Identification of the cpdA gene encoding cyclic 3',5'-adenosine monophosphate phosphodiesterase in Escherichia coli.

We have identified a gene, cpdA, located at 66.2 min of the chromosome of Escherichia coli that encodes cyclic 3′,5′-adenosine monophosphate phosphodiesterase (cAMP phosphodiesterase, EC3.1.4.17). The expression of β-galactosidase, which is a product of the lacZ gene, was repressed in cells that harbored multiple copies of the plasmid carrying the cpdA gene. Northern blotting showed that the transcription of the lacZ gene was inhibited in these cells. Multiple copies of the cpdA gene decreased the intracellular concentration of cAMP, which is a positive regulator for transcription of the lacZ gene. We found that the purified CpdA protein repressed in vitro transcription from the lacP1 promoter by decreasing cAMP. In addition, we showed that the CpdA protein hydrolyzed cAMP to 5′-adenosine monophosphate and that its activity was activated by iron. Our results suggested that regulation of intracellular concentration of cAMP is dependent not only on synthesis of cAMP but also on hydrolysis of cAMP by cAMP phosphodiesterase.

cAMP is an important cellular mediator in Escherichia coli. The role of cAMP in mediating glucose effects has been well investigated (1). The cAMP receptor protein (CRP) 1 is a regulatory protein, which binds cAMP and mediates transcriptional regulation at several promoters (2). The CRP-cAMP complex is a positive transcriptional regulator of a number of catabolic operons, including the lac operon in E. coli, and as such, plays a role in catabolite repression, whereby secondary carbon sources are not catabolized in the presence of glucose (3,4). This complex is involved not only in positive regulation of several catabolic functions but also in regulation of flagellum synthesis (5), toxin production (6), minicell production (7), coupling of DNA replication and cell division (8), and many other functions that are not directly related to catabolism. CRP forms an active conformation only when it binds cAMP. Therefore, the concentration of active CRP-cAMP complex and the biological responses mediated by active CRP-cAMP are influenced by the intracellular concentration of cAMP.
Another role of cAMP, which is independent of CRP transcription, has been reported. cAMP interacts directly with the DnaA protein and plays a role in the re-activation of DnaA, which is an essential element for initiation of DNA replication from the chromosome origin, oriC (9). Thus the cellular level of cAMP has some effects on controlling the initiation of DNA replication. Moreover, cAMP has a role in regulation of cell division in E. coli. Filamentation of growing cells is induced by elevated levels of cAMP (10). cAMP may affect cell division only indirectly through unidentified cAMP-dependent functions that are not obligatory (11).
In E. coli, the intracellular concentration of cAMP has been thought to be mainly controlled by its own synthesis. Synthesis of cAMP is catalyzed by adenylate cyclase, encoded by the cya gene, and its activity is regulated transcriptionally (12,13) and post-translationally (14,15). However, the existence of phosphodiesterase (EC 3.1.4.17), which hydrolyzes cAMP, had been reported in E. coli (16). Nielsen et al. (17) partially purified the enzyme. The gene encoding cAMP phosphodiesterase remained unidentified.
In the present study we have discovered the E. coli gene encoding cAMP phosphodiesterase, and we have purified this enzyme. We present here the nucleotide sequence of the gene (named cpdA) and properties of the enzyme.
Construction . 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) D16557.
DNA Sequencing-We determined nucleotide sequence of the BamHl-PstI chromosome segment between the tolC gene and the parE gene, according to the dideoxy method of Sanger et al. (25). The nucleotide sequences of the BamHI-PvuI and BglII-PstI chromosome segments already had been published (26,27) and deposited in the EMBL/ GenBank/DDBJ sequence data libraries under the accession numbers M37833 and X54049, respectively.
Assay for the ␤-Galactosidase Activity-Bacterial strains were grown in L medium with or without 1 mM IPTG. After the cell density reached 60 Klett units, the activity of ␤-galactosidase was measured according to Miller (19). The mean values of enzyme activity and their standard deviations were calculated from four independent experiments.
