Polyphosphates from Mycobacterium bovis – potent inhibitors of class III adenylate cyclases Ying Lan Guo 1 , Hermann Mayer 2 , Waldemar Vollmer 3 , Dorothea Dittrich 4 , Peter Sander 4,5 , Anita Schultz 1 and Joachim E. Schultz 1 1 Pharmazeutisches Institut, Universita ¨ tTu ¨ bingen, Germany 2 Institut fu ¨ r anorganische Chemie, Fakulta ¨ tfu ¨ r Chemie und Pharmazie, Universita ¨ tTu ¨ bingen, Germany 3 Institute for Cell and Molecular Biosciences, Newcastle University, UK 4 Institut fu ¨ r Medizinische Mikrobiologie, Universita ¨ tZu ¨ rich, Switzerland 5 Nationales Zentrum fu ¨ r Mykobakterien, Zurich, Switzerland cAMP is a key signaling molecule in virtually all living organisms. This ubiquity is mirrored by the abundance and diversity of the synthetic enzymes, adenylate cyc- lases (ACs). Currently, six classes of ACs exist, which share no identifiable sequence similarities. Here we deal with ACs grouped together in class III, which contains by far the most AC isozymes. Among these are all cyclases from eukaryotes and the overwhelming majority of those from bacteria [1]. In eukaryotic cells, ACs are typically pseudoheterodimeric, i.e. the result of a gene duplication [2]. Both pseudomonomers con- tribute amino acids to a single catalytic center [3,4]. Bacterial class III ACs are generally monomers that must dimerize to form two catalytic centers that are essentially identical to those of eukaryotic class III ACs [5,6]. Usually, cAMP is generated intracellularly in response to extracellular signals such as hormones, changes in ion compositions, pH or nutrients, and a variety of stress conditions. Although stimulatory conditions often persist for considerable periods of time, cAMP formed in vivo is mostly short-lived. This requires activated ACs to be quickly returned to a basal activity state [7–9]. In eukaryotic cells, GTP hydrolysis and dissociation of the activated Keywords adenylate cyclase; cAMP; Mycobacterium; polyphosphate; stress response Correspondence J. E. Schultz, Pharmazeutisches Institut, Universita ¨ t, Tu ¨ bingen, Auf der Morgenstelle 8, 72076 Tu ¨ bingen, Germany Fax: +49 7071 295952 Tel: +49 7071 2972475 E-mail: joachim.schultz@uni-tuebingen.de (Received 7 October 2008, revised 7 November 2008, accepted 10 December 2008) doi:10.1111/j.1742-4658.2008.06852.x cAMP generation in bacteria is often stimulated by sudden, but lasting, changes in extracellular conditions, whereas intracellular cAMP concen- trations quickly settle at new levels. As bacteria lack G-proteins, it is unknown how bacterial adenylate cyclase (AC) activities are modulated. Mycobacterium tuberculosis has 15 class III AC genes; therefore, we exam- ined whether mycobacteria contain a factor that may regulate AC activi- ties. We identified mycobacterial polyphosphates with a mean chain length of 72 residues as highly potent inhibitors of dimeric class IIIa, class IIIb and class IIIc ACs from M. tuberculosis and other bacteria. The identity of the inhibitor was established by phosphatase degradation, 31 P-NMR, acid or base hydrolysis, PAGE and comparisons with commercial standards, and functional substitution by several polyphosphates. The data indicate that each AC dimer occupies 8–15 phosphate residues on a polyphosphate strand. Other polyionic polymers such as polyglutamate, polylysine and hyaluronic acid do not affect cyclase activity. Notably, the structurally unrelated class I AC Cya from Escherichia coli is unaffected. Bacterial polyphosphate metabolism is generally viewed in the context of stress- related regulatory networks. Thus, regulation of bacterial class III ACs by polyphosphates could be a component of the bacterial stress response. Abbreviations AC, adenylate cyclase; poly-P, polyphosphate. 1094 FEBS Journal 276 (2009) 1094–1103 ª 2009 The Authors Journal compilation ª 2009 FEBS AC–G-protein complex terminates signaling, possibly also with the involvement of secondary modifications such as phosphorylation [10,11]. Termination of acti- vation of bacterial class III ACs has not been investi- gated. G-proteins or G-protein-like mediators of AC stimulation are unknown in bacteria. For example, in the cyanobacterium Anabaena, the ACs CyaB1 and CyaB2 are stimulated by the enzymatic product cAMP via an N-terminal tandem GAF domain [12,13]. Theoretically, this would result in perpetual self-activa- tion until ATP is exhausted, an unlikely physiological situation. Recently, it has been shown that sodium ions may be involved in the process of autoinactivation of CyaB1 AC [14]. In Mycobacterium tuberculosis, 15 class III AC genes have been identified [15,16], and for at least 10, AC activity has been demonstrated [1,17,18]. Deletion of single AC isoforms in M. tuberculosis did not result in obvious phenotypes [17,19]. The stressor conditions employed may have been insufficient and ⁄ or functional replacement could compensate for the loss of an indi- vidual AC isoform. Therefore, two major questions exist: how are the bacterial ACs stimulated, and how is stimulation terminated while stimulatory conditions continue to prevail? To address these questions, we examined whether mycobacteria contain endogenous factors that modulate AC activities. We used a cell homogenate from Mycobacterium bovis BCG, the live vaccine strain against tuberculosis, to search for the presence of such modulators. We have isolated poly- phosphate (poly-P) from M. bovis BCG, which is a well-known bacterial constituent [20,21], and