Carbont - Journal of Bacteriology

0 downloads 0 Views 925KB Size Report
source, butit could use L-serine as an auxiliary carbonsource with glucose, L- alanine, or pyruvate and could derive energy from L-serine to support oxygen ... serine is not the natural substrate of L-SD, it can .... ty of only 0.026 U/100 K.U. When grown with. TABLE 1. Sparing of glucose .... higher rates, and glucose still better.
Vol. 151, No. 2

JOURNAL OF BACTERIOLOGY, Aug. 1982, p. 777-782 0021-9193/82/080777-06$02.O0/0

L-Serine Degradation in Escherichia coli K-12: a Combination of L-Serine, Glycine, and Leucine Used as a Source of Carbont E. B. NEWMAN* AND C. WALKER Department of Biological Sciences, Concordia University, Montreal, H3G IM8 Canada

Received 16 December 1981/Accepted 12 April 1982

Escherichia coli K-12 strain CU1008 cannot use L-serine as the sole carbon source, but it could use L-serine as an auxiliary carbon source with glucose, Lalanine, or pyruvate and could derive energy from L-serine to support oxygen uptake. CU1008 grew with L-serine if it was also provided with glycine and leucine. These may act by increasing the available activity of L-serine deaminase; other explanations are also explored.

Although L-serine deaminase (L-SD) has been studied in Escherichia coli since 1955 (13), its physiological function remains unknown (11). The assay for L-SD activity depends on the formation of pyruvate from L-serine (7, 13). It has not been shown, however, that E. coli actually carries out this reaction in vivo or, if it does deaminate L-serine, that it uses this enzyme to do so. Mutants lacking L-SD activity have so far not been reported, despite considerable effort in our laboratory to find them (Newman, unpublished data). It is not even known whether L-serine is in fact the physiological substrate of the activity being studied in vitro. The very high Km for Lserine suggests that the enzyme may have some other compound as its natural substrate (10). LSD has been exceedingly hard to purify from E. coli or from other microorganisms (reviewed in reference 10). This has made it impossible to define the range of substrates upon which L-SD acts, although attempts have been made with crude extracts (7, 13). E. coli K-12 cannot use L-serine as the sole carbon source (12). In this paper we examine the ways in which E. coli K-12 does use L-serine, so that we may understand what limits its use as a carbon source. The induction of L-SD via addition of its inducers, glycine and leucine, allowed E. coli K-12 to grow with serine. In the presence of these inducers, L-serine was converted to pyruvate or at least to a compound at the same oxidation level. This indicates that even if Lserine is not the natural substrate of L-SD, it can be degraded by that enzyme activity in a physiologically significant reaction. t This paper is dedicated to M. Cohen, M. Singer, and R. Roy.

MATERIALS AND METHODS Cultures. The following strains of E. coli K-12 were used: CU1008, an ilvA derivative obtained from L. S. Williams; 5570, a tpi mutant obtained from the Coli Genetic Stock Center (CGSC), Yale University; three prototrophs, CU1 and CU4, obtained from H. E. Umbarger, and K-10, obtained from A. Garen; and W4977, a proline-requiring auxotroph obtained from R. L. Soffer. All media, assays, and methods for determining growth rates and yields were as previously described (8). Isoleucine and valine were included in all media which contained L-serine except those used for testing L-serine sensitivity. L-SD assays. L-SD activity is expressed as units of enzyme produced in the standard assay by 0.1 ml of a suspension of cells of 100 Klett units (K.U.) measured with a 540 filter. One unit of L-SD was taken to be the amount of enzyme which would catalyze the formation of 1 ,umol of pyruvate in 35 min. L-Serlne sensitivity tests. To test for inhibition of growth by L-serine, cells of a stationary-phase slant of the test strain were suspended in distilled water to about 5 x 10' cells per ml. A 0.1-ml amount of this suspension was used to inoculate tubes containing 2 ml of glucose-containing minimal medium and concentrations of L-serine varying from 0 to 500 ug/,ml. The minimum inhibitory concentration given is the concentration at which no turbidity was visible after 16 to 24 h at 3rC. All cultures grew after 48 h. Oxygen uptake rates. Oxygen uptake rates were determined with a Clark-type oxygen electrode. Exponential-phase cells were chilled quickly, centrifuged at 4°C, washed with and suspended in cold minimal medium without a carbon source, and kept cold until used. To 2 ml of minimal medium preincubated at 37°C in the oxygen electrode, samples of cells (50 to 300 ,ul) were added. After an endogenous rate was determined, substrate was added. This method, using nonstarved exponential-phase cells, gives somewhat high endogenous rates. In a typical experiment with nine separate readings, values for the endogenous rate varied from 7.9 to 24.6, with a mean of 13.2 ,umol of 02 777

