Logo of jbacterPermissionsJournals.ASM.orgJournalJB ArticleJournal InfoAuthorsReviewers
J Bacteriol. 2003 Nov; 185(21): 6415–6424.
PMCID: PMC219403

Why Is Carbonic Anhydrase Essential to Escherichia coli?


The can (previously yadF) gene of Escherichia coli encodes a β-class carbonic anhydrase (CA), an enzyme which interconverts CO2 and bicarbonate.Various essential metabolic processes require either CO2 or bicarbonate and, although carbon dioxide and bicarbonate spontaneously equilibrate in solution, the low concentration of CO2 in air and its rapid diffusion from the cell mean that insufficient bicarbonate is spontaneously made in vivo to meet metabolic and biosynthetic needs. We calculate that demand for bicarbonate is 103- to 104-fold greater than would be provided by uncatalyzed intracellular hydration and that enzymatic conversion of CO2 to bicarbonate is therefore necessary for growth. We find that can expression is ordinarily required for growth in air. It is dispensable if the atmospheric partial pressure of CO2 is high or during anaerobic growth in a closed vessel at low pH, where copious CO2 is generated endogenously. CynT, the single E. coli Can paralog, can, when induced with azide, replace Can; also, the γ-CA from Methanosarcina thermophila can at least partially replace it. Expression studies showed that can transcription does not appear to respond to carbon dioxide concentration or to be autoregulated. However, can expression is influenced by growth rate and the growth cycle; it is expressed best in slow-growing cultures and at higher culture densities. Expression can vary over a 10-fold range during the growth cycle and is also elevated during starvation or heat stress.

A large proportion of the open reading frames (ORFs) located on fully sequenced chromosomes are of unknown or unverified function. To gain functional information about uncharacterized Escherichia coli K-12 ORFs, we have been replacing selected ORFs with reporter cassettes (23). One of our targets, yadF, has been demonstrated to be a carbonic anhydrase (CA) of the β-class. Its X-ray crystal structure has been solved, and its biochemical properties have been characterized (10). Our initial attempts to delete yadF were unsuccessful, suggesting that it might encode an essential gene product. Although CA has been reported to be involved in cell processes such as photosynthesis, respiration, and CO2 transport in other organisms and in cyanate metabolism in Escherichia coli (see reference 33 for a recent review), none of these processes is expected to be necessary for the viability of E. coli.

CAs are zinc metalloenzymes that catalyze the interconversion of carbon dioxide (CO2) and bicarbonate anion (HCO3). Why should this activity be needed? The well-understood function of CynT, the only E. coli paralog of YadF, suggests a reason. CynT is a normally repressed CA that is induced during cyanate metabolism. Its role is to prevent the depletion of intracellular bicarbonate which accompanies the cyanase-catalyzed bicarbonate-dependent hydrolysis of cyanate (16). CynT rehydrates the CO2 which is produced and thus prevents its loss by rapid diffusion from the cell. Bicarbonate can be regarded as a coenzyme of cyanase (CynS), and CynT provides for its regeneration.

E. coli requires a supply of bicarbonate/CO2 as a metabolic substrate during normal growth. It is needed not only for biosynthesis of various small molecules but also for fatty acid biosynthesis and in central metabolism. The small molecules, which include arginine, pyrimidines, and purines, can be provided as supplements and, interestingly, certain mutants limited in the production of these nutrients can be suppressed with increased levels of CO2 (8). However, the need for CO2 in central metabolism cannot be replaced with supplements. The only known supply pathway for bicarbonate is via the hydration of CO2. Although it has long been known that E. coli requires CO2 to grow (7), under most conditions an adequate supply is generated endogenously. CO2 reacts spontaneously with water to produce bicarbonate, but we calculate here (see Discussion) that, during growth in minimal glucose medium in air, the demand for bicarbonate is 103- to 104-fold greater than would be provided by uncatalyzed hydration at the prevailing steady-state concentration of CO2. We therefore propose that the function of yadF, which we show here to be expressed at all times, is to meet this shortfall.

Three classes of CA (α, β, and γ) have been reported to occur in bacteria. They are unrelated in sequence and widespread among organisms of all three kingdoms of life. Indeed, E. coli YadF can correct the oxygen-sensitive phenotype of a β-CA deletion mutant of Saccharomyces cerevisiae (10, 15). In addition to the β-CAs, YadF and CynT, E. coli also encodes several gene products (CaiE, PaaY, and YrdA) which are clearly homologous to γ-group CAs, with fully conserved active-site residues. CaiE and PaaY are, like CynT, included in inducible operons that are each concerned with a particular process: CaiE in carnitine metabolism (11) and PaaY in phenylacetic acid degradation (13). Eichler et al. (11) proposed that CaiE is involved in the generation or regeneration of a coenzyme for CaiB/CaiD. The function of these two proteins has been clarified recently (12) and CaiB shown to be an acyl coenzyme A transferase. We suggest that the biochemistry of this pathway may be analogous to that of fatty acid biosynthesis and that bicarbonate may be a coenzyme for this transferase activity.

The Ralstonia eutropha homolog of YadF, Can, has recently been characterized (20) and found to be essential. Unpublished results of these authors (quoted in reference 20) show that Can is able to suppress an E. coli yadF mutant and that YadF can suppress a can mutant in R. eutropha. In view of the close homology, demonstrated activities, and likely interchangeability of the two proteins, we suggest that YadF also be assigned the name Can (and its encoding gene, can).

We both demonstrate and calculate here that CA is required for the growth of E. coli at the CO2 concentration characteristic of air. CA activity is normally supplied by the product of can, which is always expressed. Can, however, can be replaced by CynT, if induced, by the archael α-CA, Cam, or by cultivation under sufficiently elevated levels of CO2.


Strains and media.