demon- strated that it is a most powerful inhibitor of class III ACs from M. tuberculosis and other bacteria. In test- ing ACs from the class IIIa, IIIb and IIIc subfamilies, we found that all were strongly inhibited by this cell constituent, whereas a class I AC from Escherichia coli was not affected. Results In an exploratory experiment, we homogenized 1 g (wet weight) of M. bovis BCG cells (grown as a settling culture under hypoxia) with a French press, and prepared a 100 000 g supernatant (60 min). The super- natant strongly inhibited the activity of recombinant Rv1625c, a membrane-bound, mammalian-like AC from M. tuberculosis [22]. The virtual IC 50 was 1.6 lg of protein (data not shown). The suspended pellet from the above centrifugation was inactive. To assess the specificity of inhibition, several controls were used: (a) suspended cell pellets and supernatants from Myco- bacterium smegmatis or E. coli (BL-21) grown without stress in well-oxygenated rich media had only very low inhibitory potency; (b) M. bovis BCG cells were washed extensively with 50 mm Tris ⁄ HCl buffer con- taining 150 mm NaCl, yet this treatment did not detach an AC inhibitor; and (c) we tested unused as well as spent medium after harvesting of BCG. No inhibition (or activation) was observed. This led us to believe that M. bovis BCG produced a soluble intra- cellular inhibitor of Rv1625c. What is the chemical nature of the inhibitor? Boiling removed 98% of protein, but AC inhibition was unim- paired. Similarly, extended digestion with trypsin did not abolish inhibition, virtually excluding a protein. The inhibitory factor was not DNA or RNA. This was verified by nuclease digestion and controlled by agarose gel electrophoresis. After ether extraction, the inhibitor remained in the aqueous phase, excluding lipids. Next, we incubated the enriched inhibitory fraction with lyso- zyme or cellosyl, a promiscuous bacterial muramidase from Streptomyces coelicolor. The results were equivo- cal. After digestion with cellosyl, AC inhibition was retained, whereas incubation with lysozyme resulted in loss of inhibition. An HPLC analysis failed to identify muropeptides, the expected products of lysozyme or cellosyl digestion (data not shown). Lysozyme has an isoelectric point of 11. Therefore, we hypothesized that the inhibitor might be an acidic compound such as poly-P and bind to lysozyme and not to cellosyl. This would explain the contradictory results with both muraminidases. When we incubated a sample with 1.5 units of acid phosphatase at pH 5.5, AC inhibition was lost. Similarly, inhibition was almost completely lost upon treatment with HCl or NaOH at room temperature, indicating hydrolysis of a poly-P. For final inhibitor purification, the heat-treated sam- ple was bound to DEAE–Sephacel, released with 400 mm NaCl, and fractionated by Superose 6 chro- matography. Fractions were assayed for activity using recombinant Rv1625c (Fig. 1A), and analyzed by elec- trophoresis [23]. In active fractions, poly-Ps with a chain length of about 70 residues were stained, whereas inactive fractions did not stain (Fig. 1B). A dose–response curve with the combined fractions established the high inhibitory potency (Fig. 1C). This indicated that the inhibitor was poly-P. Final chemical identification was carried out by 31 P- NMR spectroscopy of concentrated fractions 50–64 from the Superose 6 column (Fig. 2, trace E). The intense resonance at )21.8 p.p.m. is characteristic for interior phosphate residues, and the weak, broad peaks at )6.2 and )20.2 p.p.m. are indicative of terminal and penultimate phosphate groups, respectively. Thus, the 31 P-NMR spectrum identified the sample as linear Y. L. Guo et al. Polyphosphate inhibition of adenylate cyclases FEBS Journal 276 (2009) 1094–1103 ª 2009 The Authors Journal compilation ª 2009 FEBS 1095 poly-Ps [24–26]. This was confirmed by comparisons with 31 P-NMR spectra of commercial polyphosphate samples (Fig. 2 and Table 1). The cyclic triphosphate (trimetaphosphate) showed a singlet at )21.5 p.p.m., due to exclusively interior residues (Fig. 2, trace A). The two terminal groups and the interior residue of the linear triphosphate gave two resonances at )7.1 and )21.4 p.p.m. at an integrated area ratio of 2 : 1 that were split by each other into a doublet and a triplet (Fig. 2, trace B). Extending the chain length to 25 phosphate residues allowed detection of the terminal groups (d = )5.8 p.p.m.), the penultimate groups (d = )20.8 p.p.m.), and the interior residues (d = )21.8 p.p.m.). The resonance of the terminal groups displayed characteristic doublets and doublets of doublets (Fig. 2, trace C). Further increases in chain length led to broadening of terminal phosphate reso- nances, which agreed with a reduced T 2 for polymeric systems (Fig. 2, trace D, poly-P75). Moreover, calcula- tion of the peak areas of the resonances allowed an estimate of the chain length. Calculated chain lengths for poly-P25 and poly-P75 were in agreement with those given by the supplier (Table 1). The mycobacte- rial sample yielded an average chain length of 72, in agreement with the electrophoretic analysis (Fig. 1B). Finally, commercial poly-Ps, which were further char- acterized by 31 P-NMR and hydrolysis by acid or base, inhibited class III ACs identically to the material from M. bovis BCG. With 167 nm Rv1625c in the assay, 200 nm poly-P75 inhibited enzyme activity completely. Owing to the scarcity of poly-P