β-Alanine Biosynthesis in Methanocaldococcus jannaschii

Abstract

One efficient approach to assigning function to unannotated genes is to establish the enzymes that are missing in known biosynthetic pathways. One group of such pathways is those involved in coenzyme biosynthesis. In the case of the methanogenic archaeon Methanocaldococcus jannaschii as well as most methanogens, none of the expected enzymes for the biosynthesis of the β-alanine and pantoic acid moieties required for coenzyme A are annotated. To identify the gene(s) for β-alanine biosynthesis, we have established the pathway for the formation of β-alanine in this organism after experimentally eliminating other known and proposed pathways to β-alanine from malonate semialdehyde, l-alanine, spermine, dihydrouracil, and acryloyl-coenzyme A (CoA). Our data showed that the decarboxylation of aspartate was the only source of β-alanine in cell extracts of M. jannaschii. Unlike other prokaryotes where the enzyme producing β-alanine from l-aspartate is a pyruvoyl-containing l-aspartate decarboxylase (PanD), the enzyme in M. jannaschii is a pyridoxal phosphate (PLP)-dependent l-aspartate decarboxylase encoded by MJ0050, the same enzyme that was found to decarboxylate tyrosine for methanofuran biosynthesis. A K_m of ∼0.80 mM for l-aspartate with a specific activity of 0.09 μmol min⁻¹ mg⁻¹ at 70°C for the decarboxylation of l-aspartate was measured for the recombinant enzyme. The MJ0050 gene was also demonstrated to complement the Escherichia coli panD deletion mutant cells, in which panD encoding aspartate decarboxylase in E. coli had been knocked out, thus confirming the function of this gene in vivo.

INTRODUCTION

Coenzyme A (CoA) is an important coenzyme in all known living organisms where it functions as an acyl carrier for amide-, ester-, and thioester-forming reactions as well as activating carbonyl groups for Claisen condensation reactions. Its biosynthetic pathway from the primary metabolites l-aspartate, α-ketoisovalerate, the methylene group of 5,10-methylene-tetrahydrofolate, cysteine, and ATP has now been completely defined in bacteria and is known to require nine enzymes (Fig. 1) (1). Recently, an alternate pathway for the biosynthesis of d-4′-phosphopantothenate was identified in Thermococcus kodakaraensis (2). Here pantoic acid is first phosphorylated to form 4-phosphopantoic acid that in turn is coupled with β-alanine to form d-4′-phosphopantothenate (Fig. 1, most archaea). In most archaea, the order of condensation of the β-alanine and phosphorylation reactions have been found to be reversed compared to the classical pathways in bacteria (2). This discovery led to the identification of the pantoate kinase (MJ0969) and the phosphopantothenate synthetase (MJ0209) as the genes encoding these enzymes in Methanocaldococcus jannaschii. At present, the activities of the bifunctional phosphopantothenoylcysteine synthase encoded by MJ0913 (3), the phosphopantotheine adenylyltransferase (MJ1030) from Pyrococcus abyssi (4), the pantoate kinase (MJ0969) from Thermococcus kodakaraensis (2, 5), the phosphopantothenate synthetase (MJ0209) from both Thermococcus kodakaraensis (6) and Methanospirillum hungatei (7) have been confirmed experimentally.

FIG 1.

Pathways for the biosynthesis of coenzyme A in most archaea and bacteria. H₄MPT, tetrahydromethanopterin; PoK, pantoate kinase; PPS, phosphopantothenate synthetase.

With this inventory of genes we are left with at least three genes/enzymes for CoA biosynthesis that needed to be identified in the methanogens. These include the enzymes for β-alanine and pantoic acid production. PanD, a pyruvoyl-containing enzyme, is responsible for catalyzing the decarboxylation of aspartate to β-alanine in bacteria (8). Although pyruvoyl-containing enzymes are known in the methanogens (9, 10), no PanD homologs are present in methanogenic genomes. The canonical enzymes for pantoic acid biosynthesis require 5,10-methylene-tetrahydrofolate, which is absent in most archaeal methanogens (11–13). A likely alternate substrate for pantoic acid biosynthesis in the methanogens would be 5,10-methylene-tetrahydromethanopterin, since this cofactor can function in an analogous manner as 5,10-methylene-tetrahydrofolate as demonstrated by its involvement in serine metabolism using the archaeal version of serine hydroxymethyltransferase (14).