Strains used in the present study are described in Table Table1.1. Cultures were grown at 37°C in either LB broth (Lennox) (1% tryptone [Difco], 0.5% yeast extract [Beta Lab], 1% NaCl) or VB minimal medium (37) supplemented with thiamine (10 μg/ml) and 0.2% glucose, unless otherwise stated. Kanamycin, ampicillin, and chloramphenicol were used at concentrations of 50, 100, and 20 μg/ml, respectively. Arabinose, when required for PBAD induction, was added at concentrations ranging from 0.2 to 1%. Azide (Sigma) was used at 0.1 mM for cynT induction. Cyanate (Aldrich) was added to VB medium-glucose agar at 0.5 mM to score CynT activity or functional suppression of cynT by other CAs. To obtain variations in generation time, supplements to minimal medium were used at the following concentrations: Casamino Acids, 0.5%; asparagine, 100 μg/ml; and tryptophan, adenine, uracil, or cytosine, 20 μg/ml. Anaerobic growth in jars with CO2 was achieved with the Oxoid gas generating kit AN35, which produces an atmosphere with 4 to 10% CO2 and converts O2 to water through a palladium-catalyzed reaction with H2 supplied by the kit. BBL Dry Anaerobic Indicator Strips were used to confirm that anaerobiosis had been attained. For growth under H2, air was exhausted and replaced with H2 from a cylinder; a Pd catalyst ensured that any remaining O2 was converted to H2O. For liquid cultures bubbled with mixtures of CO2, N2, and O2, gas composition and delivery were controlled from the Bio Console ADI1035 of an Applikon fermentor. Anaerobic liquid cultures were grown without a gas-phase in syringes to allow maintenance of anaerobiosis during sample collection; glass beads and gentle shaking were used to ensure a uniform distribution of cells.

Strains used in this study

Growth and expression experiments.

Overnight cultures were grown at 37°C in 5 ml of medium in 10-ml bottles or in shaking flasks filled to 10% of their volume, with appropriate selection, inducer, and supplementation. Cultures were diluted 100-fold, with centrifugation and washing when a change of medium was to follow, into the desired growth medium, and the optical density at 600 nm (OD600) was measured at intervals. To obtain continuous exponential growth, cultures were diluted fourfold into prewarmed medium when OD600 reached 0.2. Samples for β-galactosidase assay were taken at intervals into Z-buffer with CHCl3 and assayed as described by Miller (25).

Plasmid construction.

To clone paaY, can, cynT, caiE, and yrdA into pUC18 (GenBank AN L08752), primers containing NheI and XhoI sites upstream and downstream, respectively, and 24 bases homologous to the DNA flanking the gene to be cloned were synthesized and used in PCRs with Pfu DNA polymerase to amplify each of the genes from chromosomal DNA. The resulting PCR products were digested and ligated into pUC18 which had been digested with XbaI and SalI. Constructs were verified by restriction analysis and, for caiE and yrdA, by sequencing. To clone can, cynT, caiE, and yrdA into pTRC99A (GenBank AN U13872), the same downstream primers used for pUC18 cloning and an upstream primer with an engineered NcoI site overlapping the predicted start codon of the gene, followed by 24 bases of sequence from within the gene, were used in PCRs with Pfu as described above. Products were digested and cloned into pTRC99A DNA which had been digested with NcoI and SalI. Constructs were verified by restriction digestion analysis.

DNA manipulation.

Plasmid DNA purification, genomic DNA preparation, DNA cleaning, and DNA gel extraction were performed by using the Wizard Plus SV Minipreps kit (Promega), the Quantum Prep AquaPure genomic DNA isolation kit (Bio-Rad), QIAquick PCR purification kit (Qiagen), and the QIAquick gel extraction kit (Qiagen), respectively.

Crossover PCR and analytical PCR were carried out on a PCR Sprint cycler (Hybaid) with Pfu DNA polymerase and Taq DNA polymerase, respectively, according to the manufacturer's recommendations. All primers used were purchased from MWG-Biotech.

Ligations were carried out as described in (32). Prior to ligation with T4 DNA ligase (Roche), vectors were dephosphorylated with calf alkaline phosphatase (Roche), and both vector and insert DNA were purified by using a QIAQuick PCR purification kit (Qiagen).

Transformation of bacteria with engineered plasmid DNA was done by using a classical CaCl2 preparation, and recombinant DNA was recovered in DH5α (32). After the FLK2-cassette cloning step, recombinant DNA was recovered in TOP10 by using One Shot TOP10 competent cells (Invitrogen).

Gene deletion and/or replacement.

Precise yadF deletion was performed as described in detail previously (23). The primer pairs NiYadF (CGCTCTTGCGGCCGCTTGGAACGGCACCAGCATTTTTGACCATAGTGC)-NoYadF (AAAAACTGCAGTCCCAACTGACGTTTAGCGTTCAC ) and CiYadF (CGTTCCAAGCGGCCGCAAGAGCGCTCAAGCTGAAACACGCCAACCAC)-CoYadF (AAAAAGTCGACTTTATGACCGCCTCCAGC) were used to amplify ∼400-bp segments flanking the can region.


Phylogenetic distribution of can and cynT.

We analyzed the phylogenetic distribution and chromosomal neighborhoods of can and cynT bioinformatically by using the MicroBial Genome Database (http://mbgd.genome.ad.jp) and the NCBI database BLAST resources (http://www.ncbi.nlm.nih.gov:80/BLAST/). The local genetic organization surrounding can is conserved only among the sequenced Enterobacteriacae. Traces of this organization can still be observed in Vibrio cholerae but not in other sequenced bacteria in which can homologs are found. Although phylogenetic profiling analysis of sequenced genomes did not identify other genes sharing the can phylogenetic distribution, can is conserved, and homology searches suggest its vertical inheritance at least during the evolution of the γ-proteobacteria. cynT, in contrast, appears to be absent from E. coli's closest relatives and is much more sporadic in its occurrence, suggesting recent horizontal transfer, perhaps initially from plants, where close homologs are common.

Deletion of can.

The can deletion/replacement we designed spans coordinates 3334 to 3909 on section 12 of the E. coli K-12 chromosome (GenBank AN AE000122) and removes most of the coding sequence, including the region corresponding to the conserved Zn-binding residues of the postulated catalytic site (10). Gene replacement with the removable reporter cassette (FLK2: FRT-lacZ-aph-FRT) failed several time when attempted in the uncomplemented strain EDCM367 (Table (Table1),1), and we speculated that the gene might be essential. We therefore followed the procedure described by Arigoni et al. (2) to insert a controllable second copy of can at the ara locus. Strain EDCM421 (Table (Table1)1) has a copy of can, under PBAD control, in place of araBAD on the chromosome. Construction of the deletant EDCM464 then became possible, provided that the growth medium contained sufficient arabinose to maintain the expression of can; colonies of EDCM464 are not formed on LB plates lacking arabinose. We conclude that Can is essential for growth in standard aerobic laboratory conditions.

