Fatty acid regulation of adenylyl cyclase Rv2212 from Mycobacterium tuberculosis H37Rv Amira Abdel Motaal 1 , Ivo Tews 2 , Joachim E. Schultz 1 and Ju ¨ rgen U. Linder 1 1 Abteilung Pharmazeutische Biochemie, Fakulta ¨ tfu ¨ r Chemie und Pharmazie, Universita ¨ tTu ¨ bingen, Germany 2 Biochemiezentrum der Universita ¨ t Heidelberg, Germany Adenylyl cyclases (ACs) (EC 4.6.1.1) convert ATP to the second messenger cAMP, which regulates a variety of cellular functions, including virulence in several pathogens, such as Mycobacterium tuberculosis [1–8]. Therefore, it is no surprise that ACs are subject to regu- lation by both extracellular stimuli such as hormones, availability of nutrients or osmotic pressure, and by intracellular stimuli such as changes in pH or even cAMP levels [9–11]. Currently, the catalytic domains of AC isozymes are grouped into six classes based on sequence similarities [12–14]. Class III contains by far the largest number of ACs, including all mammalian and many bacterial cyclases. All class III ACs must di- merize to be active, because the substrate-binding sites are formed at the dimer interface. On the basis of con- served sequence differences, class III ACs are further divided into subclasses IIIa–IIId [15]. The catalytic domains of the class III ACs are most often linked to additional protein domains, which in many instances appear to impart peculiar regulatory features [15]. In the M. tuberculosis genome, 15 putative class III AC genes of subclasses IIIa–IIId have been identified that possess quite different domain compositions [15– 17]. Therefore, one may assume that each of these cyc- lases participates in a distinct signalling pathway. To date, the recombinant proteins of nine mycobacterial AC genes have been shown to be catalytically active in vitro [18–25]. In all of them, the catalytic domains are associated with additionally distinct domains such as hexahelical membrane anchors (Rv1625c, Rv1318– Rv1320c, Rv3645), a pH-sensing domain (Rv1264), AAA-ATPase and helix-turn-helix DNA-binding domains (Rv0386), an a ⁄ b-hydrolase-like domain (Rv1900c) and HAMP domains (Rv1318c, Rv1319c, Keywords adenylyl cyclase; cAMP; fatty acid; Mycobacterium tuberculosis Correspondence J. U. Linder, Abteilung Pharmazeutische Biochemie, Fakulta ¨ tfu ¨ r Chemie und Pharmazie, Universita ¨ tTu ¨ bingen, Morgenstelle 8, 72076 Tu ¨ bingen, Germany Fax: +49 7071 295952 Tel: +49 7071 2974676 E-mail: juergen.linder@uni-tuebingen.de (Received 26 May 2006, revised 14 July 2006, accepted 17 July 2006) doi:10.1111/j.1742-4658.2006.05420.x Adenylyl cyclase Rv2212 from Mycobacterium tuberculosis has a domain composition identical to the pH-sensing isoform Rv1264, an N-terminal regulatory domain and a C-terminal catalytic domain. The maximal velo- city of Rv2212 was the highest of all 10 mycobacterial cyclases investigated to date (3.9 lmol cAMPÆmg )1 Æmin )1 ), whereas ATP substrate affinity was low (SC 50 ¼ 2.1 mm ATP). Guanylyl cyclase side activity was absent. The activities and kinetics of the holoenzyme and of the catalytic domain alone were similar, i.e. in distinct contrast to the Rv1264 adenylyl cyclase, in which the N-terminal domain is autoinhibitory. Unsaturated fatty acids strongly stimulated Rv2212 activity by increasing substrate affinity. In addition, fatty acids greatly enhanced the pH sensitivity of the holoenzyme, thus converting Rv2212 to a pH sensor adenylyl cyclase. Fatty acid binding to Rv2212 was modelled by homology to a recent structure of the N-ter- minal domain of Rv1264, in which a fatty acid-binding pocket is defined. Rv2212 appears to integrate three cellular parameters: ATP concentration, presence of unsaturated fatty acids, and pH. These regulatory properties open the possibility that novel modes of cAMP-mediated signal transduc- tion exist in the pathogen. Abbreviations AAA, ATPase associated with a variety of cellular activities; AC, adenylyl cyclase; CHD, cyclase homology domain; HAMP, domain first identified in histidine kinases, adenyl cyclases, methyl accepting chemotaxis proteins and phosphatases. FEBS Journal 273 (2006) 4219–4228 ª 2006 The Authors Journal compilation ª 2006 FEBS 4219 Rv1320c, and Rv3645). Although several studies have been published in recent years, the role of these regulatory domains is just beginning to be revealed. In the isoform Rv1625c, the large membrane anchor has a prominent role in protein dimerization [26]. In the four mycobacterial class IIIb ACs, the HAMP domains appear to directly act as modulators of AC activity, possibly transmitting signals that may be picked up by a receptor function of their hexahelical membrane domains [21]. The best investigated AC iso- form is Rv1264, which contains an N-terminal pH sen- sor module [25]. The structures of Rv1264 in an active and an inhibited state have been determined by X-ray crystallography. Rv1264 may enable M. tuberculosis to counteract acidification of phagolysosomes during host invasion and aid in intracellular survival [25]. Here we investigated the AC isoform Rv2212, which has the same