Mycobacterium bovis BCG Vaccines Exhibit Defects in Alanine and Serine Catabolism

Materials.All chemicals and amino acids were purchased from Sigma. Monoclonal antibody (HBT10) against l-alanine dehydrogenase of M. tuberculosis H37Rv was kindly provided by A. B. Andersen and P. Andersen (Statens Serum Institut, Copenhagen, Denmark).

Bacterial strains and culture conditions.Twelve M. bovis BCG strains were used in this study: BCG-Japan, BCG-Russia, BCG-Moreau, BCG-Sweden, BCG-Birkhaug, BCG-Frappier, BCG-Pasteur, BCG-Glaxo, BCG-Phipps, BCG-Tice, BCG-Denmark, and BCG-Prague. The identities and histories of these strains have been described in detail previously (6). Middlebrook 7H9 medium (Difco) contains the following (per liter): 0.5 g of ammonium sulfate, 0.5 g of l-glutamate, 0.1 g of sodium citrate, 1 mg of pyridoxine, 0.5 mg of biotin, 2.5 g of disodium phosphate, 1 g of monopotassium phosphate, 40 mg of ferric ammonium citrate, 50 mg of magnesium sulfate, 0.5 mg of calcium chloride, 1 mg of zinc sulfate, 1 mg of copper sulfate, and 2 ml of glycerol. The medium was supplemented with 10% ADC (5 g of bovine albumin [fraction V], 2 g of dextrose), with 0.05% Tween 80 added after sterilization. Sauton medium contains the following (per liter): 4 g of l-asparagine, 0.5 g of monopotassium phosphate, 0.5 g of magnesium sulfate, 50 mg of ferric ammonium citrate, 2 g of citric acid, 1 mg of zinc sulfate, and 60 ml of glycerol (with 0.05% Tween 80 added after sterilization). For amino acid growth experiments, a basal Sauton medium was prepared by omitting l-asparagine. Glycerol-alanine-salts (GAS) medium contains (per liter) 2 g of ammonium chloride, 1 g of l-alanine, 0.3 g of Bacto Casitone (Difco), 4 g of dibasic potassium phosphate, 2 g of citric acid, 50 mg of ferric ammonium citrate, 1.2 g of magnesium chloride hexahydrate, 0.6 g of potassium sulfate, 1.8 ml of 10 M sodium hydroxide, and 10 ml of glycerol. Tween 80 was added to 0.05% after sterilization. An l-alanine-free version of GAS medium was prepared by omitting the l-alanine. BCG cultures were grown at 37°C with constant shaking for 2 to 4 weeks.

Escherichia coli DH5α, which was used for routine manipulation and amplification of plasmid DNA, was grown in Luria-Bertani (LB) broth or agar (Difco). When necessary, the medium was supplemented with kanamycin at 25 μg/ml for mycobacteria and 50 μg/ml for E. coli.

Molecular cloning.An ordered BAC library of the M. tuberculosis H37Rv genome (obtained from S. Cole, Institut Pasteur, Paris, France) was used as the DNA template for cloning. Standard protocols were used for manipulation of DNA. Cloning of ald (Rv2780) was accomplished in two steps. First, a 4.5-kb ScaI fragment of BAC415 was ligated to Ecl136II-linearized pUC19 to generate pUC-ALD. Then mycobacterial plasmid pALD was created by ligating the 1.9-kb KpnI fragment containing the ald gene to KpnI-linearized pMD31. To clone sdaA (Rv0069c), a 9.5-kb BamHI fragment of BAC60 was ligated to BamHI-linearized pMD31 to generate pSDA1. Plasmid pSDAA was generated by cleavage of pSDA1 with PstI, followed by self-ligation of the 10.9-kb PstI fragment. To clone glnA1, a 9-kb NheI fragment of BAC214 was first inserted into shuttle vector pMD31 linearized with XbaI to generate pGLN, which contains the glnA1, glnA2, and glnE genes. To generate pGLNA1, pGLN was cut with KpnI and religated, which removes all of glnA2 and the N-terminal portion of glnE, but leaves the entire glnA1 gene intact.

The constructed plasmids and control vectors were introduced by electroporation into M. bovis BCG, and recombinant M. bovis BCG was selected on Middlebrook 7H11 agar (Difco) supplemented with 10% oleic acid-albumin-dextrose-catalase (OADC) enrichment and 25 μg of kanamycin per ml.