isolated from M. bovis BCG, we routinely used biological material for initial studies and commercial poly-Ps for in-depth 45232 40 A B C 45 kDa 30 120 AC activity (%) 46 10 20 80 48 0 D 280 nm 40 020406080100 0 Fraction number Fraction # P75 P45 P25 5650 5852 60 6462 Poly P 46 48 100 80 60 20 40 AC activity (%) 0 Mycobacterial poly-P (M) 10 –9 10 –8 Fig. 1. Purification and analysis of an AC inhibitor from M. bovis BCG. (A) Superose 6 gel filtration and inhibition of Rv1625c by indi- vidual fractions. Solid line, D 280 nm; stippled lines (d), inhibition of 167 n M Rv1625c; arrows on top denote molecular mass mark- ers. (B) Electrophoretic analysis (15% polyacrylamide gel containing 6 M urea) of inhibitory fractions from (A). Note almost empty lanes 46 and 48, which are barely inhibitory. Controls of linear poly- Ps with average chain lengths of 75 (14 lg), 45 (7 lg), and 25 (7 lg) residues are on the left. The gel was stained with toluidine blue O to detect poly-Ps [23]. The wide bands indicate chain length variations in the commercial standards and column fractions. One hundred per cent activity corresponds to 360 nmol cAMPÆmg )1 Æ min )1 . (C) Dose–response curve of Rv1625c inhibition of combined fractions from the Superose 6 fractionation in (A). A B C D E p.p.m. Fig. 2. 31 P-NMR spectra of the concentrated fractions (50–64) of the Superose 6 gel filtration and of commercial poly-P standards: (A) cyclic poly-P3; (B) linear poly-P3; (C) linear poly-P25; (D) linear poly-P75; and (E) concentrated inhibitor. PP1, terminal phosphate groups; PP2, penultimate groups; PPn, interior phosphate residues. The concentrations of the poly-P standards were calculated to be uniformly 100 m M P i , i.e. 33.3 mM poly-P3 [(A) and (B)], 4 mM poly- P25 (C), and 1.33 m M poly-P75 (D). Polyphosphate inhibition of adenylate cyclases Y. L. Guo et al. 1096 FEBS Journal 276 (2009) 1094–1103 ª 2009 The Authors Journal compilation ª 2009 FEBS characterizations. We never noted disparities in results between the different poly-P sources. Rv1625 inhibi- tion by poly-P75 was instantaneous, as demonstrated experimentally (Fig. 3A). This established that inhibi- tion was reversible [27]. This was confirmed in dilution experiments in which the enzyme–inhibitor complex was diluted eight-fold immediately prior to the start of the reaction. Reaction velocities of Rv1625c were determined in the presence of 10, 20 and 25 nm poly- P75 as a function of the substrate Mn-ATP. A Linewe- aver–Burk plot showed that K m values were unchanged whereas V max decreased, indicating noncompetitive inhibition (Fig. 3B). We excluded any chelating effect of poly-Ps on the Mn 2+ concentration, because even if the poly-P concentration was calculated in terms of molarity of orthophosphate, it did not exceed low micromolar values, whereas Mn 2+ was fixed at 2 mm; that is, poly-P could not act as a chelating agent. Next, we investigated whether phosphates of different chain length had differing inhibitory potencies. Orthophos- phate up to 11 mm had no effect on Rv1625c, pyro- phosphate inhibited it with an IC 50 of 210 lm, the linear triphosphate had an IC 50 of 20 lm, and the cyc- lic trimetaphosphate an IC 50 of 2 mm; that is, these compounds were poor inhibitors (Table 2). In contrast, all IC 50 concentrations of poly-Ps with a chain length of 16 or more were in the submicromolar range (Fig. 4 and Table 2). When we normalized the individual IC 50 concentrations of various poly-P compounds with chain lengths > 16 to the concentration of orthophos- phate, i.e. the poly-P IC 50 concentrations in Table 2 were multiplied by the average phosphate chain length, we obtained a mean IC 50 concentration of 660 ± 62.8 nm phosphate (± SEM; range 561– 825 nm). This indicated that once a critical length of the poly-P chain is reached, the increasing affinity of poly-P is linearly related to the increase in the cal- culated total phosphate concentration. For poly-P75 and cyclic poly-P17, the apparent K i values were determined with 100 nm Rv1625c, 30 and 60 lm Mn-ATP, and different inhibitor concentrations Table 1. Chemical shifts and coupling constants of the purified inhibitor and the commercial poly-Ps. d, chemical shift; 2 J pp , coupling constant of two adjacent phosphate groups; PP1, terminal phosphate groups; PP2, penultimate phosphate groups; PPn, interior phosphate residues; s, singlet; d, doublet; t, triplet; br, broad; br s, broad singlet; br d, broad doublet. Sample d (p.p.m.) Calculated chain length (average) PP1 PP2 PPn 2 J PP (Hz) P3 cyclic )21.5 (s) Not applicable P3 linear )7.1 (d) )21.4 (t) 20.77 Not applicable P25 linear )5.8 (br d) )20.8 )21.8 (br s) 18.5; 17.4 27 P75 linear )5.7 (br) )20.9 (br) )21.9 (br s) 79 Inhibitor )6.2 (br) )20.2 (br) )21.8 (br s) 72 100 A B 80 25 nM 40 60 40 nM 50 nM 20 AC activity (%) 60 nM 80 nM 0 40 80 120 160 0 7 Time of preincubation (s) 5 25 n M 4 10 20 3 1/v (µmol·min*mg) –1 0 2 1 0 0.02 1 –0.04 –0.02 0.04 0.06 0.08 1/S (µM) –1 Fig. 3. Kinetics of the inhibition of Rv1625c by poly-P75. (A) Rv1625c at 134 n M was used in the assay. The concentrations of poly-P75 added at the beginning are indicated. The first assay was started after 7 s of preincubation of protein and poly-P75 by addition of the substrate ATP. Assays were run for 4 min. One hundred per- cent activity corresponds to 448 nmol cAMPÆmg )1 Æmin )1 . (B) Double reciprocal plot (Lineweaver–Burk) of substrate kinetics of Rv1625c in the presence of three