Acetone Formation in the Vibrio Family: a New Pathway for Bacterial Leucine Catabolism

Abstract

There is current interest in biological sources of acetone, a volatile organic compound that impacts atmospheric chemistry. Here, we determined that leucine-dependent acetone formation is widespread in the Vibrionaceae. Sixteen Vibrio isolates, two Listonella species, and two Photobacterium angustum isolates produced acetone in the presence of l-leucine. Shewanella isolates produced much less acetone. Growth of Vibrio splendidus and P. angustum in a fermentor with controlled aeration revealed that acetone was produced after a lag in late logarithmic or stationary phase of growth, depending on the medium, and was not derived from acetoacetate by nonenzymatic decarboxylation in the medium. l-Leucine, but not d-leucine, was converted to acetone with a stoichiometry of approximately 0.61 mol of acetone per mol of l-leucine. Testing various potential leucine catabolites as precursors of acetone showed that only α-ketoisocaproate was efficiently converted by whole cells to acetone. Acetone production was blocked by a nitrogen atmosphere but not by electron transport inhibitors, suggesting that an oxygen-dependent reaction is required for leucine catabolism. Metabolic labeling with deuterated (isopropyl-d₇)-l-leucine revealed that the isopropyl carbons give rise to acetone with full retention of deuterium in each methyl group. These results suggest the operation of a new catabolic pathway for leucine in vibrios that is distinct from the 3-hydroxy-3-methylglutaryl-coenzyme A pathway seen in pseudomonads.

There is current interest in the role of acetone in atmospheric chemistry and in determining natural and anthropogenic sources of acetone. Acetone has been found in the upper troposphere and lower stratosphere in surprisingly large amounts and may be an important contributor to the formation of odd hydrogen radicals (OH plus HO₂) and the sequestration of nitrogen oxides as peroxyacetylnitrate (2, 38, 44). While some of the acetone found in the atmosphere results from photochemical reactions of natural and anthropogenic hydrocarbons, direct emissions from biological sources may be important (10, 45). The atmospheric oxidation of various biogenic hydrocarbons, such as 2-methyl-3-buten-2-ol and various monoterpenes, also gives rise to the secondary production of acetone (14, 27).

There are several known biological sources of acetone. Those best characterized include enzymatic decarboxylation of acetoacetate in certain bacteria, such as clostridia (16) and Bacillus polymyxa (24), and nonenzymatic decarboxylation of acetoacetate in animals (47). Other biogenic acetone sources of uncertain magnitude and mechanism include seedlings of oil seed plants (39), branches of various plants (22), conifer buds (31), and wounded pasture grasses and clover (10, 25). Recently, we described the production of acetone in marine Vibrio species (40). Acetone formation was dependent on the presence of l-leucine in the growth medium and was repressed by glucose. These findings suggested that an inducible leucine catabolic pathway, similar to that described for pseudomonads (37), might give rise to this ketone. In the pathway worked out in Pseudomonas putida (36), leucine oxidation occurs in the sequence shown in Fig. 1 (right), giving rise to acetoacetate and acetyl-coenzyme A (CoA). In P. putida, the acetoacetate is further metabolized to acetyl-CoA via succinyl-CoA transferase and acetoacetyl-CoA thiolase (49). Possibly, a similar leucine catabolic pathway may function in vibrios, but with enzymatic or nonenzymatic decarboxylation of the acetoacetate formed.

FIG. 1.

Metabolic pathway for leucine catabolism in Pseudomonas species (right, solid arrows; redrawn from reference 36) and possible pathways for leucine-dependent acetone formation in Vibrio species (left and right, dashed arrows). Pathways A and B are discussed in the text. The isopropyl moiety of leucine is shown fully deuterated to help trace the origin of acetone in each pathway as discussed in the text. Abbreviations: HIV, 3-hydroxyisovaleric acid; HMG-CoA, 3-hydroxy-3-methylglutaryl-CoA; IV-CoA, isovaleryl-CoA; α-KIC, α-ketoisocaproic acid; MC-CoA, 3-methylcrotonyl-CoA; MG-CoA, 3-methylglutaconyl-CoA; FAD, flavin adenine dinucleotide.

