore, we wished to identify nucleotides that exhibit a higher selectivity for AC relative to G-proteins than guanine nucleotides. In previous studies we showed that NTPs activate Gs-proteins in the order of potency GTP > ITP > UTP > XTP > CTP (47, 60), rendering hypoxanthine, uracil, xanthine, and cytidine nucleotides potential candidates for AC inhibitors. Because a -thiophosphoryl group was favorable for AC inhibition relative to a ,-imidophosphoryl group (37), we examined the effects of NTPSs on AC activity in S49 wt membranes (Fig. 3A), a model system for the analysis of Gs-regulation of AC (47, 61), and S49 cyc- membranes (Fig. 3B), a model system for the analysis of Gi regulation of AC (62, 63). In S49 wt membranes, NTPSs activated AC via Gs in the expected order of potency (GTPS > ITPS > UTPS > XTPS > CTPS) (47, 60). For AC inhibition via Gi, the order of potency of NTPSs was GTPS > ITPS > XTPS > UTPS CTPS.
|
S and UTPS at concentrations of >1 µM strongly inhibited AC both in S49 wt and S49 cyc- membranes. ITPS and UTPS at 100 µM almost completely abolished AC activity. In contrast to ITPS and UTPS, GTPS, XTPS, and CTPS did not exhibit biphasic effects on AC. These data suggest that ITPS and UTPS are mixed G-protein activators and AC inhibitors. At concentrations 300 µM, AC-inhibitory effects of GTPS and CTPS in S49 wt and S49 cyc- membranes became evident (data not shown), indicating that these NTPSs inhibit AC with considerably lower potency than ITPS and UTPS. To the best of our knowledge, the data shown in Fig. 3 are the first evidence that mammalian ACs are inhibited by hypoxanthine and uracil nucleotides. Inhibition of AC by UTPS is particularly remarkable because the pyrimidine ring is considerably smaller than the purine ring (47). These data support the conclusions of the modeling studies showing that the catalytic site of mammalian AC is very flexible, allowing both purine and pyrimidine nucleotides to bind.
Potent and Selective AC Inhibition by MANT-ITPS Relative to sGC and G-proteins: Comparison with MANT-GTPS—At concentrations >1 µM, ITPS exhibited direct inhibitory effects on AC in S49 wt and S49 cyc- membranes, whereas GTPS at concentrations up to 100 µM was devoid of such inhibitory effects (Fig. 3) (47). Additionally, hypoxanthine nucleotides bind to other prototypical GTP-utilizing/GTP-binding proteins such as sGC and G-proteins with lower affinity than the corresponding guanine nucleotides (Fig. 4) (39, 43, 60, 64). Considering that MANT-GTPS is a potent competitive AC inhibitor (Ki for S49 cyc- membrane AC, 53 nM) (37), we predicted MANT-ITPS to be a potent AC inhibitor with reduced affinity for sGC and G-proteins relative to MANT-GTPS. In fact, in S49 cyc- membranes, MANT-ITPS inhibited AC with a Ki value of 24 nM (Fig. 4A). MANT-ITPS inhibited sGC 5-fold less potently than MANT-GTPS (Fig. 4B), and the overall potencies of the MANT-nucleotides at sGC were considerably lower than at AC. With respect to G-proteins, the difference in selectivity between MANT-GTPS and MANT-ITPS was even more pronounced. Specifically, MANT-GTPS and MANT-ITPS inhibited the GTPase of Gs with Ki values of 450 nM and 28 µM, respectively (Fig. 4C). The corresponding Ki values for MANT-GTPS and MANT-ITPS at Gi2 (measured with the formyl peptide receptor-Gi2 fusion protein) were 250 nM and 21 µM, respectively. Thus, S49 cyc- AC binds MANT-ITPS with 1,000-fold greater affinity than Gi- and Gs-proteins, whereas the affinity difference for MANT-GTPS was just 5–10-fold. Thus, because of its increased specificity for AC relative to sGC and particularly G-proteins, MANT-ITPS is a much more suitable fluorescence probe for mammalian AC than MANT-GTPS if G-proteins are included as AC regulators.
|
S and UTPS on AC in S49 wt and S49 cyc- membranes (Fig. 3) and the differential interactions of MANT-GTPS and MANT-ITPS with S49 cyc- membrane AC, sGC, and G-proteins (Fig. 4) prompted us to conduct a systematic analysis of the structure/activity relationships of purine and pyrimidine nucleotides for inhibition of AC and sGC. We analyzed purified catalytic AC subunits (C1·C2) (12, 13), recombinant ACs 1, 2, 5, and 6 expressed in Sf9 insect cell membranes (15, 65, 66), S49 wt membranes (expressing ACs 6 and 7) (37, 67), membranes from uninfected Sf9 cells (expressing an AC of as yet unknown identity) (46, 65), and purified sGC (7, 8), sAC (3), EF, and ACT (4, 5). First, we determined the kinetics of each cyclase in substrate/saturation experiments. Determination of the Vmax values was important for experiments with Sf9 membranes expressing mammalian ACs, because the activity of the endogenous AC of the insect cells could interfere with the analysis of recombinant ACs (46, 65). Km values were required for the subsequent calculation of Ki values. It should also be noted that the AC reaction is bireactant, i.e. it requires both ATP/Me2+ and free Me2+ (9–11). As is common practice (10, 12, 15, 33), we conducted cyclase experiments with varying concentrations of NTP/Me2+ and a fixed concentration of free Me2+ (see "Experimental Procedures"). Because we did not vary free Me2+ concentrations, our kinetic data underestimate true Vmax and Km values. Accordingly, our data actually represent apparent Km and Vmax values. Because Ki value calculations are based on apparent Km values, the Ki values reported in this paper are accordingly apparent Ki values.
