Eukaryotic-like adenylyl cyclases in Mycobacterium tuberculosis H37Rv: cloning and characterization.

Screening the Mycobacterium tuberculosis H37Rv genomic library for complementation of catabolic defect for cAMP-dependent expression of maltose operon produced the adenylyl cyclase gene (Mtb cya, (1997)) annotated later as Rv1625c (Cole, S. T., Brosch, R., Parkhill, J., Garnier, T., Churcher, C., Harris, D., Gordon, S. V., Eiglmeier, K., Gas, S., Barry, C. E., III, et al. (1998) Nature 393, 537-544). The deduced amino acid (aa) sequence (443 aa) encoded by Mtb cya contains a single hydrophobic domain of six transmembrane helices (152 aa) in the amino-terminal half of the protein. Flanking this domain are an arginine-rich (17%) amino-terminal cytoplasmic tail (46 aa) and a carboxyl-terminal cytoplasmic domain (245 aa) with extensive homology to the catalytic core of eukaryotic adenylyl cyclases. Site-directed mutagenesis of Arg(43) and Arg(44) to alanine/glycine showed a loss of adenylyl cyclase activity, whereas mutagenesis to lysine restored the activity. Hence it is proposed that the formation of the catalytic site in Mtb adenylyl cyclase requires an interaction between Arg(43) and Arg(44) residues in the distal cytoplasmic tail and the carboxyl-terminal cytoplasmic domain. Mtb adenylyl cyclase activity at the physiological concentration of ATP (1 mm) was 475 nmol of cAMP/min/mg of membrane protein in the presence of Mn(2+) but only 10 nmol of cAMP/min/mg of membrane protein in the presence of Mg(2+). The physiological significance of the activation of Mtb adenylyl cyclase by Mn(2+) is discussed in view of the presence of manganese transporter protein in mycobacteria and macrophages wherein mycobacteria reside.

Mycobacterium tuberculosis H37Rv, the etiologic agent of tuberculosis, accounts for more human causalities than any other single infection (1). Pathogenesis of M. tuberculosis is not attributable to any single gene product. In other bacterial pathogens such as Bordetella pertussis and Bacillus anthrax, adenylyl cyclase, the enzyme responsible for the synthesis of adenosine 3Ј,5Ј-monophosphate (cAMP), plays a pivotal role in pathogenesis (2). The secreted form of adenylyl cyclase from B. pertussis (3) and B. anthrax (4) invades a variety of eukaryotic cells and is activated by the intracellular calmodulin of the host. This results in the unregulated conversion of ATP to a supraphysiological concentration of cAMP, which in turn has a profound effect on the metabolism and immune function of the host cell (5,6). Thus, in infections caused by B. pertussis and B. anthrax, adenylyl cyclase plays a critical role in the onset of the disease. To examine whether adenylyl cyclase in M. tuberculosis has any role in infection, we report in this paper the cloning of cya gene from M. tuberculosis H37Rv, heterologous expression in Escherichia coli, and characterization of adenylyl cyclase (Mtb AC, 1 GenBank TM accession no. AF017731). Subsequent sequencing of the M. tuberculosis H37Rv genome revealed the presence of five genes (Rv1264, Rv1318c, Rv1319c, Rv1320c, and Rv1625c) annotated as putative adenylyl cyclases (7). The Mtb cya gene we cloned is identical to Rv1625c. The protein encoded by Rv1264 was only 0.5% as active as adenylyl cyclase encoded by the Mtb cya gene (Rv1625c). The gene product of Rv1320c, which is nearly identical to that of Rv1318c and Rv1319c, had no adenylyl cyclase activity. Mtb AC (the Rv1625 gene product) has a single six helical transmembrane domain in the amino terminus and a carboxyl-terminal cytosolic domain with extensive homology to the catalytic core of mammalian adenylyl cyclase (8). This secondary structural organization is unique to a prokaryotic adenylyl cyclase.

EXPERIMENTAL PROCEDURES
Enzymes and Chemicals-Restriction endonucleases, T4 DNA ligase, and T4 polynucleotide kinase were purchased from New England Biolabs, Beverly, MA. Calf intestinal alkaline phosphatase was from Roche Molecular Biochemicals. Acrylamide, bisacrylamide, and phenol were obtained from Life Technologies Inc. Deoxynucleoside triphosphates were purchased from Amersham Pharmacia Biotech. SeaKem GTG agarose and NuSieve GTG agarose were purchased from FMC BioProducts, Rockland, ME. Dye-conjugated Dideoxyterminator DNA sequencing kits were from PerkinElmer Life Sciences. PCR products and DNA from agarose gels were purified using kits from Qiagen, Valencia, CA. Mutagenesis kit was from Bio-Rad. [␣-32 P]ATP (30 Ci/mmol) and * Certain commercial equipment, instruments, and materials are identified in this paper to specify the experimental procedure as completely as possible. In no case does such identification imply a recommendation or endorsement by the National Institute of Standards and Technology nor does it imply that the material, instrument, or equipment identified is necessarily the best available for the purpose. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This paper is dedicated to Emeritus Prof. T. A. Venkitasubramanian, University of Delhi, India, on his 75th birthday. Prof. Venkitasubramanian made seminal contributions to the understanding of the physiology of M. tuberculosis.