Figure 2 shows the six known routes or pathways for the biosynthesis of β-alanine and include the following: pathway 1, the decarboxylation of l-aspartate (15); pathway 2, the transamination between malonate semialdehyde and l-glutamate (16) or l-alanine (17); pathway 3, the hydrolysis of dihydrouracil produced by the hydrogenation of uracil (18); pathway 4, the oxidative cleavage of spermine to 3-aminopropanal followed by the oxidation of the aldehyde of this molecule to a carboxylic acid (19); pathway 5, the action of an 2,3-aminomutase on alanine (20); and pathway 6, the addition of ammonia to acryloyl-CoA, followed by the hydrolysis of the CoA thioester (21). In each of these cases, except the transamination reaction, the enzyme(s) involved in catalyzing each specific reaction is known. However, no genes in the genome of M. jannaschii are annotated as encoding enzymes to catalyze any of these reactions. As has been seen before in M. jannaschii, it is possible that a nonorthogonal gene(s) could be responsible for catalyzing the reactions.

FIG 2.

Known pathways for β-alanine biosynthesis. The six known pathways are shown.

Tomita and coworkers have recently reported a glutamate decarboxylase homolog from T. kodakaraensis that can function as an aspartate decarboxylase, producing the β-alanine necessary for CoA biosynthesis in this organism (22). One of the roles of the glutamate decarboxylase homolog from M. jannaschii (MfnA) is to produce the tyramine for methanofuran biosynthesis (23). Recently, the homolog in Methanococcus maripaludis (MMP0131) was demonstrated to be an essential gene (24). However, the involvement of MfnA or other possible pathways leading to β-alanine is still unclear in methanogens. Therefore, in order to address the source of the β-alanine moiety present in coenzyme A, each of these known pathways to β-alanine was tested in this study. Here we show that β-alanine in M. jannaschii is produced only by the α-decarboxylation of l-aspartate. We also demonstrated that the pyridoxal phosphate (PLP)-dependent tyrosine decarboxylase (MfnA) in M. jannaschii, encoded by the MJ0050 gene, was able to catalyze the α-decarboxylation of l-aspartate to produce β-alanine in both in vivo and in vitro studies, consistent with the observations from T. kodakaraensis.

MATERIALS AND METHODS

Chemicals.

Pyruvate, dihydrouracil, formaldehyde, NADH, NADPH, spermine, l-aspartic acid, propionyl-CoA, β-alanine, [¹³C₃-¹⁵N₁]β-alanine, [2,3,3-²H₃]acrylic acid, sodium [1,2,3-¹³C₃]pyruvate, ¹⁵NH₄Cl, acryloyl chloride, trifluoroacetic anhydride, methyl 3,3-dimethoxypropionate, and thiamine were obtained from Sigma-Aldrich. M9 minimal salt was purchased from Fisher Scientific. [1,2,3,4-¹³C₄]l-aspartate was obtained from Cambridge Isotopes.

Synthesis of precursors.

Malonate semialdehyde was prepared by the acid-catalyzed hydrolysis of methyl 3,3-dimethoxypropionate (25). Acryloyl-CoA was prepared by the condensation of acryloyl chloride with coenzyme A as previously described (26). [2,3,3-²H₃]acryloyl-coenzyme A was prepared by coupling the mixed anhydride formed between [2,3,3-²H₃]acrylic acid and ethyl chloroformate with coenzyme A (27). Electrospray ionization-mass spectrometry (MS) of the acryloyl-CoA (28) showed the expected MH⁺ = 822 m/z where the [2,3,3-²H₃]acryloyl-CoA showed MH⁺ = 825 m/z. [¹⁵N₁]β-alanine was prepared by heating a aqueous solution of ¹⁵NH₄Cl and acrylic acid under basic conditions for 24 h at 110°C. [2,2′-²D₂]β-alanine was prepared by heating β-alanine dissolved in 98% deuterated sulfuric acid for 24 h at 110°C.

