31 P NMR studies of energy metabolism in xanthosine- 5¢-monophosphate overproducing Corynebacterium ammoniagenes Yasushi Noguchi 1 , Nobuhisa Shimba 2 , Yoshio Kawahara 1 , Ei-ichiro Suzuki 2 and Shinichi Sugimoto 1 1 Fermentation & Biotechnology Laboratories and 2 Central Research Laboratories, Ajinomoto Co., Inc., Kawasaki-ku, Kawasaki, Kanagawa, Japan Corynebacterium ammoniagenes is an overproducer of xanthosine-5¢-monophosphate (XMP) by consuming either glcose (glc) or glutamic acid (glu). Its energy metabolism was studied in vivo using 31 P NMR spectroscopy coupled with a circulating fermentation system (CFS). CFS enabled us to validate directly the cellular dependency on carbon sources and changes in biomolecules produced according to altera- tions in the cellular energetic status. For the most efficient XMP production, the glutamic acid and glcose molar ratios (glu/glc) in the medium were adjusted to a molar ratio of 0.31. The 31 P NMR illustrated the two distinct phases of the cellular energetic status due to the availability of the sub- strates from the medium. In the earlier phase, both glc and glu were utilized, resulting in average ATP and ADP concentrations in cells of 0.50 ± 0.17 lmolÆg )1 of dry cell weight (DCW) and an undetermined level 1 , respectively. The ADP concentration in the later phase increased to 2.15 ± 1.30 lmolÆg )1 of DCW, while the ATP concentra- tion decreased to an undetectable level in association with a remarkable decrease in XMP production. This decrease in the XMP-producing ability was associated with an increase in production of the by-product hypoxanthine. Because glu was found to be consumed completely during the earlier phase, glc was the only available substrate in the later phases 2 . These findings by in vivo NMR indicate that changes in the carbon metabolism profoundly affect XMP production by C. ammoniagenes. Keywords: xanthosine-5¢-monophosphate; in vivo NMR; energy metabolism; Corynebacterium ammoniagenes. In order to understand the microbial production of purine and other nucleotides, evaluation of the cellular energy metabolism is extremely important because the nucleotide biosynthesis requires high levels of ATP [1,2]. Difficulties in obtaining the energy metabolism profile from living cells are details of the regulatory process of nucleotide biosyn- thesis which remain to be studied. NMR spectroscopy has allowed in vivo measurement of metabolite concentrations, thereby permitting assessment of the dynamic changes in metabolic pathways and cellular regulatory mechanisms [3,4]. The circulating fermentation system (CFS) that we have previously developed [5] enables us to prolong an NMR spectroscopic measurement period under various culture conditions. Corynebacterium ammoniagenes is a Gram-positive, coryneform bacterium important to the industrial pro- duction of flavor-enhancing purine nucleotides such as inosine-5¢-monophosphate and xanthosine-5¢-monophos- phate [6–8]. Several studies have reported adenine and guanine auxotrophic mutants of C. ammoniagenes ATCC6872 that excessively secrete purine nucleotides into their cultures [6–8]. In these studies, the balance between the oxidative pentose phosphate (PP) cycle for supplementation of the carbon skeleton and the tricarboxylic acid (TCA) cycle for maximization of ATP production was discussed. However, direct estimation of contribution of these two metabolic pathways in the nucleotide production was not available. In the present study, we investigate the cellular energy metabolism in XMP-overproducing C. ammoniagenes, monitor phosphate-containing metabolites by using 31 P