concentrations of poly-P75 as an inhibitor. Y. L. Guo et al. Polyphosphate inhibition of adenylate cyclases FEBS Journal 276 (2009) 1094–1103 ª 2009 The Authors Journal compilation ª 2009 FEBS 1097 (5–80 nm). Dixon diagrams gave apparent K i values of 14 nm for poly-P75 and 20 nm for poly-P17, i.e. lower than the enzyme concentration. This indicated that a single poly-P strand inhibited more than one AC mole- cule. Next, we determined IC 50 concentrations of poly- P25 at different concentrations of Rv1625c. Here, IC 50 values increased linearly with the increasing protein concentration (Fig. 5). With a slope of the curve of 0.154 (R = 0.982; Fig. 5), and considering that at IC 50 , 50% of the dimers are bound to poly-P, this indi- cated a molar ratio between the Rv1625c dimer and poly-P25 of 3 : 1, close to the results obtained with poly-Ps with different chain lengths (Table 2). Next, we examined whether poly-Ps inhibit different bacterial AC isoforms. On the basis of systematic differences in key amino acids and on small, strictly localized length variations, class III ACs have been divided into four subfamilies, IIIa to IIId [1]. Rv1625c is a class IIIa AC, as are all mammalian membrane- bound ACs. A concatenated Rv1625c homodimer with an identical domain sequence as the membranous mammalian ACs, (Rv1625c) 2 , is active [17] and inhib- ited by poly-P, just like Rv1625c (data not shown). Do poly-Ps also inhibit AC isozymes from other class III subfamilies? We examined the following ACs: cyaG from Arthrospira platensis as another class IIIa isoform containing a HAMP domain; as class IIIb ACs, myco- bacterial Rv3645, which has a HAMP domain between its membrane anchor and the catalytic domain, and CyaB1 from Anabaena sp., which has an N-terminal cAMP-binding tandem GAF domain; as a class IIIc AC, the M. tuberculosis pH sensor Rv1264 [6]. In general, all class III ACs were potently inhibited by poly-P75 (Fig. 6A). Although the IC 50 concentrations differed slightly (11, 57, 315, 102 and 31 nm for Rv1625c, CyaG, Rv3645, CyaB1 and Rv1264, respec- tively), all were in the nanomolar range (Fig. 6A). Because the catalytic centers of class III ACs appear to be highly similar [3,5,6], it is likely that poly-Ps gen- erally inhibit class III ACs. Rv1264 has been shown to be a pH sensor that is strongly activated by pH values around 5.5 [6]. This allowed testing of whether poly-Ps affected the basal and the activated states of a clas- s IIIc AC similarly. Poly-P75 inhibited the basal state at pH 8 and the activated state at pH 5.5, but with markedly different efficacies. The activated enzyme was fully inhibited at 400 nm, whereas the basal state was not yet fully inhibited at 30 lm (Fig. 6B). The Table 2. IC 50 values of poly-Ps for the AC Rv1625c. The protein concentration in the assays was 83 n M. Phosphates IC 50 values (nM) Phosphate (P i )>11· 10 6 PP i 21 · 10 4 Linear PPP i 20 · 10 3 Cyclic PPP i 20 · 10 5 Poly-P16 100 Cyclic poly-P17 33 Poly-P25 27 Poly-P30 18 Poly-P45 12 Poly-P68 12 Poly-P75 11 120 PPPi Poly-P16 80 100 Poly-P30 Poly-P75 60 AC activity (%) 20 40 –9 0 –8 –7 –6 –5 –4 –3 10 Concentration (M) 10 10 10 10 10 10 Fig. 4. Inhibition of Rv1625c by poly-Ps of different strand lengths. Rv1625c at 167 n M was used in the assays. One hundred per cent activity corresponds to 370 nmol cAMPÆmg )1 min )1 . 26 22 18 10 14 IC 50 (nM Poly-P25) 25 50 75 100 125 150 Rv1625c dimer concentration (nM) Fig. 5. Poly-P25 IC 50 values increase with increasing AC concentra- tions. Assays were carried out at different Rv1625c concentrations. IC 50 values (y-axis) were plotted against the protein concentrations at which only dimers exist [22] (regression coefficient r = 0.999). Polyphosphate inhibition of adenylate cyclases Y. L. Guo et al. 1098 FEBS Journal 276 (2009) 1094–1103 ª 2009 The Authors Journal compilation ª 2009 FEBS reported structures of Rv1264 showed that a canonical catalytic cleft of class III ACs is formed at pH 5.5, whereas at pH 8, the catalytic amino acids are far apart [6]. The differences in the IC 50 concentrations may indicate that poly-P binds in the catalytic crevice. Another question was whether poly-P specifically inhibits class III ACs or also acts on class I ACs. We expressed Cya, the class I AC from E. coli, and purified it to homogeneity by Ni 2+ –nitrilotriacetic acid chro- matography. The specific activity of the purified protein was 27.2 nmolÆmg )1 Æmin )1 with Mg-ATP as a sub- strate. This class I AC was unaffected by up to 30 lm poly-P75 in the assay (Fig. 6A, stippled line). Finally, the specificity of poly-P inhibition was investigated using other ionic polymers. We employed anionic poly- glutamate (M r ‡ 15 000, Sigma, Munich, Germany), cationic polylysine (M r 4000–15 000, Sigma), and acidic hyaluronic acid (from Streptococcus equi, Fluka, Munich, Germany). We tested these compounds at 0.2 and 2 lm with Rv1625c. None inhibited AC activity, demonstrating that the effect of poly-P was specific. Discussion cAMP in bacteria is discussed in numerous publica- tions as a second messenger involved in regulatory pathways. However, reliable studies in which intracel- lular cAMP concentrations were determined and regu- lation of cAMP biosynthesis in vivo was examined are rare. This is due to its low intracellular concentrations and secretion of up to 95% of total cAMP into the medium, which causes unusual experimental difficulties and ambiguities [28–35]. Generally, conditions