Nucleotidyl Cyclase Activity of Particulate Guanylyl Cyclase A: Comparison with Particulate Guanylyl Cyclases E and F, Soluble Guanylyl Cyclase and Bacterial Adenylyl Cyclases Cyaa and Edema Factor

Guanylyl cyclases (GCs) regulate many physiological processes by catalyzing the synthesis of the second messenger cGMP. The GC family consists of seven particulate GCs (pGCs) and a nitric oxide-activated soluble GC (sGC). Rat sGC α1β1 possesses much broader substrate specificity than previously assumed. Moreover, the exotoxins CyaA from Bordetella pertussis and edema factor (EF) from Bacillus anthracis possess nucleotidyl cyclase (NC) activity. pGC-A is a natriuretic peptide-activated homodimer with two catalytic sites that act cooperatively. Here, we studied the NC activity of rat pGC-A in membranes of stably transfected HEK293 cells using a highly sensitive and specific HPLC-MS/MS technique. GTP and ITP were effective, and ATP and XTP were only poor, pGC-A substrates. In contrast to sGC, pGC-A did not use CTP and UTP as substrates. pGC-E and pGC-F expressed in bovine rod outer segment membranes used only GTP as substrate. In intact HEK293 cells, pGC-A generated only cGMP. In contrast to pGCs, EF and CyaA showed very broad substrate-specificity. In conclusion, NCs exhibit different substrate-specificities, arguing against substrate-leakiness of enzymes and pointing to distinct physiological functions of cyclic purine and pyrimidine nucleotides.


Introduction
Guanosine 39,59-cyclic monophosphate (cGMP) is a wellestablished second messenger that is involved in the regulation of many physiological processes like blood homeostasis, neurotransmission, intestinal secretion, and vision [1]. cGMP is generated from GTP by guanylyl cyclases (GCs) that are expressed in two isoforms: the soluble (sGC) and particulate or membranebound (pGC) form. Seven pGC isoforms have yet been found (pGC-A to pGC-G) [1][2][3]. The best described pGC is the natriuretic peptide (NP) receptor A (NRP1, pGC-A) that is activated by atrial NP (ANP) or less effectively by B-type NP (BNP) [1]. The receptor is a dimer consisting of an extracellular ligand binding domain and an intracellular domain which contains a transmembrane region, a juxtamembranous protein kinase homology domain (KHD), a hinge region, and an intracellular domain with the catalytically active GC site [1]. Recently, biochemical studies with pGC-A and pGC-B revealed that both isoforms possess separate and distinct catalytic and allosteric binding sites that bind GTP and ATP, respectively, under physiological conditions [4].
pGC-B is mainly expressed in vascular endothelial cells. pGC-B mediates vascular regeneration, cell growth, and endochondral ossification through binding of C-type natriuretic peptide and by activating cGMP-dependent protein kinase II [1]. pGC-C is expressed in the intestinal epithelium [1]. This pGC is activated by the peptides guanylin and uroguanylin, and bacterial heat-stable enterotoxins regulating electrolyte and water transport in the intestine, and epithelial cell growth and differentiation. pGC-D is located in the olfactory neuroepithelium playing an important role in odor recognition and modulating the sensitivity of sensory neurons [1].
Further members of the GC family are the isoforms pGC-E and pGC-F (ROS-GC-1 and ROS-GC-2). They play a key role in vertebrate phototransduction and are localized in the rod outer segment (ROS) [5]. pGC-E and pGC-F are regulated by neuronal calcium sensor proteins, named guanylyl cyclase-activating proteins (GCAPs). These regulatory proteins sense changes in the cytoplasmic Ca 2+ -concentration during illumination and activate ROS-GCs when the cytoplasmic Ca 2+ decreases below the value in a dark-adapted cell [5]. The GCAP-dependent stimulation is inhibited by increasing Ca 2+ [5].