DNA sequence analysis of ald and sdaA.The ald and sdaA genes of BCG-Japan, -Pasteur, and -Frappier were amplified by PCR. Template genomic DNA was prepared by established protocol. Oligonucleotide primers ALD-F (5′-TGAAGCGTACAGTATCGAGAGGGGTA-3′) and ALD-R (5′-ACGTGTGCTCGGCGTAACGA-3′) were used for PCR amplification of ald. Oligonucleotide primers SDAA-F (5′-CGCTCTACCCGGAACTGCAC-3′) and SDAA-R (5′-CTCAACGGCAACGGTTGGTC-3′) were used for PCR of sdaA. The ald and sdaA PCR products were purified and sequenced from both 5′ and 3′ ends by using the 5′ and 3′ PCR primers as well as internal primers. DNA sequencing was carried out at the Core Sequencing Facility of York University, Toronto, Canada, and confirmed by independent DNA sequencing at the Montreal General Hospital, Montreal, Canada.

Western blot analysis of l-alanine dehydrogenase.Culture filtrate proteins (CFP) and cell lysate (CL) proteins of Mycobacterium marinum, BCG-Japan, -Frappier and -Pasteur were prepared from cultures grown in Sauton (containing l-asparagine) or Sauton basal medium with 5 or 96 mM ammonium chloride. To prepare CFP, bacterial cultures were centrifuged, and the supernatant was passed through a 0.45-μm-pore-diameter filter. The cell-free CFP was dialyzed against double-distilled water and lyophilized. Cell lysates were prepared as follows. Cell pellets were washed and resuspended with 50 mM Tris-HCl (pH 7.5)-10 mM β-mercaptoethanol, and then they were homogenized by a Mini-Beadbeater cell disruptor (Biospec Products, Bartlesville, Okla.). The homogenate was centrifuged twice, and the cell lysates were obtained. For Western blotting, samples of CFP or CL proteins (5 μg per lane) were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes. The membranes were then probed with anti-Ald (1:1,000 dilution) antibody. The secondary antibodies (Sigma) used were anti-mouse antibodies conjugated to alkaline phosphatase.

Nucleotide sequence accession number.The sequences of ald of M. bovis BCG-Japan, -Frappier, and -Pasteur have been submitted to the GenBank database under accession no. AF531175 to AF531177 .

Growth of BCG strains in GAS medium.During the course of our studies, we found that the BCG-Japan strain was able to grow in GAS medium, albeit more slowly than in 7H9 medium, whereas BCG-Frappier and BCG-Pasteur did not show any visible growth, despite prolonged incubation (2 months). The growth capabilities of other BCG strains in GAS, 7H9, and Sauton media were also examined. The Sauton and 7H9 media supported the growth of all BCG strains, but only BCG-Japan, -Russia, -Moreau, -Sweden, and -Birkhaug strains could grow in GAS medium (Table 1). The chemical compositions of GAS, 7H9, and Sauton media were compared to identify ingredients that might account for this differential growth. GAS supplemented with ZnSO₄ (1 mg/liter), which is present in 7H9 and Sauton media, but not in GAS medium, or sodium pyruvate (0.5%), which is required for growth of large colonies of M. bovis, did not improve the growth of BCG strains in GAS (data not shown). Next, nitrogen sources were compared. l-Asparagine (27 mM) is the main nitrogen source in Sauton medium, while ammonium chloride (38 mM) and l-alanine (11 mM) are the main nitrogen sources in GAS. When l-asparagine (at 27 mM) was added to GAS medium, BCG-Frappier, -Pasteur, -Glaxo, -Phipps, -Tice, -Denmark, and -Prague strains grew rapidly (Table 1). GAS medium supplemented with l-aspartate, l-glutamine, or l-glutamate also supported the growth of these BCG strains, but no other amino acids did (Table 1). These results suggest that BCG strains vary in their ability to catabolize amino acids.

