Natural-abundance isotope ratio mass spectrometry as a means of evaluating carbon redistribution during glucose–citrate cofermentation by Lactococcus lactis Mohamed Mahmoud, Emmanuel Gentil and Richard J. Robins Groupe de Fractionnement Isotopique de Me ´ tabolismes, Laboratoire d’Analyse Isotopique et Electrochimique de Me ´ tabolismes, Universite ´ de Nantes, France The cometabolism of citrate and glucose by growing Lactococcus lactis ssp. lactis bv. diacetylactis was studied using a natural-abundance stable isotope technique. By a judicious choice of substrates differing slightly in their 13 C/ 12 C ratios, the simultaneous metabolism of c itrate and glucose to a range of compounds was analysed. These end- products include lactate, acetate, formate, diacetyl and acetoin. All these products have pyruvate as a co mmon intermediate. With the objective of estimating the degree to which glucose and citrate metabolism through pyruvate may be differentially regulated, the d 13 C v alues of the products accumulated over a wide range of concentrations of citrate and glucose were compared. It was found that, whereas the relative accumulation of different products responds to both the substrate concentration and the r atio between the sub- strates, the d 13 C values of the products primarily refl ect the availability o f t he two s ubstrates over the entire r ange examined. It can be concluded that in actively growing L. lactis the maintenance of pyruvate homeostasis takes precedence over the redox status of the cells as a re gulatory factor. Keywords: carbon balance; isotope ratio mass spectrometry; lactic acid bacteria; m etabolic regulation; pyruvate. A r ange of simple sug ars can be c atabolized anaerobically by Lactoc occus lactis and other lactic acid bacteria (LAB) in order to obtain energy for growth. Central to this metabo- lism is the C3 compound, pyruvate [1]. This metabolite forms the link between the essentially oxidative reactions of energy p roduction and those required f or the regeneration of reducing equivalents NAD + or NADP + (Fig. 1 ). However, pyruvate is relatively toxic [1,2], necessitating strict control over the level to which it accumulates. In L. lactis it is primarily reduced to lactic ac id by L -lactate dehydrogenase (LDH, EC 1.1.1.27), thus maintaining both pyruvate homeostasis and redox equilibrium. Under appro- priate conditions, however, fermentation leads to C1 and C2 compounds, providing alternative routes for pyruvate catabolism. Thus, varying amounts of acetate, formate and ethanol can b e produced by the actions of p yruvate formate-lyase (PFL, EC 2.3.1.54) or pyruvate dehydro- genase (PDH, EC 1.2.4.1, EC 2.3.1.12, E C 1 .8.1.4). In the case of ethanol production, this provides an alternative means for NAD + regeneration. In addition, some strains of L. lactis,suchasL. lac tis ssp. lactis bv. d iacetylactis, c an metabo lize citrate [ 3,4], w hich leads to enhanced or prolonged growth [5,6]. Citrate metabolism impinges on the pyruvate pool without con- comitant participation in the redox status of the cells (Fig. 1 ). These strains are unusual in their capacity to accumulate the C4 products, d iacetyl, acetoin and butan- 2,3-diol. T his correlation led to a number o f reports that these compounds were products of citrate catabolism [7–9] but recent work has disproved this assumption [10,11]. Although the metabolism of citrate to pyruvate does not consume NAD + , acetoin and butan-2,3-diol production can contribute to NAD + regeneration. The formation of acetoin via the unstable intermediate, a-acetolactate [12,1 3], however, requires 2 mol o f pyruvate, thus providing a less efficient route to NAD + regeneration than lactate, or indeed ethanol, formation (Fig. 1 ). Although a cetoin is the most favoured C4 product, it is diacetyl that is of greater commercial interest as this compound is responsible for the ÔbutteryÕ flavour notes in fermented d airy products. Thus, metabolic regu lation that Correspondence to R. J. Robins, Isotopic Fr actionation in Metabolism Group, Laboratory for the Isotopic and Electrochemical Analysis of Metabolism, CNRS UMR6006, University of Nantes, BP 99208, F-44322 Nantes, France. Fax: +332 51 12 57 12, Tel.: +332 51 12 57 01, E-mail: richard.robins@chimbio.univ-nantes.fr Abbreviations: IRMS, isotope ratio mass spectrometry; LAB, lactic acid bacteria; LDH, L -lactate dehydrogenase; PDH, pyruvate dehydrogenase (acetyl-transferring) complex; PFL, pyruvate formate- lyase; SPME-GC-C-IMRS, solid-phase micro-extraction-GC- combustion-IRMS. Enzymes: acetaldehyde dehydrogenase (EC 1.2.1.10); a-acetolactate decarboxylase (EC 4.1.1.5); a-acetolactate s ynthase (EC 2.2.1.6); ace- tyl kinase ( EC 2.7.2.1); alcohol de hydrogenase (EC 1.1.1.1); citrate lyase (EC 4.1.3.6); diacetyl reductase (EC 1.1.1.5); L -lactate dehydro- genase (EC 1.1.1.27); phosphate acetyl transferase (EC 2.3.1.8); pyruvate dehydrogenase (acetyl-transferring) complex (EC 1.2.4.1 + EC 2.3.1.12 + EC 1.8.1.4); pyruvate formate-lyase (EC 2.3.1.54). (Received 3 0 June 2004, revised 13 A ugust 2004, accepted 23 September 2004) Eur. J. Biochem. 271, 4392–4400 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04376.x results i n diacetyl accumulation has received considerable interest [14,15] and, as has been previously argued [11], is complex. What is apparent is that the cometabolism of citrate and glucose leads to the enhanced production of these compounds by re-routeing of the metabolic through- put [3,7–9,16,17]. Because t his is not specifically due to the metabolism of citrate to the C4 compounds [10,11], it must reflect an overall shift in the balance between different routes for pyruvate catabolism. This shift could be a response to a n altered redo x status or simply an up-shift in the size o f the pyruvate supply. Currently, it is not clear which is the more important of these factors. The regulation of glycolysis in LAB has been studied extensively and a number of potential regulatory points proposed ([1,18] and r eferences therein). However, t here is compelling evidence that neither the [pyruvate] [1,18] nor the NAD + /NADH balance [19], nor the level of ldh expression [20] regulates glycolysis. Notably, genetic manipulation of key glycolytic enzymes has failed to identify one specific control point in the pathway for pyruvate p roduction from glucose [18]. Furthermore, genetic manipulation of pyruvate catabolism, such as varied expression of ldh [20–22] or enhanced a-acetolactate synthase (EC 2 .2.1.6) production [23], c an substantially alter total metabolic throughput in the alternative pathways of pyruvate catabolism. Hence, it may b e suggested that in L. lactis pyruvate throughput plays a more important regulatory role than does pyruvate input. This hypothesis