Cell extracts of M. jannaschii.

Cell extracts of Methanocaldococcus jannaschii were prepared by sonication of cell pellets under argon and stored under anaerobic conditions at −20°C as previously described (29). These cell extracts contained ∼30 mg/ml protein. The buffer used in cell extraction was 50 mM N-tris(hydroxymethyl)methyl-2-aminoethanesulfonic acid (TES)/K⁺ (pH 7.5), 10 mM MgCl₂, 20 mM dithiothreitol (DTT), kept under argon.

Incubation of M. jannaschii cell extracts with precursors.

Anaerobically prepared and stored M. jannaschii cell extracts (50 to 200 μl) were placed in 0.5-ml centrifuge tubes that were flushed with argon. Then 5- to 10-μl portions of aqueous anaerobic solutions of the precursors were added to the cell extracts. In the standard procedure, 100 μl of M. jannaschii cell extract was incubated with the indicated substrates under argon at 70°C. In order to test for reactions 1 to 5 in Fig. 2, cell extracts were incubated with 9.1 mM [1,2,3,4-¹³C₄]l-aspartate; 4.2 mM malonate semialdehyde in the presence of NADH, and NADPH; 1 mM dihydrouracil; 10 mM spermine; or 10 mM α-alanine, respectively, at 70°C for 30 min. In order to test reaction 6 in Fig. 2, cell extracts were incubated with 4.8 mM acryloyl-CoA, 4.8 mM acryloyl-CoA and 23 mM ¹⁵NH₄⁺, or 4.8 mM ²H₃-acryloyl-CoA and 2.4 mM aspartate at 70°C for 15 min.

Isolation, derivatization, and quantitation of the β-alanine present in cell extracts.

To isolate the β-alanine for gas chromatography (GC)-MS analysis, the proteins in the cell extracts were first precipitated, and the β-alanine was purified from the soluble fraction. The resulting purified β-alanine was then converted into a volatile derivative and assayed by GC-MS. In order to accurately quantify the amount of β-alanine in the sample by isotope dilution analysis, 5 μl of 10 mM [¹³C₃-¹⁵N₁]β-alanine was added to the samples at the end of the incubation and prior to the precipitation step. The proteins were precipitated either by the addition to the incubation mixture of 2 volumes of methanol or the addition of 0.1 volume of 2 M trichloroacetic acid (TCA) in water. After centrifugation, the clear extracts were separated and diluted threefold with water. The combined soluble material was then placed on a Dowex 50-8X H⁺ column (2 by 5 mm) that was first washed with 0.5 ml water, and then the amino acids were eluted with 6 M ammonia (0.2 ml). For the samples requiring hydrolysis of the CoA thioesters, the samples were made 0.2 M in NaOH and then heated for 10 min at 100°C before being purified on a Dowex 50-8X H⁺ column (0.4 by 2 cm). The β-alanine-containing sample was further purified by applying the eluted sample dissolved in water to a Dowex 50-8X H⁺ column (0.4 by 2 cm) and eluting with a hydrochloric acid gradient. This gradient consisted of 1 ml of 0.1 M HCl, 2 ml of 0.3 M HCl, another 2 ml of 0.3 M HCl, and 1 ml of 0.6 M HCl. β-Alanine was eluted in the third and fourth fraction. The position of elution was determined by the elution of a known sample of β-alanine under the same conditions. Following elution of β-alanine, the ammonia or hydrochloric acid was evaporated from the purified samples with a stream of nitrogen gas, and β-alanine was converted into either the methyl or n-butyl trifluoroacetyl derivatives as previously described (30). The derivatized samples were further purified by preparative thin-layer chromatography (TLC) prior to GC-MS analysis, using the solvent system of methylene chloride-methyl acetate (97.5:2.5 [vol/vol]). The R_f of the methyl ester trifluoroacetyl derivative (trifluoroacetyl methyl ester [TM] of l-alanine) was 0.28, and that of the butyl trifluoroacetyl derivative (trifluoroacetyl butyl ester [TB] of l-alanine) was 0.33. The derivatives of other amino acids were well separated from the β-alanine derivatives in the GC separation and did not interfere with the measurement of isotopic abundances by GC-MS. The specific procedures used in each experiment are indicated in Table 1.

