Site-specific hydrogen isotope fractionation in the biosynthesis of glycerol

Article abstract of DOI:10.1006/bioo.1999.1145

The nuclear magnetic resonance study of site-specific natural isotope fractionation (SNIF-NMR) produced in the glycolytic conversion of glucose into ethanol and glycerol provides isotopic transfer coefficients, a(ij), which relate sites i in the products to sites j in the reactants. The isotopic connection between the carbon-bound hydrogens of glycerol and those of glucose and water in fermentation reactions carried out with Saccharomyces cerevisiae has been investigated. The a(ij) coefficients provide mechanistic information on the genealogy of the glycerol hydrogens, on the relative rates of triose phosphate isomerization and reduction of dihydroxyacetone phosphate into glycerol 3-phosphate, on the stereospecificity of the reduction, on the percentages of intra- and intermolecular hydrogen transfer occurring in the course of the reaction, and on the order of magnitude of the active overall kinetic and thermodynamic isotope effects. Thus a close connection is determined between sites 3 pro-R of glycerol (stereospecific numbering) and H₆ pro-R and to a lesser extent H₂ of glucose and between site 3 pro-S of glycerol and H₆ pro-S and H₁ of glucose. Moreover site 2 of glycerol, which exhibits a strong correlation with water is also partly connected with glucose. Possible changes in the values of the isotopic transfer coefficients, as a function of the composition of the medium or of the environmental conditions, enable mechanistic perturbations such as variations in the percentage of intermolecular transfers to be detected. Analyzed in terms of stereospecificity of the reduction step the isotopic results provide a direct chemical shift assignment of the enantiomeric pairs (1(S), 3(R)) and (1(R), 3(S)) of glycerol triacetate. The influence of added bisulfite, which strongly increases the yield in glycerol is estimated. The isotopic characterization of the bioconversion producing both ethanol and glycerol has been extended to the determination of the carbon isotopic parameters by isotope ratio mass spectrometry. Although it usually occurs as a byproduct of the fermentation, glycerol can be considered as a useful complementary probe for characterizing the glycolytic pathway and for inferring various properties of the carbohydrate precursors. (C) 2000 Academic Press.

Full text of DOI:10.1006/bioo.1999.1145

Bioorganic Chemistry 28, 1–15 (2000)
doi:10.1006/bioo.1999.1145, available online at http://www.idealibrary.com on
Site-Specific Hydrogen Isotope Fractionation in the
Biosynthesis of Glycerol
Ben-Li Zhang, Stephan Buddrus, and Maryvonne L. Martin
LAIEM, Universite´ de Nantes-CNRS, 2 rue de la Houssinie`re, B.P. 92208,
44322 Nantes cedex 03, France
Received November 19, 1998
The nuclear magnetic resonance study of site-specific natural isotope fractionation (SNIF-
NMR) produced in the glycolytic conversion of glucose into ethanol and glycerol provides
isotopic transfer coefficients, aij, which relate sites i in the products to sites j in the reactants.
The isotopic connection between the carbon-bound hydrogens of glycerol and those of glucose
and water in fermentation reactions carried out with Saccharomyces cerevisiae has been investi-
gated. The aij coefficients provide mechanistic information on the genealogy of the glycerol
hydrogens, on the relative rates of triose phosphate isomerization and reduction of dihydroxyace-
tone phosphate into glycerol 3-phosphate, on the stereospecificity of the reduction, on the
percentages of intra- and intermolecular hydrogen transfer occurring in the course of the
reaction, and on the order of magnitude of the active overall kinetic and thermodynamic isotope
effects. Thus a close connection is determined between sites 3 pro-R of glycerol (stereospecific
numbering) and H6 pro-R and to a lesser extent H2 of glucose and between site 3 pro-S of
glycerol and H6 pro-S and H1 of glucose. Moreover site 2 of glycerol, which exhibits a strong
correlation with water is also partly connected with glucose. Possible changes in the values of
the isotopic transfer coefficients, as a function of the composition of the medium or of the
environmental conditions, enable mechanistic perturbations such as variations in the percentage
of intermolecular transfers to be detected. Analyzed in terms of stereospecificity of the reduction
step the isotopic results provide a direct chemical shift assignment of the enantiomeric pairs
(1S, 3R) and (1R, 3S) of glycerol triacetate. The influence of added bisulfite, which strongly
increases the yield in glycerol is estimated. The isotopic characterization of the bioconversion
producing both ethanol and glycerol has been extended to the determination of the carbon
isotopic parameters by isotope ratio mass spectrometry. Although it usually occurs as a by-
product of the fermentation, glycerol can be considered as a useful complementary probe for
characterizing the glycolytic pathway and for inferring various properties of the carbohydrate
precursors.
᭧ 2000 Academic Press
INTRODUCTION
The investigation of site-specific natural isotope fractionation by nuclear magnetic
resonance (SNIF-NMR) (1) is a powerful source of information on overall connections
between individual atoms of reaction products and their precursor atoms in the starting
materials (2). For a complex bioconversion, such as fermentation, this approach may
provide the global isotopic balance involving specific molecular sites subjected to
the whole set of successive and competitive reaction steps. Consequently the results
1
0045-2068/00 $35.00
Copyright ᭧ 2000 by Academic Press
All rights of reproduction in any form reserved.