Other pathways for leucine catabolism in bacteria result in the production of either 3-methyl-1-aminobutane or 3-methylbutanal. Proteus vulgaris is known to have a neutral amino acid decarboxylase that is very active with l-leucine yielding 3-methyl-1-aminobutane (12). However, the metabolic role of this decarboxylation pathway is uncertain (5). Studies with P. vulgaris and Streptococcus lactis have also revealed significant metabolism of l-leucine via transamination to α-ketoisocaproic acid, followed by decarboxylation to 3-methylbutanal (33, 43). Whether 3-methyl-1-aminobutane and 3-methylbutanal are further metabolized in these bacteria is uncertain. Here we have examined the underlying mechanism of acetone formation in Vibrio, Listonella, and Photobacterium isolates metabolizing l-leucine. In contrast to what was found for the leucine-degradative pathway in pseudomonads and the acetoacetate decarboxylation pathway for acetone formation in bacilli and clostridia, it appears that these marine bacteria form acetone by a new type of pathway.

MATERIALS AND METHODS

Chemicals.

l-Leucine-5,5,5-d₃ (98 atom% deuterium [D]) was obtained from Cambridge Isotope Laboratories (Andover, Mass.). (Isopropyl-d₇) l-leucine (98.4 atom% D) was obtained from CDN Isotopes (Quebec, Canada). 3-Hydroxyisovaleric acid was obtained from TCI America (Portland, Oreg.). All of the other chemicals were of analytical grade and were obtained from Sigma Chemical Co. (St. Louis, Mo.) or Aldrich Chemicals (Milwaukee, Wis.).

Bacterial strains and media.

The sources of strains used here are summarized in Table 1. Additional marine Vibrio isolates were obtained from estuary sediments at two coastal sites in southern California (Del Mar and Los Penasquitos marsh) by plating on TCBS agar (35) as described previously (40). The strains were characterized as Vibrio splendidus by the methods described earlier (40) and the methods of Ortigosa et al. (41). All strains were maintained frozen at −70°C in 7% (vol/vol) dimethyl sulfoxide.

TABLE 1.

Vibrio family isolates used in this study and their ability to produce acetone in the presence of l-leucine

Strain(s)d	Origin or referencea	Acetone productionb in:		Verified byc
		MT glucose medium	MT leucine medium
V. harveyi 7OM23	41	5 (2.8)	991 (1.7)	HPLC, GC-MS
V. harveyi 7OM2, 7OM15, 9OM1, 9OM6, 9OM10, 9OM12	41	5–17	307–2,130
V. pelagius 7OM5	41	27 (1.0)	795 (0.8)
V. pelagius 9OM11	41	5 (2.6)	425 (0.9)	HPLC
V. splendidus D2	This study	3 (4.4), 17 (4.4)	168 (1.7), 1,060 (2.4)	HPLC
V. splendidus P5	This study	3–55 (3.9–6.6; n = 4)	453–4,320 (1.4–2.8; n = 5)	HPLC, GC-MS
V. splendidus 1OM9	41	20 (3.0)	577 (1.5)	HPLC
V. splendidus 1OM13	41	74 (0.2)	888 (0.2)
V. splendidus 33125	ATCC	3 (4.5)	145 (1.5)	HPLC
V. tubiashii 9OM7	41	7 (2.7)	427 (1.7)	HPLC
V. tubiashii 9SM11	41	3 (4.2)	709 (1.6)
L. anguillarum 19264	ATCC	22 (3.2)	400 (2.1)
L. pelagia 33784	ATCC	39 (6.2)	3,130 (1.8)	HPLC, GC-MS
P. angustum 25915	ATCC	388 (4.9)	3,300 (1.4)	HPLC, GC-MS
P. angustum 33977	ATCC	208 (3.9)	2,410 (1.9)
S. alga BrY	8	5 (4.2)	6 (7.1)
S. amazonensis SB2B		8 (3.0)	23 (2.8)
S. frigidimarina ACAM 591, ACAM 600	6	10–12	15–19
S. putrefaciens MR-1, MR-4	28	1	1–2
S. woodyi MS32	34	1 (6.3)	150 (4.0)	HPLC

^aATCC, American Type Culture Collection (Manassas, Va.). 

^bMedia and the method for acetone determination are described in Materials and Methods. Levels of nutrients: glucose, 2% (wt/vol); l-leucine, 20 mM. The data shown are the averages of duplicate determinations of acetone levels (micromolar) divided by the OD₆₀₀ of the culture. For experiments with individual strains the average OD₆₀₀ of the two cultures or range of OD₆₀₀ values for multiple determinations are shown in parentheses. 

^cMethod used to verify the identity of acetone produced in the culture. 

