Nonstatistical 13C Distribution during Carbon Transfer from Glucose to Ethanol during Fermentation Is Determined by the Catabolic Pathway Exploited*

Background: Different fermentation pathways should lead to distinctive isotope patterns in the products. Results: Three pathways for glucose catabolism show discrete isotope patterns in ethanol. Conclusion: Catabolism by different enzyme sequences leads to differential isotope redistribution patterns. Significance: Learning how isotope fractionation occurs in nature is crucial for interpreting fractionation during biochemical and physical processes for traceability. During the anaerobic fermentation of glucose to ethanol, the three micro-organisms Saccharomyces cerevisiae, Zymomonas mobilis, and Leuconostoc mesenteroides exploit, respectively, the Embden-Meyerhof-Parnas, the Entner-Doudoroff, and the reductive pentose phosphate pathways. Thus, the atoms incorporated into ethanol do not have the same affiliation to the atomic positions in glucose. The isotopic fractionation occurring in each pathway at both the methylene and methyl positions of ethanol has been investigated by isotopic quantitative 13C NMR spectrometry with the aim of observing whether an isotope redistribution characteristic of the enzymes active in each pathway can be measured. First, it is found that each pathway has a unique isotope redistribution signature. Second, for the methylene group, a significant apparent kinetic isotope effect is only found in the reductive pentose phosphate pathway. Third, the apparent kinetic isotope effects related to the methyl group are more pronounced than for the methylene group. These findings can (i) be related to known kinetic isotope effects of some of the enzymes concerned and (ii) give indicators as to which steps in the pathways are likely to be influencing the final isotopic composition in the ethanol.

Efforts to understand the causes of isotopic fractionation 2 in the 13 C/ 12 C ratio during metabolism are hampered by a lack of data on the intramolecular 13 C composition of the compounds involved. In particular, it is valuable to understand how different routes to the same product can influence isotope redistribution patterns. The anaerobic fermentation of glucose to ethanol can be carried out by a number of different pathways in which, although the substrate consumed and product accumulated are the same, the intermediate compounds and enzymes involved are not. It follows, therefore, that because isotopic fractionation is associated with reaction mechanism, different pathways should lead to different isotope patterns in the final product. The other principal factor that will determine the isotopic composition of a product is the position-specific isotopic composition of the substrate.
A combination of these two phenomena, position-specific isotope ratios in the substrate and position-specific fractionation during metabolism, determines therefore the final isotopic composition in the product(s) of fermentation. In previous studies, we have shown by isotopic quantitative 2 H NMR spectrometry that the isotopic fractionation in 2 H is distinctive depending on whether the Embden-Meyerhof-Parnas (EMP) 3 pathway or the reductive pentose phosphate (RPP) pathway is used and that the hydrogen atoms in the product can be linked to the positions in the substrate (1,2).
Until recently, however, a similar approach to obtain 13 C/ 12 C isotopic ratios was not possible. To exploit 13 C NMR for the study of intramolecular 13 C distributions, several additional difficulties had to be overcome. First, the isotopic variation of 13 C in natural compounds has a range ϳ10-fold less than 2 H (ϳ50‰ and 500‰, respectively, on the ␦-scale). As a result, isotopic 13 C NMR requires a 10-fold higher precision. Second, for the effective use of quantitative 13 C NMR at natural abundance, uniform proton decoupling of 13 C-1 H interactions is required. This was achieved by the use of adiabatic decoupling (3). Third, slow relaxation times can lead to long acquisition times and associated potential instability, a difficulty largely solved by the use of relaxation agents (4). Now, this technique can be used for quantification of individual isotopomers by exploiting (i) the separation of the resonance signals from the different carbon positions caused by their degree of shielding and (ii) the direct relationship of the peak area to the amount of 13 C resonating at a given frequency (5). Acquisition conditions need to be carefully defined and rigorously controlled to obtain repeatable spectra of sufficient quality for the required precision. Parameters such as concentration of sample, temperature, stability of the lock need to be the same, and spectra with a signal to noise ratio above 750 are recorded. The NOE has to be eliminated, and the efficiency of the 1 H decoupling sequence must be uniform over the whole range of 1 H chemical shifts (12 ppm). Studies (3,(5)(6)(7) on the optimization of an NMR methodology for accurate and precise measurements of 13 C/ 12 C isotope ratios have shed light on how the decoupling conditions of 1 H in 13 C single-pulse NMR experiments strongly affect the precision of measured peak surface areas. A major breakthrough was achieved with the development of an optimized adiabatic 1 H decoupling sequence (3). Provided this is respected, individual 13 C isotopomers can be observed and the absolute intramolecular distributions of 13 C can be determined (8). A curve fitting based on a total line shape analyses is used to obtain the area under each peak, which provides the reduced molar fraction f i /F i of 13 C at each carbon position from which the ␦ 13 C i (‰) can be calculated (see "Experimental Procedures" and Ref. 9 for definitions). The extent to which individual positions are either enriched or depleted in 13 C relative to the statistical mean indicates metabolically induced isotope fractionation.
