J Nutr Sci Vitaminol, 52, 256–260, 2006

Effect of L-Serine on the Biosynthesis of Aromatic Amino Acids in Escherichia coli

Keiko TAZUYA-MURAYAMA1, Hironori ARAMAKI1, Motohiro MISHIMA1, Keiko SAITO2, Shiho ISHIDA2 and Kazuko YAMADA2

1Daiichi College of Pharmaceutical Sciences, 2–11, Tamagawa-cho, Minami-ku, Fukuoka 815–8511, Japan 2School of Pharmaceutical Sciences, Mukogawa Women’s University, 11–68, Koshien Kyuban-cho, Nishinomiya, Hyogo 663–8179, Japan (Received August 15, 2005)

Summary It is well known that some amino acids inhibit bacterial growth. L-Serine is known to inhibit the growth of Escherichia coli by inhibition of homoserine dehydrogenase (EC 1.1.1.3). It has been reported that this L-serine inhibition may be prevented by the addi- tion of L-isoleucine or L-threonine to the medium. In our study, however, recovery of the growth inhibition of Escherichia coli by L-serine occurred in the presence of several amino acids, especially L-phenylalanine. In an attempt to further elucidate this inhibition mecha- nism, different intermediates of aromatic amino acid biosynthesis were added to the growth medium. Recovery from the inhibition did not occur in the presence of prephenate but did occur when phenylpyruvate was added to the medium. The specific activity of prephenate dehydratase decreased in cells grown in the presence of L-serine. However, L-serine did not inhibit in vitro prephenate dehydratase activity, and the expression of pheA, which encodes the prephenate dehydratase, was not depressed by L-serine. We suggest that L-serine acts via another inhibition mechanism. Although this inhibition mechanism has not been fully elu- cidated, our results suggest that the addition of L-serine to the growth medium inhibits prephenate dehydratase synthesis and thus affects L-phenylalanine biosynthesis. Key Words L-serine, Escherichia coli, growth inhibition, prephenate dehydratase, pheA

Some amino acids are known to inhibit bacterial homoserine dehydrogenase I, but also to another mech- growth (1). For example, glycine is used as a food addi- anism. Commercially available L-serine is produced by tive, not only to improve nutrition or enhance flavour, fermentation using bacterial mutants resistant to L- but also as a preservative (2, 3). It has been reported serine repression (11). The further elucidation of the that the mechanism of growth inhibition by glycine is inhibition mechanism induced by L-serine may open due to inhibition of UDP-N-acetylmuramate-L-alanine the possibility to extend the usage of L-serine as a pre- ligase (EC 6.3.2.8; L-alanine adding enzyme) which servative agent and may suggest other possibilities for leads to abnormal bacterial cell wall synthesis (4, 5). its production. In the present study we have examined Another amino acid, L-valine, inhibits acetohydroxy the mechanism of growth inhibition by L-serine and acid synthase (EC 4.1.3.18) (6, 7). L-Valine not only report that it inhibits cell growth by decreasing the inhibits enzyme activity but also depresses the gene activity of prephenate dehydratase, an enzyme involved expression. It has been reported that the simultaneous in phenylalanine and tyrosine biosynthesis. addition of L-isoleucine or L-threonine with L-serine to MATERIALS AND METHODS the medium restores the cell growth inhibition caused by L-serine, and that L-serine inhibits homoserine dehy- Materials. All chemicals were of analytical grade. drogenase (EC 1.1.1.3), the enzyme involved in L-isoleu- Prephenic acid as barium salt was purchased from cine biosynthesis (8, 9). Sigma (St. Louis, MA). Thiamin hydrochloride and pyri- We previously isolated a bacterium present on the doxine hydrochloride were from Wako Pure Chemical surface of dry soy beans and identified it as Bacillus sub- Industries, Ltd. (Osaka, Japan). Riboflavin, d-pan- tilis MF920 (10). We tested 20 amino acids in a search tothenic acid calcium salt and folic acid were from for amino acids which inhibited the growth of B. subti- Nacalai Tesque, Inc. (Kyoto, Japan). Vitamin-free lis, and found L-serine to be the most effective inhibitor casamino acids were bought from Difco (Detroit, MI). (10). In our study, however, growth inhibition of B. sub- Total RNA was extracted with an RNeasy Mini Kit, tilis by L-serine was prevented by the addition of aro- RNase-Free DNase Set and RNAprotect Bacteria matic amino acids (unpublished data). The nutrients Reagent (Qiagen; Hombrechtikon, Switzerland). The required to remove the cell growth inhibition were dif- primers were purchased from Griner Japan (Tokyo, ferent from those mentioned in previous reports (8, 9). Japan). RT-PCR was carried out with the Super Script These results suggest that the mechanism of growth One-Step RT-PCR kit with Platinum Taq (Invitrogen; inhibition by L-serine is not only due to inhibition of Carlsbad, CA). A Perkin Elmer Gene Amp PCR System