778

NEWMAN AND WALKER

taken up per h per mg of protein. In the same set of experiments, the uncorrected rates, with L-serine as substrate, ranged from 26.6 to 34.8, with a mean of 29.8 ,umol of 02 per h per mg of protein for seven readings. The stimulation due to the addition of substrate ranged from 1.4 to 4.4, with a mean of 2.5. It is clear that these endogenous rates, though considerable and variable, do not prevent an assessment of whether a substrate supports oxygen uptake or not, which is the purpose for which they are quoted. However, because of this variability, both the extent of stimulation and the actual rates (not corrected for endogenous activity) are given for each experiment as performed on two different occasions.

RESULTS L-Serine used as nitrogen source by CU1008. Although E. coli CU1008 does not use L-serine as a carbon source, it did use it as a nitrogen source, growing with glucose and L-serine with an apparent doubling time (a.d.t.) of 60 min. Glucose-grown cells made the transfer from ammonium sulfate to L-serine (2 mg/ml) as the nitrogen source without a lag, although they grew more slowly (a.d.t., 92 min) than did fully adapted cells. It seems clear, then, that CU1008 is able to take up L-serine from the medium at a rate sufficient to support rather rapid growth and

J. BACTERIOL.

both cases (Table 1, experiments 2 and 3). LSerine also spared L-alanine and pyruvate. Cultures grown with alanine (400 jig/ml) reached a turbidity of 87 K.U.; cultures grown with alanine and L-serine (600 jig/ml) reached a turbidity of 147 K.U. Similar results were obtained with pyruvate (data not shown). L-Serine, with glycine and leucine, as carbon source. The provision of glycine and leucine (650 and 200 ,ug/ml, respectively), along with L-serine, permitted CU1008 to grow with an a.d.t. of 105 to 115 min. The addition of both glycine and leucine was needed to obtain rapid growth. Either amino acid alone permitted the use of Lserine, but at much lower rates (a.d.t., 340 min). The high levels of glycine and leucine in the experiment described above were chosen as the ones which would permit maximum growth with glycine and leucine as nitrogen sources (9). Much lower levels would suffice. Cells provided with 200 jig of leucine per ml grew as well with 40 as with 650 jig of glycine per ml; similarly, cells given 650 jig of glycine per ml grew as well with 20 as with 200 jig of leucine per ml, with Lserine at 2 mg/ml being provided in all cases. When the amino acids were provided singly, either amino acid provided maximum growth (though at a slow rate) at 80 to 100 ,ug/ml. The majority of cell material and energy must thus be derived from L-serine. In glucose-grown cells, glycine and leucine are known to induce the enzyme activity L-SD, particularly when the two amino acids are provided simultaneously and to a lesser extent when they are added separately (7, 13). In cells grown with L-serine, the same pattern was seen. Cells grown with serine, glycine, and leucine showed a relatively high level of L-SD activity (0.14 U/100 K.U.). Cells grown with L-serine and leucine or with L-serine and glycine showed much less activity (0.054 and 0.031 U/100 K.U., respectively). For comparison, glucose-grown cells of CU1008, without inducers, had an activity of only 0.026 U/100 K.U. When grown with