The determination of KIEs, calculated from isotope fractionation, is recognized as an important way to obtain mechanistic information about enzymes, because they are determined by the rigidity of the bonds within the substrate and the transition state structure. They may be normal (KIE Ͼ 1) or inverse (KIE Ͻ 1), leading respectively to impoverishment or enrichment in the pertinent positions of the product. The magnitude of the isotope effect is, in general, greater for primary KIEs, in which a bond to the atom under consideration is broken. However, significant secondary isotope effects also occur: for example, a change in hybridization sp 2 -sp 3 theoretically causes a normal KIE, whereas a change sp 2 -sp 3 generates an inverse KIE (10). When a metabolic pathway consisting of a cascade of enzymes is studied, the observed fractionation will be the summed effect of the KIEs of the enzymes involved. Because the enzymes are probably not active under fully saturating conditions, as each product of an enzymatic reaction is the substrate for the next reaction, k cat is not accessible (11). Nevertheless, an effect on V max /K m can be obtained and an apparent KIE ( app KIE) potentially characteristic of the overall pathway under consideration.
That photosynthesis, wherein the assimilation of carbon dioxide by plants can involve three types of metabolism: C 3 , C 4 and CAM, leads to different patterns of 13 C distribution in the accumulated hexoses has been clearly demonstrated, initially by indirect analysis following fragmentation (12) and more recently with iq-13 C NMR (13,14). This shows that the photoassimilates are composed of mixtures of isotopomers that are determined by the physicochemical processes (15) and the enzymes (16) involved in the particular metabolism exploited. These mixtures of isotopomers are determined by the presence of isotope effects in the Calvin cycle and indicate the role of certain enzymes in the 13 C isotope discrimination. For example, the aldol condensation of glyceraldehyde-3-phosphate (GAP) with dihydroxyacetone phosphate by fructose-1,6-bisphosphate aldolase (EC 4.1.2.13) to form fructose 1,6-bisphosphate could be identified as the origin of the relative 13 C enrichment at the C-3 and C-4 positions of glucose (17).
However, very much less is known of the 13 C fractionation implicit in the pathways utilizing glucose, the postphotosynthetic isotope fractionation. Three well characterized pathways of microbial metabolism: the Embden-Meyerhof-Parnas, the reductive pentose phosphate, and the Entner-Doudoroff (ED), convert glucose to ethanol and are exploited by a variety of micro-organisms to carry out the anaerobic fermentation of glucose. The first two are also active in plants.
Of these three pathways, only the EMP has been closely examined for position-specific 13 C fractionation in ethanol biosynthesis (13,14,18), whereas some data from the RPP are also available (12). These studies clearly demonstrated that fermentation by the EMP route leads to 13 C/ 12 C isotopic ratios for the methyl (CH 3 ) and methylene (CH 2 ) carbon positions that differ depending on the source glucose. However, in these studies the aim was complete degradation of glucose, so no information on KIEs during fermentation was obtained.
