• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of jbacterPermissionsJournals.ASM.orgJournalJB ArticleJournal InfoAuthorsReviewers
J Bacteriol. Apr 2006; 188(8): 2865–2874.
PMCID: PMC1447003

Conserving a Volatile Metabolite: a Role for Carboxysome-Like Organelles in Salmonella enterica


Salmonellae can use ethanolamine (EA) as a sole source of carbon and nitrogen. This ability is encoded by an operon (eut) containing 17 genes, only 6 of which are required under standard conditions (37°C; pH 7.0). Five of the extra genes (eutM, -N, -L, -K, and -G) become necessary under conditions that favor loss of the volatile intermediate, acetaldehyde, which escapes as a gas during growth on EA and is lost at a higher rate from these mutants. The eutM, -N, -L, and -K genes encode homologues of shell proteins of the carboxysome, an organelle shown (in other organisms) to concentrate CO2. We propose that carboxysome-like organelles help bacteria conserve certain volatile metabolites—CO2 or acetaldehyde—perhaps by providing a low-pH compartment. The EutG enzyme converts acetaldehyde to ethanol, which may improve carbon retention by forming acetals; alternatively, EutG may recycle NADH within the carboxysome.

Salmonella enterica can use ethanolamine (EA) as a sole source of carbon and nitrogen under aerobic and anaerobic conditions (25, 29, 30). EA is degraded in two steps (5, 34). The first enzyme (ethanolamine ammonia-lyase) requires the cofactor adenosyl cobalamin (Ado-B12) (6), which is synthesized by S. enterica under anaerobic conditions but must be imported under aerobic conditions (17, 31). In this B12-dependent reaction, EA is cleaved to ammonia plus acetaldehyde, which is then converted to acetyl-coenzyme A (acetyl-CoA) by an NAD-dependent oxidoreductase (Fig. (Fig.11).

FIG. 1.
Basic ethanolamine metabolism. EA can enter the cell by diffusion, either via EutH or by other, unidentified routes. The diagram depicts the known roles of eut enzymes in metabolism of EA. The loss of acetaldehyde to the gas phase proposed here is indicated ...

Genetic screens revealed six genes required for the use of EA—eutA, -B, -C, -D, -E, and -R—all located within the 17-gene eut operon (20, 30). All other eut genes showed no mutant phenotype under standard conditions but seemed likely to contribute to EA use, because the operon (20) is induced by EA plus Ado-B12 (28, 37). In keeping with this expectation, several of these extra genes support use of EA under specific nonstandard conditions. The eutT gene encodes an adenosyl transferase that forms Ado-B12 (the lyase cofactor) from OH- or CN-B12 (9, 36). This provides more Ado-B12 during periods of high demand and is required in the absence of the general adenosyl-transferase, CobA (36). The EutH protein facilitates ethanolamine diffusion (see below) and is required only at very low levels of EA (23).

Five of the extra genes (eutS, -M, -N, -L, and -K) encode homologues of proteins that form the carboxysome, an organelle found in cyanobacteria and sulfur-oxidizing bacteria that concentrates CO2 for fixation by ribulose-bisphosphate carboxylase (RuBisCO) (38). We will refer to the Salmonella organelles as carboxysomes, despite the lack of evidence that CO2 is concentrated or fixed. We suspect that Salmonella does both during growth on ethanolamine. These organelles have been referred to elsewhere as “metabolosomes,” “enterosomes,” “polyhedral bodies,” or “bacterial microcompartments.” Another set of carboxysome protein homologues is encoded in an operon (pdu) that supports catabolism of 1,2-propanediol (1,2-PDL) by a pathway very similar to that for EA (4). The two organelle types form independently, and each type has been visualized during growth on either 1,2-PDL (16) or EA (8, 12). Several enzymes of the 1,2-PDL pathway copurify with carboxysome structures encoded by the pdu operon (16), indicating a close association of proteins catalyzing 1,2-PDL metabolism. Carboxysomes serve an unknown purpose in S. enterica but seem likely to play the same role in the similar EA and 1,2-PDL pathways and may share mechanistic features with related organelles that support CO2 fixation in other bacteria.

It was suggested previously that carboxysomes might protect cells from toxic effects of reactive aldehydes in the EA and 1,2-PDL pathways (8, 16, 26, 27, 41). Support for this idea came from the finding that strains defective for polA (encoding a DNA repair polymerase) or for synthesis of glutathione (gsh) failed to grow on EA. It was suggested that this phenotype reflects lack of a protective reducing agent (for gsh) or inability to repair DNA damage caused by aldehyde (polA). While toxicity is plausible, inhibition by endogenous acetaldehyde was not demonstrated, and the polA and gsh mutants used may be supersensitive to levels of acetaldehyde that are not normally toxic. In previous work, we presented evidence that endogenous acetaldehyde can accumulate to toxic levels in eutE mutant cells, which lack acetaldehyde oxidoreductase, the second step in EA utilization (23). Thus, the toxicity of endogenous acetaldehyde has been demonstrated or inferred only for mutants that either accumulate very high aldehyde levels or are supersensitive to its effects. We suspect that acetaldehyde toxicity is seldom a problem for wild-type Salmonella.

It is suggested here that the primary role of carboxysomes in S. enterica is not to protect cells against the toxic effects of aldehyde (which are seen only under extreme conditions) but rather to enhance the use of EA and 1,2-PDL by retaining a volatile intermediate that would otherwise be lost to the gas phase. The aldehyde intermediates in EA and 1,2-PDL (acetaldehdye and propionaldehyde) are volatile and subject to loss as vapor. In trying to learn how carboxysomes work, it seems reasonable to seek a single mechanism that will explain how homologous organelles concentrate CO2 in support of photosynthesis (in some bacteria) (10, 19) and restrict loss of volatile aldehydes (in enteric bacteria). These aldehydes and CO2 are both gases whose volatility and loss from solution are affected by pH. A model for carboxysome function is suggested in which the organelle provides a low-pH compartment that favors conversion of aldehydes to less volatile acetals; such a compartment would serve CO2 fixation (in other bacteria) by helping maintain a high level of dissolved CO2 gas, the substrate for RuBisCO.


Strains and crosses.