NMR spectroscopy, and discuss cellular energetics in nucleotide production. Experimental procedures Chemicals MDP (methylene diphosphonic acid) was purchased from Sigma Chemical Co. (St. Louis, MO, USA). All other chemicals were commercially available and of the highest grade. Bacterial strain and cultivation conditions XMP-overproducing C. ammoniagenes is an adenine and guanine auxotrophic mutant isolated from the wild-type strain C. ammoniagenes ATCC6872. For in vivo NMR studies, the inoculum was prepared on 50 plates of the modified LB medium, which was composed of 10 gÆL )1 Correspondence to Y. Noguchi, Fermentation & Biotechnology Laboratories, Ajinomoto Co., Inc., Suzuki-cho 1-1, Kawasaki-ku, Kawasaki, Kanagawa 210-8681, Japan. Fax: + 81 44 2117609, Tel.: + 81 44 2105898, E-mail: yasushi_noguchi@ajinomoto.com Abbreviations: CFS, circulating fermentation system; DCW, dry cell weight; MDP, methylene diphosphonic acid; PP, pentose phosphate; TCA, tricarboxylic acid cycle; XMP, xanthosine-5¢-monophosphate. (Received 15 November 2002, revised 23 February 2003, accepted 25 April 2003) Eur. J. Biochem. 270, 2622–2626 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03635.x tryptone, 5 gÆL )1 yeast extract, 5 gÆL )1 NaCl, and 20 gÆL )1 bacto-agar plus 0.01 gÆL )1 adenine and 0.01 gÆL )1 guanine. For in vivo NMR experiments, fermentations were carried out in a fermentor with an initial culture volume of 700 mL at 30 °C and pH 6.5. XMP-overproducing strains had been allowed to grow in the above medium supplemented with 100 gÆL )1 glc, 5 gÆL )1 K 2 HPO 4 ,5gÆL )1 KH 2 PO 4 , 0.05 gÆL )1 adenine, 0.05 gÆL )1 guanine, 1 gÆL )1 yeast extract, 0.5 gÆL )1 NaCl, 0.25 gÆL )1 CaCl 2 ,1gÆL )1 MgSO 4 Æ4H 2 O,and0.05mgL )1 FeSO 4 Æ7H 2 O. To optimize medium glu/glc ratio to gain maximum XMP production, 0, 12.5, 25, 37.5 and 50 gÆL )1 glu, respectively, were added to the above medium. The C. ammoniagenes wild type and Escherichia coli W3110 were cultured under the same culture conditions. Fermentation system For the in vivo NMR measurements, a previously construc- ted CFS was used [5]. In our system, the agitation speed of the fermentor was automatically regulated to maintain the dissolved oxygen tension (DOT) values in the fermentation vessel. Oxygen and carbon dioxide consumption rates were measured using an exhaust gas analyzer system (Able Co., Tokyo, Japan). The temperature of the whole system was kept at 30 °C using a circulating water bath, and acidifica- tion was corrected by automated additions of 10% NH 4 OH. NMR operation Using cells at high density (> 5 g of DCWÆL )1 ), 31 PNMR signals were recorded at 161 MHz with a Bruker DSX 400 WB spectrometer, where 4-k data points were recorded with 1280 transients and a spectral width of 16 kHz. The spectra were typically acquired with the following param- eters: pulse width, 34 ls(90° flip angle); repetition time, 1.5 s. To enhance the resolution, the free induction decay was multiplied by an exponential window function prior to Fourier transformation. To quantify the intracellular metabolites, MDP was used as a concentration standard in the NMR. Partially relaxed MDP was used to estimate metabolite concentrations following reported methods [5,9]. Analysis The DCW was determined 3 by measurement of attenuance (D) at 660 nm using a Shimadzu UV260 spectroscope and comparison with an optical density vs. dry weight calibration curve (the coefficient ¼ 2.5). The concentrations of glc and glu were determined enzymatically using a biotech-analyzer (Asahi Kasei Co., Tokyo, Japan). The concentrations of XMP and hypoxanthine in the culture were