To better understand the origins of 13 C isotope fractionation during fermentation, we have fermented glucose to ethanol using the three common pathways: EMP, ED, and RPP. Ethanol is a convenient analyte for this type of study because it is readily produced in large quantities by easy to cultivate micro-organisms and can be obtained in pure form by distillation. Key features of the three target pathways are illustrated in Fig. 1A, and the origin of the carbons incorporated into ethanol is shown in Fig. 1B.
The anaerobic oxidative catabolism of glucose to pyruvate via the EMP pathway is exploited by a large range of organisms, including the yeast Saccharomyces cerevisiae, producing 2 mol of each of ethanol and CO 2 per mol of glucose (Fig. 1B). Within the present context, the critical feature is the homolytic splitting of the six-carbon unit by fructose 1,6-bisphosphate aldolase (EC 4.1.2.13) to form two three-carbon units, which isomerize and then follow the same path from GAP to pyruvate, thence, by decarboxylation, to ethanol. The ED pathway, which is unique to prokaryotes, is less frequently exploited for anaerobic oxidative catabolism, the bacterium Zymomonas mobilis being an example of an organism that uses it in a strictly fermentative sense. The ED pathway also produces 2 mol of each of ethanol and CO 2 per mole of glucose and also involves the homolytic splitting of the six-carbon unit into two three-carbon units. However, the key enzyme here is 2-dehydro-3-deoxyphosphogluconate aldolase (EC 4.1.2.14), which produces 1 mol of pyruvate and 1 mol of GAP. As a result, the pattern of allocation of carbon to ethanol is modified (Fig. 1B). The RPP, in contrast to these two pathways, involves the heterolytic cleavage of the six-carbon unit into CO 2 , a two-carbon and a three-carbon unit (Fig. 1B). Although the three-carbon unit is again GAP and is metabolized to pyruvate as in the EMP and ED pathways, it is not converted to ethanol but to lactic acid. The two-carbon unit, acetyl-phosphate, is converted directly to ethanol without the intervention of pyruvate. This pathway is used primarily by heterolactic bacteria, including species of Lactobacillus and Leuconostoc.
Early studies of 13 C fractionation concluded "that glucose does not have large differences in the isotope ratios in the individual carbon atoms and that the Embden-Meyerhof reactions do not have a large carbon isotope effect" (19). However, at that time (1961) isotope measurement by mass spectrometry gave only a value for ␦ 13 C g (‰), the global or mean isotope content, and the technique is poorly adapted to the determination of the intramolecular distribution of isotopes. Subsequently, measurement of ␦ 13 C i (‰), the position specific isotope distribution, has proved the first part of the above conclusion to be incorrect (12,18): as yet, evidence is still to be presented that glycolysis proceeds without position-specific isotopic discrimination. To probe this, we have carried out an analysis of ␦ 13 C i (‰) in ethanol from the EMP pathway. In addition, we have examined whether fermentation by alternative pathways by which the glucose is converted to ethanol by reactions involving alternative mechanisms will indicate distinct 13 C app KIEs.
Bacterial Cultures-S. cerevisiae was obtained from Martin Vialatte Oenologie, and Saccharomyces bayanus was from the Oenology Laboratories Group. Both were stored dry at ϩ4°C. Leuconostoc mesenteroides strain 19D cit ϩ was provided by Professor Hervé Prevost (Unité de Recherche Sécurité des Aliments et Microbiologie, ONIRIS, Nantes, France) and stored at Ϫ80°C in M17 medium (20) with 15% (v/v) glycerol. Z. mobilis ZM6 (DSMZ 3580) was provided by Professor Michel Rohmer (Laboratory of the Chemistry and Biochemistry of Micro-organisms, Chemistry Institute, UMR7177, CNRS-University of Strasbourg, Strasbourg, France) and stored at Ϫ80°C in ZGM (see below for composition) with 15% (v/v) glycerol.
Bacterial Culture Conditions-All fermentations were conducted using the same batch of D-glucose. Fermentations were performed using a Labfors-5 7.5-liter bioreactor equipped with a stirrer, regulated aeration rate, digital pumps for adding medium, acid, basic, or antifoam, pH and O 2 sensors, automatic control, and environmental monitoring. For each organism, two or three fermentations were carried out anaerobically at 30°C in appropriate medium with gentle mixing (100 rpm). Samples of appropriate volume (1000 to 250 ml) were taken at appropriate times to give between 10 and 75% utilization of D-glucose. Supernatant was recovered by centrifugation (4250 g; 10 min; 4°C) and kept at Ϫ20°C.