E. coli Strains and Plasmids-E. coli strains C600 lysogen (r Ϫ m ϩ cI ϩ ), which carries a wild type repressor that turns off transcription from the P L promoter, and MZ1 (cI857 t s ), which carries a temperature-sensitive repressor that allows temperature-induced transcription of genes under the P L promoter (9), were kindly provided by Dr. Donald Court of the National Cancer Institute, Frederick, MD. The pRE expression vector based on the P L promoter (10) and pRK248 plasmid harboring cI857 t s (11) have been described previously. E. coli TP 610 (12), a cya deletion strain, was obtained from Dr. Antoine Danchin, Pasteur Institute. Isogenic E. coli strains AP850 (cya deletion) and CA8445 (cya/crp deletion) were obtained from Dr. Alan Peterkofsky, National Institutes of Health (13).
Molecular Cloning of cya Gene-Twenty micrograms of M. tuberculosis H37Rv chromosomal DNA (generously provided by Drs. John Belisle and Patrick Brennan, Colorado State University) was partially digested with 1 unit of Sau3AI restriction endonuclease at 37°C for 15 min. DNA fragments, size-selected for 2-6 kbp, were excised from a 1% agarose gel (0.8% NuSieve GTG-agarose, 0.2% SeaKem GTG-agarose) and purified by phenol extraction and ethanol precipitation. DNA fragments were ligated into the BamHI site of pBR322. The cya deletion strain TP610 was electroporated (Life Technologies, Inc.) with the ligation products, and the cells were spread on MacConkey/maltose plates containing ampicillin (100 g/ml). Eight red colonies indicative of maltose fermentation, which requires cAMP production, were purified from a background of about 20,000 white colonies, and plasmids were isolated (Qiagen plasmid purification kit). The nucleotide sequence of the 5Ј and 3Ј ends of the inserts were determined using pBR322 primers upstream and downstream of the BamHI site. The length of the inserts was assessed by EcoRI and SalI digestion of the eight recombinant plasmids. Nucleotide sequence of all of the inserts was determined using a dye-conjugated Dideoxyterminator kit with primers designed by gene walking. Translation of the nucleotide sequence revealed an open reading frame (ORF), with no initiation codon but containing a stop codon, with high homology to the eukaryotic adenylyl cyclases. Thus we cloned only the carboxyl terminus of Mtb adenylyl cyclase by maltose fermentation.
From the nucleotide sequence of the longest clone (3.4 kbp), a BamHI restriction endonuclease sequence was identified in the coding sequence of the Mtb cya gene. A 22-base oligonucleotide (5Ј-dCGGTAAGTT-TCGTCC TCACGAC-3Ј) upstream to the BamHI site was synthesized, 5Ј end-labeled with [␥-32 P]ATP and polynucleotide kinase, and used as a probe in Southern hybridization (14) of the chromosomal DNA completely digested with BamHI. A 2.6-kbp fragment, identified on the Southern blot, was ligated into the BamHI site of pRE1 vector, which was developed to clone lethal genes such as E. coli adenylyl cyclase (10). E. coli C600 lysogen was transformed with the ligation products, and the colonies were screened with the oligonucleotide probe; plasmids were isolated from pure positive colonies, and the nucleotide sequence of the 2.6-kbp BamHI fragment was determined. Translation of the nucleotide sequence revealed an ORF with a GTG initiation codon in-frame with the ORF of the carboxyl terminus.
Construction of Full-length Mtb cya Gene in pRE Expression Vector-Because none of the clones described above had the full-length cya gene, we constructed a full-length cya gene under the transcriptional control of the heat-inducible P L promoter in the pRE1 expression vector as follows. A NdeI restriction endonuclease sequence, 5Ј-CATATG-3Ј, was introduced at the initiation codon of Mtb cya by PCR using a forward primer 5Ј-dATTAGGCTTTAGCATATGGCGGCAAG-3Ј. The reverse primer was 5Ј-dCGCGGATCCTTCAATTGCG-3Ј, encompassing the internal BamHI site (Fig. 1, base pairs 1842-1847) (bold letters represent mutations). The template DNA was the pRE1 plasmid containing the Mtb cya gene coding for the amino-terminal 335 aa of adenylyl cyclase, obtained by Southern hybridization. The amplified product was purified and digested with NdeI and BamHI restriction endonucleases; the NdeI-BamHI cya fragment was ligated into the pRE1 expression vector and transformed into C600 lysogen. Several recombinants were isolated from C600 lysogen and sequenced. All of the recombinants had unaltered nucleotide sequences. One of the recombinants was chosen for the construction of a full-length cya gene in the pRE1 expression vector. The BamHI fragment that codes for the carboxyl terminus of Mtb AC was purified from the 3.4-kbp insert obtained in the first step of cloning, and the fragment was cloned into the BamHI site of the pRE1-cya (335 aa) recombinant. A recombinant with the full-length Mtb cya gene containing the BamHI fragment in the correct orientation was isolated from E. coli C600 lysogen and introduced into E. coli MZ1 for protein expression.