In the presence of Mn2+, the apparent Km values of cyclases ranged between 15 and 240 µM, with sGC exhibiting the highest and sAC exhibiting the lowest substrate affinity (Table II). The low substrate affinity of sAC was noted before (3). In addition, our apparent Km values for C1·C2 and sGC fit well to literature data obtained under similar experimental conditions (12, 41). The Vmax of AC activity in membranes from uninfected Sf9 cells was 200 pmol/mg/min. This value is in excellent agreement with data published previously (65) and represents the background above which the activity of recombinant mammalian ACs is superimposed. The Vmax values of AC activity in Sf9 membranes expressing ACs 1, 2, 5, and 6 were 6–8-fold higher than in membranes of uninfected Sf9 cells. These data are also consistent with values in the literature (65, 66). Thus, the background activity of endogenous insect cell AC in Sf9 membranes expressing recombinant ACs was low and did not interfere with the analysis of mammalian ACs. The Vmax of C1·C2 is also in accord with the literature (12). The Vmax value of sGC was 3–6-fold lower than reported in the literature (44). An explanation for this difference could be that the Vmax values reported in the literature may have been obtained with fresh sGC preparations, whereas we worked with sGC that was stored at -80 °C. In fact, we observed that storage of sGC at -80 °C for up to a year decreased Vmax by 50%. However, Km values and sensitivity to inhibition by MANT-nucleotides of sGC were not altered (data not shown). The Vmax value of EF fits the published data (5, 36). Concerning ACT, we noted that our Vmax values in the presence of Mn2+ were 10-fold lower than the Vmax values in the presence of Mg2+ reported in the literature (36). In fact, the exchange of 5 mM MnCl2 against 5 mM MgCl2 increased the Vmax of ACT from 120 to 400 s-1. However, ACT was the only cyclase studied that exhibited a higher Vmax in the presence of Mg2+ than in the presence of Mn2+. In addition, the available literature evidence suggested that inhibitor sensitivity of cyclases in general is higher in the presence of Mn2+ than in the presence of Mg2+ (12, 23, 27, 30, 37, 41). Therefore, we conducted the systematic analysis of the structure/activity relationships of inhibitors in the presence of Mn2+. Nonetheless, we also examined the kinetics of representative cyclases and their sensitivity to inhibitors in the presence of Mg2+ (see below).
|
We also studied two other fluorescent nucleotides, BODIPY-FL-GTPS (compound 19) and BODIPY-FL-GMPPNP (compound 14), because those nucleotides possess more favorable optical properties than MANT-nucleotides (68). However, the BODIPY group is much bulkier than the MANT group, which property can cause steric problems in the binding of BODIPY-nucleotides to target proteins (68, 69). The BODIPY group in compound 14 is attached to the 2'(3')-O-position and undergoes spontaneous isomerization as does the MANT group in compounds 1–6, 8, 10, 11, and 16, whereas in compound 19, the BODIPY group is attached to the -thiophosphate (68, 69). We also examined the phosphorothioates ITPS (compound 13) and UTPS (compound 17) that are mixed G-protein activators and AC inhibitors (Fig. 3). To eliminate the Gs-protein stimulatory effects of compounds 13 and 17 (relevant for the analysis of ACs 1, 2, 5, and 6 and AC in S49 wt and Sf9 membranes), we maximally activated mammalian and insect cell Gs-proteins with GTPS (10 µM). Under these conditions, only monophasic inhibition curves but no stimulatory effects of 13 and 17 on AC were observed (data not shown). In addition, substrate/saturation experiments with various fixed concentrations of compounds 13 and 17 in S49 wt membranes in the presence of Mn2+ and GTPS were analyzed with non-linear regression and revealed competitive interaction between ITPS or UTPS and ATP (data not shown). We included the two naturally occurring uracil nucleotides, 2'-d-UTP (compound 20) and UTP (compound 23), in our study as well. Finally, because the focus in previous AC inhibitor research was on P-site inhibitors rather than competitive AC inhibitors (23, 25, 70), we examined the potent P-site inhibitor, 2',5,-dd-3'-ATP (compound 25), at selected AC preparations for comparison.
Inhibitor Analysis of C1·C2—We used C1·C2 as reference for our inhibition studies because the molecular modeling was performed with this enzyme (Figs. 1 and 2 and Table I). We arranged AC inhibitors in descending order of potency in Table II. Among all compounds studied, MANT-GTP (compound 1) was the most potent C1·C2 inhibitor. A Ki value of 4.2 nM is the lowest inhibitor constant for an AC inhibitor reported so far. Substrate/saturation experiments with C1·C2 and various concentrations of MANT-GTP were analyzed with non-linear regression and revealed competitive interaction of the MANT-nucleotide and ATP (data not shown). These data are in agreement with the data obtained for inhibition of AC by