h as been tested by examining the total carbon redistribution from glucose and citrate during cofermentation under a range of concentrations of both cosubstrates. A difficulty in unravelling LAB metabolic throughput is the continuous change in environment that takes p lace as cells grow and substrate is consumed. The study of the redistribution of 13 C-label in nongrowing cells helps indicate m etabolite t urnover and concentrations [24] but does not show the throughput during growth conditions [16]. Similarly, modelling of flux has been restricted to situations with only a single fermentable substrate p resent and requires a ssumptions about the steady-state levels of metabolites [25]. As two pathways are active simultaneously and both l ead to the key intermediate, pyruvate, it is crucial to understand the extent to which their i ndividual t hrough- puts are interdependent. To overcome these difficulties and to measure directly the total carbon redistribution in actively growing LAB during glucose–citrate cofermentation, we have developed an approach that exploits the small variation in n atural 13 C content between substrates de rived f rom different biological sources [10]. These small differences can b e d etermined by isotope ratio mass spectrometry (IRMS) on the relative d 13 C scale with an accuracy of at least ± 0.2& [26]. The relative d 13 C scale is used routinely to compare different 13 C/ 12 C ratios. The scale is standardized against a calibrated reference (R) of known 13 C/ 12 C r atio and the value of the unknown (S) is expressed in & according to the formula: d 13 C ¼ 13 C 12 C hi S 13 C 12 C ÂÃ R À 1 0 @ 1 A Â 100 By fermenting glucose and citrate that differ by % 15&, intermediate values of d 13 C determined for the various fermentation products can be used to calculate the proportion originating from each of these two possible substrates. B y t his means, w e h ave previously shown t hat, under one defined set of conditions, glucose an d citrate contributed to the C4 compound s and lactic acid in proportions closely represe nting the availability of the two carbon sources [10]. Thus, their metabolic equivalence at the level of pyruvate was implied. Further investigation, however, indicated that the proportional utilization varied depending on the environment; notably that the relative availability o f t he substrates and t he le vel of advancement of the f ermentation could influence the d 13 C determined f or the p roducts [11]. esoculg 5.0 etartic etavuryp etatcal α etatcaloteca- etamrof A oC-lyteca HDAN D AN + HDAN DAN + nioteca l y tecaid P-lyteca eta teca HDAN DAN + PDA PTA HDAN DAN + PTA PDA 1 2 4 a6,5 7 b6 3 8 9 31 21 01 ed y hedlatec a D A N + HD A N HDAN DAN + l o n a hte 11 HDAN D AN + l o id - 3 , 2 - n a tu b 7 Fig. 1. Schematic p athway showing the key metabolic relationships between the substrates and p roducts in glucose–citrate cofermentation. 1, glycolysis; 2, citrate lyase (EC 4.1.3.6); 3, L -lactate dehydrogenase (EC 1.1.1.27); 4, a-acetolactate synthase (EC 2.2.1.6); 5, a-acetolactate decarboxylase (EC 4.1.1.5); 6a, nonenzymatic decarboxylation; 6b, nonenzymatic oxidative decarboxylation; 7, diacetyl dehydrogenase (EC 1.1.1.5); 8, pyruvate dehydrogenase (acetyl-transferring) c omplex (EC 1.2.4.1, EC 2 .3.1.12, EC 1.8.1.4); 9, pyruvate formate-lyase (EC 2.3.1.54); 10, acetaldehyde dehydrogenase (acetylating) (EC 1.2.1.10); 11, alcohol dehydrogenase (EC 1.1.1.1); 12, phosphate acetyl transferase (EC 2.3.1.8); 13, acetyl kinase (EC 2.7.2.1). Ó FEBS 2004 Carbon redistribution in L. lactis cofermentation (Eur. J. Biochem. 271) 4393 In order to examine the relationship between the consumption of glucose and of citrate and the accumulation of the products of pyruvate catabolism, the effect of varying the glucose and citrate availability has been examined. Both substrates have been varied over a four-to-fivefold range of concentration and the relationships between the concentra- tions and d 13 C values of s ubstrates and products u sed to construct a balance sheet for carbon redistribution. It is found that the d 13 C values o f the products primarily reflect the relative input to the pyruvate pool of the two substrates, thus co nfirming experimentally the key role proposed for pyruvate in metabolic regulation in L. lactis [1,25]. Materials and methods Bacterial strains and culture conditions Lactococcus l actis ssp. lactis bv. diacetylactis strain B7/2147 was obtained from the Collection of L actic Acid Bacteria (Institute of Food Research, Norwich, UK: collection no. B7/2147). This strain has a high capacity to produce diacetyl. The culture was stored at )80 °C i n M17 medium [27] with 15% (v/v) glycerol. Routine culture was in sterile (20 min; 121 °C; 10 5 NÆm )2 ) M17 broth, appropriately supplemented with glucose or citrate–glucose, in 200 mL Duran bottles, as described previously [10]. The general fermentation c ondi- tions were: anaerobic (static, closed), 30 °C, pH initially 6.3 ± 0.1 ( HCl) and left to evolve freely. Cultures were initiated with an 8-h preculture from t he same medium and harvested after complete consumption of substrates (16–22 h), the supernatant being recovered by centrifuga- tion (4500 g,10min,4°C) and kept at )20 °C. The d 13 C initial values for the citrate and glucose fermented were 13 C glucose ¼ )10.7& and d 13 C citrate ¼ )24.7&.The concentration of these substrates was varied from 13.9 to 55.5 m M for glucose and 0 to 34.8 m M for citrate. Reference conditions used 27.8 m M for glucose and 13.9 m M for citrate. Metabolite analysis and isotopic determinations Metabolite concentrations in the culture medium were determined d irectly on the culture filtrate by 1 HNMR using an external e lectronic reference, as described previ- ously [28]. The d 13 C acetoin and d 13 C diacetyl values were determined by solid-phase micro-extraction-GC-combustion-IRMS (SPME-GC-C-IRMS) as described previously [29]. The d 13 C acetate value was d etermined under the sam e conditions. Essentially, these products were recovered from the head- space above a sample of the f ermentation broth using polydimethylsiloxane-divinylbenzene-coated fibres (Supe- lco) and introduced directly into the injector of an HP6890 gas chromatograph (Agilent T echnologies) linked on-line to a combustion interface and an IRMS (Finnegan Mat Delta S, Finnegan). Separation was accomplished using a Stabilwax column (length, 60 m; i.d., 0.32 mm; film thickness, 0.5 lm; Restek). Samples w ere i ntroduced via a split/splitless injector (splitle ss mode, 250 °C) and chroma- tographed under the following conditions: vector gas, He; flow rate 2.2 mLÆmin )1 (constant pressure); temperature gradient, 50 °C for 0.1 min, followed by an increase of 10 °CÆmin )1 to 200 °C, then 200 °C