 .

13: 1347–1356 (1997)

A Rapid and Reliable Method for Metabolite

Extraction in Yeast using Boiling Buffered Ethanol
BENJAMIN GONZALEZ1,2, JEAN FRANC
q OIS2* AND MICHEL RENAUD1
1
Laboratoire de Technologie de la Nutrition et de l’Alimentation, Clermont-Ferrand, France
2
Centre de Bioingenierie Gilbert Durand, UMR-CNRS 5504, Laboratoire Associé INRA, Toulouse, France

Received 18 February 1997; accepted 16 April 1997

A simple and reliable method for the efficient inactivation of metabolism and for quantitative metabolite extraction
from yeast cells is presented. It is based on the use of a boiling solution made of 75% ethanol (volume/final volume)
buffered with 70 m-Hepes (final concentration), pH 7·5, to guarantee the stability throughout the whole procedure
of a large variety of metabolites, including all glycolytic intermediates, nucleotides, pyridine nucleotides and organic
acids compounds. The extraction is fast, requiring only 3 min incubation of yeast cells in the ethanol-buffered
mixture maintained at 80)C. It can be carried out either directly by spraying the cells into the boiling mixture, or after
quenching the whole culture in 60% methanol kept at "40)C. Extracts are subsequently concentrated by
evaporation under partial vacuum and the residue is resuspended in a small volume of water. This concentration
step and the use of a highly sensitive analytical method allow us to quantify metabolites in less than 10 mg dry
weight cells. This method, which can be applied to other fungi, could be very helpful for the determination
of true metabolites in mutants generated through the EUROFAN programme and for metabolic flux analysis.
? 1997 John Wiley & Sons, Ltd.

Yeast 13: 1347–1356, 1997.

No. of Figures: 3. No. of Tables: 4. No. of References: 20.

  — yeast metabolism; metabolite extraction; metabolic engineering

INTRODUCTION latter makes use of recombinant DNA technology

Metabolic control analysis and metabolic engin-
eering are complementary biological disciplines
used to quantify metabolite concentration, enzyme
activity and flux distribution throughout metabolic
pathways for modifying cellular activities of
microbial systems (Bailey, 1991; Sahm, 1993;
Westerhoff and Kell, 1996). While the former is
devoted to the quantitative analysis of the enzy-
matic steps controlling the metabolic flux, the
*Correspondence to: Jean M. François, Département de Génie
Biochimique et Alimentaire, Institut National des Sciences
Appliquées, Complexe Scientifique de Rangueil, 31077
Toulouse Cedex 04, France.
Contract grant sponsor: ECC.
Contract grant sponsor: French Ministry of Education.
to restructure metabolic networks of micro-
organisms to improve their productivity, yield
or ability to use cheaper raw material or even
more specialized desirable characteristics. Since
the metabolism is accomplished by a regulated,
coupled-network of more than 1000 reactions,
reliable in vivo data on metabolite concentrations
and kinetic properties of participating enzymes are
prerequisites for effective metabolic engineering.
The determination of metabolite levels in intact
cells is possible using nuclear magnetic resonance
(NMR) spectroscopy. Although NMR provides
in vivo data for well-determined conditions (De
Graaf et al., 1992), the low sensitivity of this
method restricts its use to abundant metabolites