^dAbbreviations for genera: V., Vibrio; L., Listonella; P., Photobacterium; S., Shewanella. 

For routine growth a marine salts tryptone medium (MT) was used; this medium is identical to medium BM of Baumann and Baumann (3), but with ammonium chloride omitted and nitrogen and carbon sources provided by 1% (wt/vol) tryptone (Difco). In some experiments medium MT was supplemented with either 1% (wt/vol) glucose or 20 mM l-leucine as an additional carbon source.

For the analysis of acetone formation in a batch culture a New Brunswick fermentor (Edison, N.J.) was used. Typically, bacteria were cultured in a 1-liter volume of marine broth or medium MT leucine at 30°C with stirring (300 to 400 rpm) and aeration at 2 liters min⁻¹. For acetone analysis, 3-ml samples were removed for determination of optical density at 600 nm (OD₆₀₀), the cells were then removed by microcentrifugation (3 min at 10,000 × g), and acetone in the culture fluid was determined as described below.

For the growth of P. putida PpG2, a wild-type strain used to deduce the pseudomonad leucine catabolic pathway (36), two media were used: a mineral salts medium H (13), containing 20 mM l-leucine as the sole carbon source, and a tryptone leucine medium, containing medium H minus ammonium chloride, 1% tryptone, and 20 mM l-leucine.

Analytical methods.

Acetone in bacterial cultures was routinely assayed by a headspace method (gas chromatography-photoionization detection [GC-PID] method). Samples (0.5 to 1 ml) of cell-free medium, obtained by centrifugation, were placed in 4.8-ml vials sealed with Teflon-lined septa and allowed to equilibrate for 1 h at 30°C. A sample of the headspace (250 μl) was removed with a gas-tight syringe and injected onto a gas chromatograph (Photovac model 10S) equipped with a capillary column (CPSil-5; 6 ft) and a photoionization detector. The carrier gas was purified nitrogen (8 cm³ min⁻¹), and the column was typically operated at 23°C; under these conditions acetone typically eluted at 115 ± 5 s. For each experiment the headspace of fresh solutions of acetone ranging from 10 μM to 10 mM were sampled identically to establish a calibration curve. For verification of acetone formation, samples of culture supernatants were treated with 2,4-dinitrophenyl hydrazine (DNPH) and the resulting hydrazone derivative of acetone was analyzed by high-pressure liquid chromatography (HPLC) as described previously (40). For some cultures acetone formation was also verified by GC-mass spectrometry (MS) analysis (see below).

Acetoacetate formation was tested with commercially available acetoacetate decarboxylase (Wako Bioproducts, Richmond, Va.) by the method of Kimura et al. (23), except that acetone in the headspace of sealed vials was analyzed by the GC-PID method described above.

Bacterial conversion of l-leucine, deutero-l-leucine, and leucine metabolites to acetone.

For these experiments V. splendidus D2 or P5 cells were grown at room temperature on MT leucine medium to logarithmic phase (OD₆₀₀ of 0.5 to 0.7) and the cells were harvested by centrifugation and washed twice with marine salts (MT medium with tryptone deleted). Washed cells were suspended in approximately 1/10 the original volume of marine salts, adjusted to contain 2 mM l-leucine or potential leucine metabolites including 3-methyl-1-aminobutane, 3-methylbutanal, sodium 3-hydroxy-3-methylbutanoate, or sodium α-ketoisocaproate (each at 2 mM), and then incubated with shaking at room temperature for 1 to 3 h. Aliquots were taken for measurement of cell density (OD₆₀₀) and acetone content; for the latter, cells were removed by centrifugation, 0.5-ml samples of the supernatants were equilibrated in sealed vials for 1 h at 30°C, and then headspace acetone was determined as described above. In some experiments cells were incubated with potential inhibitors added in water, or in dimethylformamide (20 μl per 2 ml of incubation mixture) in the case of rotenone. For inhibition by a N₂ atmosphere, cell suspensions were bubbled with N₂ gas for 1 min before and after l-leucine or α-ketoisocaproate was added and the vial was quickly sealed with a Teflon-lined septum and then incubated and processed as described above.

In some experiments the incubation mixtures contained variable amounts of l-leucine and the culture fluid was analyzed to determine the degree of leucine conversion to acetone; some volatile acetone was lost during the incubation and manipulations. Acetone was quantified as described above, and l-leucine remaining in the culture media was determined by an enzymatic method (7) with leucine dehydrogenase obtained from Sigma Chemical Co.