for 2 min. Each sample was analysed at least three times and compounds were identified by reference to authentic standards. Measured d 13 C values were corrected for slight shifts to ward the negative associated with t he use of the SPME-GC-C-IRMS protocol. Corrections were based on standard solutions containing acetic acid, acetoin and diacetyl for which d 13 C values were determined by elemental analyser-IRMS. Correction factors applied were acetic acid + 0.4&,acetoin +0.2& and diacetyl +0.6&. Lactic acid was purified from culture filtrate and the d 13 C lactate values were determined by elemental analyser- IRMS (Finnegan Mat Delta E, Finnegan) on encapsulated samples as described previously [10]. Results In order for th e analysis of 13 C redistribution to b e valid, three criteria should be fulfilled. First, i t is essential that all the available s ubstrates ar e consumed. Second, it is neces- sary to account for all the available carbon among the products of ferm entation. Third, i t is preferable t hat no or little catabolism of the initial fermentation products has taken place. Thus, a d etailed quantitative analysis of the different metabolites is a prerequisite for interpreting the d 13 C values in terms of 13 C redistribution and this information is summarized below. Influence of the [glucose]/[citrate] ratio on product accumulation Fermentation was always conducted under static, closed, but not strictly anaerobic growth conditions with glucose and citrate as the only substrates and an initial pH of 6.3 ± 0.1. A range of [glucose]/[citrate] from 0.8 to 4.0 was used, with concentrations varying from 13.9 to 55.6 m M for glucose and 6.9 to 34.8 m M for citrate. With the exception of the highest [glucose], fermentation of both substrates present was complete by 16 h. For [glucose] at 41.7 and 51.6 m M , fermentation was complete by 18 h. These end points were used for all further analyses. Growth and final pH both varied considerably depending on the quantities of substrates available (Table 1). As anticipated, increasing glucose availability gave hig her cell density and high [glucose] led to elevated [lactate] (Table 1) and a concomitant low pH. Growth was slightly diminished by high [citrate] (Table 1) and the fall in pH was less evi- dent, presumably because of the n egative acidity b alance associated with citrate catabolism (3 · COO – giving 2 · COO – +CO 2 ) [5]. This growth inhibition did not prevent fermentation as all available substrate was con- sumed over the whole range of concentrations used (Table 1). H owever, it was associated with a shift in the balance of products accumulated. The fermentation o f glucose alone led entirely t o lactate and a low level of acetate, no other product being detected. In cofermentation, lactate was always the principal product and showed a strong correlation with the available glucose (Table 1). In all cases, the lactate p roduced was between 80 and 95% of the theoretical yield from glucose consumed but never exceeded 100% even at the higher [citrate]. Indeed, an 4394 M. Mahmoud et al. (Eur. J. Biochem. 271) Ó FEBS 2004 increased availability of citrate had no influence on the total lactate produced. This i s compatible w ith the tightly linked redox balance of g lucose metabolism, glucose to p yruvate producing 2 mol of NADH per mol of glucose that are then consumed by pyruvate to initiate metabolism. As citrate metabolism to pyruvate does not generate NADH, this cannot be linked to lactate production because the total lactate produced cannot exceed the maximum theoretical yield from glucose due to the redox constraints. The picture for acetate accumulation is the reverse (Table 1). As expected from the known metabolism o f citrate (Fig. 1), [acetate] showed a strong correlation with available citrate. However, in all cases, the acetate accumu- lated was significantly higher than 100% of t he theoretical yield f rom t he activity of citrate lyase (EC 4.1.3.6), indica- ting that a proportion was derived from pyruvate. No significant influence of the [glucose] was seen, the [acetate] remained constant (19.1 ± 0.4 m M ) over a fourfold change in [glucose], indicating that % 30% of the acetate was derived from pyruvate w hen [citrate] ¼ 13.9 m M .Increas- ing [citrate] led to a slight overall increase in net acetate derived from pyruvate, from 3.5 m M at 0 m M citrate to 7.5 m M at 34.8 m M citrate. Hence, it is evident that neither lactate nor acetate production from pyruvate is strongly affected by citrate catabolism. Acetate production from pyruvate can have two main consequences. It can act to diminish the pool of pyruvate under conditions in which the major mechanism – reduction to lactate – is inadequate. It can also act to provide ATP via the acti on of acetyl kinase (EC 2.7.2.1). Anaerobic conditions favour pyruvate catabolism to acetate via PFL (Fig. 1), which will result in a 1 : 1 ratio for formate/acetate. At low or zero [citrate], n o formate was detected (Table 1), even though pyruvate- derived acetate was present, whereas at 13.9 m M citrate, the [formate] was consistently 65–70% of the pyruvate- derived acetate. This indicates t hat, although PFL activity was the princ iple source of acetate, either acetate from an alternative catabolism was also being produced or formate was being degraded. As L. lactis lacks formate dehydrogenase, the latter option is ruled out. The most likely source of this additional acetate is PDH, which, although typically associated with aerobic conditions, does show some activity in anaerobic fermentation [30]. Although PDH-mediated acetyl-CoA production is unfa- vourable as it produces NADH, it fulfils both objectives of decreasing the pyruvate pool and providing acetyl-P for ATP generation. Because ethanol was not detected by 1 HNMR inthe medium from any o f these experiments ( data not shown), acetyl-CoA formation was not linked to N AD + regener- ation. This strongly suggests that both PFL and PDH are primarily involved in regulating the size of the pyruvate pool, rather than in maintaining the redox status of the cells. In effect, the production of pyruvate from citrate does not generate NADH. If this enhanced pyruvate production led to enhanced lactate production, the NADH/NAD + balance w ould be disequilibrated. Acetate production by PFL effectively utilizes pyruvate without interfering with the NADH/NAD + balance. By contrast, excess acetate production can also be detrimental to cell growth. What, then, is the role of the alternative pathways of pyruvate catabolism that lead t o the formation of t he C4 compounds, acetoin and diacetyl? The availability of citrate rather than of glucose was, as found previously, the determining factor for the accumulation of these com- pounds (Table 1). Under