Depletion of Can from EDCM464.

In order to determine why can is essential, we attempted to deplete Can by removing arabinose from growing liquid cultures (Fig. (Fig.1).1). Surprisingly, strain EDCM464 was able to grow for a complete growth cycle in LB medium-glucose, a condition in which the complementing copy should not be expressed from PBAD. Cessation of growth was observed only when cells were diluted into the same medium and incubated further (Fig. (Fig.1).1). Repeated experiments showed that EDCM464 can divide 7 to 8 times before growth stops; we believe that this reflects a high level of expression from the arabinose promoter in the overnight cultures used as inoculum (see below) and suggests that Can itself is probably not required in large quantities. Microscopic analysis showed that Can depletion does not cause any morphological abnormalities (data not shown); Can-depleted cells simply appear to stop growing.

FIG. 1.
Can is required for continued growth in air. EDCM464 and its parent EDCM421, grown overnight in LB medium-arabinose, were inoculated into LB medium-glucose at 37°C. Cultures were allowed either to grow for one complete cycle (○) or were ...

CynT can replace Can.

cynT transcription can be gratuitously induced with a sublethal concentration of azide (16). If the cynT CA can replace Can, it should be possible to maintain a can deleted strain by addition of azide. We found that the can<>FLK2 mutation could be P1 transduced from EDCM464 to EDCM637 (MG1655 ΔlacZ; Table Table1)1) to create EDCM636, provided that transductants were selected and maintained on azide-containing plates. Thus, CynT synthesized from a single gene copy and from its own promoter can satisfy the cell's need for CA activity, and Can is not specifically required for viability.

CA was depleted from EDCM636, previously grown overnight with 0.1 mM azide, by inoculating into LB medium without azide. As shown in Fig. Fig.2,2, LB medium-grown cells of EDCM636 stop dividing 1.7 generations after azide removal, indicating rapid loss of CynT. Figure Figure22 also shows that EDCM636 cultivated in VB medium-glucose achieves only a 62% mass increase. Note that whereas EDCM464 pregrown in arabinose grows to stationary phase, growth stops earlier if EDCM464 is pregrown with azide rather than arabinose. This suggests, since cultures grown overnight with arabinose can complete a full growth cycle, that Can, expressed from PBAD, reaches a stationary-phase level manyfold higher than is required for growth. Also, since EDCM464 inoculated from azide outgrows EDCM636, it appears that the PBAD expression system is leaky, at least during stationary-phase incubation. To avoid complications caused by this, EDCM636 was used, where possible, for all subsequent experiments.

FIG. 2.
Depletion of CA from can deleted strains. (Left panel) Growth curves of EDCM636 (○) and its parent EDCM367 (□) in LB medium at 37°C. Strains were grown overnight in LB medium (with 0.1 mM azide for EDCM636) and inoculated into ...

Growth without oxygen.

When either EDCM464 or EDCM636 is grown on LB plates in anaerobic jars under H2 and CO2 (see Materials and Methods), colonies indistinguishable from wild-type are formed; thus, Can appears to be dispensable for growth under these conditions. Is this because of the increased CO2 partial pressure (which also increases the rate of supply of HCO3) or because of the lack of O2? Oxygen sensitivity has been reported as a phenotype associated with deletion of the can homolog NCE103 from Saccharomyces cerevisiae (15), and can mutants of R. eutropha can grow in an atmosphere enriched with CO2 (20). To distinguish between the need to exclude O2 and a requirement for an increased partial pressure of CO2, we tested the ability of EDCM636 to grow anaerobically under H2 alone and found that it cannot, indicating that it must be the increased partial pressure of CO2 in the earlier experiment that allowed growth.

Growth in air with added CO2.

To determine whether increased CO2 partial pressure would restore growth in air, EDCM636, pregrown with azide, was grown in liquid with or without added CO2 (Fig. (Fig.3A).3A). When a mixture of 21% O2 and 79% N2 was continuously bubbled through EDCM636 in LB medium, the usual CA depletion growth pattern was seen, whereas the parental strain EDCM367 grew with a total yield about twofold greater than usual. This demonstrates that atmospheric CO2 is not needed to support growth of the parental strain. If a 2% CO2 plus 21% O2 plus 77% N2 gas mixture was bubbled through the culture medium instead, the parental strain EDCM367 grew as before, and growth of EDCM636 continued into stationary phase, albeit at a slower rate. Thus, increased CO2 in the growth environment at least partially ameliorates the growth defect resulting from lack of Can, a finding consistent with the idea that Can is required to maintain the CO2/bicarbonate concentration in the cell. Note the inflected shape of the growth curve of EDCM636 with CO2. Initial rapid growth (to 100 min: dilution of preexisting CA) is followed by growth at a reduced rate (to 250 min); growth rate then increases again. can expression was measured during growth of the CA depleted cultures and will be described below.

FIG. 3.
Growth of EDCM636 in a CO2-controlled atmosphere. (A) Growth of EDCM636 (circles) and its parent EDCM367 (squares) in LB medium at 37°C with or without added CO2. The cultures were bubbled (∼55 ml/min/flask) with a gas mixture of either ...

Growth with no gas phase.