Northern Blot Analysis-Total RNA was extracted from exponentially growing cells according to the method of Aiba et al. (28). The RNA was diluted to various concentrations, applied to a nitrocellulose filter, and hybridized with a 32 P-labeled lacZ DNA probe as described previously (29). The 161-base pair HaeII-EcoRI fragment of pUC19 was labeled with [␣-32 P]dCTP using random oligonucleotide primers and used as the lacZ DNA probe. The radioactivity of the hybridized probe DNA was measured by using a Bioimage analyzer model BAS2000 (Fuji Photo Film Co., Tokyo, Japan).
Specific Labeling of Plasmid-encoding Proteins by Maxicell Method-To detect plasmid-encoded proteins, UV-irradiated cells of CSR603 carrying a plasmid were prepared and labeled with [ 35 S]methionine according to Niki et al. (27).
Purification of the CpdA Protein-We cloned the BamHI-VspI segment from pAX908 (see Fig. 1) into the pT7-5 plasmid, which has cloning sites downstream the T7 RNA polymerase promoter (30). The resulting plasmid was named pAX943. For overproduction of the CpdA protein, pAX943 was introduced into the BL21(DE3) strain, which carries the gene for T7 RNA polymerase under control of the lacUV5 promoter. Strain BL21(DE3) harboring pAX943 was grown at 37°C in 1.5 liters of L medium containing ampicillin (25 g/ml). After the cell density reached 30 Klett units, IPTG was added at a final concentration of 0.5 mM. Exponentially growing cells were collected at 4°C by centrifugation. Harvested cells were washed with 25 mM ice-cold Hepes-KOH buffer, pH 7.6, and suspended in 50 ml of buffer A (25 mM Hepes-KOH, pH 7.6, 25 mM KCl, 0.1 mM EDTA, 2 mM dithiothreitol, and 20% glycerol) containing 0.2 mg/ml of lysozyme, 20 mM spermidine, and 0.5 mM phenylmethylsulfonyl fluoride. The cell suspension was kept on ice for 30 min and then sonicated at 20-s intervals for 10 min. The cell lysate was centrifuged at 27,000 ϫ g for 60 min to remove unbroken cells. The cleared cell lysate was centrifuged at 150,000 ϫ g for 2 h. Ammonium sulfate was added to the supernatant to a final concentration of 35% (w/v) and stirred at 0°C for 30 min. After centrifugation at 12,000 ϫ g for 30 min, ammonium sulfate was added to the supernatant to a final concentration of 55% (w/v). The supernatant was stirred at 0°C for 30 min and then centrifuged. The pellet was resuspended in 2 ml of buffer A to yield the AS fraction. 100 l of the AS fraction was loaded onto a column of FPLC Superdex 75 HR 10/30 (Pharmacia Biotech Inc.) that had been equilibrated with buffer A. Samples (0.4 ml) were collected into tubes, and proteins in each tube were analyzed by SDS-polyacrylamide gel electrophoresis. Fractions containing CpdA protein were pooled and loaded onto a column of FPLC Mono S HR 5/5 that had been equilibrated with buffer A. Fractions containing CpdA protein were collected and dialyzed against buffer B (25 mM Tris-HCl, pH 7.6, 25 mM KCl, 0.1 mM EDTA, 2 mM dithiothreitol, and 20% glycerol) and then loaded onto a column of FPLC Mono Q HR 5/5 that had been equilibrated with buffer B. The purified CpdA protein was dialyzed against buffer A.
Determination of N-terminal Sequence of CpdA-The purified CpdA protein was electroblotted onto a polyvinylidene difluoride membrane filter, ProBlott (Applied Biosystems Inc.). The N-terminal amino acid sequence of the protein was determined with a protein sequencer, Applied Biosystems 477A, according to Imamura et al. (31).