assayed by the HPLC (high-performance liquid chromatography) method as described previously [10,11]. Results and discussion XMP production by C. ammoniagenes To validate the performance of C. ammoniagenes,anXMP- overproducing strain, cells were batch-cultured as described in the Experimental procedures, where both glc and glu were used as carbon sources. Both concentrations in the growing medium were optimized to gain maximum XMP production (Table 1). As shown in Table 1, an increase in the glu/glc ratio (glu/glc) induced a significant increase in XMP production, and this increase in XMP production was coupled with a reduction in production of the by-product, hypoxanthine. XMP production attained nearly plateau levels at a glu/glc molar ratio of 0.31. Further increases in glu/glc ratio to 0.61 lowered the XMP level (Table 1). Thus, a 0.31 molar ratio of the glu/glc was standardized for the in vivo 31 P NMR measurements to validate the effect of glu on XMP production. Changes in D 4 together with glc, glu, XMP, and hypoxanthine concentrations in this culture condition are shown in Fig. 1. Under this experimental condition, 68.9 m M XMP and 28.9 m M hypoxanthine were obtained at the end of cultivation (Fig. 1B). A rapid increase in the specific XMP produc- tion rate in parallel with the cell growth was observed until 25 h of the culturing period, but the specific XMP production rate then drastically decreased when glu in the growing medium was thoroughly consumed (Fig. 1C). Contrary to XMP, hypoxanthine was not accumulated during the glu-consuming phase (phase I), but much accumulated during the glu-deficient phase (phase II), suggesting that hypoxanthine production was induced through a switch in the carbon flux. In Table 1, glu increased the XMP/Hyp ratio, being dependent on the increased level of glu/glc ratio up to 0.46, which mainly resulted from the extension of the length of phase I. Additionally, a noticeable reduction in either O 2 con- sumption or CO 2 production rates occurred during phase II, which coincided with the onset of hypoxanthine 5 production (Fig. 1D). Cumulatively, it can be concluded that glu availability determines the efficiency of energy production or TCA cycle fluxes, which may specify Table 1. Effect of glutamic acid on XMP and hypoxanthine (Hyp) production in C. ammoniagenes XMP overproducer. 6 XMP and hypoxanthine concentrations were assayed as described in Experimental procedures. The dry cell weight (DCW) was determined by comparison with a D vs. dry weight calibration curve. 7 Glu (mol) Glc (mol) Glu/glc (mol/mol) XMP (mM) Hyp (m M ) XMP/Hyp DCW (gÆL )1 ) 0.000 0.555 0.00 45.0 40.2 1.12 36.1 0.085 0.555 0.15 60.7 37.5 1.62 34.2 0.170 0.555 0.31 68.9 28.7 2.41 32.1 0.255 0.555 0.46 70.8 26.7 2.65 30.5 0.340 0.555 0.61 64.2 38.2 1.68 28.9 Ó FEBS 2003 31 P NMR studies in XMP producing C. ammoniagenes (Eur. J. Biochem. 270) 2623 whether primarily XMP or hypoxanthine production takes place. Energetic status during the XMP production phase To understand further the regulatory mechanism of nucleotide production, dependency of the cellular ener- getic on the carbon source was investigated by measuring cellular ATP and ADP by the CFS system coupled with 31 P NMR. In Fig. 2, typical NMR spectra during the phase of XMP production (phase I) are illustrated with peak assignments based on the previously published data [12,13]. Under our experimental condition, sugar phos- phate and intracellular inorganic phosphate (P in i ) signals at approximately )5–0 p.p.m. were not distinct from the accumulated XMP and inorganic phosphate added as K 2 HPO 4 and KH 2 