All strains (Table (Table1)1) are derivatives of Salmonella enterica serovar Typhimurium strain LT2. Transductional crosses were mediated by a high-frequency transducing mutant of phage P22 (HT105 int) (35) and performed as described previously (14).

Strain list

Media and growth conditions.

Rich medium (NB) was Difco nutrient broth with added NaCl (0.l mM). Minimal solid medium for growth assays was 5 mM KH2PO4, 5 mM NaNH4HPO4, 1 mM MgSO4 buffered at various pHs by 50 mM MES [2-(N-morpholino)ethanesulfonic acid; pH 5.5 or 6.0], MOPS [3-(N-morpholino)propanesulfonic acid; pH 7.0], or Bicine (pH 8.0 or 8.5) as described previously (23). Ethanolamine hydrochloride and sodium succinate (Sigma) served as carbon sources at appropriate concentrations. In the construction of eutS, eutM, eutJ, and eutG deletion mutations (see below), glycerol (20 mM; Malinckrodt) was used as a carbon source and EA as a nitrogen source in NCN minimal medium (2). Cyanocobalamin (Sigma) was provided at 150 nM and is referred to simply as B12. Rich medium was solidified by 1.5% agar (EM Science) and minimal media by 1.5% Noble agar (U.S. Biological). To measure growth rates, inocula were 3-ml nutrient broth cultures from single colonies. After overnight growth, cells were pelleted by low-speed centrifugation (Sorval RC5B+ at 4,700 rpm for 10 min), washed in pH 7.0 NCE (no citrate E) minimal medium (2), resuspended in the same medium, and used (35 μl) to inoculate 6 ml of growth medium. Growth was monitored in three parallel cultures as optical density at 650 nm.

Acetaldehyde assay.

Acetaldehyde released to the gas phase was measured using a three-part device. (i) Air was forced by an aquarium pump through water to hydrate it. (ii) The hydrated air was delivered through a manifold to the surface of the growth media of six tubes and vented from the tops of theses tube. (iii) The effluent air from each tube was delivered to the bottom of an inclined column containing 3 ml of 0.01% 3-methyl-2-benzothiazolone, which reacts with aldehyde to form an azine derivative that absorbs light from 300 to 360 nm (39). The detection solution was replaced every 30 min, and the optical density at 308 nm was read to determine aldehyde using a standard curve. Quantitative aldehyde trapping was demonstrated using pure acetaldehyde. These assays were performed on cultures growing at 37°C in a shaking water bath (240 rpm) in NCE medium at pH 7.0. The medium included EA (15 mM), B12 (150 nM), and glycerol (27 mM) except when noted. In experiments testing pdu mutants, 1,2-propanediol (80 mM), sodium pyruvate (40 mM), and B12 (150 nM) were added; incubation was at 38°C. For testing the effects of pH on acetaldehyde release (see Fig. Fig.88 and and9),9), minimal medium was as described for solid media above with reduced phosphate and 50 mM MOPS (at pH 7) or 50 mM Bicine (pH 8) as a buffer.

FIG. 8.
Effect of pH on aldehyde release. Growth (closed symbols) and cumulative aldehyde release (open symbols) were measured at pH 7 and 8. Wild-type LT2 is depicted by squares at pH 7 and diamonds at pH 8. A eutM deletion mutant is depicted at pH 7 (triangles, ...
FIG. 9.
Aldehyde release by pdu mutants. Mutants pduCDE (circles), pduP (triangles, points down), pduB (diamonds), pduJ (triangles, points up), and pduK (right triangles) were compared to wild-type LT2 (squares) during growth on pyruvate (40 mM), 1,2-propanediol ...

The cumulative aldehyde release shown below (see Fig. Fig.44 to to66 and and9)9) was calculated as follows. Aldehyde release measurements for each 30-minute period were averaged from three independent cultures. These average values described a graph that depicted aldehyde release per 30-minute period; 80% of these rate measurements had standard deviations of less than 18% of the mean value. This curve was integrated to determine the aggregate release at each point, which is depicted in each figure.

FIG. 4.
Mutations that block acetaldehyde metabolism increase aldehyde release. Mutants eutB (circles), eutD (diamonds), eutE (triangles, points up), and eutG (triangles, points down) were compared to wild-type LT2 (squares) for aldehyde release during growth ...
FIG. 6.
Effects of other eut mutations on aldehyde release. Mutants eutH (circles), eutJ (diamonds), eutQ (triangles, points up), eutT (right triangles), and eutP (triangles, points down) were compared to wild-type LT2 (squares) for aldehyde release during growth ...

Construction of in-frame eut deletion mutants.

A chromosomal deletion mutation was constructed for each eut gene using linear transformation mediated by the Red functions of phage lambda (13, 43) as described previously (23). Each deletion replaced the coding sequence with a short sequence that did not disrupt the reading frame or translational coupling. The constructed deletions were transduced by P22-mediated crosses into a strain with a wild-type genetic background, and their structures were verified by PCR amplification and sequencing of the junction.

Deletions of the eutP, eutQ, eutT, eutD, eutN, eutA, eutB, eutC, eutL, and eutK genes were constructed as described previously (13) using an inserted chloramphenicol resistance (cat) cassette. Deletions of the eutS, eutM, eutJ, and eutG genes were made by positive selection for removal of a polar mutation in the particular gene. A small linker DNA was used as a template in a PCR with two primers (P1 and P2). The 5′ end of P1 was identical to the first 40 bp of the eut gene deleted, and the 3′ end was identical to the 5′ end of the linker. The 5′ end of P2 was reverse complementary to the last 40 bp of the eut gene, and the 3′ end was reverse complementary to the 3′ end of the linker. The recipient's polar insertion prevented growth on EA because it reduced expression of the downstream eutBC genes. The polar mutation was removed by Red-mediated recombination using the donor fragment described above and selecting for growth on EA as a nitrogen source. The constructed deletion mutant allele had an open reading frame lacking the bulk of the internal coding sequence. All genes deleted in this way were dispensable for growth on EA as a carbon and nitrogen source at pH 7.0 and were located promoter proximal to the eutBC genes.


A set of nonpolar deletions of single eut genes.