Amino acids as the nitrogen source for growth of BCG strains.Next, we examined the ability of BCG strains to utilize each amino acid as the sole nitrogen source for growth. Since GAS medium contains a small amount of Bacto Casitone (0.3 g/liter), which is a complex mixture of various amino acids and peptides, we chose Sauton medium, which is a defined medium, for this study. The original Sauton medium requires 27 mM l-asparagine, which we replaced with the same concentration of each type of amino acid. Ammonium chloride at 27 or 1 mM as the sole nitrogen source in Sauton medium was also tested. Table 2 summarizes the results for three representative BCG strains: BCG-Japan, -Pasteur, and -Frappier. Consistent with the results summarized in Table 1, these three BCG strains grew rapidly when l-asparagine, l-aspartate, l-glutamine, or l-glutamate was the sole nitrogen source. BCG-Japan was also able to grow on cationic amino acids (e.g., l-arginine and l-lysine), whereas BCG-Pasteur and -Frappier could not. Interestingly, none of the BCG strains grew when l-alanine, l-serine, l-leucine, l-isoleucine, l-methionine, or glycine was the sole nitrogen source. Other Mycobacterium species, including the pathogenic species M. avium (25) and M. marinum, as well as the nonpathogenic species M. smegmatis, were able grow on most of these amino acids as the sole nitrogen source (Table 2). A previous study showed that M. tuberculosis was able to grow in Sauton medium containing l-alanine, l-asparagine, l-aspartate, l-glutamate, or l-glutamine as the sole nitrogen source (24). Although the use of other amino acids as the sole nitrogen source for the growth of M. tuberculosis has not been formally studied, l-serine, l-arginine, l-glycine, l-isoleucine, l-leucine, and l-phenylalanine were utilized and degraded, as measured by removal of these amino acids (supplied at 1 g/liter) from the culture supernatant of M. tuberculosis (23). In particular, l-serine, l-alanine, l-asparagine, and l-aspartate were depleted by more than 50%, while the other amino acids were utilized approximately 40% (23). Our results indicated that the BCG strains exhibit defects in amino acid catabolism that are not found in (or distinguish them from) other mycobacteria.

l-Alanine, d-alanine, and l-serine inhibit the growth of BCG.One surprising result from the study described above is that all BCG strains can grow on ammonium chloride as the sole nitrogen source at either low (1 mM) or high (27 mM) concentrations (Table 2). This is contrary to the results obtained with GAS medium, which does not support the growth of BCG-Pasteur or BCG-Frappier (Table 1), even though sufficient ammonium chloride (38 mM) is present (Table 1). Lowering the ammonium concentration in GAS to 5 mM still did not support the growth of these two BCG strains (data not shown). However, the presence of l-alanine (11 mM), which is not catabolized by BCG (Table 2) in GAS medium, raised the intriguing possibility that this amino acid could actually inhibit the growth of BCG. To test this hypothesis, the growth of BCG strains was examined with a modified GAS medium, in which l-alanine had been omitted. As predicted, BCG-Frappier and BCG-Pasteur, which were unable to grow in the original GAS medium, grew rapidly in the modified, l-alanine free GAS medium (Fig. 1A). BCG-Japan also grew more rapidly in this l-alanine-free medium than in the original GAS medium (Fig. 1A). The same result was obtained for the other nine BCG strains listed in this report (data not shown).

• Open in new tab
• Download powerpoint

FIG. 1.

Inhibition of BCG growth by l-alanine and d-alanine in GAS. BCG-Japan, -Frappier, and -Pasteur, grown to stationary phase in 7H9-ADC-glycerol-Tween 80 liquid medium, were inoculated to a cell density of 2 × 10⁷ cells per ml into triplicate 5-ml volumes of GAS medium (open bar), GAS medium without l-alanine (solid bar), and GAS supplemented with 27 mM l-asparagine (hatched bar) (A) or GAS in which l-alanine replaced by d-alanine (open bar), GAS without l-alanine (closed bar) and, GAS containing d-alanine (hatched bar) supplemented with 27 mM l-asparagine (B). Cultures were incubated at 37°C with constant shaking for 16 days, and then 2-ml aliquots of cell culture were centrifuged and the cell pellet was lyophilized to determine cell dry weight.

To further confirm this result, increasing concentrations of l-alanine were added to a modified Sauton medium, in which l-asparagine was replaced by ammonium chloride (96 mM), and the growth of BCG-Japan, -Frappier, and -Pasteur was examined (Fig. 2). Strikingly, very low concentrations (0.3 mM) of l-alanine completely inhibited the growth of BCG-Frappier and BCG-Pasteur. The effect on BCG-Japan was somewhat less severe, but growth was significantly reduced upon addition of l-alanine at 0.5 mM and completely inhibited by 8 to 16 mM (Fig. 2). Taken together, these results demonstrate that l-alanine does not simply fail to support the growth of M. bovis BCG—it actually inhibits the growth of these vaccine strains.