2
ZHANG, BUDDRUS, AND MARTIN
can be interpreted in terms of individual genealogy of atoms, of relative contributions
of intra- and intermolecular hydrogen transfers, and of the overall active kinetic or
thermodynamic isotope fractionation effects affecting a specific molecular position.
For a given standardized reaction, the matrix of isotopic coefficients connecting
individual sites, or clusters of sites, in the products and in the reactants, may be used
subsequently to infer isotopic properties of consumed reactants from the isotopic
fingerprint of their products (2). It should be emphasized that investigating the isotopic
behavior at natural abundance or close to natural abundance avoids the risk of per-
turbing the biochemical pathway through kinetic isotope effects in branched reaction
steps. The determination of overall molecular carbon-13 contents by isotope ratio
mass spectrometry (IRMS) has been frequently applied to the characterization of
metabolic pathways in natural conditions (3). By isolating specific carbon sites, a
more detailed interpretation of the biochemical mechanisms is accessible by IRMS
(4,5) but only at the price of appropriate and often lengthy degradations of the
investigated molecule. New information on the genealogy of the carbon skeleton of
glycerol has recently been obtained in this way (6). In contrast to IRMS, the SNIF-
NMR approach offers the advantage of simultaneously determining all the isotope
contents of the diastereotopic molecular positions. This method is particularly well
suited to the analysis of hydrogen isotope fractionation. For example, the glycolytic
products, ethanol and water, can be used as isotopic probes to characterize sugar
parents and in particular to obtain information on the metabolism of the photosynthesis
and on the botanical origin of the precursors (7). Since the fermentation probe,
ethanol–water, exhibits a strong reduction in the number of hydrogen isotope parame-
ters as compared to the starting components, glucose and water for instance, comple-
mentary access to isotopic parameters of different molecules derived from sugars is
desirable. In this perspective, the present study considers the isotopic aptitude of
glycerol to characterize both the mechanistic pathway of its biosynthesis in a fermenta-
tion reaction and the isotopic properties of its carbohydrate precursors.
EXPERIMENTAL
Materials
Commercial sugars from different origins have been used in the fermentation
experiments. Their isotopic parameters are given in the tables. Variations in the isotope
ratio at positions 1, 2 and 6, 6Ј of glucose have been obtained by adding small
quantities (34.8–90.5 mg) of ␣-D-glucose (from Aldrich) selectively labeled with
deuterium (99.8%) at position 1, 2 or 6 to 100 or 150 g of corn glucose dissolved in
water. Nantes tap water characterized by an isotopic ratio (D/H)W ϭ 150 ppm has
been used in most fermentation experiments. In order to investigate the influence of
the aqueous medium slightly enriched water (300–2000 ppm) was also prepared.
5.5 g of baker’s yeast were used in all experiments. The composition of the
fermentation medium denoted ␣ was the following, for 1 L of solution: 150 g
D-glucose, 1 g NH4Cl, 1 g KH2PO4, two crystals of MgCl2.
In some experiments the yield of glycerol was increased by adding to the solution,
15 g of Na2SO3 (␤ medium). In this medium, bisulfite interacts with acetaldehyde
and partly inhibits its transformation into ethanol.

ISOTOPE FRACTIONATION IN GLYCEROL BIOSYNTHESIS
3
Extraction of Ethanol and Glycerol
In the conventional fermentation experiments, ethanol was extracted by fractional
distillation (2). The residue was heated (about 100ЊC) to remove water until the
volume was reduced to 1/3. The pH of the residue was adjusted to a value of 9 by
adding wet Ca(OH)2. After filtration on a buchner the solution was evaporated again,
until a paste was obtained. Glycerol in the paste was extracted three times with
ethanol/ether (2/1 v/v). The extraction solutions were combined and filtered. Glycerol
was obtained, after vacuum evaporation (10 mm Hg) at 80ЊC, with a yield of 70%.
When Na2SO3 was added to the fermentation medium, the distillation and evapora-
tion steps were perturbed by formation of a large quantity of foam. Evaporation had
to be carried out by heating small portions added successively.
Synthesis of Glycerol Triacetate
Two grams (22 mmol) of glycerol, 10 ml (106 mmol) of acetic anhydride, and 1.5
g of sodium acetate were mixed in a round bottom flask and heated 30 min under
reflux and stirring. The reaction mixture was poured into 80 ml of ice water and the
solution was extracted three times with 30 ml of ether. The combined ether phases
were washed successively with 30 ml of a sodium carbonate solution (150 g.LϪ1)
and 20 ml of a half-saturated sodium chloride solution. After drying over sodium
sulfate, the ether and the impurities, acetic anhydride and acetic acid, were evaporated
under vacuum. Four grams of triacetin was obtained (yield: 85%). The purity (Ͼ99%)
1
was checked by H-NMR.
Isotopic Determinations
The overall carbon isotope ratio of every molecular species, A, is expressed on
the relative ␦-scale:
(13C/12C)A Ϫ (13C/12C)PDB
␦A(‰) ϭ
1000,
[1]
(13C/12C)PDB
where PDB denotes the international carbonate reference (Pee Dee Belemnite).
The site-specific hydrogen isotope ratios (D/H)i are defined (8) as:
NDi
Di
(D/H)i ϭ
ϭ
,
[2]
Hi PiNH
where Di and Hi are the numbers of deuterium and protium atoms at site i, NDi the
number of isotopomers monodeuterated at position i, Pi the stoechiometric number
of hydrogens at site i, and NH the number of fully protonated molecules.
Isotope Ratio Mass Spectrometry (IRMS)
The overall ␦13C parameter of glycerol, sugars, and ethanol was measured using
a Finnigan Delta E mass spectrometer coupled with a Carlo Erba NA 1500 elemental
analyzer. The precision of the determination is better than 0.3 ‰ (9).