CCC 0749–503X/97/141347–09 $17.50

? 1997 John Wiley & Sons, Ltd.
1348
and/or to high cell density cultures (Hartbrich
et al., 1996), or else needs enriched labelled sub-
strates (Weuster-Botz and de Graaf, 1996). Thus,
intracellular metabolite measurements are usually
performed with ex vivo methods which should
allow for the following problems: (i) fast cell
sampling and inactivation of metabolism, (ii) com-
plete extraction of desired metabolite and (iii)
absence of interference with analytical methods for
metabolite determination.
In order to minimize the cell sampling time from
the fermenter, most methods use a syringe partially
filled with a precooled inactivating and extracting
agent (Franco et al., 1984). Based on this principle,
Theobald et al. (1993) have developed a sampling
device using a miniature valve allowing fast sam-
pling every 5 s and inactivation of yeast metab-
olism within 0·5 s in cold perchloric acid at a final
concentration of 10%. One disadvantage of the
direct extraction methods is that cells are not
separated from the medium, and metabolites are
diluted before measurement. In conditions where
the compound of interest is present both extra-
cellularly and intracellularly, cell separation from
the medium is necessary. Direct concentration of
the cells by centrifugation or by filtration is not
recommended because the time span of these pro-
cedures is much longer than the turnover rates for
most of the intermediates. Quenching the metab-
olism into a solution of 60% methanol kept at
"40)C followed by collection of cells by centrifu-
gation (de Koning and Van Dam, 1992) or filtra-
tion (Ruijter and Visser, 1996) has been shown to
be an efficient method to concentrate cells and
to reduce alterations of metabolite concentration
during cell sampling, provided the temperature is
kept below "20)C throughout the procedure.
Another major challenge when determining the
true level of intermediates with a very efficient
extraction method is not to destroy the metabo-
lites. Conflicting values of intracellular metabolites
found in the literature might be explained in part
by the use of different extracting agents (Gancedo
and Gancedo, 1973; Weuster-Botz and de Graaf,
1996). Metabolites are often extracted at extreme
pH and well-known procedures involve either
strong acid reagent (perchloric or trichloroacetic
acid) or alkali (NaOH or KOH) for the measure-
ment of acid-stable and alkali-stable compounds,
respectively. When both acid- and alkali-stable
metabolites have to be determined, the extraction
has to be carried out on two separate samples,
making it difficult to reproduce the same cell
 . 13: 1347–1356 (1997)
.   .
sampling and quenching. Furthermore, there are
metabolites which are unstable in both acid- and
alkali-media. De Koning and Van Dam (1992)
developed an extraction method based on rapid
quenching of yeast metabolism in a 60% methanol
solution kept at "40)C followed by a neutral
extraction of metabolites using chloroform at
"40)C. However, this method appears relatively
tedious and time-consuming since the samples
have to be shaken with chloroform for 45 min at
"35)C to achieve complete permeabilization and
extraction. If not tightly controlled, this incu-
bation step could lead to erroneous data due to
enzymatic conversion during permeabilization and
extraction (de Koning and Van Dam, 1992).
Taking into consideration all the problems men-
tioned above, we have entirely revised the method
of boiling ethanol previously used but not docu-
mented by Entian et al. (1977). This method
presents the following advantages: (i) it is cheap,
easy to handle and does not require special equip-
ment, (ii) inactivation of metabolism is made
before the concentration step, allowing work with
a small amount of cells, (iii) extraction is carried
out at neutral pH, which guarantees the stability of
a large variety of metabolites, and (iv) this method
can be combined with cell quenching in cold
methanol when cell separation from the medium
is required.
MATERIALS AND METHODS
Reagents
Enzymes and biochemicals were purchased from
Sigma or Boehringer. All chemicals were of ana-
lytical grade and used as supplied. Pure ethanol
was used for the extraction procedure.
Organism, growth conditions and glucose pulses
Experiments were performed with the pro-
totrophic Saccharomyces cerevisiae CBS 8066
and CENPK133-7D (gift of K. D. Entian and
P. Koetter, Frankfurt, Germany). The yeast was
grown under vigorous agitation (200 rpm) at 30)C
on a synthetic mineral medium (Verduyn et al.,
1992) in the presence of 1% glucose. Glucose
pulses were carried out with stationary-phase cells
(100 ml of culture at 2 mg dry weight/ml) which
were harvested by centrifugation (10 min at
5000 g), washed twice with water and resuspended
in 100 ml of glucose-free synthetic mineral
? 1997 John Wiley & Sons, Ltd.
   