In other experiments, sealed vials containing washed cells were adjusted to contain 2 mM deuterated l-leucine, either (5,5,5-d₃)-l-leucine or (isopropyl-d₇)-l-leucine, or nondeuterated l-leucine as a control. After incubation for 1 to 2 h at 23°C cells were removed by microcentrifugation at 10,000 × g, 0.5 to 1 ml was transferred to sealed vials, and then the headspace of each vial was analyzed by GC-MS. In initial experiments a Hewlett-Packard 5988A GC-MS system was used. For the experiments presented here we used the apparatus described in detail by Helmig et al. (18), except that the adsorbent cartridge was removed and samples were injected in a flow of helium into the freezeout trap (−175°C). Collected volatiles were then thermally desorbed at 75°C and injected onto a DB-1 column (0.32 mm by 60 m; 1-mm film thickness; J & W Scientific, Folsom, Calif.) programmed from 0 to 72°C, and eluting analytes were analyzed by a Hewlett-Packard MSD 5970 mass spectrometer.

RESULTS

Leucine-dependent acetone formation is widespread in the Vibrionaceae.

Previously we had isolated Vibrio spp. from estuarine samples and determined that acetone formation in these isolates was dependent on the presence of l-leucine and was repressed by glucose. These cultures did not survive storage, so new isolates were obtained from similar coastal estuary sites in southern California. Again, enrichment for Vibrio isolates by plating on TCBS agar (35) led to a high frequency of isolation of colonies that exhibit l-leucine-dependent acetone formation. Isolates D2 and P5 were further examined here. They were each identified as a Vibrio sp. by the methods we described previously (40) and were most similar to nonluminescent V. splendidus (phenon 3), as determined by the taxonomic methods of Ortigosa et al. (41).

During these experiments we wondered if acetone formation is widespread in the Vibrionaceae. We obtained a variety of characterized Vibrio species, some of which were reported to utilize l-leucine as the sole carbon source (41), and also obtained a few Listonella, Photobacterium, and Shewanella isolates; these genera are generally included in the Vibrionaceae (11, 32). These isolates and their ability to produce leucine-dependent acetone are summarized in Table 1. All of the Vibrio species tested produced detectable acetone in the presence of l-leucine, although the amount detected ranged widely from 145 to 4,320 μM/OD₆₀₀ unit. As discussed below, this variation was found to be due to the effect of the growth stage on leucine-dependent acetone formation. In each case acetone formation was almost completely absent in cells grown on marine broth supplemented with glucose as the carbon source (Table 1), consistent with our earlier results (40). Two Listonella species and two Photobacterium angustum strains also produced leucine-dependent acetone. Acetone formation was much more variable in Shewanella isolates; only Shewanella woodyi formed significant amounts of acetone under these growth conditions. Acetone formation in several of the bacterial strains was verified by derivatization with DNPH followed by HPLC analysis, and for four strains GC-MS analysis also confirmed the production of acetone (Table 1).

During these experiments we also analyzed cultures of P. putida PpG2 for production of acetone during growth on l-leucine; this strain was used to elucidate the leucine catabolic pathway shown in Fig. 1 (36). When grown on l-leucine as the sole carbon source or in a tryptone medium containing 20 mM l-leucine to mimic the medium used to culture marine vibrios, strain PpG2 produced little or no acetone (data not shown).

Effects of growth conditions on acetone formation in V. splendidus and P. angustum.

Previously we found that Vibrio isolates produced acetone primarily in the stationary phase of growth (40); however these experiments were complicated by the oxygen depletion that occurred during the incubation of culture samples in sealed vials. Here we grew V. splendidus P5 and P. angustum 25915 in a stirred fermentor under controlled aerobic conditions and at various times obtained samples, removed bacteria by centrifugation, and analyzed acetone in the resulting supernatants. Figure 2A illustrates that for V. splendidus P5 grown in marine broth, acetone was produced in stationary cells; some carryover of acetone from the inoculum was seen at early times. The decline in acetone after 8 h of culture can be attributed to volatilization of the ketone by the large volume of air passed through the culture. This experiment, repeated twice, also illustrated that acetone was not produced in response to a low pH, since the pH of the culture remained in the range of 7.7 to 7.9. To examine whether acetone formation might be due to excretion of acetoacetate into the medium with accompanying nonenzymatic decarboxylation, samples throughout the fermentation were analyzed for acetone in the presence and absence of acetoacetate decarboxylase (23). There was no significant difference in acetone levels in these samples (data not shown), indicating that acetone and not acetoacetate was released from the cells.