256

Bacterial Growth Inhibition by L-Serine 257

9600 was used for RT-PCR. GAAACCCGTTACTGGCGCTG-3¢/5¢-CGCCGCAAGATGG- Organisms and growth conditions. E. coli B IFO13168, GAATAAGAACCTTTGGG-3¢ for pheA, 5¢-ATGGTTGCTG- a wild-type, phage-propagating strain, was purchased AATTGACCGCATTACGCGAT-3¢/5¢-GTTTTAAATCCTT- from the Institute of Fermentation of Osaka (Osaka, TGTCGTTTTCACTGGAG-3¢ for tyrA, and 5¢-GTGTTTCA- Japan). The cells were grown aerobically in Davis-Min- AAAAGTTGACGCCTACGCTGGC-3¢/5¢-GAGCCGCCAA- o giolli’s minimal medium at 37 C. The medium consisted GGGTTTGAATGGTTGCTACG-3¢ for tyrB. The RT-PCR of 7 g/L K2HPO4, 2 g/L KH2PO4, 0.1 g/L MgSO4·7H2O, products were electrophoresed on 3% agarose gel with 1 g/L (NH4)2SO4, 0.5 g/L sodium citrate dihydrate, and Tris-Acetate buffer. The bands were dyed with ethidium 2 g/L glucose. The cells were pre-incubated in the loga- bromide and detected with UV. The gel was photo- rithmic growth phase and inoculated to the medium graphed with Polaroid-type film. The intensity of each with or without 2% L-serine. band was analyzed using ImageJ 1.34 software (NIH). Determination of generation times. Cell growth was The RT-PCR product was detected as optical image den- monitored by measuring turbidity at 660 nm. Cultures sity units. The integrated density of the band from pheA, pre-incubated with the minimal medium in the loga- tyrA, and tyrB was calculated for the density of the rithmic phase were inoculated to the medium supple- band from 0% L-serine. mented with L-serine and/or other nutrients. The gen- RESULTS AND DISCUSSION eration times were calculated from the linear region of the growth curve by measuring the time needed for the Growth of E. coli in minimal medium with L-serine turbidity of the culture to double. The nutrition supplied The growth curves of E. coli in medium containing L- to the medium was 0.01% vitamin-free casamino acids, serine are shown in Fig. 1. The generation times in 0, 1, 0.01% amino acids, 10 mol/L vitamins (thiamin, ribo- and 2% L-serine were approximately 1, 2, and 4.3 h, flavin, pyridoxine, folic acid and pantothenic acid, respectively. The growth of E. coli was inhibited by L- respectively), or 0.01% nucleobases (adenosine and serine and the growth rates were inversely dependent cytosine, respectively). L-Serine was added at a concen- on the concentration of L-serine. Because the inhibition tration of 2%. of bacterial growth by 2% L-serine was significant, we Assay of prephenate dehydratase. E. coli was grown used the concentration of 2% L-serine in the following with or without L-serine in minimal medium aerobi- investigations. cally at 37˚C for 3 h. After cultivation, cells were har- Recovery from L-serine-induced growth inhibition by amino vested by centrifugation (5,000 g, 5 min). Cell-free acids extract was prepared by the method of Görisch (12). Effects of nucleobases, vitamins, and casamino acids The cells were suspended with 0.1 mol/L potassium on E. coli growth inhibited by L-serine were tested (Table phosphate buffer (containing 2.5104 mol/L EDTA, 1). Recovery of E. coli growth was not observed after pH 7.0) and the specific activity (nmol/min/mg protein) addition of nucleobases, adenosine or cytosine to the of prephenate dehydratase was determined according to medium containing L-serine. The vitamins, thiamin, the method of Fischer and Jensen (13). pyridoxine, riboflavin, pantothenic acid and folic acid, Detection of pheA mRNA by RT-PCR. E. coli was were also ineffective in stimulating recovery of growth. grown in the same manner as described for the assay of However, recovery did occur when casamino acids (a prephenate dehydratase. After cultivation, the cells vitamin-free preparation of hydrolyzed casein) were were harvested, centrifuged and washed with 0.9% added to the medium. The generation time of E. coli cul- NaCl solution. Total RNA was extracted from E. coli tivated in medium containing both 2% L-serine and with an RNeasy Mini Kit, RNase-Free DNase Set and 0.01% casamino acids was almost the same as for the RNAprotect Bacteria Reagent. Approximately 100 ng of control. total RNA was applied to RT-PCR. RT-PCR was carried To clarify which amino acids among the casamino out with the Super Script One-Step RT-PCR kit with acids led to recovery from the inhibition by L-serine, we Platinum Taq. The primers used were 5¢-ATGACATCG- divided 19 amino acids into six groups. Table 2 shows