all conditions, about twice the amount of acetoin was found than of diacetyl but no butan- 2,3-diol was d etected, either by 1 H N MR or by GC (data not shown). At constant 13.9 m M citrate, a fourfold increase in [glucose] (13.9–55.6 m M ) caused both [diacetyl] a nd [acetoin] to increase % twofold. Increasing the [citrate] 2.5- fold, however, led to an % 3.5-fold increase in both acetoin and diacetyl. This pathway has the potential to contribute to both the pyruvate homeostasis and t he redox status of the cells, as 2 mol pyruvate can be used for the regeneration of 1mol NAD + via diacetyl reductase (EC 1 .1.1.5) [31,32]. However, this route to NAD + regeneration appears to have been negligible in this study, as strains containing diacetyl reductase generally produce butan-2,3-diol, because of the activity of the same enzyme on acetoin [32]. Under anaerobic culture conditions, oxidation of acetoin to diacetyl is extremely unlikely [3]. Consequently, ac etoin Table 1. Product accumulation profiles for fermentation of Lactococcus l actis with diffe ring initial amounts of glucose and citrate. N, number of repeat fermentations in these conditions. A is the o ptical dispersion at 550 n m. Duration (h) N a Initial concentration Ratio G/C Final pH Final A Lactate (m M ) Acetate (m M ) Formate (m M ) Diacetyl (m M ) Acetoin (m M ) Balance (%) Glucose (m M ) Citrate (m M ) 16 4 13.9 13.9 1.0 5.9 1.02 31.33 ± 2.26 b 18.97 ± 0.60 3.62 ± 0.25 1.15 ± 0.18 2.48 ± 0.57 112.9 ± 7.9 16 5 27.8 13.9 2.0 5.1 1.53 53.26 ± 2.47 19.10 ± 0.90 3.67 ± 0.18 1.25 ± 0.13 2.92 ± 0.38 101.5 ± 8.2 18 a 1 41.7 13.9 3.0 4.6 1.64 83.96 ± 7.47 19.31 ± 0.40 4.00 ± 0.07 2.22 ± 0.19 3.56 ± 0.40 107.8 ± 3.3 18 a 1 56.6 13.9 4.0 4.6 1.66 90.63 ± 4.75 20.38 ± 0.80 4.20 ± 0.45 2.85 ± 0.19 4.16 ± 0.92 90.6 ± 4.1 16 2 27.8 0.0 – 4.4 1.32 54.90 ± 0.85 3.45± 0.78 0.00 0.00 0.00 107.6 ± 3.8 16 5 27.8 6.99 4.0 4.8 1.52 55.16 ± 1.19 11.46 ± 0.44 0.00 0.00 0.00 95.6 ± 2.0 16 4 27.8 20.9 1.3 5.4 1.38 58.60 ± 3.06 27.10 ± 0.73 1.92 ± 1.14 2.50 ± 0.58 4.70 ± 0.93 106.0 ± 7.0 16 4 27.8 34.8 0.8 6.0 1.35 56.15 ± 1.63 42.11 ± 1.88 5.96 ± 2.23 4.62 ± 0.20 9.45 ± 0.21 107.9 ± 2.8 a At high [glucose], fermentation was not complete at 16 h but no substrates remained at 18 h. b Combined SD is given for the number of fermentations and for the replicate measurements in each fermentation. Ó FEBS 2004 Carbon redistribution in L. lactis cofermentation (Eur. J. Biochem. 271) 4395 must have been produced by the d irect decarboxylation o f a-acetolactate, a conclusion confirmed by the absence of any accumulation of butan-2,3-diol. Both products accu- mulate in strains lacking a-acetolactate decarboxylase [22,33,34] and the high diacetyl/acetoin ratio observed i n our experiments s uggests that L. lactis B7/2147 has dimin- ished a-acetolactate decarboxylase activity (C. Monnet, INRA, Paris-Grignon, France, personal communication). The extent to which nonenzymatic decarboxylation of a-acetolactate lead s to acetoin or to diacetyl is strongly dependent on the p revailing c onditions of culture, notably pH [35], O 2 [14,36] and the presence of metal ions [37] or other oxidizing agents [38] in the medium. Nevertheless, irrespective of the route by which acetoin and diacetyl a re formed, their biosynthes is appears to p lay no role in the control of the redox status of the cells. Rather, it appears that, once again, the major role of this alternative pathway is to regulate the size of the pyruvate pool. If this is the case, pyruvate catabolism should be independent of the substrate supplying the p yruvate: if it is not, then a link should b e seen between the amount of each substrate being metabolized and the redistribution of carbon into th e different products of pyruvate metabolism. These alternatives can be tested by relating the d 13 Cvalues in the products accumulated to those of t he substrates supplied. Influence of the [glucose]/[citrate] ratio on d 13 C values and substrate redistribution between products Although t he final concentrations of products indicate the total throughput for different catabolic routes, these cannot discriminate between th e utilization of alternative substrates for the different products. H owever, this can be deduced from the relationship between the initial d 13 C glucose and d 13 C citrate values of the substrates and the d 13 C lactate , d 13 C acetate , d 13 C diacetyl and d 13 C acetoin values a t term. (Dat a for d 13 C formate could not be obtained, as this product was not sufficiently w ell r esolved i n t he GC-C-IRMS.) Prelim- inary data for a limited range of substrate conditions showed that some of these f actors are r elated [11] althou gh, notably, no data for acetate were presented. In Table 2 are presented values for d 13 C for cultures using various concentrations of citrate (initial d 13 C citrate ¼ )24.7&)and of glucose (initial d 13 C ¼ )10.7&). The d 13 C lactate produced in the absence of citrate had a value of )12.5&, showing a Dd 13 C lactate ¼ )2&, as f ound previously [10,11]. In the r eference conditions (27.8 m M glucose, 13.9 m M citrate), the d 13 C lactate ¼ )14.8& is also in good agreement with previous values. In contrast to the effect on [lactate], both [glucose] and [citrate] influenced the value of d 13 C lactate . Thus, as [citrate] increased, the d 13 C lactate steadily became more negative, reaching )16.2& at 34.8 m M (Fig. 2 A). In contrast, the influence of citrate was diminished as [ glucose] increased, a value of d 13 C lactate ¼ )13.6& being f ound at 56.6 m M glucose. Thus it is clear that both substrates were being used in all conditions to give rise to lactate, even though [ citrate] had no influence on [lactate]. The d 13 C acetate produced in the absence of citrate had a value of )10.9&,evenclosertothatofglucosethanthe d 13 C lactate .However,asexpected,thed 13 C acetate was strongly influenced by [c itrate] (Fig. 2B). Thus, at the lowest [citrate] of 6.9 m M ,thed 13 C acetate ()19.7&) had shifted significantly closer to d 13 C citrate , reflecting the fact that even in these conditions at least 60% of the a cetate comes from citrate (Table 1 ). As [citrate] further increased, the d 13 C acetate became more negative and a t 34.8 m M citrate the d 13 C acetate was not significantly d ifferent from the initial d 13 C citrate .In contrast, even the highest [glucose] (55.6 m M )hadno significant influence o n the d 13 C acetate value, in agreement with at least 68% of the a cetate being derived from citrate (Table 1). In comparing the data in Table 1 and Fig. 2, it appears that the availability of each substrate, t he concentrations of products and the d 13 C values o f t he products do not bear direct relationships to each other. For example, although [lactate] shows a strict