PO 4 for essential substrates for XMP production. During phase I, ATPb and ATPc plus ADPb signals were maintained at very low intensity levels. During phase II (hypoxanthine producing phase), ATPb signals were unde- tectable, but the ATPc plus ADPb intensity increased fivefold in contrast to those during phase I. Chronological changes XMP producers, i.e. cellular ATP, ADP and Fig. 1. Changes in growth, substrates, and products of C. ammoniagenes in the XMP-production phase. (A) Growth (j), residual glc (s), and residual glu (h). (B) XMP (j) and hypoxanthine (m)production.(C)Glc(s)andglu(h) consumption, and XMP production rates (j). (D) Oxygen consumption (d)andCO 2 production (s) rates. In phase I (P-I), the bacterium consumed glu in parallel with glc, and in phase II (P-II), the bacterium consumed glc as a sole substrate. Cultivation was performed as described in the Experimental procedures. Fig. 2. Representative 31 PNMRspectraof metabolites from the XMP overproducer in phase I (A) and II (B). In spectra A and B, cell density during NMR observations were approximately 8 gÆL )1 and 23 gÆL )1 , respect- ively. Abbreviations for resonances are: methylene diphosphonic acid (MDP), uridine diphosphate glc (UDP-glc), b and c phosphate of adenine nucleotide phosphates (ATPb,and ADPb plus ATPc, respectively). The spectrum consisted of 1280 scans and was acquired at 161 MHz with a spectral width of 16 kHz, a 90° pulse angle, and a recycling time of 1.5 s. Chemical shifts are given in p.p.m. from 85% H 3 PO 4 . 2624 Y. Noguchi et al. (Eur. J. Biochem. 270) Ó FEBS 2003 NADP concentrations during the production phase are shown in Fig. 3. In phase I, ATP concentration was low but above the detectable level, but ADP concentration was undetectable. ADP concentration increased in phase II, while the ATP concentration, in turn, became undetectable. Cellular NADP as shown in Fig. 3B was kept at similar levels throughout the cultivation period. Thus, substrate availability seemed not to change NADP levels. By introducing C. ammoniagenes wild-type strain (ATCC6872) and E. coli W3110, intracellular ATP and ADP concentrations were measured to confirm whether the low energetic status observed in this XMP producer could be reproduced in these two wild-type strains under the same culture condition. Comparisons are made in Table 2. Between the two wild-type strains, cellular ATP plus ADP levels and ATP/ADP ratios were not significantly different each other through phases I and II. Contrary, in the XMP producer during phase I, the average ATP plus ADP concentration, as well as ATP/ADP ratios, were much lower than those in the two wild-type strains (Table 2). ATP plus ADP concentration in the XMP producer, the mutant C. ammoniagenes corresponded to 14% of that in the wild type (Table 2). During phase II, the average ATP plus ADP concentration raised 58% of that in the wild-type strain, but ATP concentration further decreased to an undetectable level (Table 2). Thus, low ATP levels or ATP plus ADP concentrations continuously observed in the XMP-produ- cing phase is specific to this strain and this low energetic status will not be attributable to the unavailability of substrates, but to the specific metabolic characteristic of XMP production itself. A prominent enlargement in the ATP and ADP pool size from phase I to II in the XMP producer was associated with a shift from ATP to ADP production. This shift may be explained by an effect of glu deficiency on the central carbon flux, which specifically occurs during XMP production. In fact, an increase in ATP concentration and ATP/ADP ratio could be induced during phase I by increasing glu concen- tration in the culture medium (Fig. 4). Dauner et al.