FIG. 2.

Acetone formation in V. splendidus P5 and P. angustum cultured in a stirred fermentor. (A) V. splendidus P5 grown in marine broth (conditions: 400 rpm, 2 liters of air min⁻¹, 30°C). (B) V. splendidus P5 grown in MT leucine broth (conditions: 300 rpm, 2 liters of air min⁻¹, 26°C). (C) P. angustum 25915 grown in MT leucine broth (conditions same as those for panel B). In each case the liquid volume was 1 liter and acetone in the culture fluid was determined from aliquots removed and analyzed as described in Materials and Methods.

Fermentor growth experiments with V. splendidus P5 and P. angustum grown in MT leucine broth in place of marine broth are shown in Fig. 2B and C. In these cases acetone was produced throughout the log phase of growth, although the rate of production showed an initial lag. The pH of the P. angustum culture was monitored, and, as with V. splendidus P5 in marine broth (Fig. 2A), the pH remained in the range of 7.7 to 7.9 (data not shown).

V. splendidus D2 and P5 were capable of anaerobic growth in marine broth and MT leucine broth with or without supplementation with nitrate. Analysis of these cultures and aerobic cultures showed that much more acetone was produced under aerobic conditions. For example, anaerobic cultures of V. splendidus P5 in MT leucine broth produced on average <1% of the acetone found in parallel aerobic cultures of similar cell density. These results indicate that acetone formation is not a result of fermentative reactions.

Specificity and stoichiometry of l-leucine conversion to acetone.

Since leucine-degrading pseudomonads can metabolize either d- or l-leucine (36), we tested whether this was also true for Vibrio isolates. V. splendidus P5 was cultured in MT leucine medium, and log-phase cells were washed twice with marine salts and resuspended in marine salts with 2 mM leucine (d, l, or dl isomers). Acetone formation at 1 and 2.5 h was measured, and only the l isomer and the dl racemate were metabolized in this Vibrio strain. No acetone was detected in cells incubated with the d isomer.

To determine the fraction of l-leucine metabolized that was converted to acetone, washed cells were incubated with l-leucine and the disappearance of leucine was quantified with leucine dehydrogenase (7) and the appearance of acetone was quantified by the GC-PID method. The results are shown in Table 2. In the series of experiments shown, all or a large fraction of the added leucine was taken up by the bacteria by 3 h, and 45 to 84% (average of 55%) of the l-leucine metabolized was recovered as acetone. Since some acetone (about 10%) was lost by volatilization during the incubations and handling of the samples, acetone formation was underestimated. The corrected average conversion of l-leucine to acetone was 61%. It should be mentioned that sealed incubations were complicated by semianaerobic conditions that developed and inhibited acetone formation.

TABLE 2.

A large fraction of l-leucine metabolized by V. splendidus P5 is converted to acetone^a

Initial l-leucine concn (μM)	Products at:
	1 h			3 h
	Leucine uptake (μM)	Acetone produced (μM)	% Conversion	Leucine uptake (μM)	Acetone produced (μM)	% Conversion
500	500	236	47	500	257	51
500	500	242	48	500	253	51
1,000	600	267	45	1,000	494	49
1,000	480	287	60	970	555	57
2,000	380	320	84	1,610	827	51
2,000	620	319	51	1,280	829	65

^al-Leucine uptake and conversion to acetone by V. splendidus P5 is described in Materials and Methods. Two independent experiments are shown; each data point is the average of 2 to 5 determinations. 

Bacterial conversion of putative leucine metabolites to acetone.

It was expected that vibrios might degrade leucine by the well-known pseudomonad pathway, leading to the production of acetoacetic acid and acetyl-CoA (Fig. 1). However, to explain acetone formation in this case, a plausible pathway would involve acetoacetate decarboxylation to acetone rather than acetoacetic acid metabolism by the glyoxylate pathway (36). As shown above, we had ruled out the secretion of acetoacetic acid into the medium. Several attempts to demonstrate acetoacetic acid or acetoacetate decarboxylase in extracts of V. splendidus D2 or P5, by the methods of Kimura et al. (24), were unsuccessful.