TABLE 1.

Measured concentrations of β-alanine in cell extracts incubated with possible precursors

Expt no. and possible precursor(s)	Measured ratio of the 168/172 m/z ion intensities in the β-alanine peak	Concn of β-alanine in extract (μM)
1. None (control)a	0.022	10
2. 9.1 mM [1,2,3,4-13C4]l-aspartateb (pathway 1)	0.026	12
3. 4.8 mM Acryloyl-CoAc	0.14	>60
4. 4.8 mM Acryloyl-CoA + 24 mM 15NH4+c,d (pathway 6)	0.12	60
5. 4.8 mM 2H3-acryloyl-CoA + 2.4 mM l-aspartatee	0.38	>60

^aβ-Alanine was also purified by elution from a Dowex 50W-8X H⁺ column with an HCl gradient. The TM derivative of the β-alanine was not purified by preparative TLC. The derivative was the methyl ester, and the concentration was calculated from the 167 and 171 m/z ion intensities.

^bThe TM derivative of the β-alanine was not purified by preparative TLC.

^cβ-Alanine was purified by preparative TLC. The sample was incubated at 70°C for only 15 min.

^dNo evidence was observed for the incorporation of any ¹⁵N into the β-alanine on the basis of the measured intensity of the 168 + 1 m/z ion.

^eNo evidence was observed for the incorporation of any deuterium into the β-alanine on the basis of the measured intensity of the 168 + 3 m/z ion.

TLC analysis and purification of compounds.

The TLC solvent used for the analysis of amino acids consisted of acetonitrile-water-formic acid (88%) (19:2:1 [vol/vol/vol]). In this solvent system, the following compounds had the following R_fs: CoA, 0.00; aspartate, 0.30; N-(2-carboxyethyl)-l-aspartate, 0.30; α-glutamate, 0.40; β-glutamate, 0.40; β-alanine, 0.45; and pyruvate, 0.53. The amino acids were detected by spraying the plate with 0.2% ninhydrin in ethanol and heating the TLC plate for 5 min at 150°C.

GC-MS analysis of samples.

GC-MS analyses of the samples containing derivatized β-alanine were obtained using a VG-70-EHF gas chromatograph-mass spectrometer operating at 70 eV and equipped with an RTX-5M5 column (0.32 mm by 30 m) that was programmed from 80°C to 280°C at 8°C per min. Under the GC-MS conditions used, the indicated derivatives of the following compounds had the following retention times (in seconds) (shown in parentheses) and mass spectral data (shown in brackets) [molecular weight, base peak {shown underlined}, the most abundant ions with masses over 100 m/z listed in order of decreasing intensities]: trifluoroacetyl methyl ester of l-alanine (375) [199, 140, 168], trifluoroacetyl methyl ester of β-alanine (579) [199, 139, 167, 126, 168], and trifluoroacetyl butyl ester of β-alanine (1101) [241, 168, 139, 126, 186]. The ratios of the intensities for the 168, 139, 126, and 186 m/z ions for the trifluoroacetyl butyl ester derivative of β-alanine were 100, 45, 21, and 23%, respectively. The ratios of the intensities for the 139, 167, 126, and 168 m/z ions for the trifluoroacetyl methyl ester derivative of β-alanine were 100, 50, 42, and 33%, respectively.