Mutations were confirmed by DNA sequencing using dye-conjugated Dideoxyterminator kit. The NdeI-BstEII fragment in M13 was amplified by PCR using 5Ј-dGAGCTCGGTACCATATGGCGGCA-3Ј (NdeI primer) and 5Ј-dTCTAGAGGATCCCGGTTACCACGAC-3Ј (BstEII primer). The amplified fragment was purified from unreacted primers, digested with NdeI and BstEII, purified from low melting agarose, and cloned into the pRE1-Mtb cya recombinant plasmid deleted for the DNA fragment between the NdeI and BstEII sites.
Amplification conditions for Rv1264 were as follows: melting at 98°C for 45 s, annealing at 68°C for 45 s, and polymerization at 72°C for 2 min. Amplification conditions for Rv1320c were: melting at 98°C for 45 s, annealing at 60°C for 45 s, and polymerization at 72°C for 2 min. Amplification was performed for 30 cycles in the presence of 5% dimethyl sulfoxide. Amplified products were purified from unreacted primers, digested with NdeI and XbaI, purified from low melting agarose gel, and cloned into a similarly digested pRE1 expression vector. Recombinants were isolated from E. coli C600 lysogen and introduced into E. coli MZ1 for protein expression.
Expression of Adenylyl Cyclase(s)-The pRE1/cya recombinant plasmid was introduced into E. coli strain MZ1, which carries the temper-ature-sensitive repressor cI857t s (11), for the expression of adenylyl cyclase. Cells were grown in 150 ml of LB medium containing ampicillin (50 g/ml) at 32°C to an A 650 of 0.4. A 50-ml aliquot of this culture was saved for estimating the uninduced level of adenylyl cyclase. The remainder of the culture was diluted with an equal volume of fresh LB medium kept at 60°C and immediately placed in a 42°C water bath shaker for induction. A 50-ml aliquot of induced cells was withdrawn at hourly intervals for up to 4 h. Cells were collected by centrifugation at 10,000 ϫ g for 15 min and washed with 50 mM Tris-HCl buffer, pH 7.5. Cells were suspended in buffer A (50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM dithiothreitol, and 0.1 mM phenylmethylsulfonyl fluoride) and broken by passage through a French press at 10,000 p.s.i. Cell lysates were examined for adenylyl cyclase activity (16) and for protein expression by SDS-polyacrylamide gel electrophoresis (17). Protein concentration was determined by the method of Lowry (18).
Preparation of Cytoplasmic Membranes-Three-hour induced cells from a 2 -liter culture were suspended in 30 ml of buffer A and passed twice through a French press at 10,000 p.s.i. The crude homogenate was centrifuged at 10,000 ϫ g for 15 min. The supernatant from this step was centrifuged at 100,000 ϫ g for 1 h. The membrane pellet was washed with 20 ml of buffer A without resuspension and recentrifuged for 30 min. The washed pellet was homogenized in 30 ml of buffer A. All operations were carried out at 4°C. Aliquots of membranes were kept frozen at Ϫ20°C. Membranes were diluted in buffer A for adenylyl cyclase assays.
Adenylyl Cyclase Assay-Adenylyl cyclase was assayed (16) in an incubation mixture of 0.25 ml containing 25 mM Tris, pH 7.5, 20 mM MgCl 2 , 1 mM dithiothreitol, 1 mM [␣-32 P]ATP (20 -30 cpm/pmol), 1 mM cAMP, and an ATP regenerating system containing 20 mM creatine phosphate and 50 units of creatine phosphokinase. Reactions were initiated by the addition of the enzyme source and carried out at 30°C. At 10 and 20 min (or 5 and 10 min), 100-l aliquots were withdrawn into 0.2 ml 1 N perchloric acid to terminate the reaction, and cAMP was purified as described (19).
Nucleotide Sequence Accession Number-The nucleotide sequence presented in Fig. 1 has been submitted to GenBank TM under accession number AF017731.

Molecular Cloning of Mtb cya Gene-
The full-length Mtb cya gene was isolated in two steps. First, the gene corresponding to the carboxyl-terminal 365 amino acids was cloned by complementation of the catabolic defect for maltose fermentation in the cya deletion strain, E. coli TP610, which lacks adenylyl cyclase activity. Eight red colonies, indicative of putative positives, were identified on MacConkey/Maltose plates. Cyclic AMP synthesis, measured by the in vitro adenylyl cyclase assay in extracts of E. coli TP610 harboring one or the other of the eight clones, was barely above the detection level (level of detection is 1 pmol of cAMP/min/mg of protein; data not shown) but was apparently enough to "turn on" the maltose operon in vivo. Restoration of the catabolic defect with all of these clones was dependent on the cAMP receptor protein, since a cya/crp deletion strain transformed with these plasmids failed to ferment lactose and maltose (Table I). A GenBank TM search for the presence of a common ORF in all eight clones revealed that the ORF had extensive homology (see below) to the catalytic core of eukaryotic adenylyl and guanylyl cyclases.