correlation with [glucose], w ithout any influence of [citrate], the d 13 C lactate values evolve in relation to the r elative a vailability o f citrate. At first sight, this might indicate metabolic control interaction between the two pathways supplying pyruvate. However, a different interpretation emerge s when these parameters are related to the throughput of the pyruvate pool. For cultures grown on glucose alone, d 13 C lactate ()12.5&)andd 13 C acetate ()10.9&) can b e determined in the absence of c itrate. The d 13 C lactate is close t o the d 13 C glucose , the difference probably being due to pyruvate conversion to lactate, while the d 13 C acetate is insignificantly different. That these values only differ slightly from the d 13 C glucose indicates that fractionation between glucose and pyruvate is small or negligible. Thus, it is possible to model the d 13 C pyruvate values using the molar production ratios for p yruvate f rom g lucose (2 : 1) or from citrate (1 : 1), the concentration ratios, and the known d 13 C glucose and d 13 C citrate : d 13 C calc pyr ¼ 2:mol glc 2:mol glc þ mol cit  Á d 13 Cglc þ mol cit 2:mol glc þ mol cit  Á d 13 Ccit where pyr ¼ pyr uvate, glc ¼ glucose, cit ¼ citrate and calc ¼ calculated. The resulting predicted d 13 C pyruvate val- ues are plotted w ith the measured values for the products in Fig. 3. It can be seen from Fig. 3A that, even though the evolution of the d 13 C lactate follows th e same tendency as the d 13 C pyruvate , there is a divergence for high [glucose] at constant [citrate]. This indicates that t he relative input from glucose at high [glucose] is lower than theoretically expected. This could indicate that glycolytic pyruvate production is saturated or is b eing downregulated at the higher [glucose] but that simultaneous pyruvate production from citrate shows n o such constraint. This explanation is supported by therelativeevolutionofthed 13 C values i n conditions of constant [glucose] and increasing [c itrate] (Fig. 3B). Here, the tendency is for the d 13 C lactate to approach the theoretical d 13 C pyruvate as [citrate] i ncreases. Thus, e ven though [citrate] has no influence on [lactate] it influences the carbon redistribution from the common pool of pyruvate. This supports a model in which [pyruvate] does regulate g lyco- lytic input to the pyruvate pool [1]. In contrast, t he d 13 C acetate shows a lack of correlation with the d 13 C pyruvate (Fig. 3 A,B). Only w hen no citrate is 4396 M. Mahmoud et al. (Eur. J. Biochem. 271) Ó FEBS 2004 present is d 13 C acetate close to d 13 C pyruvate .Withconstant [citrate], [acetate] is unchanging irrespective of [glucose] (Table 1), t he proportion derived from pyruvate is invari- able, and the d 13 C acetate does not significantly vary (Table 2). Obviously, in t he absence of citrate the d 13 C acetate is close to the calculated d 13 C pyruvate but as [citrate] increases, there is a rapid trend towards the d 13 C citrate . In fact, however, the measured values in these two series of conditions are exactly as predicted by a proportionation model in which the d 13 C acetate is broken down into the part derived from pyruvate and that derived directly from citrate: d 13 C calc acetate ¼ mol ac À mol cit mol ac  Á d 13 C calc pyruvate þ mol cit mol ac  Á d 13 C citrate where calc ¼ calculated, a c ¼ acetate and c it ¼ citrate. The measured d 13 C acetate values are seen to follow closely the values obtained b y calculation (Fig. 3C). Hence, it c an be concluded that acetate production, even more so than lactate production, simply follows the input to th e pyruvate pool. As acetoin a nd diacetyl are not present in the absence of citrate, no values for d 13 C acetoin and d 13 C diacetyl produced exclusively from glucose could be obtained. At 6.9 m M citrate, when no acetoin or d iacetyl could be detected by 1 H N MR, the concentrating e ffect of the S PME fibre did allow d 13 C acetoin and d 13 C diacetyl values to be d etermined, although the low concentrations mean that the values should be treated with caution. However, the fact that they are both close to the initial d 13 C glucose indicates that the a-acetolactate pathway is being supplied with pyruvate from a common pool, which, as already shown for these conditions, is strongly dominated by glucose. As the [citrate] increases, so the d 13 C acetoin and d 13 C diacetyl values are displaced towards the initial d 13 C citrate (Fig. 3 B). Similarly, augmenting [glucose] leads to values that tend towards 13 C glucose (Fig. 3 A). Both values retain approxi- mately the same r elationship t o t he calculated d 13 C pyruvate , indicating that the source of carbon used for their synthesis is directly related to the metabolism of both available 6.9 13.9 20.9 34.8 13.9 41.7 –25 –20 –15 –10 δδ δδ δδ δδ 13 C Acetoin Citrate (mM) Glucose (mM) 6.9 13.9 20.9 34.8 13.9 41.7 –25 –20 –15 –10 13 C Diacetyl Citrate (mM) Glucose (mM) 0 6.9 13.9 20.9 34.8 13.9 41.7 –25 –20 –15 –10 δδ δδ 13 C Lactate Citrate (mM) Glucose (mM) 0 6.9 13.9 20.9 34.8 13.9 41.7 –25 –20 –15 –10 δδ δδ 13 C Acetate Citrate (mM) Glucose (mM) AB DC (‰) (‰) (‰) (‰) Fig. 2. The e ffect of varying the citrate and glucose c oncentrations. (A) Final d 13 C lactate , (B) final d 13 C acetate ,(C)finald 13 C acetoin ,(D) final d 13 C diacetyl .Eachd 13 C value ( &)repre- sents the mean of one to five fermentations each analysed in triplicate, for which the appropriate standard error is given in Table 2. Table 2. Values o f d 13 C(&) d etermined for products of fermentation of Lact ococcu s l ac tis with differing initia l a moun ts of glucose and citrate. ND, not determined. Glucose initial a (m M ) Citrate initial a (m M ) Ratio d 13 C lactate (&) d 13 C acetate (&) b d 13 C diacetyl (&) b d 13 C acetoin (&) b 13.9 13.9 1.0 )15.82 ± 0.24 c )23.20 ± 0.14 )17.3 ± 0.09 )21.3 ± 0.08 27.8 13.9 2.0 )14.45 ± 0.40 )23.53 ± 0.04 )15.57 ± 0.14 )17.91 ± 0.05 41.7 13.9 3.0 )13.87 ± 0.13 )22.10 ± 0.12 )12.19 ± 0.12 )17.52 ± 0.08 56.6 13.9 4.0 )13.63 ± 0.18 )23.56 ± 0.18 )12.65 ± 0.29 )15.24 ± 0.18 27.8 0 – )12.51 ± 0.61 )10.9 ± 0.19 ND ND 27.8 6.9 4.0 )13.67 ± 0.22 )19.17 ± 0.04 )8.90 ± 0.41 )13.30 ± 0.11 27.8 20.9 1.3 )15.09 ± 0.19 )23.11 ± 0.17 )13.60 ± 0.13 )19.60 ± 0.35 27.8 34.8 0.8 )16.23 ± 0.35 )23.70 ± 0.04 )14.70 ± 0.17 )19.38 ± 0.11 a Initial d 13 C, glucose ¼ )10.7&, citrate ¼ )24.7&. b d 13 C Corrections were acetate +0.4&, diacetyl +0.6&, acetoin +0.2& (see Materials and methods). c Combined SD is given for the number of fermentations and for the replicate measurements in each fermentation; for N, see Table 1. Ó FEBS 2004 Carbon redistribution in L. lactis cofermentation (Eur. J. Biochem. 