Measurement of K_m and specific activity of M. jannaschii tyrosine decarboxylase (MfnA) with l-aspartate as the substrate.

MfnA (gene product of MJ0050) was recombinantly produced and purified as previously described (23). In a 50-μl reaction mixture volume, 9.3 μg of MfnA was incubated with l-aspartate (final concentration of 0.2 to 10 mM) in 0.1 M TES buffer (pH 8.0) for 10 min at 70°C, and the reaction was stopped by the addition of 5 μl of 2 M trichloroacetic acid followed by the addition of 4 μl of 1.3 mM [¹³C₃-¹⁵N₁]β-alanine. β-Alanine was isolated from the sample as described above, and the amount of β-alanine produced at each substrate concentration was measured by isotopic dilution analysis. The ratios of the labeled and unlabeled β-alanine were measured from the GC-MS analysis of the butyl trifluoroacetyl derivative.

E. coli complementation tests.

A ΔpanD mutant of Escherichia coli from the Keio collection (JW0127-2, CGSC strain 8404) was purchased from the E. coli Genetic Stock Center at Yale University (31). Gene analysis by STRING 9.1 (http://string-db.org/) showed that the homologs of MJ1564 in Methanocella paludicola and Archaeoglobus fulgidus were fused to CoaE, which encoded dephospho-CoA kinase, indicating that MJ1564 might be involved in CoA biosynthesis. In order to test whether the gene product of MJ0050 or MJ1564 was responsible for producing β-alanine, the abilities of the MJ0050 and MJ1564 genes to complement the E. coli panD deletion mutant cells were both tested. The gene at the MJ0050 locus was amplified by PCR and cloned into the pT7-7 plasmid based on the previously described procedure (23). MJ1564 was amplified by PCR and cloned into the pT7-7 vector using primers MJ1564-Fwd (Fwd stands for forward) (5′-GGTCATATGGAAGTGATTATTAAAGCTAAG-3′) and MJ1564-Rev (Rev stands for reverse) (5′-GCTGGATCCCTAATCACCGGATG-3′). The plasmids pMJ0050, pMJ1564, and pT7-7 were used to transform E. coli ΔpanD cells by heat shock, and transformants (ΔpanD mutant plus pT7-7, ΔpanD mutant plus MJ0050, and ΔpanD mutant plus MJ1564, respectively) were selected on the basis of ampicillin and kanamycin resistance. Complementation of the E. coli ΔpanD mutant was tested by growing the ΔpanD mutant plus pT7-7 (as a control), ΔpanD mutant plus MJ0050, and ΔpanD mutant plus MJ1564 on two kinds of liquid M9 media, one including 0.4 g/ml glucose as a carbon source supplemented with 0.6 mM thiamine, 2 mM Mg²⁺, 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG), and antibiotics (50 μg/ml kanamycin and 100 μg/ml ampicillin) and the other M9 medium containing all the above supplements and 10 mM β-alanine.

Although the E. coli ΔpanD strain does not express T7 RNA polymerase, it has been observed that the protein encoded by the gene on the T7 promoter vector was expressed at a low but functional significant level in the absence of T7 RNA polymerase (32). Therefore, the cells without addition of β-alanine were grown for 24 to 48 h due to the low protein expression level; the cells supplemented by β-alanine were grown for 12 to 18 h with constant agitation (250 rpm) at 37°C.

RESULTS

Testing known routes to β-alanine from l-aspartate, malonate semialdehyde, l-alanine, spermine, dihydrouracil, and acryloyl-CoA.

All of the six known pathways to β-alanine (Fig. 2) were experimentally tested by incubating M. jannaschii cell extracts with the suspected precursor molecules and measuring the increase in the amount of free β-alanine present. The amount of β-alanine was measured by isotopic dilution analysis with [¹³C₃-¹⁵N]β-alanine serving as the internal standard. This method also allows determination of the position of label incorporation into the β-alanine when a labeled precursor was used (see Fig. S1 in the supplemental material). The cell extracts incubated in the absence of any precursor were found to contain ∼10 μM β-alanine. Functioning of one or more of these pathways shown in Fig. 2 would be indicated by an increase in the measured amount of β-alanine and the incorporation isotope if a labeled precursor was used.