apparently does so by means of a constitutive enzyme(s). This experiment seems to exclude the possibility that the failure of cells to grow with Lserine as a carbon source is due to L-serine toxicity. L-Serine is known to inhibit E. coli K12 transiently, but this inhibition is reversed by isoleucine (3, 4). Isoleucine and valine were provided in all experiments with CU1008, so this sort of inhibition would not be expected. In any case, CU1008 could clearly grow well and without a lag on exposure to L-serine (2 mg/ml), which it took up and used as a nitrogen source. We conclude, therefore, that in the presence of isoleucine, L-serine is not toxic to CU1008. L-Serine as auxiliary carbon source. When provided together with limiting amounts of glucose, L-serine greatly increased the yield obtained. CU1008 grown with glucose and subcultured TABLE 1. Sparing of glucose by L-serine' with limiting glucose produced much more cell material when L-serine was also provided (Table Carbon Ammonium Turbidity Protein 1, experiments 1 and 2). Although these cultures Expt source sulfate (K.U.) ExPt (14/ml) were clearly deriving cell mass from L-serine, 1 + 116 Glucose 89.6 they had not adapted to use serine as the sole 2 Glucose + 171 164.3 carbon source. When cultures had reached a and constant turbidity and were subcultured with LL-serine serine as the sole carbon source (in liquid or 3 Glucose 156 166.3 solid medium), no growth was seen. and L-serine L-Serine, then, spared glucose but did not itself serve as the sole carbon source. This a was grown in minimal medium with glusparing was seen whether serine was provided coseCU1008 (400 ,ug/ml) and L-serine (600 ,ug/ml). Isoleucine as the carbon source or as both the carbon and and valine (50 iLg/ml each) were added to all cultures. nitrogen sources, the yield being the same in Values given are the averages of three determinations.

VOL. 151, 1982

glycine and leucine, they had an activity of 0.18 U/100 K.U. One can therefore understand this effect of glycine and leucine by assuming that the level of L-SD activity is the primary determinant of growth with L-serine as the carbon source and that constitutive L-SD levels are for some reason insufficient to permit growth. Glucose-grown cells of CU1008, when harvested and washed, transferred to medium with serine, glycine, and leucine with a lag of less than one h, although with a slower a.d.t. (165 min) than that of fully adapted cells (110 min). This is consistent with the idea that the use of Lserine is by a constitutive system. Cells grown with L-serine, glycine, and leucine transferred to medium with glucose, glycine, and leucine without a lag. However, they did not grow on glucose without glycine and leucine, except after a prolonged lag (4 h or longer). This is very similar to a lag seen on transfer of a serine auxotroph from glucose, serine, and xanthine to glucose and glycine (11), which is not entirely understood. L-Serine as carbon source for triose phosphate isomerase mutants. A triose phosphate isomerase (tpi) mutant cannot interconvert pyruvate and dihydroxyacetone phosphate (1). If it is to be grown with pyruvate, it must also be provided with glycerol or a compound of similar oxidation level. It will then derive energy from the pyruvate via the tricarboxylic acid cycle and use glycerol as a source of hexoses and compounds produced from them. Although L-serine degradation via L-SD is well known in other bacteria (15), it is not known whether in E. coli L-serine is degraded to pyruvate or to some other product, or indeed whether L-SD is involved. By using the tpi mutant, it is possible to test whether L-serine substitutes for glycerol, for pyruvate, or even for both and whether glycine and leucine are required for its degradation. To investigate this, cells of CGSC 5570 tpi grown with gluconate were subcultured with a variety of carbon sources, including gluconate; glycerol and serine with and without glycine and leucine; pyruvate and serine with and without glycine and leucine; glycerol and pyruvate; and each of the substances provided singly. Isoleucine and valine were always added with L-serine to avoid L-serine toxicity. Cultures grew rapidly with gluconate but also grew well on glycerol and pyruvate. They grew somewhat more slowly on glycerol with serine, glycine, and leucine and exceedingly slowly on glycerol and serine. No growth was seen with L-serine, glycine, leucine, and either succinate or pyruvate, or with any substrate tested singly (except gluconate). Thus, L-serine did not substitute for glycerol, but did substitute for pyruvate if glycine and