domain composition as Rv1264. Both the recombinant Rv2212 holoenzyme and the isolated catalytic domain (also called the cyclase homology domain, CHD) were active in vitro. Unsaturated fatty acids stimulated Rv2212 AC activity. We demonstrate that the fatty acids are connected with a pH-sensing function of the holoenzyme. Furthermore, the rather low substrate affinity of Rv2212 suggests a potential role as a cellular ATP gauge, i.e. as a sensor for the prevailing energy status of the cell. Results Sequence features The predicted M. tuberculosis Rv2212 gene product has a domain composition identical to that of the AC isoform Rv1264 (Fig. 1A). The C-terminal class IIIc CHD of 177 amino acids shares 29% identity with that of Rv1264 (41% similarity; for alignments see [20]). The N-terminus of Rv2212 has 211 amino acids and is 21% identical (31% similar) to that of Rv1264, which mediates pH sensing [20,25]. Similar N-terminal domains are exclusively found in related actinobacteri- al ACs [20]. Irrespective of the identical domain organ- ization, the limited similarity of the N-termini of Rv2212 and Rv1264 suggested that Rv2212 may be regulated in a different way from Rv1264. AC activity of Rv2212 212)388 and Rv2212 1)388 The boundaries of the Rv2212 catalytic domain, S212 to the C-terminal D388, were defined by sequence comparisons with other bacterial class IIIa ACs. The catalytic domain Rv2212 212)388 and the holoenzyme Rv2212 1)388 were expressed in Escherichia coli as sol- uble proteins and purified to homogeneity by affinity chromatography (Fig. 1B). Both displayed AC activity with a pH optimum of 6.5. Activity was Mn 2+ - dependent. With up to 10 mm Mg 2+ , AC activity was below the detection limit of 0.5 nmol cAMPÆmg )1 Æ min )1 . Guanylyl cyclase activity was absent. We consis- tently observed that AC activity varied among different protein preparations. At 0.5 mm ATP, the activity of Rv2212 212)388 was 487 ± 365 nmol cAMPÆmg )1 Æ min )1 (SD, n ¼ 31, range 104–2105), and that of Rv2212 1)388 was 377 ± 357 nmol cAMPÆmg )1 Æmin )1 (SD, n ¼ 58, range 82–1735). This clearly demonstra- ted that the N-terminal domain of Rv2212 was not autoinhibitory as in Rv1264 [20]. The excessive variability in enzyme activity did not correlate with technical parameters such as the method of cell lysis, the extent of protein purification and duration or con- ditions of storage. Furthermore, individual assays were linear with respect to time and protein concentration. Protein aggregation was excluded because gel filtration chromatography reproducibly yielded a single symmet- rical peak corresponding to a dimer (not shown). Obviously, the affinity of the AC monomers was very high and only dimers existed in solution. Furthermore, there was no indication of charge heterogeneity, as evi- dent from a single symmetrical peak that was obtained upon anion exchange chromatography (data not shown). Finally, we have never observed such a fluctu- ation of AC activities in any of the recombinant myco- bacterial ACs that we have reported on previously, such as Rv1625c, Rv1264, Rv3645, Rv1318c, Rv1319c, Rv1320c, and Rv0386 [19–21,23–25]. Therefore, the Fig. 1. Mycobacterium tuberculosis adenylyl cyclase (AC) Rv2212. (A) Predicted domain composition of Rv2212. CHD, cyclase homol- ogy domain of class III ACs. (B) Purity of the recombinant ACs Rv2212 212)388 (CHD) and Rv2212 1)388 (holoenzyme) proteins. SDS ⁄ PAGE, 15%, stained with Coomassie blue. Lane 1, 2.3 lgof Rv2212 212)388 ; lane 2, 2.8 lg of Rv2212 1)388 ; molecular weight markers on the left. Mycobacterium tuberculosis AC Rv2212 A. Abdel Motaal et al. 4220 FEBS Journal 273 (2006) 4219–4228 ª 2006 The Authors Journal compilation ª 2006 FEBS most likely explanation for these highly variable activ- ities of Rv2212 in vitro is that the dimeric enzyme can exist in several, interconvertible states with different catalytic activities. Kinetic analysis of Rv2212 212)388 yielded a V max of 7.5 lmol cAMPÆmg )1 Æmin )1 (Table 1), an SC 50 of 3.3 mm ATP, and a Hill coeffi- cient of 1.8, which indicated strong positive coopera- tivity for the substrate and was consistent with the dimeric nature of bacterial class III ACs with two sub- strate-binding sites. The properties of the Rv2212 1)388 holoenzyme were similar (V max ¼ 3.9 lmol cAMPÆ mg )1 Æmin )1 ,SC 50 ¼ 2.1 mm ATP, Hill coefficient 1.8; Table 1). Thus, Rv2212 has a higher V max than any other mycobacterial AC studied to date (range 0.007– 2.1 lmol cAMPÆmg )1 Æmin )1 ). Furthermore, ATP sub- strate affinity was rather low compared to other myco- bacterial ACs, which had SC 50 concentrations in the range 0.06–1.2 mm ATP [19–25]. Stimulation of Rv2212 by fatty acids The response of Rv2212 proteins to pH changes was modest. Rv2212 1)388 had a five-fold higher activity at pH 6.5 compared to activity at pH 9, and with Rv2212 212)388 the maximal activity difference in activ- ity was three-fold between pH 6.5 and 7.6. This was in striking contrast to the pH-sensing AC Rv1264, which is stimulated 110-fold upon a shift from pH 9.0 to pH 6.0 [25]. The