[14] have simulated a maximization of the flux in riboflavin production in Bacillus subtilis, and proposed the importance of energy supplementation on which activities of the TCA cycle and the respiratory chain depend. Several reports also Fig. 3. Changes in cellular ATP, ADP and ATP/ADP of the XMP overproducer in the production phase. (A) Cellular ATP (s)andADP (d) concentrations are shown. (B) Cellular NADP (j)isshown.To quantify the intracellular metabolites, MDP was used as a concen- trationstandardintheNMR.AlldatawerenormalizedbyDCWas described in Experimental procedures. Table 2. Comparison of cellular metabolites between the XMP-overproducing and wild-type strains. Cultivation conditions are summarized in the Experimental procedures. To quantify the intracellular metabolites, NADP was used as a concentration standard in the NMR. Partially relaxed MDP 8,9 was used to estimate metabolite concentrations. ND, not determined. Values are the mean ± SD. 8,9 Strain Average concentrations (pmolÆg )1 of DCW) Phase ATP ADP ATP+ADP ATP/ADP C. ammoniagenes ATCC6872 I 3.25 ± 0.11 0.50 ± 0.21 3.70 ± 0.46 (1.00) 6.90 (1.00) II 3.22 ± 0.21 0.54 ± 0.21 3.75 ± 0.56 (1.01) 6.83 (0.99) E. coli W3110 I 2.85 ± 0.72 0.45 ± 0.23 3.30 ± 0.71 (0.89) 7.31 (1.06) II 2.73 ± 0.52 0.47 ± 0.23 3.17 ± 0.62 (0.86) 7.21 (1.04) C. ammoniagenes XMP overproducer 10 I 0.50 ± 0. 17 ND 0.51 ± 0.18 (0.14) ND (0) II ND 2.15 ± 1.30 2.18 ± 1.28 (0.58) ND (0) Fig. 4. Effect of glu on intracellular ATP and ADP concentrations in the phase I. Intracellular ATP and ADP concentrations in phase I were estimated based on 31 P NMR data. The data shown are the mean ± SD. Ó FEBS 2003 31 P NMR studies in XMP producing C. ammoniagenes (Eur. J. Biochem. 270) 2625 proposed the importance of the balance between TCA and the oxidative PP cycles in nucleotide production [14–16]. These reports suggested that an enhancement of the net flux of the oxidative PP cycle that supplied carbon skeletons led to a reduction of ATP production due to a reduced TCA cycle flux in nucleotide production [14–16]. In our experi- ment, XMP production decreased sharply during phase II, simultaneous with an increased production of hypoxan- thine, an XMP by-product (Fig. 1). In association with this reduction of XMP production, glc utilization was also reduced, indicating that the reduced XMP flux through the PP cycle affects glc oxidation. This will result in reduction of NADH generation rate from TCA cycle and a decrease in ATP pool. The reduced ATP availability seems to enhance the production of the by-product, hypoxanthine. This cause–result corresponds well with in vivo NMR results in this study, as well as previous reports [15–17]. A consider- able increase in ADP synthesis was observed during phase II; the true reason for this increase remains to be elucidated and whether biosynthesis of adenine nucleotides is speci- fically enhanced or oxidative phosphorylation is specifically reduced. This study represents the first trial of in vivo NMR observation of bacterial nucleotide production. Our results demonstrate that the control of energy metabolism is crucial for bacterial nucleotide production as, for instance, main- tenance of efficient ATP production is able to enhance XMP production. Acknowledgements We are grateful to K. Sato and T. Kazarimoto for their helpful input. References 1. Sauer, U. & Bailey, J.E. (1999) Estimation of P-to-O ratio in Bacillus subtilis and its influence on maximum riboflavin yield. Biotechnol. Bioeng. 64, 750–754. 