Fig. 1. Effect of L-serine on growth of E. coli B IFO 13168. 258 TAZUYA-MURAYAMA K et al.

Table 1. Generation times of E. coli in the presence of Table 3. Effects of L-phenylalanine, L-tryptophan and nucleobases, vitamins, and casamino acids. L-tyrosine on generation time of E. coli inhibited by L- serine. Generation time (h) Generation time (h) L-Serine 0% 2% L-Serine 0% 2% None 1.3 4.2 Nucleobases 1.5 3.2 None 1.7 4.2 Vitamins 1.2 4.5 L-Phenylalanine 1.6 1.1 Casamino acids 1.1 1.3 L-Tryptophan 1.1 2.1 L-Tyrosine 1.6 3.3 E. coli was cultivated in minimal medium. Adenosine and cytosine were added as nucleobases at a concentration of Aromatic amino acids were added to the medium at a 0.01% into the medium; folic acid, riboflavin, pyridoxine, concentration of 0.01%. pantothenate, and thiamin were added as vitamins at 10 mol/L; and vitamin-free casamino acids were added at 0.01%. Table 4. Effects of intermediates of L-phenylalanine and L-tryptophan biosynthetic pathway on generation time of E. coli inhibited by L-serine. Table 2. Effects of amino acids on generation time of E. coli inhibited by L-serine. Generation time (h)