In vessels filled with liquid, once the dissolved O2 is consumed during the early growth of the culture, further growth is necessarily anaerobic and fermentative, and CO2 equilibrates between medium and cytoplasm rather than being lost by diffusion. During fermentative growth under acid conditions, E. coli produces CO2 abundantly (in alkali, formic acid is excreted instead, see Discussion). To test whether endogenously produced CO2 is sufficient to support growth of the mutant, cultures were inoculated at pH values ranging from 5 to 9 and then incubated with (flasks) or without (in filled containers) a gas phase. Figure Figure4A4A shows that, under aerobic conditions (with gas exchange), growth of the mutant quickly stops at any pH in the range. Figure Figure4B4B shows, however, that EDCM636 can grow quite well anaerobically at an acidic pH (pH 5 or 6), although only poorly (pH 7 or 8) or hardly at all (pH 9) under more alkaline conditions. Figure Figure4C4C confirms that the parental strain can grow without air at all pH values between 5 and 9, although the growth yield is slightly reduced at lower pH values. We attribute the ability of EDCM636 to grow in bottles without air at acid pH to the fact that endogenously produced CO2, necessary for growth of the mutant, cannot escape. To further confirm this interpretation, we attempted to grow EDCM636 anerobically in an open vessel to confirm that it was not the absence of O2 that permitted growth. Flask cultures, in LB medium at pH 6, were continually flushed with argon to both exclude O2 and to facilitate the escape of endogenously generated CO2. We found (data not shown) that EDCM636 was unable to grow under these conditions, a finding consistent with the idea that CO2 is needed for anaerobic growth.

FIG. 4.
Effect of pH on growth of EDCM636. Strains were inoculated into LB medium-glucose at pHs from 5 to 9 and grown either aerobically in flasks or anaerobically in filled syringes (see Materials and Methods). (A) EDCM636 grown in air; (B) EDCM636 grown anaerobically; ...

Can expression responds to environmental changes.

In EDCM636 can is replaced by the reporter cassette FLK2, which permits the monitoring of gene expression by measurement of β-galactosidase activity. Because CA activity can be supplied by CynT, we can monitor can expression in the absence of Can protein.

(i) Expression is inversely proportional to growth rate.

Figure Figure5A5A shows the expression of can during a single growth cycle (growth without dilution) in LB medium. Initially, as exponential growth starts in the presence of azide, the specific activity (SA) drops from the higher level characteristic of overnight cultures. There is then a short period during which SA remains low and constant. After this, as the growth rate decreases (MG1655 growing in LB medium always slows when it reaches an OD600 of 0.4) the rate of expression of can increases and continues to do so as growth rate decreases further. When azide is absent, the initial growth rate is slow and specific expression of can is relatively high; expression continues even after growth has apparently ceased. To examine growth rate dependence in more detail, we grew EDCM636 with azide in several different media to vary the generation time from ∼20 to ∼150 min. Figure Figure5B5B shows that expression decreases with growth rate over a threefold range (from ∼120 to ∼40 Miller units [MU]). Although we did not achieve slower exponential growth rates, we were able to measure expression in cultures that were growing extremely slowly because azide, air, or CO2 were limiting. These cultures, with an OD600 of between 0.1 and 0.4, all expressed can at high levels (∼200 to ∼700 MU).

FIG. 5.
Variation of can expression with growth rate and cell density. (A) An overnight LB medium-azide culture of EDCM636 was inoculated into LB medium with (open symbols) or without (solid symbols) azide and growth (circles) and β-galactosidase synthesis ...

(ii) Expression can also vary with cell density.

In Fig. Fig.5A,5A, the SA does not remain constant even during the period of exponential growth. A constant SA could only be achieved during exponential growth by frequently diluting to maintain cell density within a narrow range. In the experiment shown in Fig. Fig.5C,5C, a culture of EDCM636 was divided into two; the OD600 of one culture was maintained below 0.07 and of the other between 0.1 and 0.2 by dilution as necessary. The SA of the denser is 1.5-fold greater than that of the more dilute.

(iii) RpoS has a small effect on can expression.

Increased expression at high OD or during slowed growth is often attributable to the participation of the stationary-phase transcriptional sigma factor RpoS. To see whether RpoS is involved in can expression, an rpoS derivative of EDCM636, EDCM638 was constructed. Figure Figure6A6A shows that can is expressed at the same levels in both strains at OD600 up to 2.6. At higher ODs the expression is greater, by ∼1.5-fold, in the RpoS+ strain. Since this difference is small and since RpoS controls the expression of a variety of proteins that regulate expression during stationary phase, we do not know whether the effect is direct or indirect.

FIG. 6.
Testing for other regulators of can expression. The growth medium was LB medium-azide. (A) RpoS. EDCM636 and EDCM638 (RpoS) were sampled and assayed for β-galactosidase. SA is plotted in MU as a function of OD. The data from two separate ...

(iv) can is not autoregulated.

In EDCM464 can is under the control of PBAD. Figure Figure6B6B shows that induction of Can synthesis from PBAD does not affect the activity of the native can promoter; thus, can expression is not repressed by Can. Depletion of Can obviously does not cause significant derepression, as exponentially growing azide cultures of can deletion strains show low levels of expression. We conclude that can expression is not autoregulated.

(v) can expression is increased at high temperature.

Annotation of can in the database suggests that it may have a heat shock promoter, although evidence of heat shock induction was not noted in a microarray experiment (31). We find (Fig. (Fig.6C)6C) that there is a twofold increase in can expression after transfer from 30 to 42°C, in contrast to the characteristic decrease in expression which accompanies an increase in growth rate due to improved nutrient supply. We have not tested whether this response is dependent on the heat shock induction system since mutating dnaK or rpoH reduces the growth rate, which would itself cause increased can expression. Site-directed mutagenesis of the putative rpoH binding region has not been attempted.

(vi) Expression with added CO2.

Expression from the can promoter in EDCM636 was monitored during growth under a CO2-enriched atmosphere (Fig. (Fig.3).3). Growth is initially at a normal rate; it then ceases in the absence of CO2 or proceeds more slowly in the presence of CO2. During the initial growth phase, can expression decreases from its stationary phase level and then increases as growth slows in both the CO2 supplemented and the nonsupplemented cultures. SAs (Fig. (Fig.3B)3B) are comparable for the two cultures, despite the great difference in total enzyme synthesized (Fig. (Fig.3A).3A). When the supplemented culture reached an OD600 of 0.8, the growth rate again increased and can expression decreased. It is clear that can expression in these cultures is responding to growth rate changes rather than to the presence or absence of CO2.

Suppressors of Δcan.

In order to see whether mutations which obviate the need for Can occur, suppressor mutants of EDCM636 were sought. Of 50 streaked colonies, 6 yielded progeny able to grow without azide. All of these mutations were at least 50% cotransductionally linked to lac, which is adjacent to the cyn operon; these mutants most likely produce CynT constitutively. We constructed a cynT can double mutant by selecting for transductants under CO2 in anaerobic jars, where no CA is needed. A total of 1010 cells of double mutants were screened for the ability to grow in air. About 40 slow-growing colonies which have not yet been further characterized were found and should prove interesting, since they indicate that E. coli can become, by mutation, independent of β-CA activity.