medium. After 30 min of preincubation at 30)C,
glucose was added to the cell suspension at a final
concentration of 1%.
Determination of yeast biomass and estimation of
cell viability and permeabilization
Yeast biomass was determined by gravimetric
analysis after filtration of cell samples (20 mg dry
weight) through preweighed nylon filters (45 mm
diameter, 0·45 ìm porosity) and dried to constant
weight at 60)C under partial vacuum (200 mmHg).
A calibration curve relating dry weight to optical
density at 660 nm was drawn. Survival after acidic
and ethanol treatments was estimated by plating a
serial dilution of cells on YPD agar plates and
counting the colonies after 3 days of incubation at
30)C. Cell permeabilization was estimated by
counting in a haemocytometer the proportion of
blue-coloured cells after treatment with methylene
blue (2% final concentration).
Cell sampling and direct extraction
Unless otherwise stated, 5 ml of yeast culture
was harvested 5 min after addition of glucose,
immediately dropped directly into a 35 ml glass
tube containing 21 ml of absolute boiling ethanol
buffered with 2 ml of 1 -Hepes, pH 7·5 (ethanol-
buffered solution), and incubated for 3 min at
80)C. After cooling down the mixture on ice for
3 min, the volume was reduced by evaporation
at 45)C using a rotavapor apparatus (Brüchi
Rotavapor R-134, BioBlock, France). The vacuum
was progressively adjusted to 5 mbars using a
manometrically-controlled pump. The rate of
evaporation was about 0·5 ml min "1. The residue
was resuspended to a final volume of 1 ml with
double-distilled water and centrifuged for 10 min
at 5000 g at 4)C to remove the insoluble particles.
The supernatant was stored at "25)C until use.
Acid extraction was carried out by addition of
1·4 ml HClO4 70% and 0·64 ml imidazol 1  to
5 ml of yeast cultures that had been previously
frozen in liquid nitrogen. The suspension was
allowed to thaw on ice. Optimal extraction of
acid-stable metabolites was carried out by three
freeze–thaw cycles with rigorous shaking between
each cycle, as recommended by others (Weibel
et al., 1974; Saez and Lagunas, 1976; Theobald
et al., 1993). After centrifugation (5 min at 8000 g
at 0)C), the supernatant was neutralized with 10 -
KOH and the KClO4 precipitate was removed by
centrifugation. The supernatant was stored at
? 1997 John Wiley & Sons, Ltd.
1349
"25)C until use. Preparation of alkaline extracts
was performed by addition of hot NaOH (0·1 -
final concentration) as described in François et al.
(1984).
Quenching in methanol and extraction of
metabolites
When separation of the cells from the medium
was required, 5 ml of sample was sprayed into
26 ml of cold solution containing 60% methanol
(volume/final volume) and 70 m-Hepes, pH 7·5
(methanol-buffered solution), kept at "40)C in a
dry ice/ethanol bath. The mixture was allowed to
cool down for 3 min and was centrifuged at 5000 g
for 5 min in a Sorvall RC5B set at "10)C. It was
checked that the temperature of the mixture re-
mained well below "20)C after the centrifugation
period. Metabolites from the cells pellets were
extracted in 5 ml of a solution of 75% (v/v) boiling
absolute ethanol containing 0·25 -Hepes, pH 7·5
as described above.
Analytical procedures
Metabolite measurements were performed by
coupling appropriate enzymes with a fluorometric
detection of NADH or NADPH. Emission was
measured at 450 nm after excitation at 350 nm
using a fluorescence spectrophotometer (Hitachi
F-2000). Enzymatic determinations of metabolites
were performed at 30)C in 10 m-Tris/HCl buffer
pH 7·5 containing 2 m-EDTA, 10 m-MgSO4
and 0·2 -KCl, unless otherwise stated. Samples
were added so that the amount of metabolite in the
assay was below 5 nmol/ml.
Glucose 6-phosphate (glucose-6P), fructose
6-phosphate and ATP were measured in the pres-
ence of 0·4 m-NAD + and 0·1 m-glucose by
addition of glucose 6-phosphate dehydrogenase
(4·5 U/ml), phosphoglucoisomerase (2 U/ml) and
hexokinase (1·5 U/ml), respectively. Triose-
phosphates (dihydroxyacetone-phosphate