Generation time (h) L-Serine 0% 2% L-Serine 0% 2% None 1.4 4.2 Anthranilate 1.7 4.7 None 1.3 5.7 Chorismate 1.5 4.7 L-Phenylalanine, L-tryptophan, 1.3 1.4 4-Hydroxyphenylpyruvate 1.0 2.1 L-tyrosine Phenylpyruvate 1.5 1.4 L-Arginine, L-cysteine, L-lysine, 1.2 3.0 Prephenate 1.6 3.3 L-methionine Shikimate 2.1 4.6 L-Isoleucine, L-leucine, L-valine 1.1 3.4 L-Histidine, L-proline, L-threonine 1.1 3.1 L-Asparagine, L-aspartate, 1.4 4.1 L-glutamine, L-glutamate with the former allowing complete recovery. These L-Alanine, glycine 1.5 5.5 results suggest that L-serine affects the prephenate dehydratase, which catalyzes the reaction from Amino acids were added to the minimal medium at prephenate to phenylpyruvate. 0.01%. Depression of prephenate dehydratase activity by L-serine As shown in Fig. 3, the specific activity of prephenate dehydratase in cells grown with 2% L-serine was 40% that only the group containing L-phenylalanine, L-tryp- lower than in cells grown without L-serine. The tophan and L-tyrosine induced recovery from the inhi- decrease of activity was proportional to the concentra- bition. These three amino acids were therefore added to tion of L-serine added to the medium. However, addition the medium separately to reveal which of them has the of L-serine to the reaction mixture containing the crude potential to prevent or reverse the growth inhibition. enzyme from cells incubated without L-serine did not Almost complete recovery occurred only with L-phenyl- affect the activity (data not shown). These results sug- alanine, as shown in Table 3. The other two aromatic gest that L-serine does not inhibit the enzyme reaction amino acids, L-tyrosine and L-tryptophan, were only but rather enzyme biosynthesis. partially effective in restoration of growth. The product of the pheA gene is the bifunctional Effect of L-serine on biosynthesis of L-phenylalanine, L-tryp- enzyme, chorismate mutase and prephenate dehy- tophan, and L-tyrosine dratase. Chorismate mutase catalyzes the conversion of The biosynthetic pathways of L-phenylalanine, L- chorismate to prephenate and prephenate dehydratase tryptophan and L-tyrosine are shown in Fig. 2. The gen- catalyzes the conversion of prephenate to phenylpyru- eration times of E. coli in the presence of the intermedi- vate. These steps are the final two steps in the biosyn- ates of these aromatic amino acids are shown in Table thesis of L-phenylalanine, and the prephenate dehy- 4. Recovery from the growth inhibition caused by L- dratase is the rate-limiting enzyme in the biosynthesis serine did not occur in the presence of anthranilate, of these aromatic amino acids in microorganisms (14– shikimate or chorismate, but did occur to some extent 16). The biosynthesis of phenylalanine is regulated by with prephenate. More complete recovery was observed prephenate dehydratase, whose activity is itself regu- with phenylpyruvate and 4-hydroxyphenylpyruvate, lated by the end product, phenylalanine. The enzyme is Bacterial Growth Inhibition by L-Serine 259

Fig. 2. Biosynthetic pathway of aromatic amino acids.