Particulate fractions of sea urchin sperm expressing relatively high amounts of pGC-A were shown to generate inosine 39,59cyclic monophosphate (cIMP) and 29-deoxyguanosine 39,59monophosphate beside cGMP in the presence of Mn 2+ [6]. Additionally, the recombinantly expressed catalytic core of rat pGC-A synthesizes cGMP and 29-deoxyguanosine 39,59-monophosphate in the presence of Mn 2+ [7]. In the study by Thorpe et al. [7], it was also reported that with a radiometric method combined with thin-layer chromatography, ATP, CTP and UTP were not substrates for rat pGC-A, but no experimental data were documented in a Figure or a Table (data were reported only as data not shown). Moreover, the substitution of two amino acids in RetGC-1 converts the ''guanylyl'' cyclase into an ''adenylyl'' cyclase [8]. Rat sGC a 1 b 1 generates other cyclic nucleotides like cXMP (xanthosine 39,59-cyclic monophosphate) and cCMP (cytidine 39,59-cyclic monophosphate) [9]. Since pGC-A and sGC exhibit , 50% amino acid identity in the catalytic domain [10], here, we studied the substrate-specificity of pGC-A as well as pGC-E and pGC-F of bovine ROS. We used HPLC-MS/MS for determination of kinetic parameters for generation of the following seven 39,59-cyclic monophosphates (cNMPs): adenosine 39,59cyclic monophosphate (cAMP), cCMP, cGMP, cIMP, thymidine 39,59-cyclic monophosphate (cTMP), uridine 39,59-cyclic monophosphate (cUMP), cXMP [9]. Additionally, we systematically analyzed the substrate-specificity of the adenylyl cyclase (AC) toxins CyaA from Bordetella pertussis and edema factor (EF) from Bacillus anthracis by HPLC-MS/MS because both toxins generate cCMP and cUMP as assessed by radiometric and HPLC-based methods [11]. The comparison of EF with pGC-A is also important because the toxin, like pGC-A from sea urchin sperm generates cIMP [6,11,12]. Furthermore, CyaA and EF are potently inhibited by various purine-and pyrimidine-substituted 29(39)-O-(N-methylanthraniloyl)-nucleotides [13,14] indicating a high flexibility of their catalytic sites and pointing to potentially broad substrate-specificity. Our studies reveal very distinct substrate-specificities of various NCs, ranging from very broad (bacterial AC toxins) to relatively narrow (pGCs).

Membrane Preparation of HEK293 Pgc-A Cells
HEK293 cells stably expressing FLAG-tagged rat pGC-A were kindly provided by Dr. Michaela Kuhn (Institute of Cardiovascular Physiology, University of Würzburg, Germany) and grown in DMEM high glucose 4.5 g/L (PAA, Cölbe, Germany) supplemented with penicillin at 100 U/mL, streptomycin at 0.1 mg/mL, L-glutamine at 0.292 g/L (PAA) and 10% (v/v) FBS (Lonza, Basel, Switzerland) up to 70-80% confluency [15]. For membrane preparations cells were washed twice in 16 Dulbeccos PBS (PAA) and re-suspended in homogenization buffer (50 mM HEPES pH 7.4, 100 mM NaCl, 1 mM EDTA, 10% (v/v) glycerol, 2 mg/ mL aprotinin, 5 mg/mL pepstatin, and 2 mg/mL leupeptin). Cells were lysed using a nitrogen cavitation bomb (Parr, Moline, IL, USA) after incubation at 100 bar for 30 min. Lysate was centrifuged for 5 min at 500 6 g and 4uC to remove cell nuclei. Subsequently, the supernatant fluid was centrifuged for 20 min at 30,000 6 g and the pellet was re-suspended in 1 mL homogenization buffer using a 22G-needle. Membrane preparations were aliquoted in 10-100 mL batches and stored at 280uC until analysis. Protein concentrations of membrane preparations were determined by the Bradford protein quantitation method.

NC Assay for pGC-A in HEK293 Cell Membranes
Membrane preparations of HEK293 cells stably overexpressing rat pGC-A (10-80 mg of protein per tube) were incubated in a total volume of 100 mL at 37uC in the presence of incubation buffer, 2-7,500 mM NTP/Me 2+ , plus additional 4 mM of the corresponding MeCl 2 . The latter MeCl 2 addition is not anymore indicated in legends for the sake of brevity. In preliminary studies (Figs. S1 and S2), we conducted time course experiments with 200 mM NTP/Me 2+ to identify linear conditions for determination of precise enzyme kinetics and to allow comparison with the data previously published for sGC [9]. Kinetic studies were then conducted under linear conditions previously identified in the time course experiments shown in Figs. S1 and S2. Reaction conditions for kinetic studies ensured that no substrate depletion occurred with .90% of the substrate still being present. Assays were stopped by heating for 10 min at 95uC. After cooling, mixtures were diluted by 200 mL of a solution consisting of 97/3 (v/v) water/methanol containing 50 mM ammonium acetate, 0.1% (v/ v) acetic acid, and 100 ng/mL tenofovir. Denatured protein was precipitated by centrifugation for 10 min at 20,000 6 g. Quantitation via HLPC-MS/MS was performed as described [9].