A nonpolar deletion mutation was constructed for each of the 16 eut structural genes (see Materials and Methods). Each mutant was tested for the ability to use EA as a carbon source on solid medium (40 mM EA, 150 nM CN-B12; 37°C; pH 7.0). Under these standard conditions, eutA, -B, -C, -D, or -E mutants showed a Eut phenotype, and all other eut mutants were phenotypically indistinguishable from the wild-type strain LT2. These results (Table (Table2,2, pH 7.0) confirm previous conclusions (20, 30) based on point mutations and multigene deletions.

Effect of pH on use of ethanolamine as a carbon source

Effects of pH on mutant phenotypes.

The single-gene eut deletion mutants fell into four phenotypic classes based on their abilities to use ethanolamine as a carbon source on solid media at various nonstandard pH values (Table (Table22).

Class I mutants failed to use EA at all tested pH values and eliminated the previously described enzymatic functions, EutA, -B, -C, -D, and -E. (Mutants lacking the positive regulatory protein EutR also fell into this class.)

Class II mutants grew at pH 7 or below, but not at pH 8. This class included mutants lacking the carboxysome protein homologues EutM, -N, -L, and -K or the alcohol dehydrogenase EutG. Note that the eutS deletion, which lacks another carboxysome shell protein homologue, had no mutant phenotype at high pH under these conditions.

Class III contained only the eutH mutant, which showed impaired growth only at low pH as described previously (23).

Class IV mutants behaved like the wild type and showed little or no growth impairment at pH values at or below pH 8. The group included eutS, -P, -Q, -T, and -J mutants.

These phenotypes are based on qualitative scoring of growth patches on solid minimal medium. The conclusions were verified by growth rate determinations in liquid cultures. Growth rates for several critical strains are presented below.

Conservation of a volatile metabolite: a model for carboxysome function.

In developing a testable model to explain these results, it seemed likely that external pH might affect EA degradation in two ways. First, it increases the rate of EA influx by raising the concentration of unprotonated ethanolamine, which enters cells by diffusion (23). Second, pH affects the toxicity and volatility of acetaldehyde, an intermediate in EA degradation. Low pH favors reaction of acetaldehyde with alcohols to form less reactive, less volatile acetals. If at high external pH, EA enters and is converted to acetaldehyde faster than acetaldehyde can be consumed, an accumulation is expected. This accumulation could be subject to loss to the gas phase and/or could inhibit growth due to toxic effects.

We propose that external pH affects the rate at which the aldehyde produced escapes from cells and from the medium. Cultures growing aerobically on EA smell strongly of acetaldehyde (the odor of cocktail onions), which suggested initially that this intermediate might be lost to the gas phase at a significant rate. Excessive aldehyde loss might limit the carbon and energy obtainable from ethanolamine. The pH-dependent phenotypes described above would be explained if carboxysomes minimized the loss of aldehyde from cells and therefore became essential when ambient conditions increased the rate of carbon loss above a critical level. The carboxysome might restrict loss by providing a low-pH compartment in which aldehydes are converted to less volatile acetals by combination with alcohols. The need for carboxysomes might be reduced at low outside pH, because aldehyde is retained better in the medium (for reassimilation) or because the intracellular pH drops sufficiently to allow the whole cytosol to play the role of the carboxysome. The model proposes that the principal role of the carboxysome is retention of gases rather than protection of the cell from toxic effects. Some predictions of this model are tested below.

Acetaldehyde vapor is released during growth in the presence of ethanolamine.

Air was forced across liquid cultures (see Materials and Methods) and through a trapping solution containing a reagent that reacts with aldehyde to form a colored complex. Cells grown without ethanolamine on acetate, glycerol, or glucose (glucose data not shown) showed no detectable aldehyde release (Fig. (Fig.22).

FIG. 2.
Growth and aldehyde release by wild-type cells. Growth (closed symbols) and aldehyde release (open symbols) of strain LT2 grown on minimal acetate (squares), on glycerol (triangles, points up), or on glycerol plus ethanolamine and B12 (triangles, points ...

Aldehyde release was detected when ethanolamine and B12 were added to medium containing glycerol. Mutations blocking the first step in ethanolamine degradation (eutABC) prevented this aldehyde release (Fig. (Fig.3),3), suggesting that the assay detects acetaldehyde generated endogenously from ethanolamine.

FIG. 3.
Aldehyde release requires ethanolamine ammonia lyase. Growth (closed symbols) and aldehyde release (open symbols) of LT2 (squares) and a eutB mutant (circles) on minimal glycerol, ethanolamine, and B12 medium. Ethanolamine was provided at 15 mM. The error ...

The wild type and all eut mutants grew on glycerol at the same rate with or without added EA; that is, there was no evidence that ethanolamine (or derived acetaldehyde) was toxic under these conditions (Fig. (Fig.22 and and3).3). It should be noted that the eut operon is well induced by ethanolamine plus B12 in the presence of these carbon sources; glucose reduces induction only about threefold (28). The toxicity of ethanolamine seen previously for a eutE mutant growing on succinate (23) required a higher level of EA (40 mM).

Pathway blocks increase aldehyde release.

Mutants blocked from further metabolism of acetaldehyde (eutG or eutE) released more acetaldehyde than wild-type cells (Fig. (Fig.4).4). As described above, cells were grown on glycerol with added EA and B12. A slight increase was seen for eutD mutants, suggesting some secondary acetaldehyde accumulation in a strain blocked for conversion of acetyl-CoA to acetyl-PO4 (7, 40). Growth of all strains (on glycerol) was indistinguishable from that of the wild type (Fig. (Fig.22 and and3),3), suggesting that this increased aldehyde level was not detectably toxic.

Mutants lacking carboxysome proteins released more acetaldehyde.

Mutations eliminating the eutM, -N, -L, or -K gene increased aldehyde release during growth on glycerol plus EA (Fig. (Fig.5).5). These mutants (and eutG) fell into phenotype class II above. The eutS mutant, a member of class IV, resembled other carboxysome mutants in aldehyde release even though it showed no obvious growth phenotype at pH 8. The remaining genes in the operon fell into phenotype class III or IV (described above). Mutants lacking EutT (adenosyl transferase) or EutJ (unknown) show about the same level of aldehyde release as the wild type. The EutH (transport) defect reduced release slightly. Mutants lacking EutP and EutQ proteins released more acetaldehyde than the wild type, but the functions of these proteins are unknown (Fig. (Fig.6).6). The growth of these mutants (on glycerol) was identical to that of the wild type (Fig. (Fig.22 and and3),3), so aldehyde release was not toxic at these levels.