Other bacterial pathways for l-leucine catabolism are described in the literature. For example, Proteus vulgaris is known to metabolize l-leucine by decarboxylation to 3-methyl-1-aminobutane or by transamination to α-ketoisocaproic acid, which can then be decarboxylated to 3-methylbutanal (12, 43). To test the possibility that these l-leucine metabolites might be converted to acetone by vibrios, we incubated cells with each and determined if they would give rise to acetone. In addition, as indicated from work on l-leucine catabolism in Galactomyces reesii (29), 3-hydroxyisovaleric acid is a likely intermediate (Fig. 1), so it was also tested. Of the metabolites tested, only α-ketoisocaproic acid, the product of transamination of leucine, was a precursor of acetone in V. splendidus P5, Listonella pelagia 33784, and P. angustum 25915. For example, with V. splendidus P5 acetone formation in washed cells in two duplicate experiments was 1,327 to 1,561 μM/h/OD₆₀₀ unit with l-leucine and 1,314 to 1,462 μM/h/OD₆₀₀ unit with sodium α-ketoisocaproate. Neither 3-methyl-1-aminobutane, 3-methylbutanal, nor 3-hydroxyisovaleric acid stimulated acetone formation in any of these three strains, and in addition, with V. splendidus P5, neither isovaleric acid nor 3-methylcrotonic acid, precursors of intermediates in the 3-methylcrotonyl-CoA pathway (Fig. 1), was converted to acetone (data not shown).

Effects of inhibitors of leucine-dependent acetone formation.

To gain some insight into the metabolic pathway involved in l-leucine conversion to acetone the effects of the following metabolic inhibitors were tested: azide, rotenone and a nitrogen atmosphere (to block electron transport), and arsenite (to block branched-chain keto acid dehydrogenase) (9, 17). These experiments were conducted with V. splendidus P5, P. angustum, and L. pelagia (Table 3). The most potent inhibition was seen with removal of oxygen (i.e., a nitrogen atmosphere); for all three bacterial species inhibition ranged from 92 to 97% in different experiments. Since sodium azide and rotenone, electron transport inhibitors, had little or no inhibitory effect, these findings suggest that the oxygen dependence of acetone formation may be indicative of an oxidase or oxygenase reaction. Inhibition by arsenite ranged from 41 to 65%, which is consistent with inhibition of a lipoamide-dependent keto acid dehydrogenase of the type used by P. putida PpG2 (46) in the conversion of α-ketoisocaproic acid to isovaleryl-CoA (Fig. 1). Very similar results were seen with V. splendidus P5 incubated with α-ketoisocaproic acid in place of l-leucine.

TABLE 3.

Effect of potential inhibitors on acetone formation from l-leucine in Listonella, Photobacterium, and Vibrio strains^a

Bacterial strain	Addition	Acetone formed μM/h/ OD600 unitb	% Inhibition
L. pelagia 33784	Na arsenite	369 (625)	41
	N2 atmosphere	51 (625)	92
P. angustum 25915	Na arsenite	508 (1,090)	53
	N2 atmosphere	50 (1,090)	95
V. splendidus P5	Na arsenite	258 (747)	65
	N2 atmosphere	55 (747)	93
	Rotenone (0.1 mM)	1,140 (1,100)	0
	Na azide (1 mM)	1,510 (1,430)	0
	N2 atmosphere	45 (1,430)	97

^aAll incubation mixtures included 2 mM l-leucine, and the values shown are the averages of duplicate determinations. Each experiment was repeated on a separate occasion with very similar results. 

^bThe values in parentheses show the production of acetone in control reaction mixtures without added inhibitors. The range of acetone production with V. splendidus P5 in different control experiments, 296 to 1,430 μM acetone/h/OD₆₀₀ unit, is a result of growth phase differences, as discussed in the text. 

Bacterial conversion of deutero-l-leucine to acetone.

In order to directly test whether acetone formation occurs by the pseudomonad pathway or some other pathway, (isopropyl-d₇)-l-leucine with a label in the isopropyl moiety was incubated with washed cells that had been pregrown on MT leucine broth. Deuterated acetone that accumulated in the headspace was analyzed by GC-MS; control incubation mixtures contained nondeuterated l-leucine. These experiments were conducted with L. pelagia, P. angustum, and V. splendidus strains, and the results are summarized in Table 4. The positions of the deuterium labels and expected labeled products of the pseudomonad and vibrio pathways are shown in Fig. 1. For all three bacterial strains the acetone produced by incubation with (isopropyl-d₇)-l-leucine led to >90% enrichment of the molecular ion (m/z 64) over the nondeuterated acetone (m/z 58). This result is indicative of the retention of all six methyl deuterium atoms in the acetone produced. This labeling pattern is inconsistent with metabolism of (isopropyl-d₇)-l-leucine by the pseudomonad pathway (ignoring deuterium isotope effects), which would be expected to remove a methyl deuterium during the 3-methylcrotonyl-CoA carboxylase reaction, resulting in the eventual formation of acetoacetate with five deuterium atoms (Fig. 1). Decarboxylation of acetoacetate (enzymatic or nonenzymatic) would produce acetone with five deuterium atoms (molecular ion m/z 63). In the experiments shown in Table 4 no significant amount of molecular ion at m/z 63 was seen. Analysis of the major acetone fragment ion due to loss of methyl would be expected to give m/z 43 for the nondeuterated fragment and m/z 46 for the trideuterated fragment; only the latter pattern was seen for all three strains of Vibrionaceae (Table 4).