NC Assay for pGC-A in Intact HEK293 Cells HEK293 cells stably overexpressing rat pGC-A were seeded in a 6-well-plate with 5?10 5 cells per well for 24 h in DMEM high glucose with 10% (v/v) FBS and 200 mg/mL L-glutamine, 100 U/ mL penicillin, and 0.1 mg/mL streptomycin. The next day, cells were pre-incubated for 10 min with IBMX (100 mM) followed by stimulation with 1 mM ANP for various times. To stop stimulation and cell metabolism, medium was removed and 300 mL extraction solution consisting of acetonitrile/methanol/water (2:2:1, v/v/v)) and 25 ng/mL tenofovir was added. Cell suspension was heated for 20 min at 95uC and centrifuged at 20,000 6 g for 10 min to remove protein. Supernatant fluid was evaporated completely under nitrogen atmosphere at 40uC. Residue was resolved in 150 mL water and analyzed by HPLC-MS/MS as described in Ref. 9 except the fact that separation was performed on an Aglient 1100 series (Waldbronn, Germany) and for detection the more sensitive QTrap 5500 triple quadrupole mass spectrometer (ABSCIEX, Foster City, CA, USA) was used which is, according to our own experiments, up to 5-fold more sensitive than the mass spectrometer applied in the previous study. Parameters of HPLC-MS/MS fragments are documented in Table 1. Ion source settings and collision gas pressure were manually optimized regarding ion source voltage, ion source temperature, nebulizer gas, and curtain gas (ion source voltage of 5,500 V; ion source temperature of 600uC; curtain gas of 30 psi; collisionally activated dissociation gas of 9 psi). Chromatographic data were collected and analyzed with Analyst 1.5.1 (ABSCIEX). Quantitation was performed with nitrogen as collision gas. For determination of protein concentration, cell pellets of extraction procedure were dried at room temperature and resolved in 0.1 M NaOH at 95uC for 20 min. Ten microliters of protein solution were removed for quantitation of protein concentration by means of bicinchoninic acid protein assay.

NC Assay with ROS Membrane Preparations
Bovine ROS was prepared as described in Ref. [16] under very dim red light using sucrose density centrifugation in the presence of 115 mM NaCl, 2.5 mM KCl, 1 mM MgCl 2 , 10 mM HEPES/ KOH pH 7.5 and 1 mM DTT. For GC assays ROS membranes (81 mg rhodopsin per tube) were incubated for 5 min at 30uC with 1 mM GTP/Mg 2+ , 1 mM UTP/Mg 2+ or 1 mM UTP/Mg 2+ . Buffer contained 30 mM MOPS pH 7.2, 60 mM KCl, 4 mM NaCl, 3.5 mM MgCl 2 , 0.3 mM ATP, 0.16 mM zaprinast, 1 mM DTT, and 2 mM EGTA or 2 mM CaCl 2 , as indicated. Reactions were stopped by heating at 95uC for 10 min and analyzed as described in [17]. In brief, concentrations of cGMP, cCMP, cUMP, GMP, CMP, and UMP were determined by a LC/MS system consisting of a binary HPLC pump (1100 Series, Agilent, Waldbronn, Germany) directly coupled to a single quadrupole mass spectrometer (LC-MSD SL, Agilent) operating in negative ion mode. Fifty mL of sample solution were loaded onto a Hypercarb column (30 6 4.6 mm, 5 mm particle size) (Thermo-Fisher, Dreieich, Germany) preceded by a column saver (2 mm, Supelco Analytical) and a C 18 security guard (AJO-4286, Phenomenex). Compounds were separated by means of a linear gradient from 96% (v/v) eluent A consisting of 10 mM ammonium acetate (pH 10.0) up to 60% eluent B (acetonitrile) within 8 min and a flow rate of 0.4 mL/min. Subsequently, reequilibration to 96% eluent A was conducted for 5 min. NC Activity Assay with Purified Cyaa and EF E. coli cells were transfected with plasmid pProExH6-EF and pExCyaA-N. Full-length EF and the