4
ZHANG, BUDDRUS, AND MARTIN
Nuclear Magnetic Resonance (SNIF-NMR)
The (D/H)i ratios of glycerol (investigated in the form of its triacetate) were
measured by 2H-NMR using a calibrated reference, tetramethylurea (TMU) (8), distrib-
uted by the Institute for Reference Materials and Measurements (IRMM) in Brussels
[(D/H)TMU ϭ 123.4 ppm]. The (D/H)i values were calculated from Eq. [3]:
PTMU mTMU
SiA
MA
(D/H)i ϭ
(D/H)TMU
,
[3]
PiA mA MTMU STMU
where P, m, and M are, respectively, the stoechiometric number of hydrogens, the
mass and the molecular weight of the investigated compound, A, or of the reference,
TMU. SiA and STMU are the areas of signal i of A and of the methyl signal of
tetramethylurea in the deuterium NMR spectrum.
2
The H-NMR spectra were recorded, under broad band proton decoupling (Waltz-
16), on a Bruker DRX 500 spectrometer equipped with a fluorine lock device. The
experimental conditions were the following: recording frequency, 76.77 MHz; fre-
quency window, 1200 Hz; memory size, 32 K; exponential multiplication associated
with a line broadening of 0.5 Hz; acquisition time, 6.8 s; delay time, 1.2 s; scan
number, 3360; temperature, 55ЊC. The (D/H)i values were calculated from the average
over three spectra. Usually 2.5 g of sample, 0.17 g of TMU mixed with 150 mg of
C6F6 (locking material), and 1 ml of CH3CN were introduced, after filtration, into a
10-mm NMR tube.
In the case of ethanol the (D/H)i ratios were recorded on a Bruker DPX 400
spectrometer under the following conditions: frequency, 61.4 MHz; temperature, 303
K; frequency window, 1200 Hz; memory size, 32K; line broadening, 0.1 Hz; acquisi-
tion time, 6.8 s; delay, 0.1 s; scan number, 200. Three spectra were run successively
for every sample. The NMR tube was prepared in the same way as for glycerol
triacetate with the following quantities: 2.6 g ethanol, 1.3 g TMU, and 0.1 g C6F6.
The quantitative determinations were performed by using a dedicated algorithm
based on a complex least square analysis of the signal (10) (SNIF-NMR Concept
core system and Interliss program from EUROFINS SCIENTIFIC, Nantes, France).
This theoretical treatment involves an automatic integrated management of all the
experimental parameters, including the phases of the individual resonances and the
baseline parameters.
RESULTS
Since the deuterium content of the hydroxyl groups is averaged by chemical ex-
change with the aqueous medium, only the carbon-bound hydrogens of glycerol will
be considered. At natural abundance the five monodeuterated isotopomers, denoted
1S, 1R, 2, 3S, 3R (Fig. 1) are present, in proportions which depend on the overall
mechanistic pathway and on the resulting isotope effects.
Isotopic Probes of Glycerol
Unfortunately, due to insufficient chemical shift separation and relatively short
transverse relaxation times the glycerol molecule itself is not a convenient probe for

ISOTOPE FRACTIONATION IN GLYCEROL BIOSYNTHESIS
5
FIG. 1. Isotopomers of glycerol monodeuterated at the carbon-bound positions. The carbon atom
numbered 3 (stereospecific numbering) was formerly carbon-3 of sn-glycerol 3-phosphate, which pre-
viously bore the phosphate group in the dihydroxyacetone phosphate (DHAP) precursor. Hydrogens at
the pro-R and pro-S positions on carbon 1 and 3 (denoted 1S, 1R and 3R, 3S) have different origins.
From a stereochemical point of view the four isotopomers 1S, 1R, 3S, 3R correspond, respectively, to the
four species (1S,2S), (1R,2S), (3S,2R), (3R,2R) resulting from a stereospecific reduction of the carbonyl
site of DHAP. Two pairs of enantiomers (1S,3R) and (1R,3S) are therefore produced.
2H-NMR. Derivatives, such as glycidal (11), obtained by a series of stereospecific
enzymatic reactions could exhibit good discriminating potential. However, it should
be emphasized that carrying out a large number of reaction steps, some of which
are characterized by relatively low yields, is likely to introduce cumulated isotope
fractionation effects, which should be carefully controlled at every step of the chemical
pathway. As illustrated in Fig. 2, the simple triacetate derivative of glycerol exhibits
reasonable chemical shift separation. For symmetry reasons only three carbon-bound
sites, A, B, and 2, are distinguished. The assignment: A ϵ 1S ϩ 3R and B ϵ 1R ϩ
3S is confirmed later in this paper. Alternatively five carbon-bound hydrogen positions
can be distinguished in the 2H-spectrum of the derivative 2,2-dimethyl-1,3-dioxolane-
4-methanol (glycerol acetonide), which is easily prepared. However, due to the lack
of chiral specificity of the chemical conversion the corresponding prochiral positions
2
on carbon 1 and 3 of glycerol have been scrambled and the H spectrum exhibits
two pairs of equal signals, one pair originating from ex-(1S ϩ 3R) and the other from
ex-(1R ϩ 3S). Consequently, in spite of an increase in the number of diastereotopic
sites with respect to the triacetate, the isotopic discrimination is not improved. Never-
theless this compound can be proposed as a useful complementary probe since it is
characterized by relatively small line-widths.
Isotopic Ratios
The NMR determinations have been combined with the measurement, by IRMS,
of the overall carbon-13 contents of both the starting materials and the glycerol and
ethanol products (Table 1).
The site-specific hydrogen isotope parameters determined in fermentation experi-
ments involving glucose samples from different botanical origins and isotopically
modified maize glucose are collected in Table 1. Several series of fermentation

6
ZHANG, BUDDRUS, AND MARTIN
FIG. 2. 2H-NMR spectra (at 76.8 MHz) of triacetates, prepared from glycerol obtained by fermentation
of maize glucose (150g. LϪ1) by Saccharomyces cerevisiae. The fermentation medium used in experiment
a is described as ␣ in the experimental section. Experiment b has been carried out, in the fermentation
medium denoted ␤, with glucose slightly enriched (327 ppm) at position 2. TMU denotes the natural
abundance deuterium signal of the isotopic reference, tetramethylurea (delivered by IRMM in Brussels).
Spectra a and b have been registered at 55ЊC in CH3CN. An exponential multiplication corresponding to
a line broadening of 0.5 Hz has been applied. The signals are assigned in Fig. 1.
experiments have been carried out, in otherwise identical experimental conditions,
with maize glucose samples either normal or enriched on sites 1, 2, 6, 6Ј, at levels
ranging from 327 to 585 ppm. The results are compared for fermentations carried