modest pH sensitivity of Rv2212 indi- cated that its N-terminus probably connects to differ- ent or additional regulatory inputs. It has been reported that the AC from Brevibacteri- um liquefaciens, which has an identical domain compo- sition to Rv1264 and Rv2212, is strongly activated by pyruvate, other a-ketocarbonic acids, glycine, alanine and lactate [27]. Therefore, we examined the effects of various metabolites on Rv2212 activity. At 1 mm concentrations, d-galactose, d-mannose, l-arabinose, l-rhamnose, d-glucose, d-fructose, fructose 1,6-bis- phosphate, glucose 6-phosphate, dl-threonine, l-iso- leucine, l-valine, l-asparagine, l-histidine, l-aspartic acid, d-alanine, l-alanine, l -cysteine, l-leucine, glycine, sodium chloride, potassium chloride, sodium citrate, sodium acetate, sodium bicarbonate, NADH, glyoxylic acid, a-ketoglutarate, pyruvate and phosphoenolpyru- vate did not significantly affect Rv2212 1)388 activity. Three-fold stimulation was obtained with 100 lm pal- mitic acid (not shown). Oleic, linoleic, linolenic and arachidonic acids at 100 lm produced strong activa- tion (Table 2). To determine whether the activation was specific for the Rv2212 AC, we established dose– response curves for oleic and linoleic acids using the class IIIc ACs Rv2212 and Rv1264, and the class IIIa AC Rv1625c, a membrane-bound isoform of unrelated Table 1. Kinetic analysis of Rv2212. Values are means ± SD. Numbers of experiments are in parentheses. *P < 0.001 compared to the respective control value with Rv2212 1)388 . Parameter Rv2212 1)388 (n ¼ 8) Rv2212 212)388 (n ¼ 6) Rv2212 1)388 (n ¼ 8) 100 l M linoleic acid Rv2212 1)388 (n ¼ 4) 100 l M oleic acid Rv2212 1)388 (n ¼ 4) 100 l M arachidonic acid Rv2212 1)388 (n ¼ 2) 170 l M polidocanol Rv2212 1)388 (n ¼ 2) 100 l M Brij 35 Rv2212 1)388 (n ¼ 2) 100 l M Triton X-100 V max (lmol cAMPÆmg )1 Æmin )1 ) 3.9 ± 0.6 7.5 ± 0.1* 4.1 ± 0.7 3.2 ± 0.9 2.9 ± 0.7 3.8 ± 0.2 3.1 ± 0.1 2.9 ± 0.1 SC 50 (mM) 2.1 ± 0.5 3.3 ± 0.2* 0.9 ± 0.1* 1.0 ± 0.2* 0.7 ± 0.3* 0.3 ± 0.1* 0.3 ± 0.1* 0.8 ± 0.1* Hill coefficient 1.8 ± 0.1 1.8 ± 0.1 1.3 ± 0.1* 1.2 ± 0.2* 1.0 ± 0.1* 1.1 ± 0.2* 1.1 ± 0.1* 1.7 ± 0.1 Table 2. Stimulation of Rv2212 by fatty acids and detergents. The numbers of experiments are in parentheses. NT, not tested. To be com- parable, ‘fold stimulation’ was measured at 100 l M of each fatty acid and at 119 lM polidocanol, 105 lM Triton X-100 and 119 lM Nonidet P-40, respectively. Values are means ± SD. All stimulations were highly significant (P<0.005). Rv2212 1)388 Rv2212 212)388 EC 50 (lM) Fold stimulation EC 50 (lM) Fold stimulation Linoleic acid 56 ± 6 (2) 6.0 ± 3.3 (13) 78 ± 4 (4) 2.6 ± 1.1 (9) Oleic acid 41 ± 4 (2) 7.5 ± 2.5 (3) 65 ± 1 (2) 2.2 ± 0.1 (2) Arachidonic acid 10 ± 1 (4) 8.4 ± 0.8 (4) 10 ± 1 (2) 2.4 ± 0.4 (2) Polidocanol 16 ± 1 (2) 11.0 ± 0.4 (2) 30 ± 3 (2) 7.0 ± 0.3 (2) Triton X-100 16 ± 1 (2) 5.0 ± 0.7 (2) NT NT Nonidet P-40 15 ± 1 (2) 4.3 ± 0.2 (2) NT NT A. Abdel Motaal et al. Mycobacterium tuberculosis AC Rv2212 FEBS Journal 273 (2006) 4219–4228 ª 2006 The Authors Journal compilation ª 2006 FEBS 4221 domain composition (Fig. 2A,B). Only Rv2212 was sti- mulated 9–13-fold by fatty acids with EC 50 concentra- tions around 50 lm (Fig. 2; Table 2), whereas Rv1625c was totally unresponsive and Rv1264 was stimulated a mere two-fold. This established a specificity of the effect of unsaturated fatty acids for the Rv2212 iso- form. Furthermore, the activation was specific to the Rv2212 holoenzyme, because the catalytic domain Rv2212 212)388 was stimulated only 2–3-fold (Table 2; Fig. 2A,B). Thus the stimulation was dependent on the presence of the N-terminal domain of Rv2212. Because the high variability in the AC activity of Rv2212 was reflected in a large variability of the extent of stimula- tion, we examined the effect in more detail. Fatty acids are prototypical detergents. We there- fore investigated whether the activation of Rv2212 was due to the detergent-like properties or whether it was the result of specific molecular interactions between the protein and the unsaturated fatty acids. Indeed, nonionic detergents such as polidocanol sti- mulated the Rv2212 holoenzyme 4–11-fold (Table 2; Fig. 2C). However, the effect of the detergents was identical for the catalytic domain alone and for the Rv2212 holoenzyme, and hence was unspecific (Table 2; Fig. 2C). We conclude that the stimulation of Rv2212 by unsaturated fatty acids only partly results from their amphipathic nature and distinct dif- ferences exist when compared to the effects of non- ionic detergents. Biochemical properties of Rv2212 activation by fatty acids and polidocanol Because of the unusually large variability in basal and linoleic acid-stimulated activities, we investigated whe- ther the extent of activation was dependent on the pre- set basal activity that each preparation of recombinant protein displayed (Fig. 3). In 13 assays, the activities of Rv2212 1)388 were 0.37 ± 0.34 and 1.43 ± 0.51 lmol cAMPÆmg )1 Æmin )1 in the absence and presence Fig. 2. Effect of fatty acids and polidocanol on