It is noteworthy that the orientation of all eight of the inserts was identical and the DNA sequence of all the clones had a start point within about 100 base pairs of each other, but the end points were significantly different, ranging from about 1 to 3.4 kbp. Because there is no apparent translation initiation signal corresponding to the ORF, we reasoned that these clones encode a fusion protein of the 97 amino acids of the tetracycline resistance protein at the BamHI cloning site in pBR322 and an adenylyl cyclase catalytic core capable of synthesizing cAMP. Indeed, sequence analysis at the fusion revealed that Trp 97 of the tetracycline resistance protein was fused to Ile 79 in cya clones 1, 3, and 7; Ile 89 in cya clones 2, 6, and 8; and Ile 113 in cya clones 4 and 5. The absence of the translation initiation signal corresponding to this ORF in these eight clones also led us to believe that the full-length cya gene could not be obtained either under the transcriptional control of the constitutive tet promoter in pBR322 or its own transcriptional unit in a multicopy plasmid. This observation is in accord with our previous finding that uncontrolled expression of adenylyl cyclase with concomitant overproduction of cAMP is lethal to E. coli cells, which led to the development of the pRE expression vector for controlled lethal gene expression (10). Consequently, Southern hybridization was used to obtain that portion of the cya gene coding for the amino terminus of Mtb AC in the pRE vector as described under "Experimental Procedures." The complete coding sequence of the Mtb cya gene was assembled from the overlapping sequences of the clones obtained in pBR322 and pRE1 vectors. The nucleotide sequence and the deduced amino acid sequence of the Mtb cya are presented in Fig. 1 and have been deposited in GenBank TM (accession no. AF017731). A putative open reading frame of 443 amino acids, beginning with GTG as an initiation codon at the nucleotide triplet 841-843 and terminating with the nucleotide triplet TGA at 2170 -2172, was identified. From the M. tuberculosis genome sequence published later (7), the putative start codon for the Mtb cya gene was suggested to be the GTG triplet at 916 -918 in the same reading frame as above. The ribosome binding sequences for both of the GTG initiators are poor matches to the consensus sequence. However, visual examination of the consensus sequence suggested that the GTG initiator at nucleotides 841-843 has a relatively better ribosome binding sequence (nucleotides 826, 829, 830, and 831, underlined in Fig. 1). Hence we have expressed adenylyl cyclase in the E. coli MZ1/pRE1 expression vector system beginning at nucleotide 841. The Mtb cya gene that we cloned earlier (Gen-Bank TM accession no. AF017731 (1997)) and a gene with annotation Rv1625c from M. tuberculosis genome sequence (7) published later are identical except for the difference in the position of initiation codon.
Putative Secondary Structure of Mtb AC-A hydropathy plot (not shown) of the Mtb adenylyl cyclase protein sequence suggests that the amino-terminal domain (aa 47-199) of the enzyme contains six transmembrane helices (Fig. 2), which is consistent with the observation that the Mtb adenylyl cyclase expressed in E. coli is membrane-bound (see below). With the exception of one negatively charged amino acid in the fourth TABLE I Phenotypic properties of cya genes from M. tuberculosis A cya deletion strain AP850 (deleted for adenylyl cyclase) and a cya/crp double deletion strain CA8445 (deleted for both adenylyl cyclase and cAMP receptor protein) were transformed by electroporation with pRE1 (amp R ) harboring Rv1625c, Rv1264, or Rv1320c. Simultaneously, pRK248 plasmid (tet R ) that carries temperature-sensitive mutation in cI857 was introduced into the cells. Cells were plated on MacConkey agar containing 1% sugar, 100 g ampicillin/ml, and 10 g tetracyclin/ ml. Plates were incubated at 30°C for 16 h and shifted to 42°C for up to 4 h to induce the cya gene under the control of heat-inducible P L promoter. The color of the colonies was inspected visually. AP850 and CA8445 cells transformed with Mtb cya clones 1-8 (in pBR322) were plated on MacConkey agar containing 1% sugar and 100 g ampicillin/ml and incubated directly at 37°C for 16 h. The strength of fermentation of sugars is recorded as ϩϩϩ for strongly positive to Ϯ for weakly positive. Negative fermentation is indicated by Ϫ.
cya gene (Asp) and fifth (Glu) helices, the other helical residues are all hydrophobic. The amino-terminal cytoplasmic tail is rich in arginine residues (17%), and their importance in adenylyl cy-clase activity has been elucidated (see below).