271) 4397 substrates. I n t his, these c ompounds mimic the pyruvate- derived acetate. However, as noted previously [10,11], the d 13 Cvalues are influenced not only by the [citrate]/[glucose] r atio (Table 2), but also by the availability of citrate. Thus, at 13.9 : 56.6 m M theyaremorenegativethanat 6.9 : 27.8 m M , despite the r atio of 4 : 1 being maintained. The d 13 C lactate , in contrast, shows the same value for both sets of concentrations. This indicates that, although there is a strong influence of the inputs to the pyruvate pool, there is a secondary influence of the [citrate]. This could result from the production rate of diacetyl and acetoin varying throughout the fermentation, reflecting variations in the rate of citrate metabolism relative to t hat of g lucose. Such variation could be induced, for example by c hanges in pH, as citrate transport (but not glucose transport) is sensitive to this factor [39]. F urther analysis of a range of ratios and of the kinetics of the evolution of the d 13 C acetoin and d 13 C diacetyl values is required to define this effect. The d 13 C diacetyl value is consistently 3–4& more positive than the d 13 C acetoin (Table 2), a difference varying only slightly with changes in the availability of g lucose and citrate. This difference has a lso b een found to be retained throughout the time-course of the fermentation for L. lactis B7/2147 [11]. F urthermore, d 13 C diacetyl is consisten tly close to the theoretical d 13 C pyruvate value, whereas d 13 C acetoin is always 3– 6& more negative. Under anaerobic c onditions, diacetyl is produced only through the nonenzymatic decarboxylation of a-acetolactate, whereas acetoin may be derived by either the nonenzymatic or the enzymatic decarboxylation of a-acetolactate (Fig. 1). The high accu- mulation of diacetyl and the lack of butan-2,3-diol indicates that diacetyl dehydrogenase activity is negligible. It is proposed that the strain B7/2147 accumulates unusually high levels of diacetyl because of a deficiency in a-aceto- lactate d ecarboxylase ( C. Monnet, INRA, Paris-Grignon, France, personal communication). The discrepancy in the d 13 C values may indicate, however, t hat L. lactis strain B7/ 2147 has diminished, rather than deleted, a-acetolactate decarboxylase activity because strains characterized as lacking a-acetolactate decarboxylase [22] do not show a similar large D(d 13 C diacetyl –d 13 C acetoin ) [40]. Nonenzymatic decarboxylation shows a range of isotope effects [ 41], whereas enzymatic decarboxylation g enerally selects against 13 C [42,43]. Furthermore, previous evidence indicates t hat biologically produced acetoin, as opposed to chemically synthesized acetoin, is impoverished in 13 C in the hydroxy- methylene g roup relat ive to the k eto g roup [4 4]. H ence, the data (Table 2) support t he hypothesis that the acetoin is derived by both enzymatic and n onenzymatic decarboxyla- tion of a-acetolactate, whereas the diacetyl is produced only nonenzymatically. Further work is required to clarify this aspect of pyruvate metabolism. Discussion The role of pyruvate and the regulation of pyruvate metabolism have b een much discussed in terms of the overall regulation of LAB metabolism [1,3]. By following the simultaneous cometabolism of glucose and citrate in actively growing c ells of L. lactis, our data show that the accumulation of pyruvate-derived metabolites depends principally on the t hroughput of the pyruvate pool. With glucose as sole substrate, throughput is apparently regu- lated with reference to the maximal glycolytic capacity. Thus, the pool of pyruvate is limited by glycolysis and only small amounts of products other than lactate are observed. This is in agreement with the known relative affinities of LDH, PDH, PFL and a-acetolactate synthase ([45] and refs therein). As recently suggested, the role of glycolysis is almost exclusively t o supply ATP and throughput is probably maximal in rapidly growing anaerobic c ultures [18], the utilization of pyruvate by LDH being balanced by its supply. Thus, pyruvate does not accumulate and problems with its toxicity are avoided. A –25.0 –20.0 –15.0 –10.0 –5.0 Glucose (mM) δδ δδ 13 C (‰) δδ δδ 13 C (‰) B –25.0 –20.0 –15.0 –10.0 –5.0 Citrate (mM) C –25.0 –20.0 –15.0 –10.0 –5.0 Concentration (mM) δδ δδ 13 C acetate (‰) Fig. 3. The relationship between calculated d 13 C pyruvate values and reaction products. Th e relationship b etween calculated d 13 C pyruvate and measured d 13 C lactate , d 13 C acetate , d 13 C acetoin ,andd 13 C diacetyl (A) at constant [citrate] and variable [glucose], (B) at constant [glucose] and variable [citrate]. The c alculated d 13 C pyruvate values are obtained from the data in Tables 1 and 2 and the known m olar participation of each substrate to pyruvate formation. Legend: c alculated pyruvate (r), lactate (j), d iacetyl (m), acetoin (n), acetate (d). (C) The relationship between the c alculated d 13 C acetate (line) and the measured d 13 C acetate (symbol). The calculated d 13 C acetate values are obtained from the data in Table 1 and the calculated d 13 C pyruvate values. Legend: at constant [glucose] and variable [citrate] (broken lin e, d); constant [citrate] a nd variable [glucose] ( solid line, j). 4398 M. Mahmoud et al. (Eur. J. Biochem. 271) Ó FEBS 2004 Recent studies have indicated that none of a number of proposed control factors – the NAD + /NADH ratio [19], the glyceraldehyde-phosphase dehydrogenase activity [46], the LDH activity [20] or the phosphofructokinase activity [18] – actually controls glycolytic flux. That glycolysis is essentially unregulated under low to moderate [glucose] is shown by our data, wh ich demonstrate that the availab- ility of citrate leads to a net increase in the pyruvate productive capacity without a ny concomitant inhibition of glycolytic input. When the d 13 C values of the products of these pathways – lactate, acetate, diacetyl and acetoin – are examined, it is clear that they primarily reflect the relative input to the pyruvate p ool, in this case governed by the relative availability of glucose and citrate. While there is some i ndication of limited feedback regulation on glycolysis at high [glucose], no Ôcross-talkÕ between citrate and glucose metabolism was detected. Rather, pyruvate production is essentially unchecked and alternative path- ways of pyruvate catabolism are required to maintain pyruvate homeostasis and p revent pyruvate toxicity. T his directly supports the p ropositions of Koebmann et al.