Rv2212. Basal activ- ity is set to 100%. (A) Effect of oleic acid examined at respective optimal pH conditions. Rv2212 1)388 , filled squares, pH 6.5; Rv2212 212)388 , open squares, pH 6.5; Rv1264, circles, pH 5.7; Rv1625c, triangles, pH 7.5. Starting at 70 l M, stimulation by oleic acid is highly significant (P<0001) for Rv2212 1)388 , Rv2212 212)388 and Rv1264. For each fatty acid dilution, controls were carried out with the solvent alone. There were no significant solvent effects. (B) Effect of linoleic acid [symbols as in (A)]. All stimulations at 70 l M linoleic acid and higher were highly significant (P<0001) for Rv2212 1)388 , Rv2212 212)388 and Rv1264. (C) Effect of polidocanol on Rv2212 1)388 (squares) and Rv2212 212)388 (circles) at pH 6.5. Sti- mulations at 50 l M polidocanol and above were highly significant (P<0001) for both proteins. SD values are indicated by vertical bars (n ¼ 2). Mycobacterium tuberculosis AC Rv2212 A. Abdel Motaal et al. 4222 FEBS Journal 273 (2006) 4219–4228 ª 2006 The Authors Journal compilation ª 2006 FEBS of 100 lm linoleic acid, respectively (± SD). Thus, the variabilities were 92% and 35% in basal and stimula- ted AC activities, respectively. A plot of the stimula- tion factors versus basal activities illustrated this correlation (Fig. 3A). Rv2212 1)388 was stimulated most when the basal AC activity was below 0.4 lmol cAMPÆmg )1 Æmin )1 , whereas the potency of linoleic acid was rather modest in those instances where basal AC activity was already above 1 lmol cAMPÆmg -1 Æmin -1 (see insert in Fig. 3A). Under standard assay condi- tions, AC activity stimulated by 100 lm linoleic acid was in the range 0.9–2.3 lmol cAMPÆmg )1 Æmin )1 and thus fairly independent of basal activity, which varied by more than an order of magnitude (0.09–1.18). One interpretation is that the mycobacterial Rv2212 holo- enzyme, when heterologously expressed in E. coli, was isolated as a mixture of low-activity and high-activity states, and that the activation by linoleic acid was actually a conversion of the protein to a uniformly high-activity state. Thus, linoleic acid activation was only modest in cases where the purification a priori yielded an AC protein with a large fraction already in an activated state. Obviously, this applied exclusively to the Rv2212 1)388 holoenzyme, because the Rv2212 212)388 catalytic domain displayed fundament- ally different features of activation by linoleic acid. In nine assays, the AC activities of Rv2212 212)388 were 0.373 ± 0.206 and 0.924 ± 0.488 lmol cAMPÆmg )1 Æ min )1 (SD) in the absence and presence of 100 lm linoleic acid, respectively; that is, the variability of basal and stimulated activities in individual protein preparations were identical (54%; Fig. 3B). Thus, the lesser activation of the catalytic domain Rv2212 212)388 by linoleic acid was due to a totally different mode of activation compared to stimulation of the holoenzyme Rv2212 1)388 . Fatty acid activation of Rv2212 1)388 was then exam- ined kinetically (Fig. 4; Table 1). Stimulation was caused by an increase in substrate affinity and a con- comitant loss of cooperativity, whereas V max remained unaffected. Consequently, fatty acid stimulation decreased with increasing substrate concentration. At 5mm ATP, hardly any stimulation was observed (Fig. 4; Table 1). The kinetic basis of stimulation by polidocanol was identical (Table 1). We also tested the kinetics of activation by Brij 35, a lauryl alcohol deriv- ative, which has a 50% higher mean molecular mass. The effect of Brij 35 was identical to that of polidocan- ol, demonstrating that chain length did not matter (Table 1). Triton X-100, which contains the bulkier octylphenyl moiety, also increased substrate affinity but did not reduce cooperativity (Table 1). Obviously, diverse classes of detergents operated by different mechanisms, possibly reflecting different binding sites. Next, we determined the pH dependence of Rv2212 1)388 activation. pH profiles of basal and stimu- lated activities showed marked differences (Fig. 5). The basal AC activity of Rv2212 1)388 decreased five-fold from pH 6.5 to pH 9 (n ¼ 6). However, in the pres- ence of 100 lm oleic, linoleic or arachidonic acid, it decreased 20-fold, 19-fold and 26-fold, respectively. Obviously, the unsaturated fatty acids considerably bolstered the pH sensitivity of Rv2212 1)388 . Further- more, in a kinetic analysis at pH 9, we did not obtain substrate saturation even at 8 mm ATP, irrespective of the absence or presence of fatty acids (not shown). This demonstrated that the pH increase led to a Fig. 3. Correlation analysis of adenylyl cyclase (AC) stimulation by linoleic acid versus corresponding basal activity of the holoenzyme Rv2212 1)388 (A) (n ¼ 13) and the catalytic domain Rv2212 212)388 (B) (n ¼ 9). Open circles represent the average of four data groups. Group 1, basal activity < 0.3 lmolÆmg )1 Æmin )1 ; group 2, 0.3–0.5; group 3, 0.5–0.8; group 4, > 0.8). SD values for both dimensions are shown whenever they exceed the symbol size. A. Abdel Motaal et al. Mycobacterium tuberculosis