TABLE 1
Isotopic Parameters of the Reactants: (Sugar ϩ Water) and of the Products (Glycerol ϩ Ethanol ϩ Water) of Fermentation Reactions Conducted with
Saccharomyces cerevisiae in Two Different Fermentation Media
Starting materials (a)
Isotopic parameters
Glucose
Products (b) Isotopic parameters
Glycerol (c)
(D/H)B
Ethanol
(D/H)I
⌬(D/H)
␦13C
␦13C
(D/H)A
(D/H)2
13C
(D/H)
D
II
Origin
(ppm)
(‰)
(‰)
(ppm)
(ppm)
(ppm)
(‰)
(ppm)
(ppm)
Beet sugar
ϩ Na2SO3 (d)
Cane sugar
ϩ Na2SO3 (d)
Wine
Maize glucose (M)
—
—
—
—
—
—
Ϫ23.6
—
—
—
—
—
128.6
124.1
132.5
136.3
137.3
136.1
(0.8)
142.1
(0.7)
160.2
(1.8)
150.9
(1.3)
197
(0.3)
186.7
(0.8)
244.9
(2.2)
250.3
(2.4)
129.2
98.8
103.0
86.6
103.4
105.7
100.2
(1.0)
92
(1.9)
101.9
(2.8)
97.2
(1.9)
103.5
(2.8)
87.8
Ϫ25.6
92.1
125.6
—
123.9
—
Ϫ23.6
Ϫ10.7
Ϫ10.7
—
122.8
146.5
141.2
139.2
152.1
(0.7)
151.8
(0.4)
280.4
(1.2)
229.9
(2.1)
158.2
(0.6)
149.8
(0.8)
244.8
(3.6)
—
Ϫ11.9
—
Ϫ27.3
—
—
109.0
—
97.9
125.0
125.3
(0.6)
120.7
(0.9)
126
(0.5)
121.5
(0.3)
126.7
(0.5)
—
Ϫ10.6
Ϫ16.8
110.4
(0.3)
108.3
(0.3)
187.6
(0.5)
151
(0.4)
140.2
(0.5)
—
ϩ Na2SO3 (d) (e)
(M) H1 enriched
ϩ Na2SO3 (d)
—
—
—
—
—
—
—
—
Ϫ16.9
Ϫ17.6
—
—
—
—
—
—
—
—
585
327
341
337
376
376
(M) H2 enriched
ϩ Na2SO3 (d) (e)
(M) H66 enriched
ϩ Na2SO3 (d)
—
Ϫ16.6
Ϫ17.0
—
(1.8)
103.0
(2.2)
100.4
(2.4)
—
—
219.0 (f)
(0.8)
214.0 (f)
(0.6)
129.0
(0.8)
122.0
(0.3)
247.1
(2.2)
Note. (a) The initial concentration of glucose was 150 g.LϪ1. ⌬(D/H) represents the specific enrichment. The starting fermentation water is Nantes tap water which
is characterized by an isotope ratio (D/H)SW ϭ 150 ppm. The overall isotope ratio of the carbon bound (nonexchangeable) hydrogens of the maize glucose pool used for
producing the isotopically modified reactants is (D/H)GNE ϭ 155 ppm. (b) The isotope ratio of the end fermentation water, (D/H)QW was only slightly modified with respect
to the starting values (D/H)SW. Values of 150.3 (1.1) have been measured in experiments performed with maize glucose. The standard deviations (within parentheses) are
calculated over three repetitions. (c) The isotopic parameters of the carbon-bound hydrogens of glycerol are determined on the triacetate derivative. (d) The fermentation
medium (␤) contains a high concentration of Na2SO3 which significantly enhances the production of glycerol (Ӎ15%). (e) The starting concentration of glucose was 100
2
g.LϪ1. (f) These values integrate both monodeuterated and bideuterated species which are observed separately on the H spectrum.

8
ZHANG, BUDDRUS, AND MARTIN
out in a conventional medium and in a medium containing a relatively large amount
of Na2SO3 with a view to increase the production of glycerol.
In order to avoid isotopic perturbations introduced by exchanges of the hydroxylic
hydrogens, the overall deuterium content of the glucose skeleton was determined by
IRMS on the trinitrate derivative (7). The starting sample of maize glucose used in
most experiments is characterized by an overall isotope ratio of the carbon bound
hydrogens (D/H)GNE ϭ 155 ppm. Both glycerol and ethanol are significantly depleted
(Table 1) with respect to glucose, but, due to the high level of dilution, the isotope
ratio of the aqueous medium is only slightly increased.
In order to evaluate the hydrogen connectivity with water the same maize glucose
was also fermented in aqueous media characterized by different values of the isotope
ratio (D/H)SW. The site-specific isotope parameters of both glycerol and ethanol are
given in Table 2. The influence of a higher yield in glycerol produced by added
Na2SO3 has also been considered.
Isotopic Balance
As expected, both the carbon and hydrogen isotope parameters of glycerol vary as
those of ethanol as a function of the botanical origin of the sugar parent (Table 1).
TABLE 2
Isotopic Parameters of Glycerol and Ethanol Resulting from Fermentation of Maize Glucose
(150g. LϪ1) by Saccharomyces cerevisiae Carried Out in Water Characterized by
Different Values of the Hydrogen Isotope Ratio (D/H)WS
Fermentation medium (a)
Glycerol (b)
Ethanol (c)
(D/H)S
(ppm)
(D/H)A
(ppm)
(D/H)B
(ppm)
(D/H)2
(ppm)
(D/H)t
(ppm)
(D/H)I
(ppm)
(D/H)II
(ppm)
W
medium
150
(␣)
136.1
(0.8)
142.1
(0.7)
284.1
(3.1)
277.2
(2.3)
586.0
(5.0)
565.5
(2.0)
1119.8
(1.9)
1079.0
(2.7)
152.1
(0.7)
151.8
(0.4)
270.0
(2.7)
255.0
(0.7)
526.6
(1.2)
514.3
(2.6)
977.9
(3.5)
944.1
(5.4)
100.2
(1.0)
92.0
(1.9)
230.0
(2.8)
215.4
(1.9)
524.3
(4.1)
514.7
(1.8)
1051.9
(0.9)
135.3
(1.0)
136.0
(0.8)
267.6
(2.9)
256.0
(0.8)
549.9
(3.2)
534.9
(1.9)
1049.5
(2.3)
1016.4
(3.9)
110.4
(0.3)
108.3
(0.3)
151.7
(0.3)
157.3
(0.4)
258.9
(0.3)
268.8
(0.9)
434.0
(1.4)
460.4
(1.7)
125.3
(0.6)
120.7
(0.9)
304.8
(0.4)
298.3
(1.0)
747.9
(1.8)
712.7
(1.0)
1486.9
(4.9)
1416.9
(5.1)
(0.1)
389.9
(4.5)
ϩ Na2SO3 (␤)
(␣)
ϩ Na2SO3 (␤)
(␣)
976.4
(9.1)
ϩ Na2SO3 (␤)
(␣)
1978.5
(19.6)
ϩ Na2SO3 (␤)
1035.6
(3.0)
Note. (a) The fermentation media are described in the experimental section. (b) The isotope ratio of
the carbon bound hydrogens of glycerol have been determined on its triacetate derivative. Signals A, B,
and 2 are identifided in Fig. 1. (D/H)t is the overall isotope ratio of glycerol. The standard deviations
(within parentheses) are calculated over three repetitions. (c) (D/H)I and (D/H)II correspond, respectively,
to the methyl and methylene sites of ethanol.