Expression and Activity Analysis of Adenylyl Cyclase-The expression profile (Fig. 3) and activity (Table II)  lyl cyclase (Rv1625c) are presented. The uninduced cell extract had a very low level of cAMP-synthesizing activity from the chromosomally encoded adenylyl cyclase in E. coli MZ1 strain. The induced cell extract showed about 200-fold more cAMPproducing activity, indicating that the expressed gene product is indeed adenylyl cyclase. Coincident with the high activity, an increase in the Coomassie stain intensity corresponding to the molecular weight of adenylyl cyclase (45 kDa) was observed in the induced cell extract (Fig. 3, lane 1). A protein similar in molecular weight to that encoded by the E. coli chromosome is present at low intensity in the induced cells harboring a control plasmid (Fig. 3, lane 4). Quantitation of adenylyl cyclase by gel scan indicated that the enzyme represented about 5% of the total protein. Because it is unlikely that 5% of a six helical transmembrane protein could all be incorporated into the membrane space, we investigated the localization and the associated activity of the enzyme in the membrane and cytosolic fractions (Table II). About 65% of adenylyl cyclase activity was detected in the cytoplasmic membrane, whereas essentially all of the expressed protein with low activity was localized in the cytosolic fraction. This suggests that the portion of adenylyl cyclase incorporated into the membrane has an ordered structure and is a high activity form but the soluble enzyme is misfolded.
Mn-ATP Is a Better Substrate for Mtb AC Than Mg-ATP-We investigated the divalent metal ion requirement for adenylyl cyclase activity. Cytoplasmic membranes enriched with the Mtb AC were isolated from E. coli MZ1 as described under "Experimental Procedures." Kinetic studies using membranes containing Mtb AC revealed that the Michaelis-Menten constant (K m ) for adenosine triphosphate in the presence of 20 mM Mg 2ϩ was 3.5 mM with a V max of 210 nmol of cAMP/min/mg of protein, whereas the K m for adenosine triphosphate in the presence of 2 mM Mn 2ϩ was 50 M with a V max of 475 nmol of cAMP/min/mg protein (Table III). This amounts to a 70-fold lower K m for Mn-ATP 2Ϫ compared with Mg-ATP 2Ϫ . The rate of the reaction with 1 mM Mg-ATP 2Ϫ and 1 mM Mn-ATP 2Ϫ , at the physiological concentration of ATP, was 10 and 475 nmol of cAMP formed/min/mg of membrane protein, respectively, which amounts to a 47-fold activation by Mn 2ϩ .
Site-directed Mutagenesis of Amino-terminal Arginine Residues-Based on a putative secondary structure (Fig. 2), the arginine-rich amino-terminal cytoplasmic tail contributes to the net positive charge of this region. Although these arginine residues satisfy the "positive inside" rule in the membrane protein topology (20), we investigated whether these arginine residues have any role in AC activity in view of the ability of arginine residues to bind ATP, the substrate for adenylyl cyclase. A mutant version of Mtb AC was constructed wherein all eight arginine residues were changed by cassette mutagenesis to four alanine and four glycine residues. This mutant Mtb cya (R4G, R18A, R19G, R27A, R31G, R43A, R44G, R46A) was cloned into the pRE1 expression vector, and adenylyl cyclase was expressed in E. coli MZ1. Although the level of expression of the mutant adenylyl cyclase was the same as for the wild type adenylyl cyclase (Fig. 4, lanes 12 and 3, respectively), the mutant had no adenylyl cyclase activity over and above the uninduced level with Mg-ATP 2Ϫ or Mn-ATP 2Ϫ as the substrate (Table IV). This result suggests an important role for the amino-terminal cytoplasmic tail in the catalysis of ATP to cAMP by the cytoplasmic domain. However, we cannot discount the possibility that topological perturbations might have occurred in the disposition of membrane helices upon a drastic change of

TABLE II Expression and cellular localization of Mtb AC in E. coli
Mtb AC (Rv1625c) in pRE1 plasmid was expressed in E. coli MZ1 for 3 h at 42°C and the 100,000 ϫ g cytosolic and membrane fractions were isolated. AC assays were carried out using French press extract (78 g of protein), 100,000 ϫ g cytosolic fraction (46 g of protein), and cytoplasmic membranes (12 g of protein) for each time point as described under "Experimental Procedures." Assays were carried out for 5 and 10 min at 30°C. The enzyme activity was linear up to 30 min. Concentration of ATP was 1 mM, and that of MgCl 2 was 20 mM.  Cytoplasmic membranes of the wild type AC were prepared as described in "Experimental Procedures." The concentration of creatine phosphate for the ATP regenerating system in the AC assay was 20 mM in the presence of 20 mM MgCl 2 , whereas it was 2 mM in the presence of 2 mM MnCl 2 (at the higher Mn 2ϩ concentration, creatine phosphate was precipitated). The range of ATP concentration was 1 M to 10 mM for Mg 2ϩ and 1 M to 2 mM for Mn 2ϩ . The amount of membrane protein was 2.72 g in the Mg-ATP assays and 0.272 g in the Mn-ATP assays. Assays were carried out for 10 min at 30°C. The results are the average of two experiments.