For L. mesenteroides, 2 liters of M17 medium (20), pH 6.3, was sterilized by autoclaving (121°C, 20 min) and transferred aseptically to the culture vessel. Just before initiating the fermentation, D-glucose (300 g in 1 liter of water sterilized by autoclaving (110°C for 20 min)) was added aseptically. A preculture was prepared by suspending bacteria stored at Ϫ80°C in 2 ϫ 200 ml of M17 medium and incubating without agitation for 24 h. Following harvesting, the cells were washed twice with  MP17 (50 ml), resuspended in M17 (50 ml), and transferred aseptically to the fermenter.
Extraction of Ethanol from Fermentation Medium-Ethanol was recovered from the fermentation medium by distillation using a Cadiot column equipped with a Teflon turning band. Care was taken to ensure that at least 90% of the ethanol present was recovered so as to avoid isotopic fractionation (21).
Quantification of Ethanol-Fermentation medium was diluted to give an ethanol concentration in the range 1-10 mg/ml. To 1 ml of this, 0.5 ml of ethyl acetate was added. After vigorous agitation (3 min), the phases were separated by centrifugation (13,500 ϫ g for 1 min), and the ethyl acetate phase was recovered. Ethanol was quantified by gas chromatography on an Agilent 7820A gas chromatograph fitted with a PTA-5 (30 m ϫ 0.32 mm; film thickness, 0.5 m) with helium as a carrier gas at a constant flow (1.2 ml/min) and an injector temperature of 250°C. Elution conditions were as follows: 70°C for 4 min, 20°C/min to 200°C, and 200°C for 1.9 min. Detection was by flame ionization at 250°C. Calibration was with known concentrations of ethanol (0.67-8.34 mg/ml) treated exactly like the samples.
Quantification of D-Glucose-D-Glucose concentration was determined by HPLC on a Lichrosphere 100-NH 2 (250 ϫ 46 mm, 5 m) column eluted isocratically with acetonitrile/water (80:20) at 1 ml/min. Detection was by refractive index. Calibration was with pure D-glucose in the range 10 -250 mg/ml. Isotope Ratio Measurement by Mass Spectrometry-The global value for the whole molecule, ␦ 13 C g (‰), is the deviation of the carbon isotopic ratio R s relative to that of the international standard Vienna Pee Dee Belemnite, R v-PDB . It is determined by isotope ratio measurement by mass spectrometry and calculated from Equation 1.
C NMR Acquisition Conditions-Relaxation agent Cr(Acac) 3 solution (0.1 M) was prepared by dissolving 34.9 mg of Cr(Acac) 3 in 1 ml of DMSO-d 6 in a 4-ml vial. To this was added 600 l of ethanol and 100 l of DMSO-d 6 to act as lock. Following mixing, the solution was left 2-4 h, filtered to remove undissolved relaxation agent, and transferred to a 5-mm NMR tube. Quantitative 13 C NMR spectra were recorded at 100.6 MHz using a Bruker 400 NMR spectrometer fitted with a 5-mm 1 H/ 13 C dual ϩ probe. The temperature of the probe was set at 303 K. The offsets for both 13 C and 1 H were set at the middle of the frequency range. Inverse-gated decoupling was applied, and the repetition delay between each 90°pulse was set at 10ϫ T 1 max of ethanol to avoid the NOE and to achieve full relaxation of the magnetization. The decoupling sequence used adiabatic full passage pulses with cosine square amplitude modulation ( 2 max ϭ 17.6 kHz) and offset independent adiabaticity with optimized frequency sweep (3). Each measurement consisted of the average of five independently recorded NMR spectra.