2. Dauner, M. & Sauer, U. (2001) Stoichiometric growth model for riboflavin-producing Bacillus subtilis. Biotechnol. Bioeng. 76, 132–143. 3. Barrow, K.D., Collins, J.G., Norton, R.S., Rogers, P.L. & Smith, G.M. (1984) 31 P nuclear magnetic resonance studies of the fer- mentation of glcose to ethanol by Zymomonas mobilis. J. Biol. Chem. 259, 5711–5716. 4. Castro,C.D.,Koretsky,A.P.&Domach,M.M.(1999)Perfor- mance trade-offs in in vivo chemostat NMR. Biotechnol. Prog. 15, 185–195. 5. Noguchi, Y., Shimba, N., Toyosaki, H., Ebisawa, K., Kawahara, Y.&Suzuki,Ei.&Sugimoto,S.(2002)In vivo NMR system for evaluating oxygen-dependent metabolic status in microbial culture. J. Microbiol. Methods 51, 73–82. 6. Dulyaninova, N.G., Podlepa, E.M., Toulokhonova1, L.V. & Bykhovsky, V.Y. (2000) Salvage pathway for NAD biosynthesis in Brevibacterium ammoniagenes: regulatory properties of triphos- phate-dependent nicotinate phosphoribosyltransferase. Biochim. Biophys. Acta 1478, 211–220. 7. Han, J.K., Chung, S.O., Lee, J.H. & Byun, S.M. (1997) 6¢-Mercaptoguanosine-resistance is related with purF gene encoding 5¢-phosphoribosyl-1¢-pyrophosphate amidotransferase in inosine-5¢-monophosphate overproducing Brevibacteirum ammoniagenes. Biotechnol. Lett. 19, 79–83. 8. Usuda, Y., Kawasaki, H. & Utagawa, T. (2001) Characterization of the cell surface protein gene of Corynebacterium ammoniagenes. Biochim. Biophys. Acta 1522, 138–141. 9. Neves, A.R., Ramos, A., Nunes, M.C., Kleerebezem, M., Huge- nholtz, J., de Vos, W.M., Almeida, J. & Santos, H. (1999) In vivo nuclear magnetic resonance studies of glycolytic kinetics in Lactococcus lactis. Biotechnol. Bioeng. 64, 200–212. 10. Crosse, A.M., Greenway, D.L. & England, R.R. (2000) Accu- mulation of ppGpp and ppGp in Staphylococcus aureus 8325–4 following nutrient starvation. Lett. Appl. Microbiol. 31, 332–337. 11. Meyer, S., Noisommit-Rizzi, N., Reuss, M. & Neubauer, P. (1999) Optimized analysis of intracellular adenosine and guanosine phosphates in Escherichia coli. Anal. Biochem. 271, 43–52. 12. Lundberg, P., Harmsen, E., Ho, C. & Vogel, H.J. (1990) Nuclear magnetic resonance studies of cellular metabolism. Anal. Biochem. 191, 193–222. 13. Greiner, J.V., Kopp, S.J. & Glonek, T. (1985) Phosphorus nuclear magnetic resonance and ocular metabolism. Surv. Ophthalmol. 30, 189–202. 14. Dauner, M., Bailey, J.E. & Sauer, U. (2001) Metabolic flux ana- lysis with a comprehensive isotopomer model in Bacillus subtilis. Biotechnol. Bioeng. 76, 144–156. 15. Sauer, U., Hatzimanikatis, V., Hohmann, H.P., Manneberg, M., van Loon, A.P. & Bailey, J.E. (1996) Physiology and metabolic fluxes of wild-type and riboflavin-producing Bacillus subtilis. Appl. Environ. Microbiol. 62, 3687–3696. 16. Dauner, M., Sonderegger, M., Hochuli, M., Szyperski, T., Wuthrich, K., Hohmann, H.P., Sauer, U. & Bailey, J.E. (2002) Intracellular carbon fluxes in riboflavin-producing Bacillus subtilis during growth on two-carbon substrate mixtures. Appl. Environ. Microbiol. 68, 1760–1771. 17. Kovarova-Kovar,K.,Gehlen,S.,Kunze,A.,Keller,T.,Daniken, R.V., Kolb, M. & van Loon, A.P. (2000) Application of model- predictive control based on artificial neural networks to optimize the fed-batch process for riboflavin production. J. Biotechnol. 79, 39–52. 2626 Y. Noguchi et al. (Eur. J. Biochem. 270) Ó FEBS 2003 . was investigated by measuring cellular ATP and ADP by the CFS system coupled with 31 P NMR. In Fig. 2, typical NMR spectra during the phase of XMP production. not available. In the present study, we investigate the cellular energy metabolism in XMP -overproducing C. ammoniagenes, monitor phosphate-containing metabolites