AC Rv2212 FEBS Journal 273 (2006) 4219–4228 ª 2006 The Authors Journal compilation ª 2006 FEBS 4223 decrease in substrate affinity. The specificity of this effect with regard to fatty acids was emphasized by comparison with the detergent polidocanol. In the presence of polidocanol, the activity of Rv2212 1)388 was almost unchanged between pH 5.5 and pH 9. Although polidocanol activated Rv2212, this effect was pH-independent, i.e. the detergent abrogated the pH response of the AC, corroborating unequivocally that the effects of fatty acids and nonionic detergents on Rv2212 are mediated by distinctly different mecha- nisms. Discussion The class IIIc AC Rv2212 is the 10th AC isoform with proven enzyme activity out of 15 putative cyclases from M. tuberculosis, and like the other nine isoforms uses only Mn 2+ as a cofactor [18,21,24]. By domain composition and sequence similarity it is closely related to the mycobacterial Rv1264 isoform, yet the bio- chemical properties and regulation are different. The main differences between Rv2212 and Rv1264 are the lack of autoinhibition by the N-terminal domain of Rv2212, its remarkable response to fatty acids, and the high V max of the holoenzyme at its pH optimum [20,25]. The biochemical analysis of the Rv2212 iso- form was severely complicated by the exceptional vari- ability in basal enyzme activities. Although in the holoenzyme this variability was greatly reduced by addition of linoleic acid, the extent of stimulation was diminished in protein preparations with high basal activity. We excluded the possibility that the variability in AC activity was due to different levels of copurified fatty acids or detergents, because treatment of the recombinant protein with Bio-Beads SM-2 to remove lipids did not affect enzyme activity (data not shown). Therefore, we assume that (a) heterologous expression in E. coli resulted in a mixture of conformations of different specific activities, and that (b) fatty acids Fig. 4. Kinetic analyses of the stimulation of Rv2212 1)388 by 100 l M linoleic acid (A) and by 170 lM polidocanol (B). Squares, basal activity; circles, stimulated activity. SD values are shown (n ¼ 2–8). Stimulation by both compounds is highly significant (P<0.001) for 0.1–1.0 m M ATP and significant (P<0.05) at 2 mM ATP. Fig. 5. pH dependence of Rv2212 212)388 basal activity (open tri- angles), Rv2212 1)388 basal activity (open squares), and Rv2212 1)388 in the presence of 100 lM linoleic acid (circles), arachidonic acid (closed rhombi), oleic acid (closed triangles), and 170 l M polidocanol (open rhombi). The buffer systems used were: acetic acid ⁄ NaOH (pH 5 and 5.6), Bistris ⁄ HCl (pH 5.6–7.3) and Tris ⁄ HCl (pH 7.3–9). Vertical lines show SD values (n ¼ 2–5). The stimulations at pH 6.5 were compared to the respective activities at pH 9. Significance (P<0.05) was observed for basal and stimulated Rv2212 1)388 . Mycobacterium tuberculosis AC Rv2212 A. Abdel Motaal et al. 4224 FEBS Journal 273 (2006) 4219–4228 ª 2006 The Authors Journal compilation ª 2006 FEBS induced a more uniform high-activity state. The fatty acids, however, exerted a second, important effect in sharpening the pH profile of Rv2212 1)388 , i.e. in indu- cing pH sensing. In the presence of unsaturated fatty acids, Rv2212 showed a 20-fold increase in activity from pH 9 to pH 6.5, a response that is comparable to the 110-fold stimulation of the pH-sensing isoform Rv1264 by a similar pH shift [25]. This stimulation was not due to the detergent-like properties, because a nonionic detergent such as polidocanol stimulated Rv2212; however, it failed to promote pH sensing. We acknowledge that two mechanisms are possible to explain the pH dependence of the effect of fatty acids. One possibility is that fatty acids bind at any pH but activate only at lower pH values. The other possibility is that fatty acids neither bind nor activate at high pH. Currently, it is impossible to experimentally distinguish these possibilities. Because in the related AC Rv1264 oleic acid appears to be a protein constituent at any pH (see below), we favour pH-independent binding of the fatty acids to Rv2212. The kinetic bases of the fatty acid and pH-sensing responses of Rv2212 were similar. In both instances, activation was mainly due to an increase in ATP sub- strate affinity. In contrast, the pH response of Rv1264 is predominantly mediated by an increase in V max [25]. Prima facie this would argue for different mechanisms of pH sensing in the two isoforms. The structural transition upon acidification of Rv1264 has been eluci- dated (see scheme in Fig. 6A). In the active state (pH 6), the catalytic domains align as a closed dimer capable of binding ATP and of catalysis. In the in- active state (pH 8), the catalytic domains are drawn apart by extended a-helices such that they are neither able to bind ATP nor to catalyse cAMP formation. In Rv2212, the pronounced change in substrate affinity may be explained by an ability of ATP itself to shift the equilibrium towards an active state even at high pH and in the absence of