A γ-CA can substitute for Can.

It has been shown that a human α-CA can substitute for the R. eutropha Can (20). The ability of γ-CAs to substitute for bacterial β-CAs has not been tested. We therefore obtained a strain, BL21(λDE3)/pcam-AC, with a copy of the Methanosarcina thermophila cam gene (1) cloned under the control of a T7 promoter from the laboratory of J. G. Ferry, where it had been demonstrated to produce large quantities of active protein. We introduced the can deletion into this strain (in the presence of azide) to create SM1. When IPTG (isopropyl-β-d-thiogalactopyranoside) was added to induce T7 RNA polymerase, SM1 was able to grow on LB plates in the absence of azide, although the colonies were small, indicating that this enzyme can at least partially substitute for a β-CA.

At least three native E. coli proteins—YrdA, CaiE, and PaaY—share significant similarity with Cam. The gene for each of these, along with cynT and can, was amplified by using PCR and cloned into the high-copy-number vector pUC18. Of the three putative γ-CAs, only PaaY was visibly expressed, although CynT and Can supported growth in the absence of azide. Conjecturing that the expression signals of the others might be weak or that massive overexpression might be toxic, we cloned them instead into pTRC99A, which provides all expression signals up to and including an initiation codon and from which expression of the cloned genes is IPTG inducible. Because pTRC99A carries the lacIq gene, the cloned genes are not expected to be expressed until IPTG is added. The plasmids were then tested for their ability to suppress can mutations. A single colony of each type of EDCM636 transformant was streaked onto LB medium plus IPTG with or without azide and also used to inoculate a broth culture, supplemented with azide and IPTG, which was used, after overnight growth, to prepare material for the protein gel shown in Fig. Fig.7.7. Each of the proteins, other than CynT, is clearly visible as an overexpressed protein band of about the expected size. The large amounts of PaaY obtained presumably reflect the high copy number of the pUC18 vector. Although the cynT and can genes cloned in parallel efficiently complemented and/or suppressed can, plasmids containing the possible γ-CA genes did not suppress the need for Can. If the expressed proteins are enzymatically active, we conclude that they are not likely to have significant CA activity.

FIG. 7.
Overproduction of β-CAs and possible γ-CAs. Overnight cultures of EDCM636 containing plasmids expected to overexpress the indicated proteins were grown overnight on LB medium plus azide plus IPTG, and 25 μl/lane was prepared and ...

Phenylacetic acid (PAA) is an inducer of the enzymes of the Paa pathway, which allows it to be used as a sole carbon source; EDCM 637 is able to form colonies on plates containing no other source of carbon. EDCM636, however, does not grow on PAA plates or on glucose or LB plates containing PAA, indicating that induction of PaaY from its chromosomal gene does not provide a CA activity that can support growth.


The experiments described here show that E. coli cannot grow in air in the complete absence of a functional CA (either Can or CynT when induced) and that this inhibition is relieved by added or endogenous high concentrations of carbon dioxide. The reactions catalyzed by CA are the forward and reverse processes in the formation of bicarbonate, which is at rapidly established equilibrium with carbonic acid:

equation M1

The uncatalyzed rates of the hydration and dehydration of CO2 were first accurately measured by Mills and Urey (26). These authors showed that the forward pseudo-first-order rate constant k1 is 0.0275 s−1 at 25°C and estimated the backward rate constant k−1 to be 80 s−1.

CO2/bicarbonate is both produced and consumed by E. coli metabolism. If hydration or dehydration were the only rapid process involved, the system would approach equilibrium with a half-time of ca. 10 ms, there would be no reason to distinguish between the metabolic roles of CO2 and its hydration products, and CA would not be required. However, molecular CO2 is rapidly lost from the cell by passive diffusion and, if this causes the steady-state to remain far from equilibrium, the rates of production and consumption of CO2 and bicarbonate have to be kept individually in balance.

The main metabolic reactions that generate CO2/bicarbonate in E. coli during aerobic growth are those catalyzed by the enzymes pyruvate dehydrogenase, 2-oxoglutarate dehydrogenase, isocitrate dehydrogenase, 6-phosphogluconate dehydrogenase, and β-ketoacyl[acyl carrier protein] synthase. The mechanisms of these and related enzymes have been extensively studied (29, 30, 34, 38, 39), and in all cases the decarboxylation mechanism generates molecular CO2. During fermentation, the formate dehydrogenase component of formate-hydrogen lyase provides another potential source of CO2/bicarbonate. Although less conclusive in this case, mechanistic studies (3, 6) strongly suggest that molecular CO2 is the true product of formate oxidation.

The principal reactions in which CO2/bicarbonate is consumed are those catalyzed by the enzymes phosphoenolpyruvate carboxylase carbamoyl phosphate synthetase, 5-aminoimidazole ribotide carboxylase, and biotin carboxylase. The mechanisms of these enzymes have been extensively studied (9, 18, 27, 35, 36), and in every case it has been shown that the true substrate is bicarbonate.

In summary, endogenous CO2/bicarbonate is generated in E. coli predominantly and perhaps wholly in the form of molecular CO2, both during aerobic growth and in fermentation. A steady-state concentration of this species is established in the cell, such that the rate of loss by diffusion is equal to the net rate of production. The only significant source of the bicarbonate required as a substrate of metabolism is the hydration of this CO2. In the absence of CA, the rate of generation of bicarbonate depends on the steady-state concentration of CO2 and on the rate constant for the uncatalyzed hydration reaction. Under the conditions in which CA is an essential enzyme for growth of E. coli, the rate of the uncatalyzed reaction at the prevailing intracellular concentrations of CO2 is too low to meet the demand for bicarbonate.