Incubation of a cell extract with [1,2,3,4-¹³C₄]l-aspartate increased the level of β-alanine to ∼12 μM with 21% of the total β-alanine containing a ¹³C₃ unit (79% from the cell extract and 21% from the labeled aspartate), thus confirming that the β-alanine could be derived from l-aspartate by a α-decarboxylation reaction (reaction 1 in Fig. 2).

No increase in the amount of the 10 μM background level of β-alanine was observed when cell extracts were incubated with dihydrouracil, spermine, or α-alanine to test for reactions 3, 4, and 5, respectively, in Fig. 2. Similarly, no change in β-alanine concentration was observed when cell extracts were incubated with malonate semialdehyde, NADH, and NADPH to test for reaction 2 in Fig. 2.

The addition of ammonia to acryloyl-CoA followed by hydrolysis of the resulting HS-CoA ester would also produce β-alanine via reaction 6 in Fig. 2. The involvement of acryloyl-CoA in such a reaction has precedent in the β-alanyl-CoA:ammonia lyases characterized from Clostridium propionicum (21). Incubation of a cell extract with acryloyl-CoA, acryloyl-CoA and ¹⁵NH₄⁺, or [2,3,3-²H₃]acryloyl-CoA and aspartate showed that there was an increase in the amount of detected β-alanine to >60 μM in each case (Table 1). In the case of the acryloyl-CoA and ¹⁵NH₄⁺ incubation, no incorporation of ¹⁵N into the β-alanine was observed, indicating that the amino group of the β-alanine was not generated from ammonia. These results showed that β-alanine cannot be produced through the direct reaction of acryloyl-CoA with ammonia. Moreover, incubation of cell extracts with [2,3,3-²H₃]acryloyl-CoA and aspartate, which would produce N-(2-carboxyethyl)-l-aspartate (see Discussion), also increased the level of β-alanine; however, no deuterium was incorporated on the basis of the 168 plus 3 m/z ion intensity. However, the large increase in the amount of unlabeled β-alanine was produced only in experiments where CoA or acryloyl-CoA were added. We have established that the large increase in the concentration of β-alanine in these experiments was the result of the hydrolytic release of the β-alanine from the coenzyme A. On the basis of the above data, we eliminated the involvement of acryloyl-CoA in β-alanine biosynthesis.

Enzymatic activity of the MJ0050 gene product.

The PLP-dependent tyrosine decarboxylase (33), encoded by the MJ0050 gene, was found to generate tyramine for methanofuran biosynthesis in the methanogens (23). A homolog of this enzyme from Pyrococcus horikoshii was reported to decarboxylate l-aspartate (34). We reinvestigated the kinetic constants for this enzyme from M. jannaschii with l-aspartate as the substrate and measured a K_m of around 0.80 mM for l-aspartate and a specific activity of 0.09 μmol min⁻¹ mg⁻¹ for the decarboxylation of l-aspartate. Notably, this K_m value was close to the concentration of aspartate measured in the M. jannaschii cell extracts (35). In contrast, this enzyme had a specific activity of 1.1 ± 0.12 μmol min⁻¹ mg⁻¹ and a K_m of 1.6 ± 0.29 mM for the decarboxylation of l-tyrosine (23).

MJ0050 supplemented an E. coli ΔpanD mutant.

Although gene knockout systems had been developed in some methanogens (36, 37), none have been developed for M. jannaschii (38). In addition, producing a knockout of the MJ0050 gene required for the generation of tyramine for methanofuran biosynthesis (23) would be lethal, since methanofuran is essential for methanogenesis. Therefore, the ability to produce β-alanine catalyzed by the gene product of MJ0050 was tested by complementation of E. coli panD deletion mutants.