Spectral Data Processing-The positional isotopic distribution in ethanol was obtained from the iq-13 C NMR spectrum essentially as described previously (9,22). To obtain S i , the area under the 13 C signal for the C-atom in position i (in this case the methyl and methylene positions), curve fitting based on a total line shape analyses (deconvolution) is carried out with a Lorentzian mathematical model using Perch TM NMR Software. In this procedure, line shape parameters are optimized in terms of intensities, frequencies, line width, and line shape (Gaussian/Lorentzian, phase, asymmetry) by iterative fitted to a minimal residue. All line fitting was performed by the same operator.
Each S i has to be corrected to compensate for the slight loss of intensity caused by satellites ( 13 C- 13   F i is the statistical mole fraction for a carbon site i, that is the molar fraction for the carbon position i in the theoretical situation where there is a homogeneous 13 C distribution within the molecule, where c is the number of carbon equivalents in the molecule resonating at a given frequency, and C is the total number of carbon atoms in the molecule (for ethanol F i ϭ 1/2 for both the methyl and the methylene positions). Hence, the site-specific reduced molar fraction can be defined as f i /F i . From the reduced molar fraction ␦ 13 C i , the specific isotope composition of the carbon position i is obtained from the isotope composition of the whole molecule (␦ 13 C g ) measured by isotope measurement by mass spectrometry as follows.
A i (%), the isotopic abundance for carbon position i, is obtained from Equation 5, FEBRUARY 13, 2015 • VOLUME 290 • NUMBER 7

JOURNAL OF BIOLOGICAL CHEMISTRY 4121
where A g is the isotopic abundance of a whole molecule and is obtained from the following, where R V Ϫ PDB ϭ 0.0112372. Then Calculation of Isotope Effects-Kinetic isotope effects were calculated using a modified form of the Biegeleisen equation, where k L and k H are the rates of overall reaction with light ( 12 C) and heavy ( 13 C) isotopomers, respectively; f eth is the fraction of product formed ([ethanol]/[ethanol theoretical ]) determined by gas chromatography for ethanol and HPLC for the initial [glucose]; R eth, t is the isotopic ratio for ethanol at time t, determined by iq-13 C NMR; and R g, 0 is the mean isotopic ratio for the relevant positions in glucose at t ϭ 0, determined by iq-13 C NMR as described previously (23).

RESULTS
To calculate the app KIE values, the position-specific 13 C i content of the accumulated product, ethanol, was determined. To obtain the greatest degree of isotopic fractionation between substrate and product, fermentation to a low f eth is advanta-geous. Fermentations in anaerobic conditions adapted to each micro-organism were sampled in the range 10 -70% of advancement of the reaction, ethanol was recovered by distillation, and the iq-13 C NMR spectra were recorded (Fig. 2).
From this, the ␦ 13 C i (‰) values were calculated for the methyl and methylene positions (see "Experimental Procedures"). The mean for the f i /F i for each fermentation (5 spectra) and the mean Ϯ S.E. values of ␦ 13 C i (‰) are given in Table 1.
From the calculated ␦ 13 C i (‰) data, the position-specific KIEs were calculated using the modified Biegeleisen equation. These are given in Table 2.

DISCUSSION
Unique Features of the EMP Pathway-The position-specific app KIE observed for the fermentation of D-glucose by S. cerevisiae/S. bayanus culture is not significant for the CH 2 position and only shows a small normal effect for the CH 3 . Thus, the previous proposal that the Embden-Meyerhof reactions do not have a large carbon isotope effect (19) is substantiated by direct measurement. However, because the app KIE determined on the final product (ethanol) reflects the sum of counteracting normal and inverse KIEs, this does not exclude the possibility of fractionation having occurred at several steps in the pathway, which consists of a series of thirteen enzymatic reactions. In addition, each value is the mean of the isotope effects on the C-2 ϩ C-5 and C-1 ϩ C-6 positions of D-glucose for the CH 2 and CH 3 , respectively.