TABLE 4.

GC-MS analysis of acetone formed by Listonella, Photobacterium, and Vibrio from l-leucine and d₇-(isopropyl-d₇)-l-leucine

Bacterial strain	Leucine added	Relative abundance (%) of ionsb at:
		m/z 43 (CH3CO+)	m/z 45 (CHD2CO+)	m/z 46 (CD3CO+)	m/z 58 (CH3COCH3+)	m/z 63 (CHD2COCD3+)	m/z 64 (CD3COCD3+)
L. pelagia 33784	l-Leucine	100 (2.98 × 106)	NDc	ND	ND	ND	ND
L. pelagia 33784	d7-l-Leucinea	ND	ND	100 (2.34 × 106)	0.2	ND	17.9
P. angustum 25915	l-Leucine	100 (2.48 × 106)	ND	ND	17.5	ND	ND
P. angustum 25915	d7-l-Leucine	1.0	ND	100 (2.06 × 106)	ND	ND	17.3
V. splendidus P5	l-Leucine	100 (0.87 × 106)	ND	ND	17.5	ND	ND
V. splendidus P5	d7-l-Leucine	1.0	ND	100 (2.26 × 106)	ND	ND	17.3

^ad₇-l-leucine, d₇-(isopropyl-d₇)-L-leucine. 

^bValues in parentheses are the ion counts for the most abundant ion. 

^cND, not detected. 

In separate experiments, incubation of washed V. splendidus D2 or P5 cells with (5,5,5-d₃)-l-leucine with three deuterium atoms in one methyl group led to the formation of acetone with a primary molecular ion of m/z 51, indicative of retention of all methyl deuteriums (data not shown). This result is also consistent with formation of acetone from l-leucine by a pathway distinct from that seen in pseudomonads (Fig. 1).

DISCUSSION

We have been interested in identifying new biological sources of the volatile organic compound acetone, which, as described in the introduction, has a significant impact on the chemistry of the atmosphere. Here we have extended our earlier finding that marine vibrios produce acetone in the presence of l-leucine (40). Now we show that this metabolic trait is widespread in the Vibrio genus and other Vibrionaceae, such as Listonella and Photobacterium. However, Shewanella isolates, which some include in the Vibrionaceae (11), produced little or no acetone in the presence of l-leucine. Perhaps these results are indicative of differences in metabolism in aerobic versus anaerobic environments. The metabolism of l-leucine was shown here to be an aerobic process. Shewanella grows naturally in anaerobic environments (6, 28).

Given these results it is possible that leucine catabolism in marine environments gives rise to some of the acetone found in seawater. Small amounts of acetone at concentrations of 3 to 50 nM have been measured in open seawater (50); however these levels of the compound might be derived from partitioning with atmospheric acetone (4). Measurement of acetone in estuaries which contain abundant Vibrio populations (41) is complicated by the presence of anthropogenic sources of the ketone (50). At this point it is not possible to predict whether bacterial conversion of leucine in marine systems would be a significant contributor to atmospheric acetone.

When acetone formation in vibrios was discovered to be dependent on the presence of l-leucine, we assumed that the pathway was similar to those seen in animals, plants, and pseudomonads, involving 3-methylcrotonyl-CoA as an intermediate (Fig. 1) (1, 37). In animals l-leucine is a ketogenic amino acid that is metabolized to acetoacetate, which can break down nonenzymatically to produce acetone (47). However, we could not detect substantial pools of acetoacetate or acetoacetate decarboxylase in V. splendidus cultures. To further determine if the 3-methylcrotonyl-CoA-type pathway is operative in vibrios, we incubated cells with a deuterated l-leucine substrate, (isopropyl-d₇)-l-leucine, containing fully deuterated methyl groups. If acetone is derived from the isopropyl end of the leucine molecule, one would expect to see acetone with five deuterium atoms since a proton is removed at the 3-methylcrotonyl-CoA carboxylase step (Fig. 1). This was not the case for L. pelagia 33784, P. angustum 25915, or V. splendidus P5, and instead the acetone produced from (isopropyl-d₇)-l-leucine retained all six methyl group deuterium atoms (Table 4).