that stimulate cAMP formation in bacteria, e.g. changes in pH, ion or nutrient concentrations, are related to stress conditions [28,36]. In this study, we could not remedy the lack of knowledge of bacterial cAMP metabolism, and the physiological relevance of bacte- rial class III AC inhibition by poly-P merits further experiments. At the outset, we asked whether M. bovis BCG con- tains endogenous factors that regulate AC activities. The isolated AC inhibitor was unequivocally identified by chemical means ( 31 P-NMR, acid and base hydro- lysis, SDS ⁄ PAGE), biochemical means (phosphatase degradation), and full functional substitution by com- mercial poly-P. In addition, other bacterial constitu- ents were excluded, such as DNA, RNA, proteins and peptides, and peptidoglycans of the cell wall. Although poly-P is a metabolic staple that is present in cells from all the kingdoms of life – bacteria, fungi, plants and animals – it has never been studied in conjunction with regulation of ACs [20,37]. Mycobacteria produce poly-Ps under a variety of stress conditions, and pos- sess two poly-P kinases [21,38]. Bacterial poly-Ps vary in size and solubility, and intracellular concentrations of poly-P fluctuate considerably (2–15 ngÆmg )1 protein) [20,37,39,40]. Depending on the organism, growth and 100 A B 80 40 60 AC activity (%) 20 0 –8 –7 –6 –5 Poly-P75 (M) 10 10 10 10 100 pH 8.0 80 pH 5.5 40 60 20 AC activity (%) –8 0 –7 –6 –5 10 Poly-P75 ( M) 10 10 10 Fig. 6. Inhibition of bacterial ACs from class III and class I by poly-P75. (A) The following amounts of protein were used: Rv1264(1–397), 660 n M (d); CyaG(370–672), an N-terminal truncated version with only HAMP and catalytic domains, 640 n M (s); CyaB1(1–859), 22 n M ( ); Rv3645(1–549), 153 lg of total membrane proteins (h); Cya(1–446) from E. coli, 876 n M ( , stippled line). (B) Inhibition of basal and activated states of 660 n M Rv1264 by poly- P75. The protein concentrations used were based on the linear sec- tions of the protein dependency of the respective AC reactions. This ensured that dimerization of respective monomers was complete. Y. L. Guo et al. Polyphosphate inhibition of adenylate cyclases FEBS Journal 276 (2009) 1094–1103 ª 2009 The Authors Journal compilation ª 2009 FEBS 1099 physiological conditions, poly-P may amount to up to 20% of bacterial dry weight [40,41]. For poly-P in bacteria, several physiological functions in many locations have been proposed, e.g. poly-P accumulation in stress sensing [37,39,42–46]. Furthermore, poly-P is discussed as a primordial pre- cursor of ATP, a flexible scaffold for the assembly of macromolecules, a cellular phosphate store, a buffer system in pH regulation, being involved in chelation of cations such as Ca 2+ ,Mn 2+ and Mg 2+ [20,37,39,47]. Our data suggest a new potential role for poly-P as an inhibitor of bacterial class III ACs. Because cAMP as a bacterial alarmone is a global signaling molecule, the reported actions of poly-P on bacterial class III ACs may affect several cellular functions simultaneously. It was most interesting to note that the class I AC from E. coli was not inhibited by poly-P. This may indicate that the sequence dissimilarities between different clas- ses of ACs reflect functional differences. The presence in some bacteria of ACs from different classes would then enable different modes and levels of regulations. Actually, it may be useful for certain bacteria to con- tain AC isoforms from different classes (such as Pseu- domonas aeroginosa, which has AC isoforms from classes I, II and III), because this would broaden the modes of cellular regulation and pathogenicity [48]. Because the catalytic folds of a bacterial class IIIc and a mammalian class IIIa AC are superimposable [3,6], it is reasonable to assume that mammalian ACs will be inhibited by poly-P as well. In preliminary experiments using membranes prepared from a rat brain homogenate, we found that brain AC activity was inhibited by poly-P with an IC 50 of 10 lm (data not shown). As far as the mechanism of inhibition is concerned, the data obtained with Rv1264 indicate binding in the catalytic fold. Possibly, poly-P bridges and occludes the substrate-determining lysine (Lys296 in Rv1625c) and the arginine (Arg376), which stabilizes the transition state. Therefore, poly-P may be helpful in attempts to crystallize and characterize other bacte- rial ACs. To the best of our knowledge, poly-P metabolism has never been studied in conjunction with cAMP metabolism. The concentration of poly-P in stressed bacteria appears to be higher than is needed for AC inhibition. Under the hypoxic growth conditions of M. bovis BCG used in this study, the concentration of poly-P would probably silence all class III ACs. Several possibilities exist to explain this fact. One is that at a low level of stress conditions, such as modest oxygen deprivation or nutrient depletion, cAMP formation is elicited as an initial response. Poly-P bio- synthesis is then initiated, and this turns off cAMP production of class III ACs. Another possibility is that the availability of poly-P is locally restricted. The neg- atively charged polyanionic compound may bind to positively charged carrier molecules or it may be neu- tralized by inorganic cations. The availability of poly-P for termination of class III activation would then be tied to competition for different intracellular binding sites. Finally, the possibility exists that poly-P accumu- lation