Five enzymes have the potential to generate an isotopic fractionation on positions C-2 G and/or C-5 G on their route to become the CH 2 of ethanol (Fig. 3). Two of these enzymes, glucose-6-phosphate (G6P) isomerase (EC 5.3.1.9) and triose phosphate isomerase ( 4) are common to at least two of the three pathways studied (Fig. 1). It should be noted that any primary KIE associated with fructose-1,6-bisphosphate aldolase will not manifest itself, because the C-3 and C-4 are both lost as CO 2 . G6P isomerase causes an sp 3 -sp 2 change in hybridization at the C-2, a modification commonly associate with a normal secondary kinetic isotope effect (24). That this is manifest is highly probable, because glucose isomerase (EC 5.3.1.5) from Streptomyces murinus has an inverse isotope effect in the direction fructose to glucose (18), which would act to impoverish the C-2 of F6P derived from G6P, by ϳ15‰ at equilibrium. This depletion is much higher than that measured in ethanol, implying that further steps compensate. Although triose phosphate isomerase is a potential candidate, because during the reaction an sp 2 -sp 3 change in hybridization occurs at the C-2 position (Fig. 3A), it can be reasonably argued that this is unlikely, because the limiting step of the reaction (proton transfer from His-95 of the active site) does not involve a primary 13 C isotope effect (25). The C-2 position is involved in further reactions, notably enolase, pyruvate kinase, and pyruvate dehydrogenase, which are all susceptible to introducing isotopic fractionation. These enzymes are all located in the part of the EMP that is common to other pathways and are discussed below.

TABLE 1
Data obtained by iq-13 C NMR for the 13

C distribution in ethanol produced by fermentation of glucose
For definitions of terms, see "Experimental Procedures." All fermentations were made using the same batch of D-glucose obtained from Sigma-Aldrich (batch 071M014552V). FEBRUARY 13, 2015 • VOLUME 290 • NUMBER 7

JOURNAL OF BIOLOGICAL CHEMISTRY 4123
Unique Features of the ED Pathway-The fermentation of D-glucose by Z. mobilis gives values of app KIE similar to those obtained for the fermentation of D-glucose by S. cerevisiae/ S. bayanus culture (Table 2), with no significant isotope effect at the CH 2 but a larger normal effect for the CH 3 of app KIE CH 3 ϭ 1.0035 Ϯ 0.0002. As with the G6P to F6P transformation, the C-2 undergoes an sp 3 -sp 2 transition in the conversion of 6-phosphogluconate to 2-keto-3-deoxy-6-phosphogluconate (KDPG) by phosphogluconate dehydratase (EC 4.2.1.12). Although the mechanism has been defined (26) and the C-2 G undergoes cleavage of the C-H bond and ketonization (Fig. 4), the lack of any KIE indicates that, as with triose phosphate isomerase, this is not the kinetically limiting step. However, the lack of any KIE supports the suggestion that the G6P to F6P isomerization in the EMP pathway is indeed responsible for a small fractionation, because this step is absent in the ED pathway, and the C-2 G , unlike the C-5 G , does not undergo the steps common to the three pathways (Fig. 1).
The normal app KIE CH 3 is considerably greater than for the EMP pathway. This may reflect the origin of the CH 3 being the C-3 G ϩ C-6 G in this pathway. The C-3 G position is potentially subjected to a primary KIE during the fission of the C-3 G -C-4 G bond by the action of KDPG aldolase, which cleaves KDPG to yield directly pyruvate and GAP. Thus, the C-3 G enters the CH 3 of ethanol, in marked contrast to the EMP pathway. Although no 13 C KIE studies have been done on the KDPG aldolase, its mechanism (27) is closely similar to that of fructose 1,6-bisphosphate aldolase (28), for which a normal KIE of 1.0254 has been determined (17). This value is in consensus with the app KIE CH 3 found here.