Fig. 3 Prephenate dehydratase activity in E. coli grown with L-serine. E. coli was incubated with 0, 1 or 2% L- serine in the medium. The specific activity (nmol/min/ mg protein) of prephenate dehydratase was measured in the crude extract from the cell. The reaction mixture contained the extract and 1 mM prephenate in 50 mM phosphate buffer at pH 7.5. The reaction was stopped by addition of 2.5 M NaOH, then the mixture was mea- sured spectrophotometrically at 320 nm. The data were presented as relative activity based on the specific activ- Fig. 4 Effect of L-serine on the pheA transcription. RNAs were extracted from E. coli incubated with 0, 1, or 2% ity with 0% L-serine. ** p0.01, compared with control by Dunnett’s test. L-serine for 4 h. The RT-PCR products were electro- phoresed in 3% agarose gel. The RT-PCR products of the pheA gene were quantified by using tyrA and tyrB products as internal standards. encoded by pheA. The expression of pheA is regulated by pheU (17, 18). The product of pheU is tRNAPhe which acts as a protein repressor of pheA (19). As L-serine had intensities of RT-PCR product of pheA from E. coli incu- no effect on the activity of prephenate dehydratase in bated with 0, 1, and 2% L-serine are 100.016.6, vitro, we assumed that L-serine might inhibit the 93.313.8, and 99.715.8%, respectively. The rela- expression of pheA. We therefore investigated the tive intensities of RT-PCR product of tyrA from E. coli expression of the mRNA of pheA. incubated with 0, 1, and 2% L-serine are 100.016.6, Influence of L-serine on the expression of pheA 101.614.9, and 97.212.8%, respectively. The rela- The result of electrophoresis of the RT-PCR products tive intensities of RT-PCR product of tyrB from E. coli is shown in Fig. 4. The RNA was extracted from cells incubated with 0, 1, and 2% L-serine are 100.0 16.6, incubated with or without L-serine. The RT-PCR prod- 98.512.4, and 94.99.1%, respectively. The ratios of ucts of pheA gene were quantified by using tyrA and pheA product and the internal standard, tyrA product tyrB products as internal standards. The tyrA gene or tyrB product, were almost the same compared with encodes prephenate dehydrogenase and tyrB gene those of 0% L-serine. The mRNA of pheA was not encodes aromatic amino acid aminotransferase. The decreased by incubation with L-serine, which suggests density of each band was calculated from the density of that L-serine plays no role in the transcription of pheA. the band derived from the corresponding RT-PCR prod- However, the total amount of prephenate dehydratase uct of E. coli incubated with 0% L-serine. The relative measured by the specific activity was decreased by L- 260 TAZUYA-MURAYAMA K et al. serine addition (Fig. 3). The mechanism of growth inhi- riol 89: 1124–1127. bition by L-serine therefore remains unclear. 5) Hishinuma F, Izaki K, Takahashi H. 1971. Inhibition of Hama et al. reported that homoserine dehydrogenase L-alanine adding enzyme by glycine. Agric Biol Chem 35: was inhibited by L-serine and suggested that this is the 2050–2058. 6)Jackson JH, Herring PA, Patterson EB, Blatt JM. 1993. A basic mechanism behind the inhibitory effect of L-serine mechanism for valine-resistant growth of Escherichia (8, 9). However, from the results of our investigation, we coli K-12 supported by the valine-sensitive acetohydroxy do not believe that the inhibition of homoserine dehy- acid synthase IV activity from ilvJ662. Biochimie 75: drogenase by L-serine is the only reason for growth 759–765. inhibition. We cannot explain the recovery of growth in 7) DeFelice M, Squires C, Levinthal M, Guardiola J, Lam- the presence of phenylalanine by inhibition of homo- berti A, Iaccarino M. 1977. Growth inhibition of Escher- serine dehydrogenase. Nor does the inhibition of homo- ichia coli K-12 by L-valine: a consequence of a regula- serine dehydrogenase by L-serine explain the fact that tory pattern. Mol Gen Genet 156: 1–7. the addition of phenylpyruvate removes the growth 8) Hama H, Sumita Y, Kakutani Y, Tsuda M, Tsuchiya T. inhibition caused by L-serine. There must therefore be 1990. Target of serine inhibition in Escherichia coli. Bio- another mechanism for growth inhibition by L-serine. chem