L-SERINE AS CARBON SOURCE

779

leucine were also provided. This is consistent with the idea that L-serine degradation involves L-SD, as is the pattern of L-SD induction. CGSC 5570 grown with gluconate had the same low level of L-SD (0.02 U/100 K.U.) as did CU1008 grown on either glucose or gluconate. These levels were unchanged during growth with glycerol and pyruvate (0.015 U/100 K.U.). However, cultures grown with glycerol, serine, glycine, and leucine had a much higher level (0.15 U/100 K.U.), which is close to the level of a mutant which uses L-serine as the sole carbon source (0.3 U/100 K.U.; reference 12). The preceding interpretation is based on the assumption that glucose-grown CGSC 5570 regulates L-SD activity in the same way that CU1008 does. We therefore obtained glucoseusing transductants from CGSC 5570 and measured L-SD levels in cells grown with glucose (0.046 U/100 K.U.) and in cells grown with glucose, glycine, and leucine (0.11 U/100 K.U.). We conclude that CGSC 5570 is very similar to CU1008 in the regulation of L-serine degradation. L-Serine could be converted to pyruvate if induced levels of L-SD were present; it could not be converted to any compound in metabolic equilibrium with glycerol. L-Serine as energy source for oxygen uptake. The use of L-serine as a major source of carbon, as in the sparing of glucose, is consistent with Lserine serving as an energy source, as a source of assimilable carbon, or both. The following experiments showed that cells grown without the inducers were in fact able to derive energy by oxidation of L-serine. Their failure to grow with L-serine is therefore not likely to be due to an absolute inability to derive energy from Lserine, although it may be due to a quantitative problem. Oxygen uptake by cells grown in the presence of L-serine. As one might suppose, cells grown with L-serine, glycine, and leucine derived energy from the oxidation of L-serine. With such cells, L-serine stimulated oxygen uptake markedly (Table 2, experiment 1). The addition of glycine and leucine neither supported oxygen uptake nor stimulated the uptake seen with L-serine (data not shown). Stimulation of oxygen uptake by L-serine was also seen in cells grown without inducers, though to a lesser extent. This was true of cells grown with excess glucose with L-serine as the nitrogen source (Table 2, experiment 2) and of cells grown with glucose and L-serine as joint carbon sources, each at a low level (0.6 mg/ml) (Table 2, experiment 3). The significance of the rates of L-serine oxidation may be assessed by comparing them with rates stimulated by glucose. The cultures compared in Table 2 all used glucose at rapid rates,

780

J. BACTERIOL.

NEWMAN AND WALKER

TABLE 2. Oxygen uptake by cells grown in the presence of L-serine Expt

Growth medium

1

L-Serine, glycine, and L-leucine

2

L-Serine and glucose

(NH4)2S04 +

-

Substrate

Stima

Rateb

L-Serine Glucose

2.5 2.6

14.2 12.3

L-Serine Glucose

1.9 5.5

9.5 12.2

5.5 1.7 L-Serine 14.6 Glucose 4.5 a Stim, Stimulation by substrate, expressed as the ratio of the rate of oxygen uptake after addition of the substrate to the initial rate before addition of substrate. b Rate of oxygen uptake with substrate, not corrected for the endogenous rate, expressed as micromoles of 02 taken up per hour per milligram of protein. Values are given for a typical experiment. 3

L-Serine and glucose

+

from 11.9 to 25.8 pumol of 02 per mg of protein per h. These rates are similar to those reported for E. coli by Harrison (6) (5.0 to 16.0 mmol/g [dry weight] per h, which would be equivalent to 10.0 to 32 ,umol/mg of protein per h. Cells grown with L-serine as the carbon source used L-serine even more readily than they used glucose (Table 2, experiment 1). L-Serine was used less well than glucose in cells grown with glucose and serine (Table 2, experiments 2 and 3) but was clearly usable to a metabolically significant extent. We conclude that cells which are using Lserine, for whatever metabolic purpose, have no difficulty in deriving energy from it. The rates of oxygen utilization again paralleled the levels of L-SD activity. Cells grown with L-serine as the nitrogen source (Table 2, experiment 2) or with serine as an accessory carbon source (Table 2, experiment 3) showed only basal levels of L-SD (0.03 and 0.026 U/100 K.U., respectively). Cells grown with L-serine and inducers had considerably more L-SD activity (0.14 U/100 K.U.). This correlates well with the difference in oxygen uptake rates. Oxygen uptake by cells grown in the absence of L-serine. The preceding experiments were all performed with cells grown under conditions in which one could suppose that L-serine must be entering the cells and being metabolized. Since the oxygen electrode experiments involved suspensions of intact cells, the rates seen must have been affected by both the rate of entry of Lserine and the rate at which it could be metabolized. This makes it difficult to determine whether L-serine can be used as an energy source by cells grown without exposure to L-serine. It is in fact possible to show energization of oxygen uptake by L-serine in cells not previously exposed to it (Table 3). Cells grown with glucose used L-serine at low rates, pyruvate at higher rates, and glucose still better. One would suppose, then, that L-serine at the concentration used (2 mg/ml) must enter glucose-grown cells and be used by them. However, one cannot tell