catalytic domain of CyaA-N (amino acids 1 to 373) were purified from E. coli [18]. CaM was purified from calf brain [19]. The activity of the extracted enzymes was tested by assays with [a-32 P]ATP [13]. [a-32 P]ATP (3,000 Ci/ mmol) was purchased from Perkin Elmer Life Sciences (Boston, MA, USA). For Michaelis-Menten kinetics of EF and CyaA-N with the substrate ATP 100 pM enzyme was used, whereas 100 nM of AC toxins was applied when determining k cat and K M of other NTPs. The NTP/Mg 2+ concentrations were varied between 5 mM and 1 mM. The CaM concentration was adjusted to a stoichiometry of 1:10 for AC toxin to CaM. The assay tubes contained final concentrations of 5 mM Mg 2+ , 10 mM Ca 2+ , 10 mM TRIS?HCl pH 7.5, and 0.1 wt% BSA. The temperature for the assay was set to 37uC and reaction times were varied according to the NTP/Mg 2+ used (GTP/Mg 2+ : 30 min; TTP/ Mg 2+ : 60 min; CTP/Mg 2+ , UTP/Mg 2+ : 10 min; ATP/Mg 2+ : 5 min; XTP/Mg 2+ , ITP/Mg 2+ : 90 min). The total volume was 50 mL per tube. The reaction was stopped by heat-inactivation at 95uC for 5 min. The precipitated proteins were removed by centrifugation. Fourty mL of the supernatant fluid was combined with 40 mL of internal standard (IS) solution. As internal standard cXMP was used at a concentration of 200 ng/mL when determining the turnover of all NTPs except for XTP/Mg 2+ . When the Michaelis-Menten kinetic of XTP/Mg 2+ was determined, 200 ng/mL cIMP was applied as internal standard. Quantitation of generated cNMPs was performed by HPLC-M/ MS. The chromatographic separation of cAMP, cCMP, cUMP, cIMP, and cTMP and the internal standard cXMP was performed on a LC-10ADVP HPLC system (Shimadzu, Kyoto, Japan) equipped with a binary pump system. A combination of Supelco column saver (2.0 mm filter, Supelco Analytical, Bellafonte, CA, USA), security guard cartridge (C 18 , 462 mm) in an analytical guard holder KJO-4282 (Phenomenex, Aschaffenburg, Germany) and an analytical Nucleodur C 18 Pyramid RP column (5063 mm, 3 mm particle size, Macherey-Nagel, Düren, Germany) temperature controlled by a convenient HPLC column oven at 25uC was used. Eluent A consisted of 5 mM ammonium acetate and 0.1% (v/v) acetic acid in water and eluent B was methanol. The injection volume was 50 mL and the flow rate was 0.4 mL/min throughout the chromatographic run. Eluent A (100%) was used from 0 to 5 minutes followed by a linear gradient from 100% A to 70% A until 9 min. Eluent A (70%) was kept until 11 min and reequilibration of the column at 100% A was achieved from 11 to 15 min. The retention times of analytes were as follows: cAMP: 9.7, cCMP: 5.0, cIMP: 8.7, cTMP: 8.8, cUMP: 7.9 and cXMP: 8.4 min. The analyte detection was performed on an API 2000 triple quadrupole mass spectrometer (Applied Biosystems) using selected reaction monitoring (SRM) analysis in positive ionization mode. The following SRM transitions (m/z) using a dwell time of 40 ms were detected: cAMP: +330/136; cCMP: +306/112; cIMP: +331/137; cTMP: +305/127; cUMP: +307/97; cXMP: +347/ 153. The mass spectrometer parameters were as follows: ion source voltage: 5500 V, temperature: 350uC, curtain gas: 40 psi, collisionally activated dissociation (CAD) gas: 5 psi.

Statistics
Presented data are the mean with range of two independent experiments or the means 6 standard error of the mean (SEM) of six independent experiments, as indicated. GraphPad Prism software version 5.01 (San Diego, CA, USA) was used for nonlinear regression and calculation of mean, SEM, s 0.5 , V max , n Hill , and IC 50 values.