• Open in new tab
• Download powerpoint

FIG. 2.

Inhibition of BCG growth by increasing concentrations of l-alanine in Sauton medium containing NH₄Cl (96 mM). BCG-Pasteur, -Frappier, and -Japan, grown to stationary phase in 7H9-ADC-glycerol-Tween 80 liquid medium, were washed and resuspended in Sauton basal medium (no nitrogen source). Resuspended cells of each strain were inoculated into triplicate 5-ml volumes of Sauton medium containing NH₄Cl (96 mM) with increasing concentrations of l-alanine. Cultures were incubated at 37°C with constant shaking for 30 days, and cell dry weight was determined.

We further found that d-alanine and l-serine also inhibit the growth of BCG strains. The presence of d-alanine (11 mM) in the aforementioned l-alanine-free GAS medium inhibited the growth of BCG-Pasteur and BCG-Frappier and significantly reduced the growth of BCG-Japan (Fig. 1B). Similarly, addition of l-serine (11 mM) to l-alanine-free GAS medium inhibited the growth of BCG-Frappier, BCG-Pasteur, and, to a lesser degree, BCG-Japan (data not shown).

Expression of ald in BCG relieves the growth inhibition by l-alanine and d-alanine. l-Alanine can be used as the sole nitrogen source for growth of most Mycobacterium species, including M. tuberculosis, M. avium, and M. smegmatis (24, 25). l-Alanine can be broken down to ammonium and pyruvate by l-alanine dehydrogenase. d-Alanine can be converted via racemization to l-alanine and then acted upon by l-alanine dehydrogenase. The l-alanine dehydrogenase of M. tuberculosis H37Rv is encoded by the ald gene (Rv2780). The biochemical activity of the M. tuberculosis enzyme has been demonstrated in vitro. Ald converts l-alanine to pyruvate and ammonium, and the activity is highly specific for l-alanine (22). Interestingly, this enzyme was detected in the culture supernatant fraction of M. tuberculosis, but not in M. bovis BCG-Japan nor BCG-Copenhagen, even though DNA hybridization (Southern blotting) showed that the ald gene is present in both BCG strains (2). To determine if the ald gene is disrupted in BCG, the ald alleles of BCG-Japan, -Pasteur, and -Frappier were amplified by PCR, and the DNA sequences were determined. DNA sequencing revealed that all three BCG alleles contain an identical point deletion (nucleotide A at 266) in ald, which results in a frameshift mutation and prevents expression of a full-length l-alanine dehydrogenase protein (data not shown). Genome sequence data obtained from the Sanger Center revealed that the M. bovis type strain AF2122/97 (spoligotype 9) contains the same point deletion. This suggests that the ald mutation was already present in the M. bovis strain employed by Calmette and Guérin to derive the BCG vaccine.

We postulated that the failure of BCG strains to catabolize l- or d-alanine is caused by the absence of a functional l-alanine dehydrogenase. To test this, the ald gene of M. tuberculosis was cloned into shuttle vector pMD31 and transformed into the BCG-Frappier, -Pasteur, and -Japan strains. Western blot analysis confirmed that the l-alanine dehydrogenase protein was produced in the recombinant strains (see below). These recombinant BCG strains were tested for their ability to grow in GAS medium containing l-alanine or d-alanine. BCG-Pasteur and -Frappier strains expressing M. tuberculosis ald grew rapidly in GAS medium containing either l-alanine or d-alanine, whereas control strains containing pMD31 did not (Fig. 3). Although BCG-Japan transformed with either ald or the control vector grew in GAS medium containing l-alanine or d-alanine, the BCG-Japan strain expressing ald consistently grew more rapidly (Fig. 3). These results confirm that the inhibitory effects of alanine are due to the absence of a functional alanine dehydrogenase and indicate that the accumulation of undegraded alanine inhibits growth of BCG strains in GAS medium.

• Open in new tab
• Download powerpoint

FIG. 3.