is controlled locally, such that ACs are partially maintained in an inhibited, inactive state. Release from inhibition could occur by poly-P degradation by tightly regulated phosphatases. The latter would thus attain regulatory significance. Experimental procedures Mycobacterial strain and growth M. bovis BCG 1721, a streptomycin-resistant derivative of BCG Pasteur, carrying a non-restrictive rpsL mutation (K42R) [49], was grown as a settling culture in tissue culture flasks in Middlebrook 7H9 medium supplemented with oleic acid, albumin, dextrose, catalase (Difco, Heidelberg, Germany) and Tween-80 (0.05%) [50]. Flasks were shaken once daily by hand; that is, cells were grown under hypoxic stress. E. coli was grown in LB medium, and M. smegmatis in LB + 0.05% Tween-80 under constant shaking for oxy- genation (210 r.p.m. shaking speed). Bacteria were harvested at an attenuance (D 600 nm ) of 0.5–0.7 by centrifugation (10 min, 4400 g at 4 °C) and stored at )80 °C until use. Expression and purification of Rv1625c Rv1625c(1–443) in pQE60 was expressed in E. coli BL- 21(DE3)(pRep4) and purified to homogeneity using 0.6% n-dodecyl-b-d-maltoside for solubilization as previously described [17]. The purified protein could be stored at )80 °C without loss of activity for at least 6 months. The catalytic domain Rv1625c(204–443) and other bacterial ACs used were expressed and purified to homogeneity as previously reported [12,17,22,51,52]. AC assay AC activity was measured for 10 min in a volume of 100 lLat30°C [53]. The reactions contained 22% glycerol, 50 mm Tris ⁄ HCl (pH 7.5), 2 mm MnCl 2 or 2 mm MgCl 2 , the indicated concentrations of ATP with 25 kBq of [ 32 P]ATP[aP] and 2 mm cAMP with 150 Bq of [2,8- 3 H]cAMP to monitor yield during product isolation. For determination of kinetic constants, ATP was varied from 14 lm to 100 lm, with constant 2 mm Mn 2+ . The reaction was started by addition of enzyme. ATP conver- sion was limited to < 10%, to guarantee linearity. ATP Polyphosphate inhibition of adenylate cyclases Y. L. Guo et al. 1100 FEBS Journal 276 (2009) 1094–1103 ª 2009 The Authors Journal compilation ª 2009 FEBS was separated from product cAMP by sequential chroma- tography [53]. Purification and analysis of poly-Ps from M. bovis BCG Cells were suspended in 50 mm Tris ⁄ HCl (pH 7.5) and broken with a French press. The homogenate was centri- fuged (100 000 g for 1 h), and the supernatant was heated at 95 °C for 30 min. Coagulated protein was removed (100 000 g for 1 h). The supernatant was centrifuged through Sephadex G50 spin columns to remove small con- taminants; the inhibitory capacity was in the eluate. Next, DEAE–Sephacel was added, and the material bound to the anion exchanger. The matrix was poured into a col- umn, washed, and eluted with 400 mm NaCl. The eluate was applied to a Superose 6 column (30 · 1 cm), and frac- tions were examined by electrophoresis and for inhibitory activity (Fig. 1). Electrophoresis of poly-Ps was carried out as previously described [23]. A 15% acrylamide ⁄ bisacryla- mide gel with 6 m urea was prepared with TBE buffer (89 mm Tris ⁄ borate, 2 mm EDTA, pH 8.3). The gel was prerun at 200 V for 60 min. Poly-Ps with average chain lengths of 25, 45 and 75 residues were used as markers (Sigma). Probes that contained 25% of sample buffer (50% sucrose, 0.125% bromophenol blue and 450 mm Tris ⁄ borate at pH 8.3, 13.5 mm EDTA) were loaded and electrophoresed for 25 min. Gels were stained with 0.05% toluidine blue O in 25% methanol and 5% glycerol for 20 min. Destaining was performed with the same solvent, lacking the dye. The concentration of poly-P in mycobac- terial preparations were assessed using standard curves with poly-P75. 31 P-NMR measurement The 31 P-NMR spectra were obtained on Bruker Avance 400 and Bruker Avance 500 spectrometers operat- ing at 161.98 and 202.45 MHz, respectively. The spectra were recorded by applying 30° pulses with a repetition time of 1 s at 21 °C, and referenced against external 85% H 3 PO 4 . Multipuls decoupling sequences were applied to remove any proton phosphorus interactions. 31 P peak areas were obtained by fitting the spectra to a set of Lorenzian line shapes using the Bruker topspin 2.0 software package. The pH of all samples was adjusted to 7.5. The fractions of Superose 6 that showed 95% inhibitory activity against Rv1625c in 50 mm Tris ⁄ HCl and 100 mm NaCl were concentrated with an Amicon Ultra-4 5K cen- trifugal filter device, and subsequently diluted with D 2 Oto reduce noise (50% final D 2 O in the water). A total volume of 0.5 mL was transferred into a Norell 507-HP sample tube; 50 mm linear poly-P3, poly-P25 and poly-P75, and cyclic poly-P3 with sodium cations in 50% D 2 O, were used as standards. Muropeptide determination by HPLC The concentrated Superose 6 fractions (see above) were analyzed for the presence of muropeptides according to Glauner [54]. Briefly, the sample was incubated with cello- syl or lysozyme at pH 4.8 at 37 °C overnight, deactivated (100 °C, 10 min), and centrifuged (14 000 g, 8 min). Eighty microliters of the supernatant was mixed with 80 lLof sodium borate (0.5 m, pH 9.0) and 1–2 mg of sodium boro- hydride. The excess of borohydride was destroyed after incubation at room temperature for 30 min. Separation of muropeptides occurred on a 250 · 4.6 mm 3 lm Hypersil ODS column at 55 °C using a 135 min gradient from buffer A (50 mm sodium phosphate, pH 4.3) to buffer B (75 mm sodium phosphate, pH 4.9, 15% methanol) at a flow rate of 0.5 mLÆmin )1 . Detection of muropeptides occurred at 205 nm. We confirmed that cellosyl degraded mycobacterial cell wall material in respective controls with mycobacterial homogenates. Acknowledgements This work was supported by the Deutsche Forschungs- gemeinschaft (SFB 766). P. Sander is in part supported by the Swiss National Science Foundation (contract: 3100A0_120326) and the European Union (LSHP-CT- 2006-037217). References 1 Linder JU & Schultz JE (2003) The class III adenylyl cyclases: multi-purpose signalling modules. Cell Signal 15, 1081–1089. 