To make this description more quantitative, it is necessary to propose an explicit model for the steady-state concentration of CO2. The diffusion coefficient for a dilute solution of CO2 in water at 20°C is 1.78 × 10−9 · m2 · s−1 (22), and this provides the principal constraint on the overall diffusion rate, unless cell boundary structures provide a significant extra barrier. No direct measurements of the diffusivity of molecular CO2 across the cell boundary of E. coli cells appear to have been made, but in the case of red blood cells the most recent and sensitive measurements (14) have shown that the cell membrane provides only a small additional diffusive barrier. Diffusion equations (4) can be combined with the aqueous diffusion coefficient and the shape and size of a typical E. coli cell (19) to estimate that if the net rate of production of CO2 in a cell is F mol · s−1, the excess steady-state concentration (above the equilibrium concentration corresponding to the gas phase composition) will be about F × 1011 mol · liter−1.

The results of careful measurements of the flux rates of individual processes in the central metabolic pathways of E. coli growing under a variety of precisely defined conditions have been presented (17). For E. coli ML308 growing at 0.94 h−1 aerobically on glucose, the overall net production rate is 14.78 mol of CO2 · kg−1(dry weight biomass) · h−1. An average cell (28) has a dry weight of 2.8 × 10−13 g and would therefore produce 1.15 × 10−18 mol of CO2 · s−1 under these conditions. If the proposed model for diffusion provides a reasonable approximation, the expected excess intracellular concentration of CO2 is ca. 10−7 mol · liter−1. If the gas phase concentration is zero, this will support a maximum uncatalyzed flux from CO2 to bicarbonate of ca. 3 × 10−24 mol · s−1 in a single cell. Air normally contains 0.03% by volume of CO2, and this will produce a concentration of 10 μM in the cell; in the absence of CA, this could support a flux of ca. 3 × 10−22 mol · s−1.

The potential demand for bicarbonate can be estimated by considering the following three classes of pathway in which it is required.

(i) Incorporation into E. coli cell mass.

During growth on simple substrates, one gram atom in each mole of arginine or nucleotide base is derived from bicarbonate ion, incorporated by the reactions catalyzed by carbamoyl phosphate synthetase and 5-aminoimidazole ribotide carboxylase. In the cells described by Holms, generation of 14.78 mol of CO2 · kg−1 · h−1 accompanied a growth rate of 0.94 h−1. From the analysis of 1 g (dry weight) of E. coli (see reference 28, p. 135), it may be calculated that this corresponds to a requirement for 7.4 × 10−20 mol · s−1 of bicarbonate for a single cell. In growth on complex media, biosynthesis of amino acids and nucleic acid precursors is not necessary, and these pathways should contribute no demand for bicarbonate under these conditions.

(ii) Bicarbonate/CO2 cycles.

Bicarbonate is incorporated into malonyl coenzyme A, via biotin carboxylase, in the first committed step of fatty acid biosynthesis and is released as CO2 in a subsequent condensation step. One molecule of bicarbonate is required for each two-carbon unit incorporated into lipid. Under the conditions outlined in the previous paragraphs, this corresponds to a flux of 1.79 × 10−19 mol · s−1 in a single cell.

Bicarbonate/CO2 is cycled in a similar way during the metabolism of cyanate, except that incorporation and release occur during the single step catalyzed by cyanase. Cyanate can be used by E. coli as its sole source of nitrogen and, under these conditions, there would be an additional demand for 7.46 × 10−19 mol of bicarbonate · s−1 in a single cell (normally provided by CynT).

(iii) Carboxylation in central metabolism.

Bicarbonate is the substrate for phosphoenolpyruvate carboxylase; the flux through this reaction varies considerably when E. coli is grown aerobically on a variety of simple substrates (17). For E. coli ML308 growing at 0.94 h−1 aerobically on glucose, the measured flux corresponds to the use of 2.01 × 10−19 mol of bicarbonate · s−1 in a single cell.

Thus, during growth at a moderate rate even on complex media, E. coli has a demand for ca. 2 × 10−19 mol of bicarbonate · s−1 per cell. This demand will increase somewhat if biosynthesis of amino acids and nucleic acid bases is required and will increase by a factor of up to 3 to 4 if nitrogen is derived solely from cyanate. During aerobic growth, the diffusion calculation suggests that the intracellular steady-state concentration of CO2 will be 103- to 104-fold, which is too low to meet this demand in the absence of CA. These estimates provide a quantitatively convincing explanation of why CA is essential during aerobic growth of E. coli.

Anaerobic conditions provide a more complex situation because most of the growth substrate is converted into fermentation products rather than being incorporated into bacterial cell mass. Growth still requires bicarbonate—most unavoidably to be cycled in fatty acid biosynthesis—but the generation of CO2 by dehydrogenases is eliminated or strongly constrained by the need to recycle the NADH produced concomitantly by these enzymes. Formic acid is a major fermentation product, and this may be disproportionated to CO2 and H2 by the formate hydrogen lyase reaction. This activity is constrained by the need for pH homeostasis (5). Loss of CO2 by diffusion raises the pH, so the enzyme is only active at low pH. In a fermenting culture that is effervescing CO2, the intracellular concentration is ca. 20 mM. Under anaerobic conditions, E. coli may generate enough bicarbonate for growth, even in the absence of CA, provided the pH does not rise unduly.

The results of growth experiments with Can-depleted E. coli match expectations based on this analysis of can function. No growth occurs in aerobic cultures at any pH (pH 5 to 9), but slow growth is restored at pH 7 by bubbling 2% CO2 through the culture medium. This concentration is 66-fold higher than the normal concentration in air and can provide a correspondingly increased flux of bicarbonate. Anaerobic growth is possible without added CO2 but only in acid conditions, when there is abundant endogenous generation of CO2 and when culture vials are closed (preventing CO2 loss by diffusion out of the medium). This analysis also provides a somewhat different perspective on the role of CynT. Through the activity of the cyanase pathway, E. coli is able to grow with cyanate as its sole source of nitrogen. Since bicarbonate is dehydrated in the cyanase reaction, metabolism in this case requires the supply of one mole of bicarbonate for each gram atom of nitrogen incorporated into growing cell mass. This implies a 20- to 30-fold increase in the bicarbonate requirement, and it is this extra need that is met by the induction of CynT.