Time-Courses of pGC-A
In order to identify suitable experimental conditions for determination of enzyme kinetics, we first investigated the timedependent generation of cNMPs (Figs. S1 and S2). Membranes (10-80 mg protein per tube) from HEK cells stably expressing pGC-A were activated by 1 mM ANP in the presence of 200 mM NTP/Me 2+ plus additional 4 mM MnCl 2 or MgCl 2 , respectively, followed by incubation at 37uC for 2-100 min. In the presence of Mn 2+ pGC-A showed a linear generation of cGMP for up to 100 min (Fig. S1A). However, when Mn 2+ was replaced by Mg 2+ , we could hardly detect GC activity ( Fig. 1) (10). Therefore, we added ATP/Mg 2+ (500 mM) that reduces the Hill coefficient and decrease the K M [4]. Under these conditions, a linear production of cGMP up to 15 min was observed (Fig. S2), with the GC activity being even higher than in the presence of Mn 2+ (Fig. 1). We also examined the influence of ATP/Mn 2+ (0.5 mM) on GC activity in the presence of additional 4 mM MnCl 2 . Under these conditions, an inhibitory effect on GC activity was reported [7,20,21]. Indeed, we observed a reduction of cGMP production, in parallel with cAMP generation (Fig. 1).
In the presence of Mn 2+ , pGC-A also generated cAMP and cXMP (Figs. S1C-D), although much lower catalytic activities than for cGMP formation were observed. Moreover, we observed cIMP formation in the presence of Mn 2+ as well as in the presence of Mg 2+ (Figs. S1 and S2). The formation rate in presence of Mn 2+ was cIMP.. cGMP.. cAMP. cXMP. In contrast to sGC, no catalytic activity for the generation of cyclic pyrimidine nucleotides was observed (Figs. S1E and F). In the presence of Mg 2+ , the formation rate of cGMP was significantly higher than for cIMP (Figs. S2A and B), and generation of cAMP, cXMP, cCMP, cUMP, and cTMP could not be detected even after an incubation of 60 min (Fig. S2C). The detergent Triton X-100 largely enhanced the catalytic activity of pGC-A with ATP and GTP as substrates. However, Triton X-100 did not render CTP, TTP or UTP pGC-A substrates (Fig. S3).

Enzyme kinetics of pGC-A in Membranes
To determine accurate kinetic parameters of pGC-A, all further experiments were performed within linear reaction times (5-10 min) and non-depleting substrate conditions (Figs. S1 and S2). Fig. 2 and Table 2 summarize the kinetic parameters obtained from substrate saturation experiments for pGC-A in the presence of Mn 2+ or Mg 2+ . In the presence of Mn 2+ , a very robust GC activity with a s 0.5 of 231 mM and a V max of 9.8 nmol min 21 mg 21 was observed. The Hill coefficient (n Hill ) of 1.5 indicated strong cooperativity. When Mn 2+ was replaced by Mg 2+ , GC activity could only be detected when ATP/Mg 2+ was added. Under these conditions, affinity and cooperativity towards GTP were slightly decreased and V max was reduced to 7.3 nmol min 21 mg 21 . Besides GC activity, an inosinyl cyclase (IC) activity could also be detected. In the presence of Mn 2+ , V max , s 0.5 , and n Hill of IC activity were within the same range (14.4 nmol min 21 mg 21 , 232 mM, and 1.8, respectively) as for GTP. With Mg 2+ , IC activity strongly decreased to a V max of 1.0 nmol min 21 mg 1 and affinity decreased. However, Hill coefficient hardly changed after cofactor substitution. In the presence of Mn 2+ , we could also detect very low V max values for adenylyl (AC) and xanthosinyl cyclase (XC) activities with ,0.2 nmol min 21 mg 21 . ATP and XTP showed nearly the same affinities towards pGC-A as GTP with a s 0.5 of 224 and 333 mM, respectively. AC activity showed only a small cooperativity with n Hill of 1.1. XTP was best fitted using a substrate inhibition model. However, determination of K i was ambiguous. pGC-A is regulated by ATP decreasing the s 0.5 of the cyclase [4,22,23]. For XTP, no regulation of pGC-A has yet been described. Therefore, we analyzed the influence of XTP on GC activity. Fig. 3 shows that in contrast to ATP, XTP reduced GC activity with an IC 50 of 145.361.2 mM. A simultaneous generation of cXMP was not observed.