2 Krupinski J, Coussen F, Bakalyar HA, Tang WJ, Fein- stein PG, Orth K, Slaughter C, Reed RR & Gilman AG (1989) Adenylyl cyclase amino acid sequence: possi- ble channel- or transporter-like structure. Science 244, 1558–1564. 3 Tesmer JJ, Sunahara RK, Gilman AG & Sprang SR (1997) Crystal structure of the catalytic domains of ade- nylyl cyclase in a complex with Gsalpha.GTPgammaS. Science 278, 1907–1916. 4 Tesmer JJ, Sunahara RK, Johnson RA, Gosselin G, Gilman AG & Sprang SR (1999) Two-metal-ion catalysis in adenylyl cyclase. Science 285, 756– 760. 5 Sinha SC, Wetterer M, Sprang SR, Schultz JE & Linder JU (2005) Origin of asymmetry in adenylyl cyclases: structures of Mycobacterium tuberculosis Rv1900c. EMBO J 24, 663–673. 6 Tews I, Findeisen F, Sinning I, Schultz A, Schultz JE & Linder JU (2005) The structure of a pH-sensing myco- bacterial adenylyl cyclase holoenzyme. Science 308, 1020–1023. Y. L. Guo et al. Polyphosphate inhibition of adenylate cyclases FEBS Journal 276 (2009) 1094–1103 ª 2009 The Authors Journal compilation ª 2009 FEBS 1101 7 Schultz J & Daly JW (1973) Cyclic adenosine 3¢,5¢- monophosphate in guinea pig cerebral cortical slices. 3. Formation, degradation, and reformation of cyclic adenosine 3¢,5¢-monophosphate during sequential stimu- lations by biogenic amines and adenosine. J Biol Chem 248, 860–866. 8 Imashimizu M, Yoshimura H, Katoh H, Ehira S & Ohmori M (2005) NaCl enhances cellular cAMP and upregulates genes related to heterocyst development in the cyanobacterium, Anabaena sp. strain PCC 7120. FEMS Microbiol Lett 252, 97–103. 9 Schultz JE, Klumpp S, Benz R, Schurhoff-Goeters WJ & Schmid A (1992) Regulation of adenylyl cyclase from Paramecium by an intrinsic potassium conductance. Science 255, 600–603. 10 Romero-Avila MT, Flores-Jasso CF & Garcia-Sainz JA (2002) alpha1B-Adrenergic receptor phosphorylation and desensitization induced by transforming growth factor-beta. Biochem J 368, 581–587. 11 Rondard P, Iiri T, Srinivasan S, Meng E, Fujita T & Bourne HR (2001) Mutant G protein alpha subunit activated by Gbeta gamma: a model for receptor activa- tion? Proc Natl Acad Sci USA 98, 6150–6155. 12 Kanacher T, Schultz A, Linder JU & Schultz JE (2002) A GAF-domain-regulated adenylyl cyclase from Anaba- ena is a self-activating cAMP switch. EMBO J 21, 3672–3680. 13 Bruder S, Linder JU, Martinez SE, Zheng N, Beavo JA & Schultz JE (2005) The cyanobacterial tandem GAF domains from the cyaB2 adenylyl cyclase signal via both cAMP-binding sites. Proc Natl Acad Sci USA 102, 3088–3092. 14 Cann MJ (2007) Sodium regulation of GAF domain function. Biochem Soc Trans 35, 1032–1034. 15 Cole ST, Brosch R, Parkhill J, Garnier T, Churcher C, Harris D, Gordon SV, Eiglmeier K, Gas S, Barry CE III et al. (1998) Deciphering the biology of Mycobacte- rium tuberculosis from the complete genome sequence. Nature 393, 537–544. 16 McCue LA, McDonough KA & Lawrence CE (2000) Functional classification of cNMP-binding proteins and nucleotide cyclases with implications for novel regula- tory pathways in Mycobacterium tuberculosis. Genome Res 10, 204–219. 17 Guo YL, Kurz U, Schultz A, Linder JU, Dittrich D, Keller C, Ehlers S, Sander P & Schultz JE (2005) Inter- action of Rv1625c, a mycobacterial class IIIa adenylyl cyclase, with a mammalian congener. Mol Microbiol 57, 667–677. 18 Shenoy AR & Visweswariah SS (2006) New messages from old messengers: cAMP and mycobacteria. Trends Microbiol 14, 543–550. 19 Dittrich D, Keller C, Ehlers S, Schultz JE & Sander P (2006) Characterization of a Mycobacterium tuberculosis mutant deficient in pH-sensing adenylate cyclase Rv1264. Int J Med Microbiol 296, 563–566. 20 Kornberg A (1995) Inorganic polyphosphate: toward making a forgotten polymer unforgettable. J Bacteriol 177 , 491–496. 21 Winder FG & Denneny JM (1957) The metabolism of inorganic polyphosphate in mycobacteria. J Gen Micro- biol 17, 573–585. 22 Guo YL, Seebacher T, Kurz U, Linder JU & Schultz JE (2001) Adenylyl cyclase Rv1625c of Mycobacterium tuberculosis: a progenitor of mammalian adenylyl cyclases. EMBO J 20, 3667–3675. 23 Robinson NA & Wood HG (1986) Polyphosphate kinase from Propionibacterium shermanii. Demonstration that the synthesis and utilization of polyphosphate is by a processive mechanism. J Biol Chem 261, 4481–4485. 24 MacDonald JC & Mazurek M (1987) Phosphorus magnetic resonance spectra of open-chain linear poly- phosphates. J Magn Reson 72, 48–60. 25 Bental M, Pick U, Avron M & Degani H (1991) Poly- phosphate metabolism in the alga Dunaliella salina stud- ied by 31P-NMR. Biochim Biophys Acta 1092, 21–28. 26 Castrol CD, Koretsky AP & Domach MM (1999) NMR-observed phosphate trafficking and polyphos- phate dynamics in wild-type and vph1-1 mutant Saccha- romyces cerevisae in response to stresses. Biotechnol Prog 15, 65–73. 27 Bisswanger H (2002) Enzyme Kinetics Principles and Methods, 3rd edn. Wiley-VCH, Weinheim. 