In the present study we have also monitored can expression under a variety of conditions in order to see how it might be controlled. We find that its level is not autoregulated, nor does it respond, as might be expected, to the partial pressure of carbon dioxide in the gas phase. On the other hand, it is responsive to increasing cell density, to temperature upshift and, during late stationary phase, to the availability of RpoS. In each case expression increases by ∼2-fold. Transcription after temperature shift, with an overshoot followed by reduced transcription, is characteristic of σ32 responsive promoters, supporting the correctness of the sequence-based annotation, and may indicate an increased need for HCO3 during one or more of the stresses that increase σ32 availability.

However, the most notable conditions affecting expression are changes in growth rate and growth phase (although it is difficult to separate these two variables). What we observe is that the slower the growth, the greater the expression of can, with a 10- to 20-fold range of expression observed. We have not attempted to use genetics to further investigate the molecular mechanism of this variation (although RpoS may be involved), because mutations in genes such as relA, involved in many cases of growth rate regulation, are themselves characterized by slow growth, making it difficult to distinguish specific effects from indirect effects due to growth rate changes.

Why, however, should E. coli need more Can per mass when it is growing least quickly? One factor, of course, is that during growth on minimal medium, bicarbonate is needed to synthesize essential amino acids and nucleotides, which are supplied in broth. However, another possibility is that slow-growing cells, which are smaller, have a higher surface/volume ratio than faster-growing cells and therefore require proportionately more fatty acid biosynthesis. Cells entering stationary phase remain metabolically active, although cell mass no longer increases, and undergo major shifts in protein and lipid composition, as well as decreasing in size. Although we do not yet know why more Can should be required by slow-growing or starved cells than by rapidly growing ones, among the processes requiring bicarbonate, lipid synthesis appears to be the one most likely to make a relatively increased demand during slowed growth. It is also possible that an elevated level of Can is needed during slow growth within host animals for as-yet-unknown reasons.


We thank Sabrina Zimmerman for providing and retesting pcamAC and Bruce Ward and Pamela Beattie for assistance with anaerobic growth and elevated CO2 testing.

We are grateful to the Biotechnology and Biological Sciences Research Council (United Kingdom) for financial support for this study.