whether the rate is limited by entry or metabolism. Cells grown with pyruvate, on the other hand, did not use L-serine at all and used pyruvate more rapidly than they used glucose. It seems, then, that entry or metabolism of glucose is to some extent inducible. This was confirmed by growing cells with alanine. These cells used both alanine and pyruvate rapidly. They also used glucose well, but again less rapidly than did glucose-grown cells. DISCUSSION The phenomenon reported here is that a strain of E. coli K-12, CU1008, would not grow with Lserine as the sole carbon source but would grow if also provided with glycine and leucine. The strain would not grow with glycine and leucine as carbon sources. Moreover, the requirement for leucine and glycine was low enough that the cells must have been deriving most of their carbon from L-serine. Why, then, did they not use L-serine in the absence of glycine and leucine? CU1008 did use two compounds closely related to L-serine, i.e., pyruvate and L-alanine. TABLE 3. Oxygen uptake by cells grown in the absence of L-serine Carbon source Ratea Substrate stima during growth 24.2 4.4 Glucose Glucose 9.7 3.3 Pyruvate 2.3 3.3 L-Serine

Glucose Pyruvate L-Serine

2.7 7.9 1.0

9.5 22.2 0

Glucose Pyruvate Alanine a As explained in Table 2.

2.0 3.6 3.8

13.6 22.5 35.2

Pyruvate

Alanine

VOL. 151, 1982

Moreover, it contained in considerable amounts activity, L-SD, which converts Lserine to pyruvate (9). One would therefore expect that the strain would be able to convert Lserine to pyruvate and ammonia and grow on the pyruvate. This would be very similar to growing with L-alanine by converting that amino acid to pyruvate and ammonia, which the cell does readily. The cell had no difficulty in deriving material from L-serine. L-Serine was used as a secondary carbon source, sparing glucose, alanine, and pyruvate. Moreover, CU1008 could derive energy from L-serine, even when it had been grown without glycine and leucine. In cells grown with L-serine and glucose (total carbon limiting), Lserine energized oxygen uptake at substantial rates. This might not be surprising, since these cells probably derived some of their energy for growth from L-serine. However, cells grown with excess glucose with L-serine as the nitrogen source also used L-serine to energize oxygen uptake. Such cells most likely convert L-serine to pyruvate to derive their nitrogen; they can grow with pyruvate and can derive energy from L-serine, but nonetheless they cannot grow with L-serine alone. The conversion of L-serine to pyruvate might be limited quantitatively, either because cells exposed only to L-serine do not have high enough levels of L-SD or because they maintain what they have in an inactive form (10). This quantitative explanation for the failure of E. coli K-12 to grow with L-serine is supported by several lines of evidence. Since glycine and leucine are inducers of L-SD (7, 13), their effect may be due to the fact that the activity of L-SD in uninduced cells is insufficient. The fact that mutants which do use L-serine have higher L-SD levels (12) is in accord with this idea. Similarly, the fact that the tpi mutant could use L-serine only to replace pyruvate, and then only when glycine and leucine were present, suggests that L-serine can be degraded to a quantitatively significant extent only when the inducers are present. The role of L-SD in serine degradation has been documented in other microorganisms (15). However, the experiments presented here are the clearest evidence available at present that L-SD can be physiologically useful in Lserine degradation in E. coli. L-Serine could substitute in the tpi mutant for pyruvate but not for glycerol. This suggests that L-serine degradation is entirely via pyruvate or a related compound and therefore that L-serine biosynthesis is irreversible. L-Serine is derived from phosphoglyceric acid via a three-enzyme pathway (14). The first two steps are clearly reversible. It seems likely, then, that the cell has no reaction to phosphorylate L-serine, which