28 Makman RS & Sutherland EW (1965) Adenosine 3¢,5¢- phosphate in Escherichia coli. J Biol Chem 240, 1309– 1314. 29 Lee CH (1977) Identification of adenosine 3¢,5¢-mono- phosphate in Mycobacterium smegmatis. J Bacteriol 132, 1031–1033. 30 Padh H & Venkitasubramanian TA (1977) Adenosine 3¢,5¢-monophosphate in mycobacteria. Life Sci 20, 1273–1280. 31 Padh H & Venkitasubramanian TA (1976) Adenosine 3¢,5¢-monophosphate in Mycobacterium phlei and Myco- bacterium tuberculosis H37Ra. Microbios 16, 183–189. 32 Pastan I & Perlman R (1970) Cyclic adenosine mono- phosphate in bacteria. Science 169 , 339–344. 33 Peterkofsky A & Gazdar C (1971) Glucose and the metabolism of adenosine 3¢:5¢-cyclic monophosphate in Escherichia coli . Proc Natl Acad Sci USA 68, 2794– 2798. 34 Peterkofsky A & Gazdar C (1973) Measurements of rates of adenosine 3 ¢:5 ¢-cyclic monophosphate synthesis in intact Escherichia coli B. Proc Natl Acad Sci USA 70, 2149–2152. 35 Bettenbrock K, Sauter T, Jahreis K, Kremling A, Leng- eler JW & Gilles ED (2007) Correlation between growth rates, EIIACrr phosphorylation, and intracellular cyclic Polyphosphate inhibition of adenylate cyclases Y. L. Guo et al. 1102 FEBS Journal 276 (2009) 1094–1103 ª 2009 The Authors Journal compilation ª 2009 FEBS AMP levels in Escherichia coli K-12. J Bacteriol 189, 6891–6900. 36 Maeda M, Lu S, Shaulsky G, Miyazaki Y, Kuwayama H, Tanaka Y, Kuspa A & Loomis WF (2004) Periodic signaling controlled by an oscillatory circuit that includes protein kinases ERK2 and PKA. Science 304, 875–878. 37 Brown MR & Kornberg A (2004) Inorganic polyphos- phate in the origin and survival of species. Proc Natl Acad Sci USA 101, 16085–16087. 38 Suzuki H, Kaneko T & Ikeda Y (1972) Properties of polyphosphate kinase prepared from mycobacterium smegmatis. Biochim Biophys Acta 268, 381–390. 39 Kornberg A, Rao NN & Ault-Riche D (1999) Inorganic polyphosphate: a molecule of many functions. Annu Rev Biochem 68, 89–125. 40 Rao NN & Kornberg A (1996) Inorganic polyphos- phate supports resistance and survival of stationary- phase Escherichia coli. J Bacteriol 178, 1394–1400. 41 Rao NN, Roberts MF & Torriani A (1985) Amount and chain length of polyphosphates in Escherichia coli depend on cell growth conditions. J Bacteriol 162, 242– 247. 42 Sureka K, Dey S, Datta P, Singh AK, Dasgupta A, Rodrigue S, Basu J & Kundu M (2007) Polyphosphate kinase is involved in stress-induced mprAB-sigE-rel signalling in mycobacteria. Mol Microbiol 65, 261– 276. 43 Reusch RN, Huang R & Bramble LL (1995) Poly-3-hy- droxybutyrate ⁄ polyphosphate complexes form voltage- activated Ca2+ channels in the plasma membranes of Escherichia coli. Biophys J 69, 754–766. 44 Kim KS, Rao NN, Fraley CD & Kornberg A (2002) Inorganic polyphosphate is essential for long-term survival and virulence factors in Shigella and Salmonella spp. Proc Natl Acad Sci USA 99, 7675–7680. 45 Jahid IK, Silva AJ & Benitez JA (2006) Polyphosphate stores enhance the ability of Vibrio cholerae to overcome environmental stresses in a low-phosphate environment. Appl Environ Microbiol 72, 7043–7049. 46 Fraley CD, Rashid MH, Lee SS, Gottschalk R, Harri- son J, Wood PJ, Brown MR & Kornberg A (2007) A polyphosphate kinase 1 (ppk1) mutant of Pseudomonas aeruginosa exhibits multiple ultrastructural and func- tional defects. Proc Natl Acad Sci USA 104, 3526–3531. 47 Tanaka S, Lee SO, Hamaoka K, Kato J, Takiguchi N, Nakamura K, Ohtake H & Kuroda A (2003) Strictly polyphosphate-dependent glucokinase in a polyphos- phate-accumulating bacterium, Microlunatus phosphovo- rus. J Bacteriol 185, 5654–5656. 48 Smith RS, Wolfgang MC & Lory S (2004) An adenylate cyclase-controlled signaling network regulates Pseudo- monas aeruginosa virulence in a mouse model of acute pneumonia. Infect Immun 72, 1677–1684. 49 Sander P, Meier A & Bottger EC (1995) rpsL+: a dom- inant selectable marker for gene replacement in myco- bacteria. Mol Microbiol 16, 991–1000. 50 Sander P, Rezwan M, Walker B, Rampini SK, Krop- penstedt RM, Ehlers S, Keller C, Keeble JR, Hagemeier M, Colston MJ et al. (2004) Lipoprotein processing is required for virulence of Mycobacterium tuberculosis. Mol Microbiol 52, 1543–1552. 51 Linder JU, Schultz A & Schultz JE (2002) Adenylyl cyclase Rv1264 from Mycobacterium tuberculosis has an autoinhibitory N-terminal domain. J Biol Chem 277, 15271–15276. 52 Linder JU, Hammer A & Schultz JE (2004) The effect of HAMP domains on class IIIb adenylyl cyclases from Mycobacterium tuberculosis . Eur J Biochem 271, 2446–2451. 53 Salomon Y, Londos C & Rodbell M (1974) A highly sensitive adenylate cyclase assay. Anal Biochem 58, 541–548. 54 Glauner B (1988) Separation and quantification of muropeptides with high-performance liquid chromatog- raphy. Anal Biochem 172, 451–464. Y. L. Guo et al. Polyphosphate inhibition of adenylate cyclases FEBS Journal 276 (2009) 1094–1103 ª 2009 The Authors Journal compilation ª 2009 FEBS 1103 . polyphosphates with a mean chain length of 72 residues as highly potent inhibitors of dimeric class IIIa, class IIIb and class IIIc ACs from M. tuberculosis and other bacteria. The identity of the. Polyphosphates from Mycobacterium bovis – potent inhibitors of class III adenylate cyclases Ying Lan Guo 1 , Hermann Mayer 2 , Waldemar Vollmer 3 ,. activity (%) –8 0 –7 –6 –5 10 Poly-P75 ( M) 10 10 10 Fig. 6. Inhibition of bacterial ACs from class III and class I by poly-P75. (A) The following amounts of protein were used: Rv1264( 1–3 97), 660