FIG. 5.
Aldehyde release increases in carboxysome mutants. Mutants eutN (circles), eutM (diamonds), eutL (triangles, points up), eutK (right triangles), and eutS (triangles, points down) were compared to wild-type LT2 (squares) for aldehyde release during growth ...

The growth phenotype of carboxysome mutants is corrected by restricting diffusion.

In the plate growth experiments described above, class II mutants (eutMNLKG) failed to use EA on media with a high pH. These tests were made in a standard 37°C incubator with loose petri dish lids, allowing free exchange of gases with the surrounding atmosphere. The effects of this exchange were tested in two ways. First, exchange was limited by incubating plates as described above, but at 30°C. Second, gas exchange was limited (at 37°C) by incubating plates in a closed air-tight box (2.5 liters). The container provided sufficient oxygen to support growth but restricted the loss of gases (such as acetaldehyde).

As seen in Table Table3,3, both reduced temperature and containment allowed wild-type cells to extend their range of growth from pH 8 up to pH 8.5. Restricted exchange also allowed mutants of classes III and IV to grow at pH 8.5 (data not shown). Control experiments (not shown) demonstrated that oxygen was sufficient (full growth on glycerol) and that pH differences did not affect the ability of the wild type to grow on ethanolamine (Table (Table2)2) or glycerol (not shown).

Growth on ethanolamine with restricted gas exchange

More importantly, diffusion limitation corrected the high-pH phenotypes of carboxysome-defective mutants (eutM, -N, -K, and -L) and the other mutant of class II, eutG. That is, containment extended the growth of these mutants from pH 7 to pH 8.5, and low temperature extended their growth from pH 7 to pH 8. Raising the temperature to 40°C reduced the growth of wild-type cells at pH 6.0 and prevented their growth at pH 7.0 and 8.0 (data not shown). A growth improvement caused by retaining aldehyde vapor is consistent with the mutant phenotype being due to excessive loss of aldehyde and is inconsistent with a phenotype due to aldehyde toxicity.

The eutG mutant lacks aldehyde reductase, which converts internal acetaldehyde to ethanol. This mutant had a growth phenotype like that of the carboxysome mutants (see above) and released more aldehyde than the wild type. As described below, the phenotype of eutG mutants (like that of carboxysome mutants) is corrected by providing a different carbon source, suggesting that EutG helps to conserve or use aldehyde rather than removing a toxic accumulation. The ethanol produced by EutG may promote formation of acetals within the carboxysome. Alternatively, this enzyme may stimulate EutE (oxidoreductase) by recycling NADH produced within the carboxysome.

If restricting loss of aldehyde to the outside environment improves growth, one might expect acetaldehyde vapor to support the growth of S. enterica on medium devoid of any other carbon source. To test this, a eutB and a eutM mutant were grown in adjacent wells of a 96-well plate. The eutB mutant (lacking ethanolamine ammonia-lyase) was given EA plus B12, which does not support its growth but will induce the operon. When a eutM mutant (defective in carboxysomes) was grown nearby on glycerol, the neighboring eutB cultures still failed to grow. However when the eutM mutant was given EA and B12 in addition to glycerol, the eutB mutant in nearby wells was allowed to grow, presumably on the acetaldehyde liberated by the growing eutM mutant (data not shown). In addition, lyase mutants (induced for eut by EA and B12) grew on medium when aldehyde vapor was bubbled from an aldehyde solution into the growth medium. The final density of the fed culture depended on the initial concentration of aldehyde in the source solution (data not shown).

Growth on ethanolamine is impaired by increased gas exchange.

The model predicts that more rapid exchange of ambient air might increase the loss of acetaldehyde vapor and reduce growth on ethanolamine. This effect should be more severe in mutants impaired in the ability to retain acetaldehyde. To test these predictions, petri plates were incubated within a chamber that was continuously flushed with various gas mixtures. Flushing plates with 20% or 14% O2 (balance, N2) prevented the growth of class II mutants (eutMNLKG) at all pH values tested (Table (Table4).4). This rapid exchange of the gas phase also prevented the wild type from growing on ethanolamine at pH values above 7. This exchange did not reduce the growth of any strains on glycerol.

Ethanolamine use during rapid gas exchange

Acetaldehyde retention is increased by exposure of cells to CO2 or benzoate.

Addition of CO2 to the gas mixture used to flush the container (20% O2 plus 5% CO2; balance, N2) allowed the wild type and all mutants to grow across the entire pH range (Table (Table4;4; data not shown for class III and IV mutants). Several reasons for the correction were entertained.

The exposure to CO2 might reduce the pH of the plate. Over the period of these tests, CO2 reduced the pH of the medium slightly (0 to 0.9 pH units), but subsequent tests with higher buffer concentrations or adjusted pH eliminated the possibility that the effect of CO2 reflected only changes in the pH of the medium (data not shown).

The CO2 may have some direct connection to carboxysome function. This was enticing, because the growth defect of photosynthetic organisms lacking carboxysomes is corrected at high CO2 concentrations (1, 10), leading to the conclusion that the organelles in these organisms concentrate CO2. This seems unlikely to explain the effects on the use of ethanolamine, in view of the results described below.

The provided CO2 may enter cells uncharged and reduce the internal pH of the cytosol, replacing the function of the carboxysome. To test this, benzoate was added to the growth medium. Benzoate, like CO2, can reduce intracellular pH by entering in an uncharged state and releasing a proton into the cytosol (32). Addition of benzoate improved the growth of class II mutants, even when cells were flushed with a gas mixture lacking CO2. This growth improvement was achieved with no change in the pH of the medium. We suggest that CO2 (like benozate) corrects the phenotype of carboxysome mutants by reducing intracellular pH.

Suppression of ethanolamine ammonia-lyase (EutBC) mutants by limiting gas exchange.