Analysis of NC Activity of pGC-A in Intact HEK293 Cells
We also analyzed the formation of cyclic nucleotides by pGC-A after stimulation by ANP (1 mM) in intact HEK293 cells. Stimulation of intact HEK293 cells stably overexpressing rat pGC-A by ANP resulted in a very strong and long-lasting increase of cGMP-formation (Fig. 4A). Interestingly, substantial basal concentrations of cAMP, cUMP, and cCMP were observed in intact HEK293 cells (Figs. 4B-4D). However, stimulation by ANP did not increase cCMP, cUMP and cAMP concentrations in intact HEK293 cells. cXMP, cIMP and cTMP were below the detection limit under basal conditions in intact HEK293 cells and were not increased above the detection limit after ANP stimulation (Fig. 4E).

Analysis of pGC-E and pGC-F
To answer the question whether the lack of pyrimidinyl cyclase activity of GC in the presence of Mg 2+ is a general property of GCs, we analyzed preparations of ROS containing pGC-E and pGC-F. Membranes (81 mg rhodopsin per tube) were incubated for 5 min at 30uC with 3.5 mM MgCl 2 and 1 mM GTP/Mg 2+ , UTP/Mg 2+ , and CTP/Mg 2+ , respectively. As expected, we observed GC activity in samples containing GTP (Fig. 5A). Samples containing whole ROS as well as the addition of CaCl 2 resulted in a decrease to basal cGMP concentrations. No pyrimidinyl cyclase activity was observed in ROS membranes (Fig. 5A). Besides generation of cGMP, high amounts of nucleoside 59-monophosphates (NMPs) were detected (Fig. 5B) although a PDE-inhibitor was added. To evaluate if these NMPs were generated by nucleotidases or phosphatases or were contaminants of the stock solution we determined NMP levels in the absence of membrane. As illustrated in Fig. 5C the detection of GMP in Fig. 5B was largely due to enzymatic degradation of GTP. However, the presence of CMP and UMP was due to impurity of NTP standards with about 0.4%.

NC Activity of CyaA and EF
A previous study on the substrate-specificities of EF and CyaA used radiometric-and HPLC-based assays and focused on Mn 2+ as cofactor for catalysis [11]. Moreover, complete kinetic analysis was only performed for ATP, CTP and GTP as substrates. In our present study, we used a highly sensitive HPLC-MS/MS-based method and assessed complete kinetics under Mg 2+ conditions for ATP, CTP, UTP, ITP, GTP, TTP and XTP (

NC Activity of Pgcs
sGC possesses purinyl and pyrimidinyl cyclase activity [9]. Garbers and coworkers reported that sea urchin sperm pGC generates cIMP in the presence of Mn 2+ [6], but kinetic studies were not conducted. Waldman et al. studied rat kidney membranes in the presence of Mg 2+ and found that ATP is no pGC substrate [24]. Lack of generation of cCMP, cUMP, and cAMP was reported using purified preparations of the catalytic core of pGC-A in the presence of Mn 2+ [7]. However, previous studies were performed by radiometric assays that are not unequivocal in terms of chemical identification of molecules [7,21,24]. Here, for the first time, a systematic analysis of NC activity of rat pGC-A was performed using structure-based identification and quantitation by HPLC-MS/MS. Using Mg 2+ , we could confirm that pGC-A lacks AC activity [25] and that substrate specificity of pGC with respect to NTPs (29-deoxy-NTPs were not studied here) is restricted to GTP and ITP. This is most likely due to the close structural similarity of ITP and GTP. The exchange of Mn 2+ to Mg 2+ dramatically decreased formation of cIMP by pGC-A. In contrast, the exchange of Mn 2+ against Mg 2+ did not have dramatic negative effects on V max of rat sGC a 1 b 1 with ITP [9]. In order to understand these striking biochemical biochemical differences, at a molecular level it will be necessary to resolve the crystal structures of both pGC-A and rat sGC a 1 b 1 in complex with ITP analogs that are not NC substrates. A possible ITP analog for such future crystallization studies is 29(39)-O-(N-methylanthraniloyl) inosine 59triphosphate since 29(39)-O-(N-methylanthraniloyl)-substituted NTPs bind to various NCs with high affinity [11,13,14].