Step 1 was preincubated for 10 min at 37°C, and then components of step 2 were added and incubated for an additional 10 min at 37°C. After the addition of components of step 3, incubation was continued further for 30 min at 37°C to allow the formation of open complex. The final reaction volume after these three steps was 35 l, to which a substrate solution (15 l) containing 4 mM each of nucleoside triphosphates (ATP, CTP, GTP, and [␣-32 P]UTP (2 Ci)) and heparin (final concentration, 200 g/ml) was added to allow a single round of transcription for 5 min (step 4). The reaction was stopped, and the labeled transcripts were analyzed by electrophoresis on 8% polyacrylamide gel containing 8 M urea (32). The following templates were used (33,34): the 205-base pair PvuII-XbaI fragment of plasmid pUC19 carrying the wild-type lac promoters (lacP1 and lacP2) and the 512-base pair Hin-dIII fragment of plasmid pWT101 carrying the wild-type trp promoter.
Determination of cAMP Concentration-cAMP was assayed using the cAMP enzyme immunoassay kit (Cayman Chemical Company, Ann Arbor, MI) according to the method specified by the manufacturer. We incubated CpdA and cAMP in 50 mM Tris-HCl, pH 7, and 0.1 mM dithiothreitol at 37°C for 1 h. After incubation, samples were boiled for 5 min and centrifuged. The supernatants were diluted properly and assayed concentration of cAMP. To determine the intracellular cAMP concentration, culture samples equivalent to 1.0 A 600 were taken, boiled for 5 min, and centrifuged at 1200 ϫ g for 5 min at 4°C. The supernatants were used for this assay.

Identification of the Novel Inhibitor of Expression of ␤-Ga-
lactosidase Independently of LacI Repressor-We sequenced the 66.2 min region of the E. coli chromosome located between the parE and the tolC (mukA) genes and found that there were four open reading frames (Fig. 1). The wild-type E. coli cells (W3110) harboring plasmid pAX903, which has the BamHI-PstI segment (3229.50 -3232.50-kb coordinate in the physical map), formed pale blue colonies on peptone-agar plates containing X-gal and the inducer IPTG of the lac operon. In contrast, the wild-type cells harboring the vector plasmid pACYC184 formed dark blue colonies. This suggested that when present at high doses, one product or a combination of products from three open reading frames in this segment inhibited the synthesis of ␤-galactosidase. We subcloned the segment and found that plasmids pAX908 and pAX923, both carrying ORF3, led to the formation of pale blue colonies, but pAX904, pAX906, and the vector pACYCl84, none of which carry ORF3, led to the formation of dark blue colonies (Fig. 1). We then assayed the activity of ␤-galactosidase in liquid culture of the wild-type strain transformed with the pAX923 plasmid and the pACYC184 vector plasmid. Cells harboring pAX923 showed only about 20% of the activity of cells harboring pACYC184 ( Fig. 2A, a). The results suggest that the ORF3 protein is an inhibitor of ␤-galactosidase synthesis.
To analyze the inhibitory mechanism of ␤-galactosidase synthesis, we measured the amount of lacZ mRNA in the presence and the absence of plasmid pAX923. The induced level of lacZ mRNA in the pAX923-harboring strain was 26% of that in the control strain (Fig. 2B). Thus we concluded that high levels of the ORF3 protein inhibited transcription of the lac operon.
The inhibitory effect of multiple copies of the ORF3 protein on expression of ␤-galactosidase was also observed in a ⌬(lacIZYA) deletion strain harboring the lacZ ϩ plasmid (pXX747) ( Fig. 2A, b). This indicated that the inhibitory effect was independent of the LacI repressor.
Purification of the Inhibitor Protein-To identify the ORF3 protein responsible for the inhibitory action of ␤-galactosidase synthesis, we labeled plasmid-encoded proteins with [ 35 S]methionine in a maxicell system and identified an apparently 30-kDa protein as the ORF3 gene product (Fig. 3A). The molecular mass of this protein was deduced from the DNA sequence to be 30,937 Da (Fig. 4). We purified the ORF3 protein from cell extracts of an overproducing strain (Fig. 3B). The purified protein was analyzed for the amino acid sequence from the N terminus. The resulting amino acid sequence was MESLLTLPXAGEARVRIL (X represents an unidentified amino acid residue), which agreed with the deduced amino acid sequence from the nucleotide sequence (Fig. 4). The result also indicated that a TTG codon for Leu was used as the initiation codon for the first Met. At least 31 genes of E. coli have been described as using the TTG start codon (35). A homologous polypeptide was found in the genome data base of Haemophilus influenzae Rd, and the similarity of the ORF3 protein with the predicted polypeptide, HI0399, was 71% (36). We called the gene encoding the ORF3 protein cpdA.