In the course of the experiments described above, it was surprising to note that some class I mutants, which fail to use ethanolamine under standard conditions, were allowed to grow on EA at a rate approaching that of the wild type under conditions that restrict loss of acetaldehyde. The corrected mutants lack either the first degradative enzyme, ethanolamine ammonia-lyase (eutABC), or phosphotransacylase (EutD). A mutant lacking the second step (EutE) was not corrected. The correctable mutants used ethanolamine when the plates were incubated in a confined space, when pH was reduced, when CO2 was added to air flushed over the plates, or when benzoate was included in the medium (Table (Table55).

Suppression of class I mutants by CO2

This suggested that ethanolamine can be converted to acetaldehyde without EutABC at a rate that supports growth when loss of acetaldehyde is minimized. Perhaps in these mutants acetaldehyde is lost excessively because the alternative enzymes responsible are not associated with the carboxysome. We know little about the inferred alternative pathway except that it requires the proteins AdhP (a homologue of alcohol dehydrogenase) and Ppk (polyphosphate kinase), both encoded outside the operon. This pathway allows ethanolamine to serve as a carbon source, but not as a nitrogen source (22), and is discussed below.

Lack of carboxysomes simultaneously decreases the growth rate and increases acetaldehyde release.

In the liquid growth tests described above, aldehyde release was detected while cells grew on an alternative carbon source. To correlate growth impairment with acetaldehyde release, a eutM mutant was tested while EA served as the sole carbon source. In liquid medium, this mutant grew more poorly on EA than wild-type cells at pH 7 (Fig. (Fig.7).7). (Note that on solid medium, growth of the eutM mutant failed completely only at a slightly higher pH.) To quantify released aldehyde under conditions of reduced growth, release was normalized to the cell number. The growth-impaired eutM mutant released more acetaldehyde per cell than the wild type.

FIG. 7.
Aldehyde release during growth on ethanolamine as a carbon source; growth (closed symbols) and aggregated aldehyde release normalized to cell growth (open symbols). Wild-type LT2 (squares) was compared to a eutM deletion mutant (triangles, points down). ...

Aldehyde release, like growth phenotype, is affected by pH.

The effect of pH on aldehyde release correlated with its effect on growth phenotypes scored on plates (described above). The eutM mutant was grown in liquid medium on glycerol with added EA/B12, and its aldehyde release was compared with that of wild-type cells at pH 7 and 8. While higher pH increased the aldehyde release of both wild-type and mutant strains, the increase was more pronounced in the mutant. The growth rates were only minimally affected by the pH difference (Fig. (Fig.88).

Providing an alternative carbon source corrects all growth phenotypes of carboxysome mutants.

Evidence was provided above that acetaldehyde derived from ethanolamine is released during growth on glycerol and has little or no toxic effect even in mutants lacking carboxysome proteins. This suggested that the main role of carboxysomes is to retain acetaldehyde as a carbon source rather than to prevent its toxicity. In further support of this conclusion, the growth defect of class II eut mutants at high pH is corrected by the addition of succinate, an alternative carbon source. Succinate also corrects the growth defects caused on solid medium by increased gas exchange (data not shown).

Cells growing on 1,2-propanediol release aldehyde.

The model for carboxysome function proposed for EA use would be expected to apply to the parallel propanediol (1,2-PDL) pathway. Growth on 1,2-PDL is associated with production of a carboxysome, and the pathway includes an aldehyde. The first enzyme (PduCED) produces propionaldehyde, which is converted to propionyl-CoA by the second enzyme (PduP). We measured aldehyde release during growth on 1,2-PDL, testing wild-type cells and various mutant strains, including three lacking homologues of carboxysome shell proteins (PduB, -J, and -K). All of the mutants tested were in-frame deletions constructed and generously provided by Thomas Bobik. In these tests, strains were grown on pyruvate plus 1,2-PDL and B12. (Glycerol could not be used as a carbon source because it inhibits propanediol dehydratase [42]).

Mutant and wild-type strains grew similarly on pyruvate with and without 1,2-PDL, indicating no growth inhibition by 1,2-PDL or its derived aldehyde (not shown). The wild-type strain released aldehyde only in the presence of 1,2-PDL, and that release was prevented by the pduCDE deletion mutation (Fig. (Fig.9),9), which eliminated propanediol dehydratase. As predicted by the model, three carboxysome mutants tested (pduJ, -K, and -B) released significantly more aldehyde than wild-type LT2. Unexpectedly, a pduP mutant, blocked after propionaldehyde, showed less aldehyde release than the wild type rather than more. This result is discussed below.


Acetaldehyde, an intermediate in ethanolamine degradation, is both volatile and toxic. Evidence has been presented that the primary function of the carboxysome encoded by eut is to minimize acetaldehyde loss and improve the efficiency of growth on EA. This function becomes critical under conditions that increase the rate of aldehyde loss. Such conditions include high pH of the medium and rapid exchange of ambient gases. Aldehyde loss is minimized when intracellular pH is reduced by providng CO2 or benzoate. As expected, phenotypes seen in media with high pH are corrected by restricting gas exchange. All phenotypes attributed to the loss of aldehyde are corrected by providing an alternative source of carbon. These results suggest that the phenotypes reflect a lack of carbon source (due to loss of acetaldehdye) rather than toxic effects of aldehyde. These conclusions are summarized in Fig. Fig.1010.

FIG. 10.
Proposed role of the carboxysome. It is proposed here that the primary function of a carboxysome is to retain and prevent loss of a volatile metabolite, acetaldehyde in the case of the ethanolamine pathway. Retention of volatile aldehydes (in Salmonella ...

While acetaldehyde is toxic when added at high concentrations to cell cultures (26), the toxicity of acetaldehyde produced endogenously from ethanolamine has been seen only in EutE mutants (23), which lack acetaldehyde oxidoreductase (EutE), and in polA and gsh mutants, which may be supersensitive to toxic effects of aldehyde (26, 27).