activators. Structural analysis also suggests a binding site for fatty acids in Rv2212. In a recent high-resolution crys- tallographic study, we investigated the N-terminal reg- ulatory domain of Rv1264 and identified a binding site for fatty acids. The dimer of N-terminal domains contains two hydrophobic tunnels like cul-de-sacs. A number of different crystal forms (protein databank codes 2EV1, 2EV2, 2EV3 and 2EV4), as well as MS analysis, have allowed us to show that a fatty acid may be an intrinsic constituent of the Rv1264 N-ter- minus. We found that oleic acid is present in the Rv1264 crystal at any pH (unpublished results). This would explain the rather small stimulation of Rv1264 by added fatty acids (see above). In light of the data presented here, the physiological ligand for the N-ter- minal domain of Rv2212 is probably also an unsatur- ated fatty acid. To test whether the N-terminal domain of Rv2212 could also contain a tunnel for the uptake and binding of fatty acid molecules, we attempted modelling (Fig. 6B,C). The results demon- strate that fatty acid binding is compatible with the three-dimensional model of Rv2212, but they do not explain the apparent differences in biochemical responses between the two enzymes. Such an analysis would clearly require an experimentally determined structure of Rv2212. The activity of Rv2212 is governed by unsaturated fatty acids, pH and ATP concentration in vitro, but Fig. 6. Models of stimulation of Rv2212 by fatty acids. (A) Sche- matic representation of Rv2212 holoenzymes in the inhibited and active state according to crystal structures of Rv1264 [25]. In the inhibited state, the catalytic domains are recruited by the N-terminal domains and unable to bind ATP. In Rv2212, fatty acids and a pH shift synergize to cause activation. Alternatively, a high ATP con- centration may substitute for this. (B) Homology model of the dimeric Rv2212 N-terminal domain, viewed from the top. N and C indicate the N-terminus and C-terminus, respectively. Oleic acid is modelled with carbon atoms in green and oxygen in red. The arrows indicate the access sites of the ligand-binding tunnels. (C) Side view of (B). A. Abdel Motaal et al. Mycobacterium tuberculosis AC Rv2212 FEBS Journal 273 (2006) 4219–4228 ª 2006 The Authors Journal compilation ª 2006 FEBS 4225 what is the physiological relevance of these parameters? Viewing the results together, a significant response to fatty acids and consequently pH can only be expected if the ATP concentration is equal to or lower than the SC 50 value (0.9–2.1 mm; Table 1). The ATP content of M. tuberculosis is 130, 270 and 520 pg per 10 6 viable cells during chronic infection, during acute infection and in vitro, respectively [28]. Taking into account the reported average volume of a mycobacterium of 0.96 femtoliters [29], this translates into intracellular ATP concentrations of 0.27, 0.55 and 1.1 mm. Thus, it is conceivable that Rv2212 operates as an ATP sensor in vivo, integrating two other signals, the presence of certain fatty acids and the intracellular pH. The cyto- plasmic concentration of free fatty acids in M. tuber- culosis is unknown. Free fatty acids are the substrate for fatty acyl-AMP ligases, which are crucial for the synthe- sis of certain complex lipids of the cell envelope [30]. Thus the presence of free fatty acids in the mycobacte- rial cytoplasm appears certain. Furthermore, mycobac- teria have been shown to accumulate triacylglycerols as intracellular inclusions [31] that may contain some free fatty acids. Analysis of these lipid inclusions showed palmitic, stearic, oleic, palmitoleic and lignoceric acids to be major constituents. Thus these fatty acids are at least transiently present in the cytoplasm, and triacyl- glyerol synthesis actually is thought to serve detoxifica- tion of free fatty acids [31–33]. Taken together, the results suggest that Rv2212 is a biochemical integrator of three different signals, with cAMP as the output. Its properties are distinct from those of the other nine AC isoforms investigated to date. Some functional redundancy may exist concern- ing the isoform Rv1264, because both enzymes are able to respond to low pH. The results presented here are one further step towards understanding signal transduction through cAMP in M. tuberculosis. Experimental procedures Materials Genomic DNA from M. tuberculosis was a gift of E Boett- ger (University of Zu ¨ rich, Medical School). Radiochemicals were from Hartmann Analytik (Braunschweig, Germany). All enzymes were purchased from either Roche Diagnostics (Mannheim, Germany) or New England Biolabs (Frank- furt, Germany). pQE30 and Ni-nitrilotriacetic acid ⁄ agarose were from Qiagen (Hilden, Germany). Fine chemicals were purchased from Merck (Darmstadt, Germany), Roche Diagnostics, Roth (Karlsruhe, Germany) or Sigma (Taufkirchen, Germany). Bio-Beads SM-2, a removal agent for organic compounds including fatty acids and detergents, were from Bio-Rad (Munich, Germany) and were used according to the manufacturer’s instructions. Rv2212 constructs The annotated ORF