The Inhibitor Represses lacZ Transcription in Vitro through Decreasing the Amount of cAMP-We tested possible effects of the purified CpdA protein on lacZ transcription using an in vitro transcription system. For this purpose, we used a DNA fragment carrying the wild-type lacP1 and lacP2 promoters as template. In the presence of both cAMP and CRP, a transcript from the lacP1 promoter was observed (Fig. 5, lane 2), whereas a transcript from the lacP2 promoter was observed in the absence of cAMP and CRP (Fig. 5, lane 1). Because the lacP2 promoter is a weak promoter in vivo, the lacZ transcription in vivo is negligible in the absence of cAMP or CRP (37). When 50 pmol of the CpdA protein was preincubated with cAMP and CRP before the addition of template DNA and RNA polymer-ase, transcription from the lacP1 promoter was partially inhibited, and instead a small amount of the lacP2 transcript was detected (Fig. 5, lane 4). In the absence of CpdA protein but with cAMP and CRP added before the addition of template DNA and RNA polymerase, transcription from the lacP1 promoter was observed, and the lacP2 transcript was not detected (Fig. 5, lane 3). When 100 pmol of the CpdA protein was preincubated with cAMP and CRP, transcription from the lacP1 promoter was completely repressed (Fig. 5, lane 5), and the level of the lacP2 transcript was as high as that in the absence of cAMP and CRP (compare Fig. 5, lanes 1 and 5). The CpdA protein showed no inhibitory effect on the transcription from the mutant lacP1 promoter of lacUV5 DNA, which did not require CRP-cAMP for transcription (data not shown). The CpdA protein had no effect on transcription from the trp promoter used as an internal control (compare Fig. 5, lanes 3 and  5). Furthermore, the same extent of inhibition of transcription from lacP1 promoter was observed when cAMP alone was incubated with the CpdA protein (100 pmol) before the addition of CRP (Fig. 5, lane 6). These results suggest that the CpdA protein decreases the amount of cAMP in this assay solution. In fact, this inhibitory effect was not found when a 10-fold excess of cAMP was added (Fig. 5, lane 7).
We also examined the effect of the CpdA protein on transcription from the gal promoter, which consists of CRP-depend-  2. Effect of the cpdA gene on expression of ␤-galactosidase and the amount of lacZ mRNA. A, the multicopy cpdA ϩ plasmid (pAX923) or its vector plasmid (pACYC184) was introduced by transformation into the following E. coli strains: the wild-type strain W3110 and CSH50 harboring the lacZ ϩ plasmid (pXX747). The activity of ␤-galactosidase was measured in the presence or the absence of IPTG. B, the radioactivity of a lacZ DNA probe that hybridized to the total RNA extracted from the wild-type strain W3110 harboring pACYC184 or pAX923 was measured as described under "Experimental Procedures." Photo-stimulated luminescence (PSL) is a unit for radioactivity that has been established by the Fuji Photo Film Company and is directly proportional to cpm units (41). ent galP1 and CRP-independent galP2 promoter. Preincubation of the CpdA protein with cAMP and CRP caused repression of transcription from the galP1 promoter and produced instead a transcript from the galP2 promoter (data not shown).
Identification of the Inhibitor as cAMP Phosphodiesterase-We measured the concentration of cAMP after the incubation with the CpdA protein using the cAMP enzyme immunoassy system. This assay is based on competition between free cAMP and cAMP tracer linked to acetylcholinesterase molecules for a limited number of cAMP-specific rabbit antiserum binding sites. The amount of cAMP tracer that is able to bind to the rabbit antiserum will be inversely proportional to the concentration of free cAMP in the sample and will be measured by the activity of acetylcholinesterase.