The results presented do not shed light on the mechanisms by which carboxysomes might conserve acetaldehyde—a formidable problem. It seems reasonable that the mechanism used may also explain the ability of similar organelles to concentrate CO2 in support of photosynthesis (10, 19). Both aldehydes and CO2 are gases whose volatility and loss from solution are affected by pH. When alcohols are available, low pH should favor the conversion of aldehydes to less volatile acetals, thereby minimizing loss. Low pH favors conversion of carbonate to CO2 gas, required by RuBisCO in the dark reactions of photosynthesis. The ability of the organelle to maintain a low-pH interior might explain both conservation of aldehyde in S. enterica and provision of CO2 gas in other organisms. An alternative possibility is that the organelle allows entry of CO2 gas but not carbonate. Organisms that fix CO2 could use this dissolved gas and help maintain it using carbonic anhydrase, while S. enterica might take advantage of the pH drop due to internal carbonate formation. We understand that there is a serious mechanistic problem with this theory in that the organelles are bounded by proteins, not lipids, and appear freely permeable when tested in vitro (10). This problem is faced by all models that propose concentration of substrates or cofactors, retention of intermediates, or sequestration of toxic compounds (8, 16, 41). We can only speculate that the purified organelles behave differently from those inside cells.

The phenotypes reported here parallel phenotypes reported for carboxysome mutants in photosynthetic bacteria, which photosynthesize normally if provided with a higher level of CO2 (1, 10). In S. enterica, ethanolamine phenotypes are also corrected by increased CO2. We argue here that CO2 acts in S. enterica by reducing the intracellular pH. This may also be true for cyanobacteria, where carboxysome mutant phenotypes might be corrected by providing benzoate rather than CO2.

It is surprising that the growth phenotype of a eutG (ethanol dehydrogenase) mutant is like that of carboxysome mutants—deficient in harvesting carbon and energy from EA. However, eutG mutants showed aldehyde excretion like that of eutE (aldehyde oxidoreducase) mutants. EutG may simply serve to help recycle electron carriers and balance redox reactions during fermentation by converting excess acetaldehyde to ethanol and NADH to NAD. Recycling of NADH would be particularly important during anaerobic fermentation, which might be expected to occur by conversion of acetyl-CoA to acetate with the production of one ATP and the release of acetate. However, we have been unable to demonstrate anaerobic fermentation of ethanolamine (use as a sole carbon and energy source without an external electron acceptor). It is possible that the reduced nature of ethanolamine and the high flux required for growth might generate more NADH than can be consumed even aerobically. It also seems possible that the NAD available to EutE within the carboxysome is limited and EutG is needed to recycle NADH produced locally by EutE. The phenotype would also be explained if ethanol produced by EutG helped trap acetaldehyde as an acetal within the carboxysome.

Recent results of Starai et al. (40) demonstrate that considerable acetate is produced even during aerobic growth on EA and that EutD (an acetyl phosphotransacylase) contributes to this (7, 8). In this work, eutD mutants were found to have no growth defect on ethanolamine, a reasonable finding, since EutD directs carbon toward acetate and out of the cell (Fig. (Fig.1).1). The presence of this gene in the eut operon was attributed to the need to recycle CoA when pools of acetyl-CoA are excessive. In contrast, we report here that a similar eutD in-frame deletion mutant failed to use ethanolamine under any standard conditions. We attribute the need for this enzyme to a requirement to prevent accumulation of acetyl-CoA and secondarily of acetaldehyde. (EutD mutants excrete as much acetaldehyde as carboxysome mutants.) We suggest that during growth on high levels of ethanolamine, cells convert accumulated excess acetyl-CoA to acetate (which is excreted for later use) rather than risk secondary accumulation of acetadehyde, which is subject to irretrievable loss to the gas phase or conversion to ethanol (which cannot be used). The difference in observed eutD phenotypes needs to be resolved but seems likely to reflect the use of different genetic backgrounds.

The similar pathways for degradation of EA and 1,2-PDL both include an aldehyde, and both operate with the use of carboxysomes. In both systems, aldehyde vapor was detected for wild-type cells and increased in mutants defective for carboxysomes. A key difference in the behaviors of the two pathways was that the oxidoreductase mutant (eutE) released more aldehyde (as expected for our model) while the analogous pdu mutant (pduP) released less. The behavior of the pduP mutant might be explained if PduP not only converts propionaldehyde to the CoA derivative, but also interacts directly with dioldehydratase to enhance its activity or prevent its inactivation by aldehyde. This might cause a pduP mutation to secondarily reduce dioldehydratase activity in vivo.

The eutS gene encodes a distant homologue of carboxysome proteins. Unlike mutants lacking other carboxysome shell proteins, the eutS mutant showed no growth defect at high pH on solid medium. Like the other mutants, it did show increased loss of acetaldehyde during growth on glycerol plus acetaldehyde in liquid medium. The EutS protein may make a more subtle contribution to carboxysome structure that is required for ethanolamine use at more extreme conditions. Consistent with this, work on the propandediol (pdu) operon (16) has shown that some carboxysome shell proteins can be lost without preventing the formation of an observable carboxysome, while the lack of others leads to structural abnormality or complete absence. Particular shell proteins may provide a specific function without being essential to the formation of the organelle.

A second pathway for ethanolamine use was inferred from the observation that eutBC mutants (lacking ethanolamine ammonia-lyase) grow on ethanolamine when high levels of CO2 or benzoate are provided. We suggest that some other enzyme (perhaps AdhP) initiates the degradation of ethanolamine (outside of carboxysomes), forming acetaldehyde by some disproportionation reaction in which two molecules of ethanolamine yield one acetaldehyde and a diamine or aminoamide that cannot be degraded. The requirement of the second pathway for polyphosphate kinase (Ppk) may reflect the fact that Ppk helps cells tolerate growth with reduced internal pH (24) and therefore may be required under the conditions used here to activate the second pathway.

Assigning a function to carboxysomes helps solve the broader question of why so many genes are required to degrade EA and 1,2-PDL. Seventeen genes support the conversion of ethanolamine (two carbons) to acetyl-CoA and 26 support the conversion of 1,2-PDL (three carbons) to propionyl-CoA. The need for 17 genes to degrade a two-carbon compound seems surprising, especially when one compares the pathway for ethanolamine to that for acetate. Both are two-carbon compounds that are converted to acetyl-CoA, yet acetate use involves only three proteins (Acs, Ack, and Pta), and only one (Acs) is required.