ISOTOPE FRACTIONATION IN GLYCEROL BIOSYNTHESIS
9
In spite of a lower accuracy of the (D/H)i ratios measured on glycerol as compared
to ethanol, similar trends in the values of (D/H)A, (D/H)B on one hand and (D/H)i
(methyl site of ethanol) on the other hand, are observed. Thus site B of glycerol is
enriched in deuterium to the same extent (about 17 ppm) as the methyl site of ethanol
when going from the cane C4 precursor to the beet C3 precursor. This behavior can be
explained by the role of partly identical hydrogen parents in the carbohydrate material.
Since glycerol is only a minor product of the fermentation its isotopic parameters
may be very sensitive to kinetic isotope effects intervening in the first steps of the
glycolytic reaction. Thus the strong 13C depletion already observed (6) in glycerol as
compared to its glucose precursor must be compensated for by a corresponding
enrichment of the other products. Since ethanol is also slightly depleted, the isotopic
balance requires that carbon dioxide issued from sites 3 and 4 of glucose be enriched
in 13C. This behavior has been checked in independent experiments.
The isotopic balance is more difficult to establish in the case of hydrogen as a
consequence of the participation of the large aqueous pool. Indeed since hydrogens
are partly transferred from water with kinetic isotope effects significantly higher
than unity (12–14) the deuterium depletions observed, at complete carbohydrate
consumption, in ethanol and to a lesser extent in glycerol, must be accompanied by
a small, hardly detectable, enrichment of the aqueous medium.
Significant differences in the isotope contents of the products are observed when
Na2SO3 is added to the fermentation medium. The hydrogen isotope ratio of site 2
of glycerol in particular is modified with respect to fermentation in the absence of
Na2SO3. This behavior probably depends partly on variable fractionation effects
associated with higher yields of glycerol production (about 15 against 5%).
Coefficients of Isotopic Transfer Between Reactants and Products
A set of linear equations of type 4 is expected to relate the isotopic ratios of the
fermentation products, i, to those of the starting materials, j (2).
(D/H)i ϭ aij(D/H)j ϩ Ci
[4]
The aij coefficients connecting sites i ϭ A ϭ (1S ϩ 3R), B ϭ (1R ϩ 3S) and 2 of
glycerol to sites j ϭ 1 to 6 of glucose and j ϭ w of water are accessible from the
determination of the isotopic ratios, (D/H)i, of glycerol obtained in fermentation
experiments carried out with glucose samples and water characterized by different
deuterium distributions, (D/H)j. In principle such experiments could be performed
by using only natural reactants with different isotopic distributions. However, in the
present state of accuracy of the isotopic determinations performed on glucose (23) it
was preferred to resort, in addition, to artificial modifications of the isotopic contents
2
through addition of very small quantities of selectively H-enriched glucoses (Table
1). The coefficients aij of the investigated correlations are gathered in Table 3.
DISCUSSION
Although detailed investigations of the individual reaction step of glycolysis have
been carried out and various kinetic isotope effects have been determined (12 13,16–