Biophys Res Commun 168: 1211–1216. Moreover, the expression of the mRNA of pheA, which 9) Hama H, Kayahara T, Tsuda M, Tsuchiya T. 1991. Inhi- bition of homoserine dehydrogenase I by L-serine in encodes prephenate dehydratase, was not affected by Escherichia coli. J Biochem 109: 604–608. the addition of L-serine. We suggest that L-serine has the 10) Saito K, Tanaka K, Yamada K. 2001. Growth inhibition ability to inhibit the translation of the protein synthesis by L-serine in Bacillus subtilis MF920. Nippon Shokuhin of prephenate dehydratase. However, there is a possibil- Kagaku Gakkaishi (Jpn J Food Chem) 8: 117–120 (in Japa- ity that L-serine was metabolized to another substance nese). and the metabolite inhibited the bacterial growth. The 11)Peters-Wendisch P, Stolz M, Etterich H, Kennerknecht mechanism and details are unknown. N, Sahm H, Eggeling L. 2005. Metabolic engineering of In conclusion, we confirmed the bacteriostatic effect corynebacterium glutamicum for L-serine production. of L-serine on the growth of E. coli, which could be Appl Environ Microbiol 71: 7139–7144. removed by addition of L-phenylalanine. These findings 12) Görisch H. 1987. Chorismate mutase from Streptomyces aureofaciens. In: Methods in Enzymology (Kaufman S, indicate that the bacteriostatic effect of L-serine is due to Colowick SP, Kaplan NO, eds), Vol 142, p 463–472.Aca- inhibition of the biosynthetic pathway of L-phenylala- demic Press, New York. nine, in particular to a decrease in the activity of one of 13) Fischer R, Jensen R. 1987. Prephenate dehydrogenase the enzymes, prephenate dehydratase. The effect of L- (monofunctional). In: Methods in Enzymology (Kauf- serine on prephenate dehydratase has not been fully man S, Colowick SP, Kaplan NO, eds), Vol 142, p 507– explained, but the PCR result indicates that L-serine 512. Academic Press, New York. does not depress the expression of mRNA associated 14) Nelms J, Edwards RM, Warwick J, Fotheringham I. with tyrosine and phenylalanine biosynthesis. How- 1992. Novel mutations in the pheA gene of Escherichia ever, we need to investigate further the effect of L-serine coli K-12 which result in highly feedback inhibition- on prephenate dehydratase. Commercial L-serine is pro- resistant variants of chorismate mutase/prephenate duced by fermentation of bacterial mutants which are dehydratase. Appl Environ Microbiol 58: 2592–2598. 15) Zhang S, Pohnert G, Kongsaeree P, Wilson DB, Clardy J, resistant to L-serine repression. Prephenate dehydratase Ganem B. 1998. Chorismate mutase-prephenate dehy- inhibition by L-serine may therefore be due to a point dratase from Escherichia coli. Study of catalytic and reg- mutation for L-serine inhibition. If the mechanism of ulatory domains using genetically engineered proteins. transient inhibition by L-serine were fully elucidated, it J Biol Chem 273: 6248–6253. is possible that L-serine could be an effective preserva- 16) Gjetting T, Petersen M, Guldberg P, Guttler F. 2001. Mis- tive like glycine. sense mutations in the N-terminal domain of human phenylalanine hydroxylase interfere with binding of REFERENCES regulatory phenylalanine. Am J Hum Genet 68: 1353– 1) Amos H, Cohen GN. 1954. Amino acid utilization in 1360. bacterial growth. Biochem J 57: 338–343. 17) McFall E. 1964. Genetic structure of the D-serine deam- 2)Komagata K, Ogawa H, Fukushima K. 1968. Inhibitory inase system of Escherichia coli. J Mol Biol 65: 746–753. effect of glycine on microorganisms. Shokuhin Eiseigaku 18)Gowrishankar J, Pittard J. 1982. Regulation of phenyl- Zasshi (J Food Hygiene Soc Jpn) 9: 289–294 (in Japa- alanine biosynthesis in Escherichia coli K-12: Control of nese). transcription of the pheA operon. J Bacteriol 150: 1130– 3) Ri JK, Tatsuguchi K, Tsutsumi M, Watanabe T. 1985. 1137. Combined antimicrobial effects of glycine and several 19)Gavin N, Davidson BE. 1991. Regulation of pheA expres- reagents. Shokuhin Eiseigaku Zasshi (J Food Hygiene Soc sion by the pheR product in Escherichia coli is mediated Jpn) 26: 279–284 (in Japanese). through attenuation of transcription. J Biol Chem 266: 4) Strominger JL, Birge CH. 1965. Nucleotide accumula- 7750–7753. tion induced in Staphylococcus aureus by glycine. J Bacte-