The need for more genes may be explained in part by the volatile intermediates in the EA and 1,2-PDL pathways, which requires the carboxysome structural genes—five in eut and seven in pdu (15, 20). In addition, the two alcohols are more reduced than acetate, making it necessary to get rid of excess reducing equivalents in some way. The use of a B12-catalyzed first step minimizes this problem by converting an alcohol to an aldehyde with release of H2O (and NH3 in the case of EA) and no reducing equivalents. However, B12 is both complex to produce and inherently reactive. Two eut genes (eutT and eutA) and four pdu genes contribute to the management of B12 (3, 18, 21, 33). Thus, the additional eut and pdu genes may be required to contain a volatile aldehyde intermediate and to allow the use of a reactive cofactor.


This work was supported in part by NIH grant GM34804.

We thank Thomas Fazzio, whose work led to the experiments presented here; Tom Bobik for supplying strains; and our laboratory colleagues Peter Anderson, Julianne Grose, Douglas Huseby, Chris Mace, and Marian Price-Carter, who contributed caveats, critiques, and camaraderie.


1. Badger, M. R., and G. D. Price. 2003. CO2 concentrating mechanisms in cyanobacteria: molecular components, their diversity and evolution. J. Exp. Bot. 54:609-622. [PubMed]
2. Berkowitz, D., J. M. Hushon, H. J. Whitfield, Jr., J. R. Roth, and B. N. Ames. 1968. Procedure for identifying nonsense mutations. J. Bacteriol. 96:215-220. [PMC free article] [PubMed]
3. Bobik, T., Y. Xu, R. Jeter, K. Otto, and J. R. Roth. 1997. Propanediol utilization genes (pdu) of Salmonella typhimurium: three genes for the propanediol dehydratase. J. Bacteriol. 179:6633-6639. [PMC free article] [PubMed]
4. Bobik, T. A., G. D. Havemann, R. J. Busch, D. S. Williams, and H. C. Aldrich. 1999. The propanediol utilization (pdu) operon of Salmonella enterica serovar Typhimurium LT2 includes genes necessary for formation of polyhedral organelles involved in coenzyme B12-dependent 1,2-propanediol degradation. J. Bacteriol. 181:5967-5975. [PMC free article] [PubMed]
5. Bradbeer, C. 1965. The clostridial fermentations of choline and ethanolamine. I. Preparation and properties of cell-free extracts. J. Biol. Chem. 240:4669-4674. [PubMed]
6. Bradbeer, C. 1965. The clostridial fermentations of choline and ethanolamine. II. Requirement for a cobamide coenzyme by an ethanolamine deaminase. J. Biol. Chem. 240:4675-4681. [PubMed]
7. Brinsmade, S. R., and J. C. Escalante-Semerena. 2004. The eutD gene of Salmonella enterica encodes a protein with phosphotransacetylase enzyme activity. J. Bacteriol. 186:1890-1892. [PMC free article] [PubMed]
8. Brinsmade, S. R., T. Paldon, and J. C. Escalante-Semerena. 2005. Minimal functions and physiological conditions required for growth of Salmonella enterica on ethanolamine in the absence of the metabolosome. J. Bacteriol. 187:8039-8046. [PMC free article] [PubMed]
9. Buan, N. R., S. J. Suh, and J. C. Escalante-Semerena. 2004. The eutT gene of Salmonella enterica encodes an oxygen-labile, metal-containing ATP:corrinoid adenosyltransferase enzyme. J. Bacteriol. 186:5708-5714. [PMC free article] [PubMed]
10. Cannon, G. C., C. E. Bradburne, H. C. Aldrich, S. H. Baker, S. Heinhorst, and J. M. Shively. 2001. Microcompartments in prokaryotes: carboxysomes and related polyhedra. Appl. Environ. Microbiol. 67:5351-5361. [PMC free article] [PubMed]
11. Court, D. L., J. A. Sawitzke, and L. C. Thomason. 2002. Genetic engineering using homologous recombination. Annu. Rev. Genet. 36:361-388. [PubMed]
12. Czymmek, K., J. Penrod, E. Kofoid, and J. R. Roth. Unpublished data.
13. Datsenko, K. A., and B. L. Wanner. 2000. One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc. Natl. Acad. Sci. USA 97:6640-6645. [PMC free article] [PubMed]
14. Davis, R. W., D. Botstein, and J. R. Roth. 1980. Advanced bacterial genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.
15. Havemann, G. D., and T. A. Bobik. 2003. Protein content of polyhedral organelles involved in coenzyme B12-dependent degradation of 1,2-propanediol in Salmonella enterica serovar Typhimurium LT2. J. Bacteriol. 185:5086-5095. [PMC free article] [PubMed]
16. Havemann, G. D., E. M. Sampson, and T. A. Bobik. 2002. PduA is a shell protein of polyhedral organelles involved in coenzyme B12-dependent degradation of 1,2-propanediol in Salmonella enterica serovar typhimurium LT2. J. Bacteriol. 184:1253-1261. [PMC free article] [PubMed]
17. Jeter, R., B. M. Olivera, and J. R. Roth. 1984. Salmonella typhimurium synthesizes cobalamin (vitamin B12) de novo under anaerobic growth conditions. J. Bacteriol. 159:206-216. [PMC free article] [PubMed]
18. Johnson, C. L. V. J., E. Pechonick, S. D. Park, G. D. Havemann, N. A. Leal, and T. A. Bobik. 2001. Functional genomic, biochemical, and genetic characterization of the Salmonella pduO gene, an ATP:cob(I)alamin adenosyltransferase gene. J. Bacteriol. 183:1577-1584. [PMC free article] [PubMed]
19. Kaplan, A., and L. Reinhold. 1999. CO2 Concentrating mechanisms in photosynthetic microorganisms. Annu. Rev. Plant Physiol. Plant Mol. Biol. 50:539-570. [PubMed]
20. Kofoid, E., C. Rappleye, I. Stojiljkovic, and J. Roth. 1999. The 17-gene ethanolamine (eut) operon of Salmonella typhimurium encodes five homologues of carboxysome shell proteins. J. Bacteriol. 181:5317-5329. [PMC free article] [PubMed]