1. Alber, B. E., and J. G. Ferry. 1994. A carbonic anhydrase from the archaeon Methanosarcina thermophila. Proc. Natl. Acad. Sci. USA 91:6909-6913. [PMC free article] [PubMed]
2. Arigoni, F., F. Talabot, M. Peitsch, M. D. Edgerton, E. Meldrum, E. Allet, R. Fish, T. Jamotte, M. L. Curchod, and H. Loferer. 1998. A genome-based approach for the identification of essential bacterial genes. Nat. Biotechnol. 16:851-856. [PubMed]
3. Axley, M. J., and D. A. Grahame. 1991. Kinetics for formate dehydrogenase of Escherichia coli formate-hydrogen lyase. J. Biol. Chem. 266:13731-13736. [PubMed]
4. Berg, H. C. 1983. Random walks in biology. Princeton University Press, Princeton, N.J.
5. Böck, A., and G. Sawers. 1996. Fermentation, p. 262-282. In F. C. Neidhardt, R. Curtiss III, J. L. Ingraham, E. C. C. Lin, K. B. Low, B. Magasanik, W. S. Reznikoff, M. Riley, M. Schaechter, and H. E. Umbarger (ed.), Escherichia coli and Salmonella: cellular and molecular biology, 2nd ed. ASM Press, Washington, D.C.
6. Boyington, J. C., V. N. Gladyshev, S. V. Khangulov, T. C. Stadtman, and P. D. Sun. 1997. Crystal structure of formate dehydrogenase H: catalysis involving Mo, molybdopterin, selenocysteine, and an Fe4S4 cluster. Science 275:1305-1308. [PubMed]
7. Brown, O. R., and H. F. Howitt. 1969. Growth inhibition and death of Escherichia coli from CO2 deprivation. Microbios 3:241-246.
8. Charles, H. P., and G. A. Roberts. 1968. Carbon dioxide as a growth factor for mutants of Escherichia coli. J. Gen. Microbiol. 51:211-224. [PubMed]
9. Chollet, R., J. Vidal, and M. H. O'Leary. 1996. Phosphoenolpyruvate carboxylase: a ubiquitous, highly regulated enzyme in plants. Annu. Rev. Plant Physiol. Plant Mol. Biol. 47:273-298. [PubMed]
10. Cronk, J. D., J. A. Endrizzi, M. R. Cronk, J. W. O'Neill, and K. Y. Zhang. 2001. Crystal structure of Escherichia coli β-carbonic anhydrase, an enzyme with an unusual pH-dependent activity. Protein Sci. 10:911-922. [PMC free article] [PubMed]
11. Eichler, K., F. Bourgis, A. Buchet, H. P. Kleber, and M. A. Mandrand-Berthelot. 1994. Molecular characterization of the cai operon necessary for carnitine metabolism in Escherichia coli. Mol. Microbiol. 13:775-786. [PubMed]
12. Elssner, T., C. Engemann, K. Baumgart, and H.-P. Kleber. 2001. Involvement of coenzyme A esters and two new enzymes, an enoyl-CoA hydratase and a CoA-transferase, in the hydration of crotonobetaine to l-carnitine by Escherichia coli. Biochemistry 40:11140-11148. [PubMed]
13. Ferrandez, A., B. Minambres, B. Garcia, E. R. Olivera, J. M. Luengo, J. L. Garcia, and E. Diaz. 1998. Catabolism of phenylacetic acid in Escherichia coli: characterization of a new aerobic hybrid pathway. J. Biol. Chem. 73:25974-25986. [PubMed]
14. Forster, R. E., G. Gros, L. Lim, Y. Ono, and M. Wunder. 1998. The effect of 4,4′-diisothiocyanato-stilbene-2,2′-disulfonate on CO2 permeability of the red blood cell membrane. Proc. Natl. Acad. Sci. USA 95:15815-15820. [PMC free article] [PubMed]
15. Gotz, R., A. Gnann, and F. K. Zimmermann. 1999. Deletion of the carbonic anhydrase-like gene NCE103 of the yeast Saccharomyces cerevisiae causes an oxygen-sensitive growth defect. Yeast 15:855-864. [PubMed]
16. Guilloton, M. B., A. F. Lamblin, E. I. Kozliak, M. Gerami-Nejad, C. Tu, D. Silverman, P. M. Anderson, and J. A. Fuchs. 1993. A physiological role for cyanate-induced carbonic anhydrase in Escherichia coli. J. Bacteriol. 175:1443-1451. [PMC free article] [PubMed]
17. Holms, H. 1996. Flux analysis and control of the central metabolic pathways in Escherichia coli. FEMS Microbiol. Rev. 19:85-116. [PubMed]
18. Knowles, J. R. 1989. Mechanism of biotin-dependent enzymes. Annu. Rev. Biochem. 58:195-221. [PubMed]
19. Koppes, L. J. H., C. L. Woldringh, and N. Nanninga. 1978. Size variations and correlations of different cell cycle events in slow-growing Escherichia coli. J. Bacteriol. 134:423-433. [PMC free article] [PubMed]
20. Kusian, B., D. Sültemeyer, and B. Bowen. 2002. Carbonic anhydrase is essential for growth of Ralstonia eutropha at ambient CO2 concentrations. J. Bacteriol. 184:5018-5026. [PMC free article] [PubMed]
21. Lange, R., and R. Hengge-Aronis. 1991. Growth phase-regulated expression of bolA and morphology of stationary-phase Escherichia coli cells are controlled by the novel sigma factor sigma S. J. Bacteriol. 173:4474-4481. [PMC free article] [PubMed]
22. Mark, H. F., D. F. Othmer, C. G. Overberger, and G. T. Seaborg (ed.). 1978. Kirk-Othmer encyclopedia of chemical technology, 3rd ed., p. 83. Wiley-Interscience, New York, N.Y.
23. Merlin, C., S. McAteer, and M. Masters. 2002. Tools for characterization of Escherichia coli genes of unknown function. J. Bacteriol. 184:4573-4581. [PMC free article] [PubMed]
24. Merlin, C., G. Gardiner, S. Durand, and M. Masters. 2002. The Escherichia coli metD locus encodes an ABC transporter which includes Abc (MetN), YaeE (MetI), and YaeC (MetQ). J. Bacteriol. 184:5513-5517. [PMC free article] [PubMed]
25. Miller, J. M. 1992. A short course in bacterial genetics: a laboratory manual and handbook for Escherichia coli and related bacteria. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.
26. Mills, G. A., and H. C. Urey. 1940. The kinetics of isotopic exchange between carbon dioxide, bicarbonate ion, carbonate ion and water. J. Am. Chem. Soc. 62:1019-1026.
27. Mueller, E. J., E. Meyer, J. Rudolph, V. J. Davisson, and J. Stubbe. 1994. N5-carboxyaminoimidazole ribonucleotide: evidence for a new intermediate and two new enzymic activities in the de novo purine biosynthetic pathway of Escherichia coli. Biochemistry 33:2269-2278. [PubMed]
28. Neidhardt, F. C., J. L. Ingraham, and M. Schaechter. 1990. Physiology of the bacterial cell: a molecular approach. Sinauer Associates, Sunderland, Mass.
29. O'Donovan, C., M. J. Martin, A. Gattiker, E. Gasteiger, A. Bairoch, and R. Apweiler. 2002. High-quality protein knowledge resource: SWISS-PROT and TrEMBL. Brief. Bioinform. 3:275-284. [PubMed]
30. Price, N. C., and L. Stevens. 1999. Fundamentals of enzymology, 3rd ed., p. 291. Oxford University Press, Oxford, United Kingdom.
31. Richmond, C. S., J. D. Glasner, R. Mau, H. Jin, and F. R. Blattner. 1999. Genome-wide expression profiling in Escherichia coli K-12. Nucleic Acids Res. 27:3821-3835. [PMC free article] [PubMed]
32. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Laboratory Press, Cold Spring Harbor, N.Y.
33. Smith, K. S., and J. G. Ferry. 2000. Prokaryotic carbonic anhydrases. FEMS Microbiol. Rev. 24:335-366. [PubMed]
34. Stoddard, B. L., and D. E. Koshland. 1993. Structure of isocitrate dehydrogenase with α-ketoglutarate at 2.7-Å resolution: conformational changes induced by decarboxylation of isocitrate. Biochemistry 32:9317-9322. [PubMed]
35. Thoden, J. B., G. Wesenberg, F. M. Raushel, and H. M. Holden. 1999. Carbamoyl phosphate synthetase: closure of the B-domain as a result of nucleotide binding. Biochemistry 38:2347-2357. [PubMed]
36. Thoden, J. B., T. J. Kappock, J. Stubbe, and H. M. Holden. 1999. Three-dimensional structure of N5-carboxyaminoimidazole ribonucleotide synthetase: a member of the ATP-grasp protein superfamily. Biochemistry 38:15480-15492. [PubMed]
37. Vogel, H. J., and D. M. Bonner. 1956. A convenient growth medium for Escherichia coli and some other microorganisms (medium E). Microb. Genet. Bull. 13:43-44.
38. Walsh, C. 1979. Enzymatic reaction mechanisms, p. 914. W. H. Freeman & Co., San Francisco, Calif.
39. Zhang, L., L. Chooback, and P. F. Cook. 1999. Lysine 183 is the general base in the 6-phosphogluconate dehydrogenase-catalyzed reaction. Biochemistry 38:11231-11238. [PubMed]

Articles from Journal of Bacteriology are provided here courtesy of American Society for Microbiology (ASM)
PubReader format: click here to try


Save items

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • Compound
    PubChem chemical compound records that cite the current articles. These references are taken from those provided on submitted PubChem chemical substance records. Multiple substance records may contribute to the PubChem compound record.
  • Gene
    Gene records that cite the current articles. Citations in Gene are added manually by NCBI or imported from outside public resources.
  • GEO Profiles
    GEO Profiles
    Gene Expression Omnibus (GEO) Profiles of molecular abundance data. The current articles are references on the Gene record associated with the GEO profile.
  • MedGen
    Related information in MedGen
  • Nucleotide
    Primary database (GenBank) nucleotide records reported in the current articles as well as Reference Sequences (RefSeqs) that include the articles as references.
  • PubMed
    PubMed citations for these articles
  • Substance
    PubChem chemical substance records that cite the current articles. These references are taken from those provided on submitted PubChem chemical substance records.

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...