Mg 2ϩ
Mn 2ϩ all the arginine residues to glycine and alanine residues such that adenylyl cyclase was rendered inactive. To examine which of the arginine residue(s), if any, may play a critical role in the activity, site-directed mutagenesis of each arginine residue was performed. An adenylyl cyclase activity profile of the wild type and site-directed mutants is presented in Table IV. Adenylyl cyclase mutants R4G, R18A, R19G, R18A/R19G, R27A, R31G, and R46A are all nearly as active (70 -120%) as the wild type enzyme with either Mg-ATP 2Ϫ or Mn-ATP 2Ϫ as the substrate. It is clear from these data that Arg 43 and Arg 44 have a significant role in the activity of the enzyme. The R43A and R44G mutants are only about 35% as active as the wild type enzyme with Mg-ATP 2Ϫ , whereas the double mutant R43A/R44G is nearly inactive having only about 0.2% activity of the wild type enzyme. With Mn-ATP 2Ϫ as the substrate, the R43A mutant has 60% activity and the R44G mutant has 40% activity, reflecting the higher affinity of adenylyl cyclase for this substrate (Table III). The double mutant R43A/R44G, however, has only 1% activity even with Mn-ATP 2Ϫ as the substrate. This interesting observation suggests that Arg 43 and Arg 44 vicinal residues have critical role in the catalysis of ATP to cAMP by the cytoplasmic domain. We tested whether a conserved substitution like lysine for arginine would restore the activity. We found that lysine restored the activity that was lost by alanine/ glycine substitutions at Arg 43 and Arg 44 . The double lysine mutant R43K/R44K resulted in an ϳ30-fold activation of adenylyl cyclase activity compared with the R43A/R44G mutant (0.2 3 7% with Mg-ATP 2Ϫ and 1.3 3 39% with Mn-ATP 2Ϫ as substrate). These results suggest that the formation of the catalytic site requires an interaction between the arginine-rich distal amino-terminal cytoplasmic tail and the carboxyl-terminal cytoplasmic domain.
Protein expression for the wild type (Fig. 4, lane 3) and the site-directed mutants R43A, R44G, R46A, R43A/R44G, R43K, R44K, R46K, and R43K/R44K (lanes 4 -11, respectively) of adenylyl cyclase showed that mutant protein(s) is expressed just as well as the wild type enzyme. This observation rules out the possibility that the differences in the activity of various mutants of adenylyl cyclase is not a reflection of the level of the enzyme expression but is due to the mutation itself.
Expression and Activity Analysis of Other Putative Adenylyl Cyclases-The M. tuberculosis H37Rv genome sequence revealed five genes (Rv1264, Rv1318c, Rv1319c, Rv1320c, and Rv1625c) annotated as putative adenylyl cyclases (7). The Mtb cya gene we cloned is identical to Rv1625c. Genes Rv1318c, Rv1319c, and Rv1320c have about 70% identity among themselves, whereas Rv1264 and Rv1625c have about 20% identity to this group of genes and to each other. We cloned Rv1320c and Rv1264 to represent all putative cya genes, expressed in E. coli (Fig. 3) and tested for adenylyl cyclase activity (Table V). The Rv1320c gene product (62 kDa) represented about 5% of the E. coli protein (lane 3) but had no adenylyl cyclase activity under the standard assay conditions. Although the Rv1264 gene product (42 kDa) was expressed to about 10% of E. coli protein (lane 2), it was only 0.5% as efficient as Rv1625c (Mtb cya) in synthesizing cAMP. Phenotypic analysis of Rv1264 to ferment lactose/maltose was positive, and that of Rv1320c was negative (Table I), consistent with adenylyl cyclase activity profile (Table V).
Calmodulin, Forskolin, and Mammalian G s ␣ Have Marginal Effects on Mtb AC-It is well established that cAMP synthesis by the chimera of the cytoplasmic catalytic domains of eukaryotic adenylyl cyclase is activated by G s ␣-GTP␥S (50-fold), forskolin (150-fold) and the combination of G s ␣-GTP␥S and fors-

TABLE IV Adenylyl cyclase activity in the site-directed mutants
French press extracts were prepared from 3-h heat-induced cells as described under "Experimental Procedures." Adenylyl cyclase assays were performed with 25-30 g of crude extract protein of the wild type and mutants with Mg-ATP (20 mM Mg 2ϩ /5 mM ATP) and with Mn-ATP (2 mM Mn 2ϩ /1 mM ATP). Assays were carried out for 6 and 12 min at 30°C. The results are the average of two experiments with Mg-ATP and three experiments with Mn-ATP. kolin with a synergistic effect (600-fold) (21,22). Because the Mtb AC shows extensive homology with the catalytic core in the cytoplasmic domains of eukaryotic adenylyl cyclase (Fig. 5), we tested the effect of Gs␣-GTP␥S, forskolin, and calmodulin on the membranous Mtb AC. Only a modest stimulation (25-50%) of the Mtb AC by these effectors alone or in combination was observed (data not shown). However, it is interesting to note that the basal activity of Mtb AC expressed in E. coli (2.0 nmol of cAMP/min/mg, Table I) is as high as the G s ␣-GTP␥S and forskolin stimulated activity of eukaryotic chimeric adenylyl cyclase also expressed in E. coli. (1.1 nmol of cAMP/min/mg) (21). The crystal structure of the eukaryotic adenylyl cyclase in complex with forskolin and Gs␣-GTP␥s identified the essential amino acid residues involved in the activation of the enzyme by these effectors (23,24). Based on this structural information and sequence alignment (Fig. 5), it is clear that two critical residues for activation by forskolin, Thr 426 and Ser 927 in the eukaryotic type I adenylyl cyclase CI and CII domains, respectively, are replaced by Asn 372 and Asp 300 in the Mtb AC. Although the other predicted contacts, Phe 254 , Trp 367 , and Val 371 , are conserved in Mtb AC, the absence of Thr at position 372 and Ser at position 300 in the Mtb AC might explain the diminished activation of this enzyme by forskolin.