21. Mori, K., T. Tobimatsu, T. Hara, and T. Toraya. 1997. Characterization, sequencing, and expression of the genes encoding a reactivating factor for glycerol-inactivated adenosylcobalamin-dependent diol dehydratase. J. Biol. Chem. 272:32034-32041. [PubMed]
22. Penrod, J., T. Fazzio, C. Mace, E. C. Kofoid, and J. R. Roth. Unpublished data.
23. Penrod, J. T., C. C. Mace, and J. R. Roth. 2004. A pH-sensitive function and phenotype: evidence that EutH facilitates diffusion of uncharged ethanolamine in Salmonella enterica. J. Bacteriol. 186:6885-6890. [PMC free article] [PubMed]
24. Price-Carter, M., T. G. Fazzio, E. I. Vallbona, and J. R. Roth. 2005. Polyphosphate kinase protects Salmonella enterica from weak organic acid stress. J. Bacteriol. 187:3088-3099. [PMC free article] [PubMed]
25. Price-Carter, M., J. Tingey, T. A. Bobik, and J. R. Roth. 2001. The alternative electron acceptor tetrathionate supports B12-dependent anaerobic growth of Salmonella enterica sevrovar Typhimurium on ethanolamine or 1,2-propanediol. J. Bacteriol. 183:2463-2475. [PMC free article] [PubMed]
26. Rondon, M. R., A. R. Horswill, and J. C. Escalante-Semerena. 1995. DNA polymerase I function is required for the utilization of ethanolamine, 1,2-propanediol, and propionate by Salmonella typhimurium LT2. J. Bacteriol. 177:7119-7124. [PMC free article] [PubMed]
27. Rondon, M. R., R. Kazmierczak, and J. C. Escalante-Semerena. 1995. Glutathione is required for maximal transcription of the cobalamin biosynthetic and 1,2-propanediol utilization (cob/pdu) regulon and for the catabolism of ethanolamine, 1,2-propanediol, and propionate in Salmonella typhimurium LT2. J. Bacteriol. 177:5434-5439. [PMC free article] [PubMed]
28. Roof, D. M., and J. R. Roth. 1992. Autogenous regulation of ethanolamine utilization by a transcriptional activator of the eut operon in Salmonella typhimurium. J. Bacteriol. 174:6634-6643. [PMC free article] [PubMed]
29. Roof, D. M., and J. R. Roth. 1988. Ethanolamine utilization in Salmonella typhimurium. J. Bacteriol. 170:3855-3863. [PMC free article] [PubMed]
30. Roof, D. M., and J. R. Roth. 1989. Functions required for vitamin B12-dependent ethanolamine utilization in Salmonella typhimurium. J. Bacteriol. 171:3316-3323. [PMC free article] [PubMed]
31. Roth, J. R., J. G. Lawrence, and T. A. Bobik. 1996. Cobalamin (coenzyme B12): synthesis and biological significance. Annu. Rev. Microbiol. 50:137-181. [PubMed]
32. Salmond, C. V., R. G. Kroll, and I. R. Booth. 1984. The effect of food preservatives on pH homeostasis in Escherichia coli. J. Gen. Microbiol. 130:2845-2850. [PubMed]
33. Sampson, E. M., C. L. Johnson, and T. A. Bobik. 2005. Biochemical evidence that the pduS gene encodes a bifunctional cobalamin reductase. Microbiology 151:1169-1177. [PubMed]
34. Scarlett, F. A., and J. M. Turner. 1976. Microbial metabolism of amino alcohols. Ethanolamine catabolism mediated by coenzyme B12-dependent ethanolamine ammonia-lyase in Escherichia coli and Klebsiella aerogenes. J. Gen. Microbiol. 95:173-176. [PubMed]
35. Schmieger, H. 1971. A method of detection of phage mutants with altered transducing ability. Mol. Gen. Genet. 110:378-381. [PubMed]
36. Sheppard, D., J. Penrod, T. Bobik, E. Kofoid, and J. Roth. 2004. Evidence that a B12-adenosyl transferase is encoded within the ethanolamine operon of Salmonella enterica. J. Bacteriol. 186:7635-7644. [PMC free article] [PubMed]
37. Sheppard, D. E., and J. R. Roth. 1994. A rationale for autoinduction of a transcriptional activator: ethanolamine ammonia-lyase (EutBC) and the operon activator (EutR) compete for adenosyl-cobalamin in Salmonella typhimurium. J. Bacteriol. 176:1287-1296. [PMC free article] [PubMed]
38. Shively, J., and R. English. 1991. The carboxysome, a prokaryotic organelle: a mini review. Can. J. Bot. 69:957-963.
39. Soda, K. 1967. A spectrophotometric microdetermination of keto acids with 3-methyl-2-benzothiazolone hydrazone. Agric. Biol. Chem. 31:1054-1060.
40. Starai, V. J., J. Garrity, and J. C. Escalante-Semerena. 2005. Acetate excretion during growth of Salmonella enterica on ethanolamine requires phosphotransacetylase (EutD) activity, and acetate recapture requires acetyl-CoA synthetase (Acs) and phosphotransacetylase (Pta) activities. Microbiology 151:3793-3801. [PubMed]
41. Stojiljkovic, I., A. J. Baumler, and F. Heffron. 1995. Ethanolamine utilization in Salmonella typhimurium: nucleotide sequence, protein expression, and mutational analysis of the cchA cchB eutE eutJ eutG eutH gene cluster. J. Bacteriol. 177:1357-1366. [PMC free article] [PubMed]
42. Toraya, T., T. Shirakashi, T. Kosuga, and S. Fukui. 1976. Substrate specificity of coenzyme B12-dependent diol dehydratase: glycerol as both a good substrate and a potant inactivator. Biochem. Biophys. Res. Commun. 69:475-480. [PubMed]
43. Yu, D., J. A. Sawitzke, H. Ellis, and D. L. Court. 2003. Recombineering with overlapping single-stranded DNA oligonucleotides: testing a recombination intermediate. Proc. Natl. Acad. Sci. USA 100:7207-7212. [PMC free article] [PubMed]

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


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...