10
ZHANG, BUDDRUS, AND MARTIN
TABLE 3
Isotopic Coefficients aij Connecting Sites i of Products to Sites j of Reactants in Fermentation
Experiments Carried Out with Saccharomyces cerevisiae
Glycerol
Ethanol
aij (a)
i
A
B
t
2
I
II
j
1
␣
␤
␣
␤
␣
␤
␣
␤
0.04
0.03
0.18
0.13
0.29
0.29
0.53
0.51
0.22
0.23
0.02
0.0
0.25
0.25
0.45
0.43
0.10
0.11
0.08
0.05
0.21
0.22
0.50
0.48
0.0
0.01
0.0
0.0
0.0
0.02
0.52
0.52
0.13
0.13
0.09
—
0.29
0.28
0.18
0.19
0.0
0.0
2
6, 6Ј (b)
W
0.02
—
0.03
0.03
0.74
0.71
Note. (a) The sites considered in the starting medium are j ϭ 1,2,6,6Ј of glucose and j ϭ W of water.
Sites i ϭ I(methyl) and II(methylene) of ethanol and sites i ϭ A, B, and 2 (Fig. 1) of glycerol are observed
in the products. Two series of experiments have been conducted in parallel with maize glucose fermented
either in a conventional medium (␣) or in a medium containing a high concentration of Na2SO3 (␤) (see
Experimental Procedures). (b) Since the starting glucose contains species bideuterated at postion 6, isotope
effects are observed on the chemical shifts corresponding to isotopomers bideuterated on carbon 3 of
glycerol and on carbon I of ethanol. This phenomenon introduces distorsions of the line shape, which
are somewhat detrimental to the accuracy of the determinations.
21) the proportions of hydrogens at the different sites of glycerol, which have been
transferred from water, in a fermentation reaction conducted at natural isotopic abun-
dance, are not known. These proportions may depend to some extent on the experimen-
tal conditions of the reaction. In this respect the SNIF-NMR method is expected to
provide an efficient source of information on the overall participations of intra- and
intermolecular transfers in biosyntheses carried out with different microorganisms
and in different media (14,22,23).
From the point of view of the hydrogen isotopes the transformation of glucose into
glycerol occurring in the course of the glycolytic pathway can be characterized by
transfer coefficients aij, which enable the carbon-bound hydrogen atoms of glycerol,
i, to be connected to those of glucose, j (Eq. [4]).
FIG. 3. Reaction mechanism for the formation of glycerol from glucose. The fate of hydrogens from
sites 1(Ⅺ) 2(⅜) and 6-pro-S of glucose (᭝) is tentatively illustrated ignoring possible exchanges with
the aqueous medium. The steps which may be responsible for intermolecular exchanges are mentioned
(H2O). In order to illustrate the genealogy of the glycerol atoms the numbering adopted for the carbon
atoms follows that of glucose. The stereospecific numbering corresponding to sn-3-glycerol phosphate
is given within circles in the last glycerol formula. G6P, glucose 6-phosphate; F6P, fructose 6-phosphate;
F1,6P2, fructose 1,6-diphosphate; DHAP, dihydroxyacetone phosphate; G3P, glyceraldehyde 3-phosphate;
sn-Gly 3 P, sn-3-glycerol phosphate; Gly, glycerol.

ISOTOPE FRACTIONATION IN GLYCEROL BIOSYNTHESIS
11

12
ZHANG, BUDDRUS, AND MARTIN
Isotopic Connection with Glucose and Hydrogen Genealogy
It has been assumed that glycerol produced in a glycolytic reaction involving
glucose originates from the two fragments C1 - C2 - C3 and C4 - C5 - C6 of the glucose
skeleton (Fig. 3). These fragments result from the splitting of fructose 1,6-biphosphate
in the aldolisation step leading to the trioses, dihydroxyacetone phosphate, DHAP,
and glyceraldehyde 3-phosphate, G3P.
If the equilibrium between DHAP and G3P mediated by triose phosphate isomerase
is relatively fast, the glycerol molecules obtained through a reversible NADH-depen-
dent reduction of DHAP catalyzed by glycerol dehydrogenase may be issued from
both parts of the starting glucose. In these conditions carbons 1, 2, and 3 of glycerol
(Fig. 1) are derived from, respectively, carbons 3 ϩ 4, 2 ϩ 5, and 1 ϩ 6 of glucose.
Similarly, on the basis of the present knowledge of the glycolysis mechanism, each
of the four hydrogens bound to carbons 1 and 3 (Fig. 1) are expected to have the
following double origins (Fig. 3): 1S, introduced from water at the aldolization step
and from H4 of glucose with possible exchange with water; 1R, from H3 and H5 of
glucose, but exchange with water intervenes at the triosephosphate isomerization step;
3R, from H2 of glucose but with partial intermolecular transfer at the glucose phosphate
isomerization step and from the pro-R H6 position of glucose; 3S, from H1 and H6 pro-S
of glucose. Hydrogen at position 2 is introduced from NADH by glycerol phosphate
dehydrogenase. Full scrambling of the two glucose moieties should be characterized
in particular by significant values of the coefficients ai1, ai2, ai 6pro-R, and ai 6pro-S
involving sites 1,2,6pro-R, 6pro-S of glucose. A maximum theoretical value for full
conversion of 0.25 is then assigned to these coefficients as a result of the fourfold
degeneracy of the structure.
In a second hypothesis where the reduction of DHAP into glycerol is faster than
the reversible isomerization of the triose phosphates, G3P i DHAP, the glycerol
molecule would result solely from the C1 - C2 - C3 fragment of glucose and the origin
of each of the four hydrogens at positions 1 and 3 would be restricted to the first of
the two sources described above. Such an hypothesis has been favoured in the case
of yeast fermentation carried out in the absence of Na2SO3 (6). Significantly different
values of the coefficients aij of the transfer matrix, [A], connecting reactants and
products are expected in this mechanistic possibility. In particular the coefficient ai6
involving both sites 6 of glucose should be now close to zero.
The isotopic coefficients aA6 and aB6 connecting sites 6, 6Ј of glucose to sites A
and B of glycerol, exhibit relatively high values (Table 2), which prove that, even in
the absence of Na2SO3, glycerol is derived from both fragments, 1-2-3 and 4-5-6, of
glucose scrambled at the triosephosphate isomerization step. The observed values of
aA6 and aB6 may be compared to the theoretical values 0.25, which would correspond
to complete transfer from positions 6 pro-R and 6 pro-S of glucose to sites 3 pro-R
and 3 pro-S, respectively, of glycerol in conditions of full scrambling of the two
moieties issued from fructose 1,6-bisphosphate. The experimental results are in agree-
ment with a situation where both the aldolase and triosephosphate isomerase steps
are close to equilibrium. This behavior is very similar to that of the coefficients aI
and aI 6pro-S, which connect sites 6, 6Ј of glucose to the methyl site of ethanol,
6pro-R
the major product of the fermentation reaction (14). In addition the absence of