of gene Rv2212 (GenBank Accession Number BX842579, NP_216728, 378 amino acids) starts with a GTG codon, although an in-frame ATG start codon is just 10 codons (30 bp) upstream. Therefore, we extended the ORF and included the N-terminal decapeptide MGVPAGTLRQ. The ORF (388 codons) was amplified by PCR using specific primers and genomic DNA as a tem- plate. BamHI and HindIII restriction sites were added at the 5¢-end and 3¢-end, respectively, and the PCR product was inserted into pQE30. This added an N-terminal MRGSH 6 GS tag. Similarly, the catalytic domain (Rv2212 212)388 ) was fitted with a 5¢ BamHI and a 3¢ HindIII site and inserted into pQE30. The fidelity of all constructs was verified by double-stranded DNA sequen- cing. Primer sequences are available on request. Expression and purification of proteins Plasmids with either Rv2212 1)388 or Rv2212 212)388 were transformed into E. coli BL21(DE3)[pREP4]. Protein expression was induced with 60 lm isopropyl-thio-b-d-gal- actoside for 4–5 h at 22 °C. Bacteria were collected by centrifugation at 2600 g for 15 min with a Centricon H-401 centrifuge, A8.24 rotor (Kontron-Hermle, Gosheim, Ger- many), washed once with buffer (50 mm Tris ⁄ HCl, 1 mm EDTA, pH 8), frozen in liquid nitrogen and stored at ) 80 °C. For purification, cells from 200 to 400 mL of cul- ture were suspended in 20 mL of lysis buffer (50 mm Tris ⁄ HCl, 0.02% a-monothioglycerol, pH 8), lysed by soni- cation, and treated with 0.2 mgÆml )1 lysozyme (30 min, 0 °C). Subsequently, 5 mm MgCl 2 and 20 lgÆml )1 DNaseI were added (30 min). After centrifugation (31 000 g for 30 min at 0 °C, Centricon H-401, A8.24 rotor), 15 mm imi- dazole (pH 8) and 250 mm NaCl (final concentrations) were added to the supernatant. Protein was equilibrated for a minimum of 60 min with 250 lLofNi 2+ -nitrilotriacetic acid ⁄ agarose slurry on ice, and then transferred to a mini- column and successively washed with 10 mL of buffer A (lysis buffer containing 15 mm imidazole, 250 mm NaCl and 5mm MgCl 2 ) and 5 mL of buffer B (lysis buffer with 15 mm imidazole and 5 mm MgCl 2 ). The protein was eluted with 0.3 mL of buffer C (lysis buffer with 150 mm imidazole and 2 mm MgCl 2 ). After addition of 20% glycerol (v ⁄ v), proteins were stored at ) 20 °C. AC assays AC activity was measured at 37 °C for 10 min in a volume of 100 lL [34]. Standard reactions contained 50 mm Mycobacterium tuberculosis AC Rv2212 A. Abdel Motaal et al. 4226 FEBS Journal 273 (2006) 4219–4228 ª 2006 The Authors Journal compilation ª 2006 FEBS Bistris ⁄ HCl (pH 6.5), 22% glycerol, 3 mm MnCl 2 , 500 l m [a 32 P]ATP and 2 mm [2,8- 3 H]cAMP. Substrate kinetics were analysed with 0.1–6 mm ATP and 10 mm MnCl 2 . Kin- etic constants were derived from Hill plots. Throughout, 100 nm Rv2212 1)388 and 200 nm Rv2212 212)388 were used. Solutions of fatty acids and detergents Fatty acids were prepared as 5 mm solutions in 1 mm Tris. Further dilutions were made with AC assay buffer. Deter- gents were prepared as 10% solutions in the AC assay buffer and further diluted with the same buffer. Homology modelling The homology model of Rv2212 was constructed from amino acids 4–186 with the high-resolution structure of the N-terminal domain of Rv1264 as a template (protein data- bank accession number 2EV1), using the program what if [35]. Briefly, a sequence alignment and the template struc- ture are input, and the program first mutates all disparate residues to glycine, and then places the side chains in reverse order of degrees of freedom for the individual resi- dues using the rotamer search procedure, usually resulting in bulky residues being placed first. The model is then manually corrected, geometrized and minimized using pro- cedures implemented in what if. Acknowledgements This work was supported by the Deutsche Forschungs- gemeinschaft. AM was supported by a scholarship of the Deutscher Akademischer Austauschdienst (DAAD). References 1 Botsford JL & Harman JG (1992) Cyclic AMP in pro- karyotes. Microbiol Rev 56, 100–122. 2 D’Souza CA & Heitman J (2001) Conserved cAMP sig- naling cascades regulate fungal development and viru- lence. FEMS Microbiol Rev 25, 349–364. 3 Gross A, Bouaboula M, Casellas P, Liautard JP & Dornand J (2003) Subversion and utilization of the host cell cyclic adenosine 5¢-monophosphate ⁄ protein kinase A pathway by Brucella during macrophage infection. J Immunol 170, 5607–5614. 4 Petersen S & Young GM (2002) Essential role for cyclic AMP and its receptor protein in Yersinia enterocolitica virulence. Infect Immun 70, 3665–3672. 5 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. 6 Wolfgang MC, Lee VT, Gilmore ME & Lory S (2003) Coordinate regulation of bacterial virulence genes by a novel adenylate cyclase-dependent signaling pathway. Dev Cell 4, 253–263. 7 Rickman L, Scott C, Hunt DM, Hutchinson T, Menen- dez MC, Whalan R, Hinds J, Colston MJ, Green J & Buxton RS (2005) A member of the cAMP receptor protein family of transcription regulators in Mycobac- terium tuberculosis is required for virulence in mice and controls transcription of the rpfA gene coding for a resuscitation promoting factor. Mol Microbiol 56, 1274– 1286. 