We noted that the catalytic activity of pGC-A in time course experiments was linear with GTP/Mn 2+ for up to 100 min, whereas for all other NTP/Mn 2+ conditions and for all NTP/ Mg 2+ conditions, catalysis was linear only for much shorter periods of time (Figs. S1 and S2). Accordingly, incubation time for kinetic studies (Fig. 2) were adjusted to short incubation times, ensuring linearity. Similar to pGC-A, we observed only short linear reaction rates for many substrate conditions at sGC [9]. These data cannot be explained by substrate depletion since the maximum substrate conversion rates were just 8%. Probably, the substrates differ from each other in their ability to stabilize sGC and pGC, GTP being the most effective NTP in this respect [9].
Since no crystallographic data of pGC-A have been reported yet, current models of the structure of pGC-A are derived from homology modeling based on the crystal structure of ACs [10]. pGC is suggested to contain two catalytic domains that act cooperatively [1]. However, a new model postulates that pGCs are asymmetric homodimers with an allosteric and a catalytic domain that bind GTP and ATP, respectively [4]. Kinetics of pGC-A activity are complex [4,10,[26][27][28][29][30][31][32]. In the presence of Mn 2+ , V max values of 300 pmol min 21 mg 21 for bovine adrenal cortex [31] and 19 mmol min 21 mg 21 for highly purified pGC-A [10] are reported. Values of s 0.5 range between 0.017 mM for highly purified receptor [10] and 0.2 mM for rat tissue preparations [32]. The Hill coefficients (n Hill ) range between 1.421.74 indicating a highly cooperative behavior [6,[26][27][28][29][30][31][32]. In the presence of Mg 2+   with a concave reciprocal plot indicating cooperative behavior [6]. We could confirm cooperativity with n Hill of 1.8 for IC activity and s 0.5 was within the reported range (232 mM). IC activity could also been detected for the first time in the presence of Mg 2+ . Here, V max value was significantly reduced to 1 nmol min 21 mg 21 and s 0.5 increased to the millimolar concentration range. We show for the first time that pGC-A generates cAMP in the presence of Mn 2+ . Although V max values were fifty-fold lower than for GTP, affinities were in the same range as GTP. Very recently, ATP has been shown to bind at an allosteric site in the catalytic domain reducing Hill coefficient and K M [4]. Moreover, ATP inhibits the catalysis of pGC-A in a competitive manner [21]. This observation was very surprising for the authors because ATP is reported to activate full-length pGC-A [21]. Here, we show that this discrepancy is due to the AC activity of pGC-A. We also found that XTP is a substrate of pGC-A in the presence of Mn 2+ . We observed a s 0.5 within the range of GTP and a substrate inhibition at millimolar concentrations indicating that XTP influences GC activity. Indeed, we observed an inhibitory effect of XTP on GC activity in the presence of Mn 2+ . In contrast to sGC that shows pyrimidinyl cyclase activity with Mn 2+ as cofactor [9], pGC-A did not accept CTP, UTP and TTP as substrates. Our present data with holo-pGC-A regarding UTP and CTP are in accord with previous data on the catalytic core of pGC-A [7,21]. The pyrimidinyl cyclase activity of sGC may be due to unique structural features of the enzyme [33]. A molecular explanation for these biochemical differences between pGC-A and sGC hinges on the resolution of crystal structures.
The focus of our present study was the analysis of the substratespecificity of wild-type pGC-A. The precise elucidation of the molecular basis for the complex kinetic properties of pGC-A and the analysis of the interaction of the catalytic and allosteric site will require extensive mutagenesis studies [4,34,35]. As an additional approach, (Rp)-and (Sp)-diastereoisomers of nucleoside 59-O-(1and 2-thio)triphosphates should be studied since the stereoisomers differ from each other in their ability to serve as substrates for pGCs [36].
Dessauer et al. [37] reported that PP i , one of the products of the NC reaction, at concentrations between 0.3-1.6 mM, can inhibit catalysis of mammalian membranous AC. Hence, we have to      [36] also argue against the hypothesis that inhibition of catalysis by PP i plays a major role for complex enzyme kinetics of pGC. Stimulation of intact HEK293 cells stably overexpressing pGC-A with ANP resulted in a massive and long-lasting increase of cGMP formation (Fig. 4A), whereas no increases in cAMP, cIMP, cXMP, cUMP, cCMP and cTMP were detected. Our failure to detect increases in cNMP concentrations in intact cells other than cGMP is unlikely due to cNMP degradation because we conducted experiments in the presence of a non-selective phosphodiesterase inhibitor. Most likely, in intact cells only very low levels of ITP and XTP are present [38]. Moreover, AC activity of pGC-A was very low, too, so that putative cAMP increases could not be detected. We could not confirm the natural existence of cTMP and cXMP in cells (Fig. 4E), although cTMP and cXMP had been tentatively identified in rat tissue [39]. Probably, the previously reported detection of cTMP and cXMP in tissues is an artifact due to insufficient sensitivity of the previously available mass spectrometers [40].