Unique Features of the RPP Pathway-The fermentation of D-glucose by L. mesenteroides shows a distinctly different isotopic fractionation compared with the EMP or ED pathways. Three features are striking: (i) the CH 2 position shows a normal app KIE CH 2 ϭ 1.0058 Ϯ 0.0007, (ii) the app KIE CH 3 ϭ 1.0057 Ϯ 0.0011 is of a similar magnitude, and (iii) both values are greater than for the other pathways ( Table 2). As in the EMP and ED pathways, cleavage of the C-3 G -C-4 G bond occurs, but crucially, it is only the (C-2-C-3) G unit that is converted to ethanol, the (C-4 -C-5-C-6) G yielding D-lactic acid (Fig. 1). It is found that this difference has a major influence on the app KIE values. An analysis of the data is simplified, especially for the CH 3 of ethanol, as the C2 G is only involved in one reaction: phosphogluconate dehydrogenase (decarboxylating) (EC 1.1.1.44), which simultaneously ketonizes the C-2 G position and cleaves the C-1 G -C-2 G bond and could therefore involve a primary KIE (Fig. 5). The CH 2 position similarly is derived from a cleavage reaction that catalyzed by phosphoketolase (EC 4.1.2.9), which cleaves the C-3 G -C-4 G bond and therefore could involve a primary KIE. The C-3 G does undergo further reactions, but the only one susceptible to any isotope fractionation is the final alcohol dehydrogenase (EC 1.1.1.1/EC 1.1.1.2), in which a change in hybridization sp 2 -sp 3 occurs (see below).
The C-1 of D-glucose is lost during the decarboxylation of 6-phosphogluconate to ribulose-5-phosphate catalyzed by 6-phosphogluconate dehydrogenase (decarboxylating). Decarboxylation is characteristically associated with a normal primary KIE (29,30). The mechanism of 6-phosphogluconate dehydrogenase has been elucidated (31,32) and a 13 C KIE ϭ 1.0209 Ϯ 0.0005 has been determined for the C-1 G during decarboxylation by the enzyme from the yeast Candida utilis (31). Although no data are available for the C-2 G position, a KIE of a similar magnitude is probable (33), so an app KIE Ϸ1.006 is compatible with the proposed mechanism. It should be noted that, because reactions are not occurring under V max /K m conditions, values of app KIE obtained in vivo can be expected to be lower than values of KIE obtained under in vitro conditions.
Considering the origin of the CH 3 of ethanol, a KIE associated with the cleavage of the C-3 G -C-4 G bond is probable (34). Although no isotopic data are available for this reaction, its similarity to the better studied fructose-1,6-bisphosphate aldolase (35) for which a normal KIE of 1.0254 (17) has been determined, making it probable that the app KIE CH 3 ϭ 1.0057 is due to the activity of this enzyme. That this value is smaller than determined for the fructose-1,6-phosphate aldolase might reflect the sp 2 -sp 3 change during the alcohol dehydrogenase step, which is likely to manifest a small secondary inverse KIE (see below).
Features Common to the EMP and ED Pathways-Both the EMP and ED pathways produce ethanol from GAP via pyruvate (Fig. 1A). Only the reaction by which 2-phosphoglycerate is converted to phosphoenolpyruvate involves a change in hybridization state of sp 3 -sp 2 for both the C-5 G and C-6 G and could therefore manifest a secondary 13 C KIE. This is catalyzed by enolase in an elimination reaction involving a carbanion intermediate (36,37). The slow step of the reaction, the abstraction of the H at the ␤C (i.e. the C-2 G or C-5 G ) is associated with a strong 2 H isotope effect: subsequent enolization of the ␥C (i.e. the C-1 G or C-6 G ) is not. Therefore, it can be concluded that this step is unlikely to contribute to the app KIE CH 3 of ethanol. Furthermore, any normal isotope effect at the ␥C is likely to be reversed by an inverse isotope effect in the subsequent reaction catalyzed by pyruvate kinase in which an sp 2 -sp 3 change in hybridization occurs. The only common step in which a primary KIE might occur is pyruvate dehydrogenase (decarboxylating) (EC 1.2.4.1), responsible for the bond cleavage within the pyruvate dehydrogenase complex. Primary 13 C KIEs have been shown for both carbon positions involved in the decarboxylation, the ␤C (i.e. the C-2 G or C-5 G of glucose) showing a 13 C KIE ϭ 1.0213 Ϯ 0.0017 and 1.0254 Ϯ 0.0016 for the enzyme from bacteria (Escherichia coli) and yeast (S. cerevisiae) respectively (29). Furthermore, a small secondary effect on the ␥C of 1.0031 Ϯ 0.0009 was found. It can therefore be proposed that this enzyme makes a major contribution to the values of app KIE CH 3 obtained of 1.0014 Ϯ 0.0002 and 1.0035 Ϯ 0.0002 for the ethanol from S. cerevisiae and Z. mobilis fermentations, respectively.