Growth of recombinant BCG (rBCG) strains expressing l-alanine dehydrogenase in GAS medium. The growth rates of BCG-Pasteur/ald, BCG-Japan/ald and the control strains BCG-Pasteur/pMD31 and BCG-Japan/pMD31 were compared. Cells of each strain grown to stationary phase in 7H9-ADC-glycerol-Tween 80 liquid medium were washed and resuspended in Sauton basal medium (no nitrogen source). Resuspended cells were inoculated into duplicate 15-ml volumes of GAS medium without l-alanine (solid diamonds), GAS medium containing l-alanine (solid squares), and GAS medium in which l-alanine was replaced by d-alanine (solid triangles) and supplemented with 0.05% Tween 80. Cultures were incubated at 37°C with constant shaking, and growth was monitored in terms of the optical density at 600 nm (OD₆₀₀). All growth experiments described in this paper were repeated multiple times (≥3) and the results are reproducible. The growth capabilities of BCG-Frappier/ald and BCG-Frappier/pMD3l were similar to those of BCG-Pasteur/ald and BCG-Pasteur/pMD31, respectively (data not shown).

Expression of sdaA in BCG relieves the growth inhibition by l-serine. l-Serine is a good nitrogen source for the growth of M. tuberculosis and M. smegmatis, but not M. bovis BCG (Table 2). We hypothesized that BCG lacks a functional enzyme for the degradation of l-serine. In E. coli, two l-serine deaminases, encoded by the sdaA and sdaB genes, catalyze the deamination of l-serine to pyruvate and ammonia (30). The genome of M. tuberculosis H37Rv contains a homolog to sdaA (Rv0069c), but not sdaB. The sdaA gene of M. tuberculosis was cloned into pMD31 and electroporated into BCG-Frappier, -Pasteur, and -Japan. The recombinant BCG-Pasteur and -Frappier strains containing pSDAA grew rapidly in alanine-free GAS medium containing l-serine, whereas the vector control strains did not grow (Fig. 4). BCG-Japan strain carrying the sdaA gene also grew more rapidly than a strain containing pMD31 vector alone (Fig. 4). This result suggests that later BCG strains contain mutations that affect either the production or level of expression of l-serine deaminase. The absence of active l-serine deaminase would explain why BCG-Pasteur and BCG-Frappier cannot catabolize l-serine. To test this, we amplified the sdaA alleles of BCG by PCR and determined the DNA sequences. Sequence analysis revealed that the sdaA alleles of all three BCG strains are identical to the M. tuberculosis sdaA gene (and also the M. bovis sdaA sequence obtained from the Sanger Center [data not shown]). Thus, it appears that the failure of BCG-Pasteur and BCG-Frappier to catabolize l-serine is caused by inadequate l-serine deaminase expression, but not a defect in the sdaA gene.

• Open in new tab
• Download powerpoint

FIG. 4.

Growth of recombinant BCG strains expressing l-serine deaminase in GAS medium containing l-serine. The growth capabilities of BCG-Pasteur/sdaA, BCG-Japan/sdaA, and the corresponding control strains were compared. Cells of each strain, grown to the stationary phase in 7H9-ADC-glycerol-Tween 80 liquid medium, were washed and resuspended in Sauton basal medium (no nitrogen source). Resuspended cells were inoculated into duplicate 15-ml volumes of GAS medium without l-alanine (solid triangles) or GAS medium in which l-alanine was replaced by l-serine (solid squares) and supplemented with 0.05% Tween 80. Cultures were incubated at 37°C with constant shaking. The growth rates of BCG-Frappier/sdaA and BCG-Frappier/pMD31 were similar to those of BCG-Pasteur/sdaA and BCG-Pasteur/pMD31, respectively (data not shown).

l-Alanine, d-alanine, and l-serine inhibit GS of BCG.We have shown that BCG-Pasteur and BCG-Frappier lack functional l-alanine dehydrogenase and l-serine deaminase activities and cannot catabolize alanine or l-serine. We now suggest that this undegraded alanine and serine prevent the growth of BCG strains by inhibiting GS activity. GS plays a central role in nitrogen metabolism in bacteria (29). Working in tandem with glutamate synthase (GOGAT), GS catalyzes the synthesis of glutamine and glutamate. GS assimilates nitrogen by aminating l-glutamate in an ATP-dependent reaction. The l-glutamine formed by GS is in turn used by GOGAT to generate l-glutamate, which together provide nitrogen for almost all amino acids, proteins, and nucleotides.