In bacteria, β-alanine is obtained through decarboxylation of l-aspartate using aspartate decarboxylase (panD) (39). In E. coli, the panD deletion mutants (ΔpanD) lack aspartate decarboxylase activity and thus are defective in β-alanine biosynthesis (39). Therefore, E. coli ΔpanD strain was the perfect tool to test the enzyme(s) that can produce β-alanine. In addition, it had been reported that the E. coli ΔpanD DV9 mutant is able to grow on a solid medium plate (including agar) unsupplemented with β-alanine but that it exhibits β-alanine auxotrophy when incubated in liquid medium (40). The same phenomenon was observed for E. coli ΔpanD JW0127 in this study. Therefore, selection for complementation was performed in liquid M9 medium in the presence of ampicillin (100 μg/μl) and kanamycin (50 μg/μl). The selected transformants, E. coli ΔpanD mutant plus pT7-7, E. coli ΔpanD mutant plus MJ1564, and E. coli ΔpanD mutant plus MJ0050, were able to grow on a liquid M9 medium supplemented by 10 mM β-alanine (Fig. 3). E. coli ΔpanD mutant plus pT7-7 and ΔpanD mutant plus MJ1564 exhibited β-alanine auxotrophy when incubated in the unsupplemented liquid M9 medium. In contrast, as shown in Fig. 3, expression of MJ0050 could complement the ΔpanD deletion cells, which were able to grow on the liquid M9 medium without β-alanine. The supplementation results were consistent with the in vitro observation that the tyrosine decarboxylase in M. jannaschii could use l-aspartate as a substrate to produce β-alanine.

FIG 3.

Complementation of E. coli ΔpanD knockout mutant with MJ0050. Complementation of E. coli ΔpanD mutant was tested by growing the ΔpanD mutant with pT7-7, MJ0050, or MJ1564 in M9 medium with and without β-alanine as described in Materials and Methods.

DISCUSSION

To determine the biosynthetic origin of β-alanine in M. jannaschii, a method for the extraction, purification, and analysis of the incorporation of stable isotopically labeled precursors into β-alanine in crude cell extracts was developed. By adding a known amount of labeled β-alanine to an incubation mixture, it was possible to simultaneously measure the concentration of free β-alanine in cell extracts by isotope dilution analysis and measure the incorporation of stable isotopically labeled precursors. The method developed demonstrates the power of using stable isotopes as tracers and internal standards in biochemical studies. We have tested all the known biochemical pathways for the production of β-alanine (Fig. 2). With the exception of incubating extracts with aspartate or acryloyl-CoA, each of the pathways showed no increase in the amount of β-alanine present in cell extracts after incubation with the different precursors. On the basis of our data, we also have eliminated the involvement of acryloyl-CoA in β-alanine biosynthesis.

In addition to the six known routes, other biochemically plausible routes to β-alanine were also examined. These include the reaction of acrylate and ammonia and the elimination of fumarate from N-(2-carboxyethyl)-l-aspartate. The acrylate and ammonia reaction produced no increase in the amount of β-alanine. Although a large increase in the amount of β-alanine in cell extracts was observed from incubating the cells with N-(2-carboxyethyl)-l-aspartate, this was likely catalyzed by adenylosuccinate lyase (MJ0929) present in the extracts. We have demonstrated that this enzyme catalyzes this reaction (data not shown). The origin of N-(2-carboxyethyl)-l-aspartate from a reaction of malonate semialdehyde with aspartate, followed by a reduction; from the reaction of acetyl-CoA, formaldehyde, and aspartate; from the reaction between acryloyl-CoA and aspartate; or from the reaction of acryloyl and aspartate could not be demonstrated (data not shown).

In the case of aspartate, [1,2,3,4-¹³C₄]l-aspartate was found to serve as the precursor for all the carbons of β-alanine, which indicates that an α-decarboxylase could be responsible for generating β-alanine from l-aspartate. The canonical enzyme for catalyzing this reaction for coenzyme A biosynthesis is a pyruvoyl-dependent decarboxylase (15), but a gene encoding this enzyme is absent in the genome of M. jannaschii.