an enzyme

L-SERINE AS CARBON SOURCE

781

would reverse the last step of the biosynthetic pathway. The evidence that the effect of glycine and leucine is a quantitative one affecting L-SD activity cannot be considered definitive. Leucine induces other enzymes in addition to L-SD (5). The mutants which use L-serine have a very pleiotropic phenotype, an increase in L-SD being only one of many effects (12). There is, moreover, a very great deal of L-SD activity in extracts of uninduced strains of E. coli K-12 (10). However, this activity is not in its most active state and has an exceedingly high Km (10). The existence of L-SD in an inactive form has been reported for all bacterial L-SD studied, as was first shown for Clostridium spp. (2). It may be, then, that the enzyme activity works very sluggishly, if at all, in uninduced cells and that the effect of glycine and leucine is to activate an already existing activity rather than to induce its synthesis. Other possible explanations of the effect of glycine and leucine appear to be less likely. There appears to be very little problem with Lserine uptake. Cells grown with glucose and ammonium ion transferred to medium containing glucose with L-serine as the nitrogen source with a lag of under 20 min. Synthesis of the transport system must, then, be either constitutive or very rapid in the presence of glucose and L-serine. However, regulation of L-serine uptake or degradation may be set so as to preclude its use. For example, L-serine uptake might be inhibited by ammonia. It is also possible that L-serine is toxic to the cells and that this toxicity is counteracted by a second carbon source or by glycine and leucine. Only one form of L-senne toxicity has been studied in detail: transient L-serine inhibition which is reversed by isoleucine (3, 4). Since CU1008 grew with glucose and L-serine (2 mg/ ml) as the nitrogen source, we can exclude the possibility that growth on L-serine is prevented by L-serine inhibition. However, there is no way to exclude the possibility that L-serine given alone might be toxic. ACKNOWLEDGMENTS This work was supported by grant A6050 of the Canadian National Science and Engineering Research Council, for which we are very grateful. We thank M. Cohen, M. Singer, and R. Roy for their assistance.

LITERATURE CITED 1. Anderson, A., and R. A. Cooper. 1969. Gluconeogenesis in E coli. The role of triose phosphate isomerase. FEBS Lett. 4:19-20. 2. Carter, J. E., and R. D. Sagers. 1972. Ferrous iondependent L-serine dehydratase from Clostridium acidiurici. J. Bacteriol. 109:757-763. 3. Cosloy, S. D., and E. McFall. 1970. L-Serine-sensitive

782

NEWMAN AND WALKER

mutants of Escherichia coli K-12. J. Bacteriol. 103:840-

4.

5. 6.

7. 8. 9.

841. Daniel, J., and A. Danchln. 1979. Involvement of cyclic AMP and its receptor protein in the sensitivity of Escherichia coli K-12 towards serine. Mol. Gen. Genet. 176:343-350. Fraser, J., and E. B. Newman. 1975. Derivation of glycine from threonine in Escherichia coli K-12 mutants. J. Bacteriol. 122:810-817. Harrison, D. E. F. 1976. The regulation of respiration rate in growing bacteria. Adv. Microb. Physiol. 14:243-309. Isenberg, S., and E. B. Newman. 1974. Studies on L-serine deaminase in Escherichia coli K-12. J. Bacteriol. 118:5358. Newman, E. B. 1970. Metabolism of serine and glycine in E. coli K-12. I. The role of formate in the metabolism of serine-glycine auxotrophs. Can. J. Microbiol. 16:933-940. Newman, E. B., G. Batist, J. Fraser, S. Isenberg, P. Weyman, and V. Kapoor. 1976. The use of glycine as nitrogen source by Escherichia coli K-12. Biochim.

J. BACTERIOL. Biophys. Acta. 421:97-105. 10. Newman, E. B., and V. Kapoor. 1980. In vitro studies on L-serine deaminase activity of Escherichia coli K-12. Can. J. Biochem. 58:1292-97. 11. Newman, E. B., and B. Maganik. 1963. The relationship of serine, glycine metabolism to the formation of single carbon units. Biochim. Biophys. Acta 78:437-448. 12. Newman, E. B., J. F. Morris, C. Walker, and V. Kapoor. 1981. A mutation affecting L-serine and energy metabolism in E. coli K12. Mol. Gen. Genet. 182:143-147. 13. Pardee, A. B., and L. S. Prestidge. 1955. Induced formation of serine and threonine deaminases by Escherichia coli. J. Bacteriol. 70:667-674. 14. Pizer, L. I., and M. L. Potochny. 1964. Nutritional and regulatory aspects of serine metabolism in Escherichia coli. J. Bacteriol. 88:611-619. 15. Wong, H. C., and T. G. Lessie. 1979. Hydroxy amino acid metabolism in Pseudomonas cepacia: role of L-serine deaminase in the dissimilation of serine, glycine, and threonine. J. Bacteriol. 140:240-245.