In gram-negative bacteria, such as E. coli, Salmonella enterica, and Klebsiella aerogenes, GS is under feedback inhibition. Purified GS is inhibited in vitro by l-alanine, l-serine, and glycine, which bind to the substrate site for glutamate (29). In M. tuberculosis and M. bovis BCG, GS has been identified as an extracellular protein encoded by glnA1 (Rv2220) (18). An inhibitor of GS, l-methionine-S-sulfoximine, blocked the multiplication of pathogenic mycobacteria, including M. tuberculosis and M. bovis BCG, demonstrating that GS is essential in these bacteria (17). This raised the possibility that undegraded alanine and serine could inhibit GS and subsequently prevent the growth of BCG. Therefore, we hypothesized that overexpression of glnA1 in BCG should alleviate the sensitivity of BCG to alanine and serine.

To test this possibility, the M. tuberculosis glnA1 gene was cloned into pMD31 and transformed into BCG-Frappier and BCG-Pasteur. Consistently, the recombinant strains containing glnA1 grew rapidly in alanine-free GAS medium supplemented with high concentrations of (11 mM) l-serine or d-alanine, whereas the vector control strains did not (Fig. 5A). However, the recombinant strains were still unable to grow in the original GAS medium, which contains l-alanine (Fig. 5A). It is possible that l-alanine has stronger inhibitory effect on GS than d-alanine and l-serine and that the level of GS is still not high enough to relieve this inhibition. Others have shown that in M. tuberculosis and M. avium, the ammonium concentration in the culture medium profoundly influences the GS activity: an increase in the concentration of ammonium from 7.6 to 76 mM results in a 10-fold decrease in GS activity (1, 18). The standard GAS medium contains a high concentration of ammonium (38 mM) that may decrease GS activity. Indeed, when the ammonium concentration in GAS medium was lowered to 5 mM, the BCG-Frappier strain overexpressing glnA1 was able to grow in the presence of l-alanine (Fig. 5B). Taken together, these results indicate that undegraded alanine and serine inhibit BCG growth through blockage of GS.

• Open in new tab
• Download powerpoint

FIG. 5.

Growth of recombinant BCG expressing glnA1 in GAS medium. BCG-Frappier/glnA1 and BCG-Pasteur/glnA1, grown to the stationary phase in 7H9-ADC-glycerol-Tween 80 liquid medium, were washed and inoculated into GAS medium without l-alanine (solid diamonds), GAS medium containing l-alanine (solid squares), GAS medium in which l-alanine was replaced by d-alanine (solid circles), and GAS medium in which l-alanine was replaced by l-serine (solid triangles) (A) or GAS medium containing l-alanine and a low concentration (5 mM) of ammonium (B). The ammonium concentration in the original GAS medium was 38 mM.

Regulation of ald and sdaA expression by ammonium.We found that the expression of ald is regulated by the ammonium concentration of the culture medium. Western blotting showed that l-alanine dehydrogenase was detected in recombinant BCG strains carrying the M. tuberculosis ald gene grown in Sauton medium, but not in Sauton medium in which asparagine was replaced by ammonium (96 mM) (Fig. 6). A similar result was obtained with M. marinum, a fish pathogen that contains a functional chromosomal ald gene (Fig. 6). These results suggest that expression of ald is suppressed by a high ammonium concentration, which would explain why, even in the presence of low concentrations of alanine, recombinant BCG containing ald was still unable to grow in Sauton medium containing ammonium (96 mM) (similar to Fig. 2 [data not shown]). Alternatively, GS may be posttranslationally modified and have very low activity at 96 mM ammonium, such that only a small amount of l-alanine is required for total inhibition of GS. Similarly, recombinant BCG strains containing sdaA did not grow in Sauton medium containing ammonium (96 mM) when l-serine was present (data not shown), presumably for the same reasons.

• Open in new tab
• Download powerpoint

FIG. 6.

Regulation of ald expression by ammonium. Cells of M. marinum, BCG-Frappier/ald, and BCG-Pasteur/ald were grown in Sauton medium and Sauton medium in which asparagine was replaced with 96 mM ammonium. The cell filtrate fractions (CFP [upper panel]) and cell lysates (CL [lower panel]) were prepared and analyzed by Western blotting with antibody against l-alanine dehydrogenase. Lanes 1, 3, and 5 show M. marinum, BCG-Frappier/ald, and BCG-Pasteur/ald grown in normal Sauton medium, respectively. Lanes 2, 4, and 6 show M. marinum, BCG-Frappier/ald, and BCG-Pasteur/ald grown in Sauton medium containing ammonium (96 mM), respectively. BCG-Frappier and BCG-Pasteur did not produce Ald under both growth conditions (data not shown).