Following on from this proposal is the possibility to estimate the effect isotopic caused by the KDPG aldolase on C-3 G by comparing the app KIE CH 3 calculated for the two types of fermentation. The app KIE CH 3 for Z. mobilis is 1.0035, and for S. cerevisiae it is 1.0014. The app KIE CH 3 observed for Z. mobilis can be estimated as the average of the app KIE CH 3 for S. cerevisiae and the KIE caused by KDPG aldolase, because no other reaction is likely to cause fractionation at that position (C-3 G ) in the upper portions of these two pathways (see Fig. 1 and above). By calculation, the deduced KIE CH 3 for KDPG aldolase on the C-3 G is 1.0021. Although only 10% the magnitude of the primary KIE determined for fructose-1,6-bisphosphate aldolase (17), this once again indicates that reactions are occurring under conditions far from V max /K m . In addition, these two enzymes differ mechanistically (27). It is notable, however, that the value obtained is of the same magnitude as for the RPP pathway (Table 2), in which a similar cleavage catalyzed by phosphoketolase, is involved.
Features Common to All Three Pathways-The only step common to all three pathways in the final step: the reduction of acetaldehyde to ethanol, during which a change of hybridization sp 2 -sp 3 occurs that could induce a small secondary inverse KIE. The yeast and liver enzymes gave intrinsic 13 C equilibrium isotope effects of 1.0164 and 1.0149, respectively, with benzyl alcohol as substrate (38). However, the value was highly dependent on the C-H bond distance in the transition state, which will not be equivalent in acetaldehyde. Only in the case of L. mesenteroides is the app KIE CH 2 of the same order. However, because any isotope effect is predicted to be inverse, it appears that this step does not influence the final 13 C value of the CH 2 group.
Conclusions-The application of iq-13 C NMR to the study of fermentation pathways has made possible the detection of weak app KIEs in the formation of ethanol from D-glucose. These isotope effects are manifest because the transformation is to an end product that accumulates. Thus, despite the potential reversibility of a number of the individual steps in each pathway, the overall vector is driven by the metabolic need to regenerate NAD ϩ in order that metabolism can continue. Hence, the pathway overall is a unidirectional closed system, and any isotope fractionation associated with reverse reactions will be negated. Although it is confirmed that glycolysis (EMP) essentially proceeds without isotopic discrimination, as previously supposed (19), the data support the previous observation that the methylene group becomes relatively enriched during fermentation and the methyl becomes relatively impoverished (13). As indicated in Table 2, a normal KIE is found for the CH 3 , which will lead to impoverishment, and a negligible or slightly inverse KIE is found for the CH 2 . Furthermore, rather more significant KIEs are seen to be present in the ED and RPP routes of D-glucose catabolism.
The relatively small size of the app KIEs observed and the absence of comprehensive mechanistic data on all the enzymes involved limits the extent to which the data can be interpreted. Nevertheless, a number of enzymes can be implicated, notably those involved in key steps characteristic to each pathway. Little can be concluded for the EMP pathway, other than that isotopic effects are relatively small and fully consistent with a series of secondary isotope effects in which normal and reverse effects counteract, giving an overall negligible effect for the app KIE CH 2 and a very small normal app KIE CH 3 . For the ED pathway, a net estimate for the KIE at C-3 G caused by KDPG aldolase can be postulated. This effect is relatively small for a primary KIE, which suggests that isotopic fractionation caused by other enzymes is also playing a contributory role. The RPP pathway shows more intense isotopic fractionation at both the methyl and methylene positions. This fractionation is satisfactorily explained by a normal KIE for both positions associated with the action of phosphoketolase. The data obtained serve to focus attention on these enzymes, for which more details of their specific properties need to be obtained.