ISOTOPE FRACTIONATION IN GLYCEROL BIOSYNTHESIS
13
noticeable isotope effects on aA6 and aB6 is consistent with purely intramolecular
transfers associated with small secondary isotope effects. High levels of connection
are also observed between sites 1 and 2 of glucose and, respectively, positions 3 pro-
S and 3 pro-R of glycerol since the coefficients aB1 and aA2 reach values of about
0.22 and 0.18, whereas the theoretical values corresponding to complete intramolecular
transfers are equal to 0.25. Partial exchange with the aqueous medium occurring when
hydrogen at position 2 of glucose 6-phosphate is transferred to position 1 of fructose
6-phosphate at the isomerization step mediated by phosphoglucose isomerase
(19,30,31) may be responsible for a lowering of the aA2 coefficient with respect to
the theoretical value.
More generally, ignoring possible fractionation contributions resulting from second-
ary kinetic isotope effects in conditions of incomplete transformation, the ratio of the
experimental to the theoretical values of the aij coefficients involving sites j ϭ 1 to
6 of glucose and i ϭ A,B,2 of glycerol would represent the proportion of hydrogens
directly transferred from glucose to glycerol at every position. The present results
therefore characterize a high degree of intramolecular transfer from, on one hand
hydrogens 6 pro-R and, to a lesser extent, 2 of glucose to position 3 pro-R of glycerol
and, on the other hand from 6 pro-S and 1 of glucose to 3 pro-S of glycerol. It may
be concluded that glycerol is potentially a good isotopic probe for characterizing
disappeared sugar precursors (case of alcoholic beverages for instance).
Stereospecificity of the Enzymatic Reduction of DHAP
The values of the coefficients aij connecting sites 1, 2, 6, 6Ј of glucose to signals
A and B of glycerol are also representative of the stereospecificity of the reduction
step by glycerol dehydrogenase.
Thus the evolution of the spectrum of triacetin prepared from glycerol, itself derived
from a glucose sample slightly enriched at site 2, is illustrated in Fig. 2. A relative
increase in the intensity of signal A accompanies the deuterium enrichment at position
2 of glucose. This behavior corroborates the stereospecific character of the reduction
step of DHAP mediated by glycerol dehydrogenase (24–27) since a nonstereospecific
reaction would lead to similar enhancements of the A and B signals. Moreover since
signal A has been tentatively assigned, on the basis of a conformational analysis
1
using H-NMR parameters (28,29), to both the pro-R position at the pro-R carbon
2
and the pro-S position at the pro-S carbon of glycerol its H-enrichment provides
direct proof of the si-face type of the reduction of DHAP. The hydrogen atom at
position 2 of glucose initially transferred to the pro-R site on carbon 1 of fructose
1,6-diphosphate (19, 30) and subsequently occupying the pro-R position on carbon-
3 of DHAP must be associated with the pro-R position on carbon-3 of a sn-glycerol
3-phosphate molecule characterized by a stereochemistry R of carbon-2 (Fig. 3).
Conversely, admitting the si-face character of the reduction of DHAP by glycerol
phosphate dehydrogenase, the present experiments corroborate the chemical shift
1
2
assignment of the H- and H-NMR signals A and B to, respectively, the 1S ϩ 3R
and 1R ϩ 3S positions (Fig. 1). Although the considered reaction had already been
indirectly investigated, this example illustrates the advantages of the method for
directly elucidating the stereochemistry of individual reaction steps occurring in
complex bioconversions.

14
ZHANG, BUDDRUS, AND MARTIN
Isotopic Connections with the Aqueous Medium
Both the aiw and aij coefficients connecting glycerol sites, i, to water, w, on one
hand and to glucose sites, j, on the other hand, are expected to provide, respectively,
direct or indirect information on the participation of water in the hydrogen genealogy
of glycerol.
Since, whatever the yield of the bioconversion, strong isotope effects may accom-
pany hydrogen transfer from the large aqueous pool, the coefficients aiw involving
water are not directly exploitable for estimating the percentage of hydrogen originating
from water. However, in principle, for a given site such as i ϭ 2 that is strongly
connected with water, the proportion of hydrogen originating from water can be
indirectly estimated from the ratios of the experimental and theoretical values of the
aij coefficients involving glucose. In fact the deficit with respect to the maximum
value for j ϭ 1 to 6 may be attributed to the participation of water. An overall value
of the kinetic isotope effect characterizing hydrogen transfer from water can then be
calculated. This approach has enabled a limit value of 1.4 to be assigned to the
global kinetic isotope effect accompanying intermolecular transfers from water to the
methylene group of ethanol in the fermentation reaction (14). Moreover it was shown
that hydrogen transfer from NADH to the pro-R methylenic position of ethanol in
the reduction step of acetaldehyde conveys only relatively low indirect connection
with site 4 of glucose (14). In the case of glycerol the absence of connection between
site 2 and sites 1,2,6,6Ј of glucose is confirmed (Table 3). This site is strongly
connected with water and a primary kinetic isotope effect of about 2 would be
estimated from the a2W coefficient in the hypothesis of a unique origin from the
aqueous medium. However, possible connection with site 4 must be considered. The
value of a24 is not directly accessible from the present results but the significant
variations observed in the D/H value of site 2 of glycerol as a function of the origin
of the sugar precursor (32) suggest a tighter connection with the glucose skeleton
than in the case of the methylene group of ethanol. This connection is corroborated
by the behavior of the C2 parameter in equation 4 : (D/H)2 ϭ 0.52(D/H)w ϩ 23.4.
In spite of the restricted precision reached on this extrapolated parameter it may be
concluded from its noticeable value that a significant proportion of hydrogens in site
2 of glycerol is not derived from the aqueous medium.
In this respect it should be noted that in the glycolytic pathway the aldehyde
hydrogen of G3P transferred to NADH after the triose phosphate isomerization step
is related to hydrogen-4 of glucose and to the pro-S hydrogen on DHAP, itself
introduced from water at the aldolization step. Another source of exchange with water,
which has been shown to intervene in the absence of G3P (16,33–35), involves the
pro-S hydrogen of the CH2OH group of DHAP. Since this exchange enzymatically
catalyzed by aldolase also affects DHAP molecules resulting from isomerization of
the G3P moiety, it may be responsible for a loss of connectivity between H-4 of
glucose and both the 1-pro-S site in signal A and the H2 position of glycerol.
It should also be emphasized that the strong connection of clusters A and B with
carbon-bound hydrogens of glucose is further confirmed by the behavior of the C2
parameter of Eq. [4], involving j ϭ water. Thus C2, which measures the average
contribution of hydrogens not issued from water, is of the order of 70 ppm for A and