The cAMP concentration was found to decrease markedly. As shown in Fig. 6A, the purified CpdA protein decreased cAMP concentration in a time-dependent manner. We found that FeCl 2 stimulates the enzymatic activity of CpdA protein. The addition of 100-fold cGMP did not inhibit the reaction of the CpdA with cAMP (Fig. 6A). Calculation of the data presented in Fig. 6B shows that the CpdA protein of E. coli has a K m of approximately 0.5 mM cAMP and a V max of 2.0 mol/min/mg.
To confirm that the reaction product is 5Ј-AMP, 5Ј-nucleotidase was added to the reaction mixture after incubation with cAMP and the CpdA. 5Ј-Nucleotidase hydrolyzes 5Ј-AMP to adenosine. The mixture was further incubated for 2 h and was separated (Fig. 7, lane 9). The mobility of the product was not the same as that of 5Ј-AMP (Fig. 7, lane 7). It showed the same mobility as adenosine (Fig. 7, lane 10). Thus, 5Ј-nucleotidase hydrolyzed the reaction products to adenosine. We therefore concluded that the CpdA protein is a cAMP phosphodiesterase (EC 3.1.4.17) of E. coli that hydrolyzes cAMP to 5Ј-AMP, and we propose to designate this gene as cpdA (cyclic AMP phosphodiesterase).
Effect of cpdA Gene Dosage on the Intracellular Concentration of cAMP-Based on the results of in vitro analysis, we believed that the inhibitory effect by multiple copies of the cpdA gene on expression of ␤-galactosidase was caused by decreasing the intracellular concentration of cAMP. To test this possibility, we measured the intracellular concentration of cAMP in the wild-type strain harboring the pAX923 plasmid or the pACYC184 vector plasmid. The intracellular concentration of cAMP in the pAX923-harboring strain was only 11% of the level of that in the pACYC184-harboring control strain (Fig. 8). In addition, the concentration of cAMP in the cpdA-disrupted strain (SH8150), was 2-fold higher than the parental cpdA ϩ strain (W3110). These results suggest that the cpdA gene product also participates in decreasing the intracellular concentration of cAMP in vivo.

DISCUSSION
The results described in this paper from in vitro and in vivo experiments showed that the gene product that inhibited the expression of ␤-galactosidase was cyclic 3Ј,5Ј-adenosine monophosphate phosphodiesterase (EC 3.1.4.17), which hydrolyzes cAMP to 5Ј-AMP. We refer this gene as cpdA, which is located at 66.2 min of the E. coli chromosome. There are different types of cyclic 3Ј,5Ј-nucleotide phosphodiesterases; some enzymes hydrolyze cGMP as well as cAMP, whereas other enzymes prefer cGMP to cAMP as a substrate. This phosphodiesterase that we purified, CpdA, is specific for decomposition of cAMP, and its activity was not affected by the addition of cGMP. Its enzymatic parameters are as follows: V max is 2.0 mol/min/mg, and K m is 0.5 mM. The occurrence of cAMP phosphodiesterase in E. coli was first described by Brana and Chytil (16). Partial purification of cAMP phosphodiesterase was performed by Nielsen et al. (17), and they purified a cAMP phosphodiesterase from a crude extract of E. coli by about 100-fold. This enzyme has a molecular weight of about 30,000 and a Michaelis constant of 0.5 mM cAMP and is activated by iron. These properties of the cAMP phosphodiesterase, as purified by Nielsen et al. (17), are consistent with our present results with the purified CpdA protein.
The cAMP phosphodiesterase activity detected by Nielsen et al. (17) is most likely the activity of the CpdA protein.