[18] and Neves et al. [19] that input to and ou tput from the pyruvate pool are regulated by factors external to the primary metabolic pathways. Increasing throughput into pyruvate from citrate leads to a progressive increase in the activity of alternative pathways. Even so, it is found that the d 13 C v alues f or all the products reflect the i nput into the pyruvate pool. That augmenting [citrate ] le ads to an increase in pyruvate- derived products i n the alter native pathways i ndicates that the LDH capacity becomes limiting. This is confirmed b y flux control analysis, which suggests that the LDH capacity in wild-type L. lactis cells is only % 70% in excess of the glycolytic rate [20], and by strains with diminished LDH activity, which a ccumulate higher levels of other products, even at low su bstrate supply or in the absence of citrate [21,22,40]. There appears to be no correlation between the activity of the given alternative pathways and a need to regenerate NAD + , as indicated by the l ack of ethanol accumu lation in the current system. Rather, it appears that lactate prod uction is sufficient to satisfy this need and the metabolism of acetate to regenerate NAD + is not required. Instead, acetate production from pyruvate can be seen as an ATP- generating process. As high [acetate] occurs, ethanol production could be inhibited [47] but, because L. lactis ldh – can accumulate ethanol even in th e presence of citrate (C. Monnet, INRA, Paris-Grignon, France, personal communication) [40], this appears improbable. Therefore, it can be argued that the most important role of the alternative c atabolic uses of pyruvate is to m aintain a low [pyruvate]. Hence, pathways in which no N ADH con- sumption occurs but in which ATP generation is possible (acetate via PDH and PFL) are favoured over those which consume NADH (ethanol and butan-2,3-diol) because t o consume NADH would t end to disequilibrate the glucose-to-lactate redox balance (Fig. 1). In these studies, neither product was found, suggesting that both acetoin a nd acetyl-CoA reduction were absent. In conclusion, this study shows t hat carbon redistribu- tion from multiple substrates can effectively be followed by IRMS by measuring 13 C at natural abundance. This approach allows insights into metabolism that are difficult to obtain b y other techniques. It enables the study of the concurrent consumption of substrates and the quantification of the orientation of their carbon towards a range of products that can arise from more than one route. Its application to following how the utilization of glucose and citrate for products other th an lactate is affected at the genetic level is the subject of current studies. Acknowledgements WearegratefultoHughGriffinandHaroldUnderwood(IFR, Norwich, UK) for supplying the Lactococcus lactis B7/2147 culture and advice on growth conditions, to Christophe Monnet (INRA, Paris-Grignon, France), Helena Santos (ITQB, Oeiras, Portugal) and a number of o ur colleagues in Nantes for advice a nd discussion, and to the Human Nutri tion Research C entre (Nantes) f or the use of the GC-C-IRMS apparatus. Mohamed Mahmoud acknowledges the financial support of the Arab Republic of Egypt for a doctoral bursary. References 1. Cocaign-Bousquet, M., Garrigues, C., Loubie ` re,P.&Lindley,N. (1996) Physiology of pyruvate metabolism in La cto coccu s lactis. Antonie Leeuwenhoek 70, 253–267. 2. Rondags, E., Halliday, E. & M arc, I. (1998) Diacetyl production mechanism b y a strain of La ctococcus lactis spp. lactis bv. diace- tylactis. Study of a-acetolactic acid extracellular accumulation under anaerobiosis. Appl. Biochem. Biotechnol. 69 , 113–125. 3. Hugenholtz, J. (1993) Citrate metabolism i n lactic acid bacteria. FEMS Microbiol. Rev. 12, 165–178. 4. van Beynum, J. & Pette, J. (1939) The decomposition of citric acid by Betacoccus cremoris. J. Dairy Res. 10, 250–266. 5. Hache, C., C achon, R., Wache, Y., Belguendouz, T., Riondet, C., Deraedt, A. & Divie ` s, C. (1999) Influence of lactose–citrate cometabolism on the differences of growth and energetics in Lactococcus lactis , Leuconostoc mesenteroides ssp. cremoris. Appl. Microbiol. 22, 507–513. 6. Magni, C., de Mendoza, D., Konings, W. & Lolkema, J. (1999) Mechanism of citrate metabolism in Lactococcus lactis:resistance against lactate to xicity at low pH. J. Bacteriol. 181, 1451–1457. 7. Boumerdassi, H., Monnet, C., Desmazeaud, M. & Corrieu, G. (1997) Effect of citrate on production of diacetyl and acetoin by Lactococcus lact is ssp. lactis CNRZ. 483 cultivated in the presence of oxygen. J. Dairy Sci. 80 , 634–639. 8. Petit, C., Vilchez, F. & Marczak, R. (1989) Influence of citrate on diacetyl a nd acetoin production by f ully grown cells of Strepto- coccus lactis subsp. diaceylactis. Curr. Microbiol. 19, 319–323. 9. Cachon,R.&Divie ` s, C. (1993) A de scrip tive model f or citrate utilization by Lactococcus l actis ssp. lactis bv. diacetylactis. Bio- technol. Lett. 15, 837–842. 10. Goupry, S., Croguennec, T., Gentil, E. & Robins, R.J. (2000) Metabolic flux in glucose/citrate co-fermentation by lactic acid bacteria as measured by isotopic ratio analysis. FEMS Microbiol. Lett. 182, 207–211. 11. Goupry,S.,Gentil,E.,Akoka,S.&Robins,R.(2003)Co-fer- mentation of glucose and citrate by Lactococcus lactis d iacetyl- actis: quantification of the relative metabolic rates by isotopic r atio analysis at natural abundance. Appl. Microbiol. Biotechnol. 62, 489–497. 12. Snoep, J., Teixeira d e Mattos, M., Starrenburg, M. & H ugen holtz, J. (1992) Isolation, characterization and physiological r ole of the pyruvate d ehydrogenase complex and alpha-acetolactate synthase Ó FEBS 2004 Carbon redistribution in L. lactis cofermentation (Eur. J. Biochem. 271) 4399 of Lactococcus lactis subsp. lactis bv diacetylactis. J. Bacteriol. 174, 4838–4841. 13. Curic, M., S tuer-Lauridsen, B., Re nault, P. & Nilsson, D. (1999) A general method for selectio n of a-acetolactate decarboxylase- deficient Lac tococcus lactis mutants to improve diacetyl forma- tion. Appl. E nviron. Microbiol. 65, 1202–1206. 14. Hugenholtz, J., Kleerebe zem, M., Starrenbu rg, M., Delcour, J ., de Vos, W. & Hols, P. (2000) Lactococcus lactis as a cell factory for high-level diacetyl production. Appl. Environ. Microbiol. 66, 4112– 4114. 15. Hugenholtz, J. & Kleerebezem, M. (1999) Metabolic e ngineering of lactic acid bacteria: overview of the approaches and results of pathway rerouting involved in food fermentations. Curr. Opin. Biotechnol. 10, 492–497. 16. Ramos,A.,Jordan,K.,Cogan,T.&Santos,H.(1994) 13 Cnuclear magnetic resonance studies of citrate a nd glu cose c ome tabolism by Lactococcus lactis. Appl. Environ. Microbiol. 60, 1739–1748. 17. Verhue, W. & Tjan, F. (1991) Study o f the c itrate metabolism of Lactococcus l actis su bsp. lactis biovar. diacetylactis by means of 13 C nuclear magnetic resonance. Appl. Environ. Microbiol. 57, 3371–3377. 18. Koebmann, B., Ande rsen, H ., Sole m, C . & Je nsen, P . (2002) Experimental determination of control of glycolysis in Lacto- coccus lactis. Antonie Leeuwenhoek 82, 237–248. 