In ROS membrane preparations expressing pGC-E and pGC-F we observed GC activities that are typical for incubations at different Ca 2+ -concentrations and using GTP/Mg 2+ as substrate (Fig. 5A). For example, chelating Ca 2+ by EGTA resulted in about eight-fold higher activity than keeping the Ca 2+ -concentration high. These results are consistent with previous reports on GC activation profiles in bovine ROS reviewed in Ref. 5. All GCs studied so far, i.e. sGC [9], and pGC A, pGC-E and pGC-F share the property that in the presence of Mg 2+ they lack pyrimidinyl cyclase activity (Figs. S1 and S2 and 5).

NC Activity of CyaA and EF
The exotoxins CyaA and EF have already been demonstrated to generate cIMP, cCMP and cUMP [11,12] and are potently inhibited by various purine and pyrimidine nucleotide analogs [13,14]. In contrast to sGC and pGC, broad substrate specificities of CyaA and EF were even detected in presence of Mg 2+ (Table 3) In fact, both CyaA and EF exhibited CC activity, but k cat values were approximately 300 times lower than for AC activity. Both toxins also produced cUMP, cGMP, cIMP and cXMP at very low catalytic rates. cTMP production could only be detected for CyaA. XTP and TTP both exhibited substrate inhibition kinetics. Thus, among all NCs studied so far, CyaA and EF possess the broadest substrate-specificity. This is most evident in the presence of Mg 2+ .

Conclusions and Future Studies
In contrast to sGC [9], pGC-A lacks pyrimidinyl cyclase activity. Since GTP and ATP have already been shown to bind at the allosteric binding domain of pGC-A [4] future studies have to investigate whether further nucleotides regulate pGC-A by binding at this domain. sGC-generated cCMP and cUMP may serve a unique signal transduction role like cCMP and cUMP produced by the bacterial ''adenylyl'' cyclase toxins edema factor from Bacillus anthracis and CyaA from Bordetella pertussis [9,11,41].
UC-and CC activity of certain NCs cannot be considered as trivial substrate leakiness of enzymes since some NCs such as the pGCs studied herein, possess a higher substrate-specificity than sGC, EF and CyaA. Future studies will have to assess the substrate-specificities of membranous and soluble ACs. At least with regard to inhibitors, these enzymes also exhibit broad basespecificity like sGC, EF and CyaA [13,42]. Moreover, we have not yet studied pGCB, pGC-C and pGC-D [1]. At the time being, we have no evidence for the existence of a mammalian NC with preferential CC-and/or UC activity [43]. It remains to be determined whether the quite considerable basal levels of cUMP and cCMP serendipitously observed in HEK293 cells (Figs. 4C and 4D) are due to the activity of sGC or an AC. We can at least exclude pGC-A as a regulator of cellular cCMP and cUMP levels. Based on the broad substrate-specificity of EF and CyaA demonstrated in our previous study [11] and herein, it is worthwhile to analyze the substrate specificity of other bacterial NCs. ExoY from Pseudomonas aeruginosa, structurally related to EF and CyaA, is an excellent candidate [44,45]. Specifically, ExoY was previously classified as AC, but a recent study revealed that the toxin possesses much higher GC-than AC activity [45]. A tentative CC-and UC activity of ExoY has still to be explored.  Figure S4 Calibration curves of cNMPs and corresponding substrate saturation experiments with CyaA-N and EF, respectively. A-C cCMP calibration and CTP saturation, D-F cAMP calibration and ATP saturation, G-I cGMP calibration and GTP saturation and J-L cUMP calibration and UTP saturation. For kinetic parameters see Table 3. (TIF) Figure S5 Calibration curves of cNMPs and corresponding substrate saturation experiments with CyaA-N and EF, respectively. A-C cIMP calibration and ITP saturation, D-F cXMP calibration and XTP saturation, G-H cTMP calibration and TTP saturation. cTMP formation was only detected in CyaA-N but not in EF. For kinetic parameters see Table 3. (TIF)