8 Spreadbury CL, Pallen MJ, Overton T, Behr MA, Mostowy S, Spiro S, Busby SJ & Cole JA (2005) Point mutations in the DNA- and cNMP-binding domains of the homologue of the cAMP receptor protein (CRP) in Mycobacterium bovis BCG: impli- cations for the inactivation of a global regulator and strain attenuation. Microbiology 151, 547–556. 9 Peterkofsky A, Reizer A, Reizer J, Gollop N, Zhu PP & Amin N (1993) Bacterial adenylyl cyclases. Prog Nucleic Acid Res Mol Biol 44, 31–65. 10 Tang WJ & Hurley JH (1998) Catalytic mechanism and regulation of mammalian adenylyl cyclases. Mol Phar- macol 54, 231–240. 11 Sunahara RK & Taussig R (2002) Isoforms of mamma- lian adenylyl cyclase: multiplicities of signaling. Mol Interv 2, 168–184. 12 Barzu O & Danchin A (1994) Adenylyl cyclases: a het- erogeneous class of ATP-utilizing enzymes. Prog Nucleic Acid Res Mol Biol 49, 241–283. 13 Sismeiro O, Trotot P, Biville F, Vivares C & Danchin A (1998) Aeromonas hydrophila adenylyl cyclase 2: a new class of adenylyl cyclases with thermophilic properties and sequence similarities to proteins from hyperthermo- philic archaebacteria. J Bacteriol 180, 3339–3344. 14 Cotta MA, Whitehead TR & Wheeler MB (1998) Iden- tification of a novel adenylate cyclase in the ruminal anaerobe, Prevotella ruminicola D31d. FEMS Microbiol Lett 164, 257–260. 15 Linder JU & Schultz JE (2003) The class III adenylyl cyclases: multi-purpose signalling modules. Cell Signal 15, 1081–1089. 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 Shenoy AR & Visweswariah SS (2004) Class III nucleo- tide cyclases in bacteria and archaebacteria: lineage- specific expansion of adenylyl cyclases and a dearth of guanylyl cyclases. FEBS Lett 561, 11–21. 18 Reddy SK, Kamireddi M, Dhanireddy K, Young L, Davis A & Reddy PT (2001) Eukaryotic-like adenylyl A. Abdel Motaal et al. Mycobacterium tuberculosis AC Rv2212 FEBS Journal 273 (2006) 4219–4228 ª 2006 The Authors Journal compilation ª 2006 FEBS 4227 cyclases in Mycobacterium tuberculosis H37Rv: cloning and characterization. J Biol Chem 276, 35141–35149. 19 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. 20 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. 21 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. 22 Shenoy AR, Sreenath NP, Mahalingam M & Viswes- wariah SS (2005) Characterization of phylogenetically distant members of the adenylate cyclase family from mycobacteria: Rv1647 from Mycobacterium tuberculosis and its orthologue ML1399 from M. leprae. Biochem J 387, 541–551. 23 Castro LI, Hermsen C, Schultz JE & Linder JU (2005) Adenylyl cyclase Rv0386 from Mycobacterium tuber- culosis H37Rv uses a novel mode for substrate selection. FEBS J 272, 3085–3092. 24 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. 25 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. 26 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. 27 Takai K, Kurashina Y, Suzuki-Hori C, Okamoto H & Hayaishi O (1974) Adenylate cyclase from Brevibacter- ium liquefaciens. I. Purification, crystallization, and some properties. J Biol Chem 249, 1965–1972. 28 Dhople AM & Ryon DL (2000) ATP content of Myco- bacterium tuberculosis grown in vivo and in vitro. Microbios 101, 81–88. 29 Cox RA (2004) Quantitative relationships for specific growth rates and macromolecular compositions of Mycobacterium tuberculosis, Streptomyces coelicolor A3(2) and Escherichia coli B ⁄ r: an integrative theoretical approach. Microbiology 150, 1413–1426. 30 Trivedi OA, Arora P, Sridharan V, Tickoo R, Mohanty D & Gokhale RS (2004) Enzymic activation and trans- fer of fatty acids as acyl-adenylates in mycobacteria. Nature 428, 441–445. 31 Garton NJ, Christensen H, Minnikin DE, Adegbola RA & Barer MR (2002) Intracellular lipophilic inclu- sions of mycobacteria in vitro and in sputum. Micro- biology 148, 2951–2958. 32 McCarthy C (1971) Utilization of palmitic acid by Mycobacterium avium. Infect Immun 4, 199–204. 33 Weir MP, Langridge WH 3rd & Walker RW (1972) Relationships between oleic acid uptake and lipid meta- bolism in Mycobacterium smegmatis . Am Rev Respir Dis 106, 450–457. 34 Salomon Y, Londos C & Rodbell M (1974) A highly sensitive adenylate cyclase assay. Anal Biochem 58, 541– 548. 35 Vriend G (1990) WHAT IF: a molecular modeling and drug design program. J Mol Graph 8, 52–56, 29. Mycobacterium tuberculosis AC Rv2212 A. Abdel Motaal et al. 4228 FEBS Journal 273 (2006) 4219–4228 ª 2006 The Authors Journal compilation ª 2006 FEBS . Fatty acid regulation of adenylyl cyclase Rv2212 from Mycobacterium tuberculosis H37Rv Amira Abdel Motaal 1 , Ivo Tews 2 , Joachim E. Schultz 1 and. 1. Mycobacterium tuberculosis adenylyl cyclase (AC) Rv2212. (A) Predicted domain composition of Rv2212. CHD, cyclase homol- ogy domain of class III ACs. (B) Purity of the recombinant ACs Rv2212 212)388 (CHD). sensor adenylyl cyclase. Fatty acid binding to Rv2212 was modelled by homology to a recent structure of the N-ter- minal domain of Rv1264, in which a fatty acid- binding pocket is defined. Rv2212