ISOTOPE FRACTIONATION IN GLYCEROL BIOSYNTHESIS
15
reaches about 85 ppm for B. This result is in agreement with the highest proportion
of intermolecular hydrogen transfer affecting hydrogen formally originating from site
2 of glucose and ending at cluster A as compared to hydrogen originating from site
1 and ending at cluster B. Another contribution to the difference between the C2
values could result from different deuterium contents at the 6 pro-R position of glucose
connected to B and at the 6 pro-S position connected to A.
ACKNOWLEDGMENTS
We are grateful to Mr. M. Trierweiler and Mrs. F. Mabon for their collaboration in the NMR determina-
tions, to Professor N. Naulet for help in mass spectrometry determinations, to Y. Yunianta for preliminary
experiments, and to Professors G. J. Martin and Dr. R. Robins for fruitful discussions.
REFERENCES
1. Martin, G. J., and Martin, M. L. (1981) Tetrahedron Lett. 22, 3525–3528.
2. Martin, G. J., Zhang, B. L., Naulet, N., and Martin, M. L. (1986) J. Am. Chem. Soc. 108, 5116–5122.
3. O’Leary, M. H. (1998) BioScience 38, 328–336.
4. De Niro, M. J., and Epstein, S. (1977) Science 197, 261–263.
5. Monson, K. D., and Hayes, J. M. (1980) J. Biol. Chem. 255, 11435–11441.
6. Weber, D., Kexel, H., and Schmidt, H. L. (1997) J. Agric. Food Chem. 45, 2042–2046.
7. Martin, G. J., Danho, D., and Vallet, C. (1991) J. Sci. Food Agric. 56, 419–434.
8. Martin, M. L., and Martin, G. J. (1990) in NMR Basic Principles and Progress, Vol. 23, pp. 1–61,
Springer-Verlag, Berlin.
9. Koziet, J., Rossmann, A., Martin, G. J., and Ashurst, P. R. (1993) Anal. Chim. Acta 271, 31–33.
10. Martin, Y. L. (1993) J. Magn. Res. 111, 1–10.
11. Rabiller, C., Mesbahi, M., and Levayer, F. (1990) Tetrahedron Asymmetry 2, 187–190.
12. Bush, K., Shiner, V. J., and Mahler, H. R. (1973) Biochemistry 12, 4802–4805.
13. Den Hollander, J. A., Brown, T. R., Ugurbil, K., and Shulman, R. G. (1979) Proc. Natl. Acad. Sci.
USA 76, 6096–6100.
14. Zhang, B. L., Yunianta, and Martin, M. L. (1995) J. Biol. Chem. 270, 16023–16029.
15. Zhang, B. L., Quemerais, B., Martin, M. L., Martin, G. J., and Williams, J. M. (1994) Phytochem.
Anal. 5, 105–110.
16. Rose, I. A., and Rieder, S. V. (1955) J. Amer. Chem. Soc. 77, 5764–5765.
17. Levy, H. R., and Vennesland, B. (1957) J. Biol. Chem. 228, 85–96.
18. Rieder, S. V., and Rose, I. A. (1959) J. Biol. Chem. 234, 1007–1010.
¨
19. Simon, H., Medina, R., and Mullhofer, Z. (1968) Z. Naturforch. 235, 59–64.
20. Simon, H., and Kraus, A. (1976) in Isotope Organic Chemistry, Ch.5, pp. 153–229.
21. Cook, P. F., Blanchard, J. S., and Cleland, W. W. (1980) Biochemistry 19, 4853–4858.
22. Vallet, C., Said, R., Rabiller, C., and Martin, M. L. (1996) Bioorg. Chem. 24, 319–330.
23. Zhang, B. L., Yunianta, Vallet, C., Martin, Y. L., and Martin, M. L. (1997) Bioorg. Chem. 25, 117–129.
24. Fischer, H. O. L., and Baer, E. (1941) Chem. Rev. 29, 287–316.
25. Bublitz, C., and Kennedy, E. P. (1954) J. Biol. Chem. 211, 963–967.
26. Swick, R. W., and Nakao, A. (1954) J. Biol. Chem. 206, 883.
27. Karnowsky, M. L., Hauser, G., and Elwyn, D. (1957) J. Biol. Chem. 226, 881–890.
28. Kosugi, Y., and Matsubara, K. (1989) J. Jpn. Oil Chem. Soc. 38, 415–420.
29. Nishida, Y., Uzawa, H., Hanada, S., Ohrui, H., and Meguro, H. (1989) Agric. Biol. Chem. 53, 2319–
2326.
30. Rose, I. A., and O’Connell, E. L. (1961) J. Biol. Chem. 236, 3086–3092.
31. Willem, R., Malaisse-Lagae, F., Ottinger, R., and Malaisse, W. J. (1990) Biochem. J. 265, 519–524.
32. Zhang, B. L., Buddrus, S., Trierweiler, M., and Martin, G. J. (1998) J. Agric. Food Chem., in press.
33. Bloom, B., and Topper, Y. J. (1956) Science 124, 982–983.
34. Rose, I. A. (1958) J. Am. Chem. Soc. 80, 5835–5836.
35. Rutter, W. J., and Ling, K. H. (1958) Biochim. Biophys. Acta 30, 71–79.