DISCUSSION
Analysis of Putative cya Genes-We report here the cloning of adenylyl cyclase gene(s) (cya) from M. tuberculosis H37Rv and biochemical characterization of the enzyme. Cole et al. (7) deciphered the nucleotide sequence of the M. tuberculosis genome and suggested five gene products as putative adenylyl cyclases based on 30 -40% homology within 200 amino acids of the catalytic core of eukaryotic adenylyl cyclase. These putative cya genes are annotated as Rv1625c, Rv1264, Rv1318c, Rv1319c, and Rv1320c. Gene products of three of these five, viz. Rv1318c, Rv1319c, and Rv1320c, are a highly homologous group with 70% identical amino acids. Gene products of Rv1625c and Rv1264 have only about 20% identity among themselves and with the gene cluster of Rv1318c, Rv1319c, and Rv1320c. In our attempt to clone the Mtb cya gene by complementation of the catabolic defect in the E. coli cya deletion strain, in work done before the genome sequence was published (25), we obtained eight clones of the same gene, Rv1625c (Mtb cya), but none of the other (GenBank TM accession no. AF017731). Hence, we cloned Rv1264 and Rv1320c (to represent Rv1318c and Rv1319c) by PCR to evaluate the function of the encoded proteins. The protein encoded by Rv1625c is undoubtedly adenylyl cyclase as observed here. The Rv1264 gene product has about 0.5% activity compared with Rv1625c, although the protein encoded by Rv1264 was expressed better than Rv1625c (Fig. 3). The protein encoded by Rv1320c did not function as adenylyl cyclase under the conditions tested. It is not surprising then that cloning by complementation assay produced only Rv1625c from the chromosomal library but none of the other genes. From this analysis, we can corroborate the annotation of Rv1264 as adenylyl cyclase but not Rv1320c or its homologues Rv1318c and Rv1319c. However, we cannot discount the possibility that Rv1318c, Rv1319c, and Rv1320c gene products may require some unidentified cofactor and function as adenylyl cyclases in their native environment.
Homology with Eukaryotic Adenylyl Cyclases-The lack of similarity of Mtb AC to other prokaryotic adenylyl cyclases but its conserved nature with respect to eukaryotic adenylyl cyclases (see below) is noteworthy. The Mtb AC contains an aminoterminal cytoplasmic tail (46 aa) followed by a single six helical transmembrane domain (152 aa) and a cytoplasmic catalytic domain (245 aa) and hence belongs to a new class of adenylyl cyclases; such an organization was found in Stigmatella aurantiaca, a Gram-negative myxobacterium (26). All of the higher eukaryotic adenylyl cyclases thus far characterized, with the one exception being soluble adenylyl cyclase (27), contain the same secondary structural organization but duplicated. Consequently, the monomeric molecular weight of the Mtb AC is about half that of eukaryotic adenylyl cyclases. The duplication of this class of Mtb AC sequence in the higher eukaryotic adenylyl cyclases is intriguing from the evolutionary point of view. In this context, we have cloned the cya gene from Mycobacterium smegmatis, a nonpathogenic species of mycobacte-   (Fig. 5). The similarity with conserved substitutions was 43% in the CI domain and the 54% in CII domain.