Current information relating to the phosphodiesterases in prokaryotic cells is not sufficient for domain analysis. However phosphodiesterases in eukaryotic cells are well characterized and have been divided into two classes, I and II (38) Fig. 1) is shown. The amino acid sequence derived from purified CpdA protein is underlined. The boundaries of cloning region of pXX904 and pXX906 are indicated by arrows above the nucleotide sequence. The nucleotide sequence data have been deposited in EMBL/GenBank/DDBJ sequence data libraries under the accession number D16557. gene (3). We confirmed that the CpdA protein was cAMP phosphodiesterase, which hydrolyzed cAMP. Cells harboring the cpdA multicopy plasmid have a higher concentration of the CpdA protein by increased gene dosage, which increases the expression of cAMP phosphodiesterase. We measured the intracellular concentration of cAMP of the wild-type strain harboring the multiple copies of the plasmid carrying the cpdA gene and found that the presence of this plasmid markedly decreased intracellular cAMP. In turn, the intracellular concentration of cAMP was decreased, the formation of the CRP-cAMP complex was reduced, and expression of the lacZ gene was repressed.
Decreasing intracellular concentration of cAMP by multiple copies of the cpdA gene would inhibit expression of other CRPdependent genes besides the lac operon. We observed that in an in vitro transcription system, expression of the gal promoter, which was dependent on CRP-cAMP, was repressed by the CpdA protein. Using MacConkey plates, which indicate the utilization of a carbon source in the medium, we observed that cells harboring multiple copies of the cpdA gene were partially inhibited for the utilization of galactose and maltose as well as lactose (data not shown). Besides expression of CRP-dependent genes, variation of intracellular concentration of cAMP by the cpdA gene would be expected to influence cell division and the initiation of DNA replication.
It is known that adenylate cyclase plays an important regulatory role in the intracellular level of cAMP (1). The activity of adenylate cyclase, which synthesizes cAMP from ATP, is regulated transcriptionally and translationally. This enzyme is encoded by the cya gene, and expression of the gene is negatively controlled by the CRP-cAMP complex (12,13). Translation of E. coli adenylate cyclase is initiated at a UUG codon, Steps 1, 2, and 3 are preincubation steps, whereas at step 4, RNA synthesis is allowed. A fixed amount of the wild-type trp DNA was mixed as a reference template. Products from these templates are shown by arrows. Almost equal amounts of the trp transcript as an internal control can be seen in all lanes. RPase represents RNA polymerase. NTP represents the mixture of four ribonucleoside triphosphates. and it was reported that the cya UUG codon limited cya expression at the level of translation (40). In addition to such gene regulation, the activity of adenylate cyclase is post-translationally regulated by phosphorylation via a phosphoenolpyruvate-dependent sugar phosphotransferase system (14,15). Although the regulation of cAMP phosphodiesterase (CpdA) that hydrolyzes cAMP remains unclear, we found that translation of cAMP phosphodiesterase also is initiated at a UUG codon, as is adenylate cyclase. Thus, expression of the cpdA gene might be also regulated translationally.
The intracellular concentration of cAMP is very critical for regulation of various cellular systems; it influences expression of a great variety of genes by forming active CRP-cAMP complexes. Indeed, an excess of CRP-cAMP disturbed cell proliferation. The cpdA-disrupted strain showed 2-fold higher intracellular cAMP than the control strain but grew normally, suggesting that the cpdA gene is not essential for growth. However, under the condition that cells had an excess of CRP-cAMP complex, the cpdA-disrupted cells did not form colonies. Cells harboring multiple copies of crp, which encoded CRP protein, did form colonies on agar plates containing 3 mM cAMP, but the cells were filamentous. On the other hand, the cpdA-disrupted cells harboring multiple copies of the crp gene were not able to grow on agar plates containing 3 mM cAMP. 2 We also observed that the level of expression of ␤-galactosidase was higher in the cpdA-disrupted cells compared with the wild-type cell in the medium containing glucose. 2 The shut-off of expression of ␤-galactosidase by the addition of glucose as carbon source, the so-called glucose effect, was not severe in the cpdA-disrupted strain because the intracellular concentration of cAMP was twice as much as that of the isogenic cpdA ϩ strain (see Fig. 8). Taken together, these results indicate that the cAMP phosphodiesterase plays an important regulatory role in determining intracellular of cAMP content.