19. Neves, A., Ventura, R., Mansour, N., Shearman, C., Gasson, M., Maycock, C., Ramos, A. & Santos, H. (2002) Is the glycolytic flux in Lactococcus lactis primarily controlled by the redox charge? Kinetics of NAD + and NADH pools determined in vivo by 13 C NMR. J. Biol. Chem. 277, 28088–28098. 20. Andersen, H., Ped ersen, M., Hammer, K . & Jensen, P. (2001) Lactate de hydrogenase has no control on lactate production but has a strong negative control on formate production in Lacto- coccus lactis. Eur. J. Bioc hem. 268, 6379–6389. 21. Neves, A., Ramos, A., Shearman, C., Gasson, M. & Santos, H. (2002) Catabolism of mannitol in Lactococcus lactis MG1363 and a mutant defective in la ctate d ehydrogenase. Microbiology 148, 3467–3476. 22.Monnet,C.,Aymes,F.&Corrieu,G.(2000)Diacetyland a-acetolactate overpro duction by Lactococcus lactis subsp. lactis biovar. d iacetylactis mutants that are deficient in a-acetolatate decarboxylase and have low lactate dehydrogenase activity. Appl. Environ. Microbiol. 66, 5518–5520. 23. Swindell,S.,Benson,K.,Griffin,H.,Renault,P.,Ehrlich,S.& Gasson, M. (1996) Genetic manipulation of the pathway for diacetyl metabolism in Lactococcus lactis. Appl. Environ. Micro- biol. 62, 2641–2643. 24. Neves,A.,Ramos,A.,Shearman,C.,Gasson,M.,Almeida,J.& Santos, H. (2000) Metabolic characteriz ation of Lactococcus lactis deficient in lactate dehydro genase using in vivo 13 C-NMR. Eur. J. Biochem. 267, 3859–3868. 25. Hoefnagel, M., Starrenburg, M., Martens, D ., Hugenholtz, J., Kleerebezem, M., Van Swam, I., Bongers, R., Weste rhoff, H . & Snoep, J. (2002) Metabolic e ngineering of lactic acid b acteria, the combined approach : kinetic modelling, metabolic control and experimental analysis. Microbiology 148, 1003–1013. 26. Brand, W. (1996) High precision isotope ratio m onitoring tech- niques in mass sp ec trometry. J. Mass Spe ctrom. 31, 225–235. 27. Terzaghi, B. & Sandine, W. (1975) Improved medium for lactic Streptococci and their bacteriophages. Appl. Microbiol. 29, 807– 813. 28. Silvestre, V., Goupry, S., Trierweiler, M., Robins, R. & Akoka, S. (2001) Determination of substrate and product concentrations in lactic acid bacterial fermentation by proton NMR using the ERETIC method. Anal. Chem. 73, 1862–1868. 29. Goupry, S., Gentil, E., Rochut, N. & Robins, R. (2000) Evalua- tion of solid-phase microextraction for the isotopic analysis of volatile compound s p rodu ced d uring fermentation b y lactic acid bacteria. J. Agric. Food Chem. 48, 2222–2227. 30. Nordkvist, M., Jensen, N. & V illadsen, J. (2003) Glucose m eta- bolism in Lactococcus l actis MG1363 under diffe rent ae ratio n conditions: requirement of acetate to sustain gro wth und er microaerobic conditions. Appl. Environ. Microbiol. 69, 3462–3468. 31. Crow, V. ( 1990) Pro perties of 2,3-butanediol dehydrogenases fro m Lactococcus lactis su bsp. lactis in relation to citrate fermentation. Appl. Microbiol. Biotechnol. 56, 1656–1665. 32. Rattray, F., Walfridsson, M . & N ilsson, D. (2 000)Purification and c haracteriza tion of a d iacet yl reductase from Leuconostoc pseudomesenteroides. Int. Dairy J. 10, 781–789. 33. Hugenholtz, J. & S tarrenburg, M. (1992) Diac etyl production by different strains of Lactococcu s lactis su bsp. la c tis var. diacetylactis and Leuconostoc spp. Appl. Microbiol. Biotechnol. 38, 17–22. 34. Monnet, C., Schmitt , P. & Divie ` s, C. (1994) D iacetyl p roduction in milk by an a-acetolactate acid accumulating strain of Lactococcus lactis ssp. lactis biovar. diacetylactis. J. Da iry Sc i. 77, 2916–2924. 35. Rondags, E., Stein, G., Germain, P. & Marc, I . (1996) Kinetic study of the c hemical reactivity of a-acetolactate as a fun ction of pH in water, and in f resh and f ermented culture media used for Lactococcus lactis spp. lacti s bv. diacetylactis cultivation. Biotechnol. L ett. 18, 747–752. 36. Boumerdassi, H., Desmazeaud, M., Monnet, C., Boq uein, C. & Corrieu, G. (1996) Improvement of diacetyl production by Lac- tococcus lactis ss p. lactis CNRZ. 483 through oxygen control. J. Dairy Sci. 79, 775–781. 37. de Man, J . (1959) The f ormation of diac etyl and a cetoin from a-acetolactic acid. Recueil des Traveaux Chimiques des Pays-Bas de Belgique 78, 480–486. 38. Park, S H., Xing, R. & Whitman, W. (1995) Nonenzymatic acetolactate oxidation to diacetyl by flavine, nicotinamide and quinone coenzymes. Biochim. Biophys. Acta 1245, 366–370. 39. Cachon, R., Daniel, S. & Divie ` s, C. (1995) Proton-dependent kinetics of cit rate uptake in growing cells of Lactococcus l actis subsp. lactis bv. d iacetylactis. FEMS Microbiol. Lett. 131, 319–323. 40. Mahmoud, M. (2004) Effets de modifications ge ´ ne ´ tiques e t e nvi- ronnementales sur le me ´ tabolismedebacte ´ ries la ctiques: e ´ tude quantitative de fl ux par SM RI et 2 H-RMN en abondance natu- relle. The ` se de Doctorat, Universite ´ de Nantes, France. 41. O’Leary, M. (1992) Physical and chemical basis of carbon isotope fractionation in plants. Plant Cell Environ. 15, 1099–1104. 42. Melzer, E. & Schmidt, H L. (1987) Carbon isotope effects on the pyruvate dehydrogenase reaction and their importance for relative carbon-13 depletion in lipids. J. Biol. C hem. 262, 8 159–8164. 43. Melzer, E. & Schmidt, H L. (1988) Carbon isotope effects on the decarboxylation of carboxylic acids. Comparison of the lactate oxidase reaction and the degradation of pyruvate by H 2 O 2 . Bio- chem. J. 252, 913–915. 44. Rinaldi, G., Meinschein, W. & Hayes, J. (1974) Intramolecular carbon isotopic distribution in biologically produced acetoin. Biomed. Mass Spectr. 1, 415–417. 45. Cachon, R. & Divie ` s, C. (1993) Kinetics of lac tate fermentatio n and c itrate bioconversion by Lactococcus lactis ssp. lactis in batch culture. J. Appl. Bacteriol. 75, 387–392. 46. Solem, C., Koebmann, B. & J ensen, P. (2003) Glyceraldehyde-3- phosphate dehydrogenase has no control o v er glycolytic flu x in Lactococcus lactis MG1363. J. Bacteriol. 185 , 1564–1571. 47. Lin, J., Schmitt, P. & Divie ` s, C. (1991) C haracterisation of a citrate-negative mutant of Leuconostoc mesenteroides subsp. mesenteroides: metabolic and plasmidic proper ties. Appl. M icro- biol. Biotechnol. 34, 628–631. 4400 M. Mahmoud et al. (Eur. J. Biochem. 271) Ó FEBS 2004 . Natural-abundance isotope ratio mass spectrometry as a means of evaluating carbon redistribution during glucose–citrate cofermentation by Lactococcus lactis Mohamed Mahmoud, Emmanuel Gentil and. 148, 3467–3476. 22.Monnet,C.,Aymes,F.&Corrieu,G.(2000)Diacetyland a- acetolactate overpro duction by Lactococcus lactis subsp. lactis biovar. d iacetylactis mutants that are deficient in a- acetolatate decarboxylase and have low lactate dehydrogenase. mass spectrometry; lactic acid bacteria; m etabolic regulation; pyruvate. A r ange of simple sug ars can be c atabolized anaerobically by Lactoc occus lactis and other lactic acid bacteria (LAB)