For this metabolic transformation to occur, an oxygen atom must be introduced at C-4 of the leucine carbon skeleton to provide the keto oxygen of the resulting acetone. There are two precedents for the transformation of leucine to metabolites with this type of oxygen substitution; both involve α-ketoisocaproic acid as an intermediate. First, transformation might occur as a result of an α-ketoisocaproate dioxygenase reaction, such as that described for rat liver (42). A putative bacterial dioxygenase would convert α-ketoisocaproic acid to 3-hydroxyisovaleric acid as shown in Fig. 1 (pathway A), and subsequently its oxidative cleavage, probably as the acyl-CoA derivative, could produce acetone and acetyl-CoA. Acetone would be released as a by-product, and the acetyl-CoA would be metabolized by the acetate assimilation pathway known to occur in most free-living vibrios (19). Consistent with pathway A is the finding that production of acetone from l-leucine or α-ketoisocaproic acid was dependent on oxygen but not blocked by electron transport inhibitors. Although an α-ketoisocaproate dioxygenase has apparently not been demonstrated in bacteria, it is noteworthy that a related enzyme, 4-hydroxyphenylpyruvate dioxygenase, has been described for the Vibrionaceae, V. cholerae, and Shewanella colwelliana (26). Both enzymes catalyze a similar reaction: the atoms of molecular oxygen are introduced into both the α-keto carbon and the side chain accompanied by decarboxylation of the keto acid (30, 42). The failure of 3-hydroxyisovaleric acid to support acetone formation in whole cells is inconsistent with pathway A, although this result might be explained by lack of uptake of the acid.

A second scheme for introduction of the requisite oxygen atom would involve production of isovaleryl-CoA, and 3-methylcrotonyl-CoA as in the pseudomonad leucine catabolic pathway (37), followed by a hydration reaction, such as that catalyzed by enoyl-CoA hydratase, resulting in the formation of 3-hydroxyisovaleryl-CoA (Fig. 1, pathway B). The latter metabolic sequence has recently been demonstrated in the yeast G. reesii, which converts isovaleric acid via isovaleryl-CoA and 3-methylcrotonyl-CoA to 3-hydroxyisovaleryl-CoA (29). As in pathway A, 3-hydroxyisovaleryl-CoA could be cleaved to produce acetone and acetyl-CoA. The operation of pathway B is supported by the finding that acetone production was inhibited by arsenite, a known inhibitor of lipoic acid-dependent enzymes such as branched-chain keto acid dehydrogenase; however, arsenite is not very specific, inhibiting a variety of enzymes (48). The role of oxygen in acetone formation in pathway B could be explained by a branched-chain acyl-CoA oxidase reaction. This type of enzyme has been described in relation to leucine catabolism in plant peroxisomes (15); in the Pseudomonas leucine degradation pathway isovaleryl-CoA is oxidized by a flavin-dependent acyl-CoA dehydrogenase (37) (Fig. 1).

Clarification of the leucine degradation pathway in vibrios will require demonstration of key enzymes such as an α-ketoisocaproate dioxygenase, isovaleryl-CoA oxidase, and the putative 3-hydroxyisovaleryl-CoA lyase. The last enzyme would use a β-lyase mechanism analogous to the well-known 3-hydroxy-3-methylglutaryl-CoA lyase, which catalyzes a Claisen-type cleavage reaction (Fig. 1) (20). However, this type of lyase has apparently not been previously described. We have noted that growth of V. splendidus P5 on leucine but not glucose led to the induction of isocitrate lyase (data not shown). This finding is consistent with the idea that leucine catabolism gives rise to an acetate equivalent that is processed by the glyoxylate pathway known to exist in Vibrio species (21).

Clarification of the vibrio leucine degradation pathway is in progress. It will be interesting to see if this new pathway is present in other biological systems, such as the buds, stems, and leaves of plants, that can produce significant amounts of acetone by an unknown mechanism (10, 22, 25, 31, 39).