Putative Roles of Arg 43 and Arg 44 in Adenylyl Cyclase Activity-The amino-terminal cytoplasmic tail (46 aa) of Mtb AC contains eight arginine residues that contribute to the net positive charge of this region of the protein and satisfy the positive inside rule in the membrane protein topology (20). Because it is also known that arginine residues bind ATP, in this case the substrate for AC, site-directed mutants of these arginine residues were created to determine whether they have any role in AC activity. Site-directed mutagenesis to alanine or glycine clearly showed an important role for Arg 43 and Arg 44 in the activity of AC, whereas Arg 4 , Arg 18 , Arg 19 , Arg 27 , Arg 31 , and Arg 46 had no apparent effect. Mutagenesis of Arg 43 and Arg 44 to conserved substitution to lysine restored the enzyme activity although to a level lower than the wild type enzyme activity, which perhaps reflects the importance of the longer side chain of arginine. Further support for the role of Arg 43 and Arg 44 in the activity of AC comes from the observation that all eight of the carboxyl-terminal Mtb cya clones in pBR322, obtained by complementation of the catabolic defect, with start points in the second and third helices ( Fig. 2; Ile 79 , Ile 89 , or Ile 113 fused to Trp 97 of tetracycline resistance protein) but devoid of the arginine-rich amino-terminal cytoplasmic tail, also had no appreciable activity. We propose that Arg 43 and Arg 44 vicinal residues bind ATP and deliver the substrate to the 26-kDa cytoplasmic domain for efficient catalysis. In higher eukaryotic adenylyl cyclases, the catalytic CII domain exhibits low adenylyl cyclase activity, but the interaction of the CII domain with the CI domain enhances the activity of the enzyme (21,22). A similar analogy may be drawn with the Mtb AC wherein the arginine-rich cytoplasmic tail may contribute to overall catalysis by the 26-kDa cytoplasmic domain.
We considered an alternative explanation of why Arg 43 and Arg 44 residues are important for activity. Although Arg 46 would be the residue that contributes to the "snorkel effect" (28), the role of Arg 43 and Arg 44 is elusive in membrane topology. The putative location of Arg 43 and Arg 44 near the membrane interface is intriguing in that these residues may serve as the topological determinants (20) to anchor AC to the membrane and contribute to the structural topology and integrity of the enzyme and thereby to full activity. Mutagenesis of these basic amino acid residues to hydrophobic residues like alanine and glycine might perturb the anchoring process, resulting in a low activity form of the enzyme. Whether or not R43A, R44G, and R43A/R44G mutations distort the membrane helical topology, the mutations clearly affected the enzyme activity. We have shown in Table II that 35% of the activity of the wild type AC is in the 100,000 ϫ g soluble fraction, and 65% of the activity is associated with the membranes. Contrary to this distribution of enzyme activity, greater than 99% of AC protein is found in the soluble fraction as judged by SDS-polyacrylamide gel electrophoresis Coomassie stain, whereas membranous AC protein could not be detected by this technique (data not shown). In this context it should be noted that the protein expression profiles for the wild type (Fig. 4, lane 3)  Given these data, even if a mutant AC failed to integrate into the membrane with proper topology, the mutant AC(s) should have had all of the activity associated with the soluble fraction if the mutation had no effect on the enzyme activity. The double mutant R43A/R44G and the mutant wherein all of the eight arginine residues are mutated (8R 3 G/A) are nearly inactive (Table V). Point mutation effects of R43A and R44G are smaller, and the double mutation effect is more than additive. Hence, it appears that Arg 43 and Arg 44 are involved in facilitating the activity of Mtb AC; a hypothetical model representing an interaction between these arginine residues and the carboxyl terminus is presented in Fig. 2.
Role of Adenylyl Cyclase in Bacterial Pathogenesis-Adenylyl cyclase has been shown to be an important virulent factor in bacterial pathogens like B. pertussis and B. anthrax (2). The soluble secreted form of adenylyl cyclase from these two species invades host cells and is activated by the host calmodulin, thereby depleting ATP pool and elevating cAMP concentration in the host cell. This unregulated conversion of ATP to cAMP compromises the bactericidal activity of the host immune system (5,6). We have suggested that Mtb AC is a six helical transmembrane protein and have shown that the active form of the enzyme is membrane-bound. Hence, the enzyme cannot be secreted into the host cell to alter the physiological concentration of ATP/cAMP. In any event, calmodulin has no effect on Mtb AC. There may exist alternative mechanisms for creating the ATP/cAMP imbalance in macrophages where mycobacteria reside. The physiological significance of the activation of Mtb AC by Mn 2ϩ cannot be ignored. The genomic sequence indicates that a homologue of the macrophage Nramp Mn 2ϩ transporter protein (29,30) is present in M. tuberculosis H37Rv (Rv0924c) (7). It is known that macrophages can import Mn 2ϩ and thereby provide a rich Mn 2ϩ environment for the potential activation of resident Mtb AC. The elevated cAMP thus obtained can be secreted into the macrophages, which would inhibit several phagocyte-associated processes including lysosomal fusion (31), and mycobacteria can propagate within the host macrophages.
Padh and Venkitasubramanian (32,33) have documented the presence of cAMP in mycobacteria. However, intracellular cAMP-mediated functions via cAMP receptor protein were not observed (34). The M. tuberculosis H37Rv genome sequence revealed eukaryotic-like protein kinases (7,35), which may be involved in the intracellular cAMP-mediated signaling pathways. An important extracellular function for cAMP from M. tuberculosis H37Rv may be the prevention of fusion between phagosomes and lysosomes. This working model needs to be tested experimentally to assign a role for adenylyl cyclase in the pathogenesis of M. tuberculosis.