Since the decarboxylation of l-aspartate is the only source for β-alanine biosynthesis in M. jannaschii, the question is which enzyme could be catalyzing this decarboxylation? Attempts to identify the gene encoding the enzyme to catalyze this reaction through a comparative genomic analysis of the genes involved in coenzyme A biosynthesis failed to find a specific gene that encodes this enzyme. Since no pyruvoyl-dependent enzyme encoding an aspartate decarboxylase was present in methanogen genomes, we investigated whether a PLP-dependent enzyme was responsible. The products of the MJ0610, MJ0684, and MJ0959 genes are predicted to encode PLP-dependent enzymes for which no function has yet been identified. Each of these genes was cloned, the recombinant protein was produced, and the function was tested. However, none of these were found to catalyze the decarboxylation of aspartate (data not shown).

The enzyme encoded by the MJ0050 gene in M. jannaschii was originally annotated as glutamate decarboxylase, which belongs to the class II decarboxylase family (41), the PLP-dependent transferase superfamily. Recently, glutamate decarboxylase-like proteins (GAD) from insects (42) and humans (43), which also belong to the class II decarboxylase family, were reported to function as aspartate decarboxylases to produce β-alanine. Specifically, GAD2 from mosquitoes was found to exclusively function in the production of β-alanine (42).

The homolog of MJ0050 in Pyrococcus horikoshii, PH0937 (41% identical to MJ0050 in sequence), was heterologously expressed in E. coli, and the resulting protein was purified (34). The P. horikoshii enzyme exhibited a broad substrate specificity and was able to decarboxylate l-glutamate, l-aspartate, cysteine, and cysteine sulfite (34). In addition, Tomita and coworkers reported a PLP-dependent glutamate decarboxylase from T. kodakarensis (the homolog of the MJ0050 gene shares 43% sequence identity with MJ0050), which is able to produce β-alanine necessary for CoA biosynthesis by α-decarboxylation of the l-aspartate (22). The catalytic efficiency was nearly 20-fold higher using aspartate as the substrate instead of glutamate (22). However, no decarboxylation of tyrosine was observed with these enzymes, consistent with the lack of methanofuran in Pyrococcus and Thermococcus.

In methanogens, the essential role of the MJ0050 gene product, MfnA, is to generate tyramine for methanofuran biosynthesis from decarboxylating tyrosine (23). In this study, a K_m of ∼0.80 mM for l-aspartate and a specific activity of 0.09 μmol min⁻¹ mg⁻¹ for the decarboxylation of l-aspartate were obtained for MfnA, indicating sufficient catalytic ability to produce the β-alanine in vitro. In addition, expression of the MJ0050 gene product was able to complement the ΔpanD mutant in E. coli cells, suggesting that the β-alanine is produced by this PLP-dependent decarboxylase, which provides the first evidence of β-alanine biosynthesis in methanogens.

It has been reported that methanogens contain much lower levels of coenzyme A than nonmethanogens (11, 44). Considering the low abundance of coenzyme A in methanogens, and hence a low requirement for its biosynthetic precursors, MfnA in M. jannaschii could easily supply the small amount of β-alanine needed for coenzyme A production. Thus, by using a promiscuous decarboxylase, the cells are able to adapt to produce the products desired for the specific cells in which it resides. In bacteria, it has been observed that the promiscuous/moonlighting enzymes are involved in CoA biosynthesis (45, 46). MfnA (gene product of MJ0050) provides another example of promiscuous enzymes in archaea (47). Generally, the genome size of archaea is much smaller (M. jannaschii has a 1.6-Mb circular chromosome [48]) than the genome size of most eukaryotes and most non-host-dependent bacteria. In addition, M. jannaschii is an autotrophic archaea; therefore, a promiscuous enzyme can provide an obvious advantage that allows the enzyme to react to a broader range of substrates to maximize catalytic versatility using limited enzyme resources (49).