[DHAP]+glyceraldehyde-3-phosphate [G3P]) were
measured in the presence of 10 ì-NADH by
addition of glycerol 3-phosphate dehydrogenase
(1·5 U/ml) and triose isomerase (1·0 U/ml). The
same reaction mixture was used for the determi-
nation of NADH except that 0·1 m-DHAP was
added in the place of NADH. Intracellular pyru-
vate was measured in the presence of 10 ì-
NADH by addition of lactate dehydrogenase
(2·0 U/ml). After completion of the reaction,
0·2 m-phosphoenolpyruvate was added and after
 . 13: 1347–1356 (1997)
1350 .   .
stabilization of the baseline, pyruvate kinase Table 1. Rate of metabolite loss during treatment with
(1·8 U/ml) was added to measure ADP. Isocitrate non-buffered and buffered ethanol solution at 75% final
was measured in the presence of 0·1 m-NADP + concentration.
by addition of isocitrate dehydrogenase (1·2 U/ml).
á-Ketoglutarate was measured in the presence of Slope (% lost/min)
0·2 m-NH4Cl, 0·1 m-ADP as activator, 10 ì- Buffered with
NADH by addition of glutamate dehydrogenase 70 m-Hepes
(2·4 U/ml). NAD + was assayed in glycine buffer as Non-buffered pH 7·5
described by Klingenberg (1974).
Glucose-6P 0·13&0·14 0·20&0·18
Fructose-6P "0·73&0·22 ND
RESULTS AND DISCUSSION ATP "0·29&0·15 0·02&0·02
ADP "0·60&0·14 ND
To our knowledge, the first paper dealing with the Pyruvate "1·38&0·20 "0·33&0·28
extraction of metabolite using boiling ethanol was NAD "4·63&0·03 "0·57&0·43
from Entian et al. (1977). In the original protocol, NADH "8·70&0·09 "0·27&0·18
no special precaution was taken as cells were Isocitrate "0·95&0·21 ND
harvested by centrifugation and metabolites were á-Ketoglutarate 0·39&0·28 ND
extracted by incubation of the pellets in boiling
ethanol for 10 min. Although this method has Commercial metabolites (1 m each) prepared in synthetic
subsequently been used by many other investiga- mineral medium, pH 5·0 were incubated at 80)C in boiling
ethanol (75% final volume) in the absence or presence of 70 m
tors (Ciriacy and Breitenbach, 1979; Gamo et al., (final concentration) Hepes buffer, pH 7·5. Aliquots of 1 ml
1993; Boles and Zimmermann, 1993), the lack of were withdrawn at times 0, 1, 3, 5, 10 and 15 min, evaporated
data about the reliability of this method prompted and resuspended in 1 ml of bidistilled water. The results are
us to revise the procedure, taking into consider- expressed as the percentage of metabolite lost per min
ation the following aspects: (i) stability of all (slope)&standard deviation of three independent experiments.
metabolites (e.g. acidic- and alkaline-stable
metabolites, organic acids) in boiling ethanol, (ii)
complete cell permeabilization for total and repro- was added at a final concentration of 70 m to the
ducible release of metabolites, and (iii) comparison solution of boiling ethanol (Table 1). Hence, we
with other sampling and extraction procedures to can assume that other intracellular metabolites
demonstrate the validity and the accuracy of this which have not been tested here, would also be
method. stable in boiling buffered ethanol. Furthermore,
we have verified that this solution had sufficient
Stability of metabolites during the extraction buffering capacity to neutralize yeast cultures
procedure ranging from pH 3·5 to 6·0.
A critical requirement when using ethanol as an
extracting agent was to ascertain the lack of chemi- Extraction, comparison with classical methods and
cal interference of this solvent with the stability of recovery of metabolites
metabolites during extraction. This problem was It has been known for more than 100 years that
first investigated with solutions of commercial ethanol kills enzymes, and thus is well suited to
metabolites which were prepared in mineral inactivate metabolism rapidly. We tested how
medium pH 5·0 (Verduyn et al., 1992), in order to much hot ethanol was needed to completely ex-
be as close as possible to the extraction of metab- tract metabolites from yeast cells. As shown in
olites from whole yeast cultures. As shown in Figure 1, more than 66% (v/v final) of buffered
Table 1, certain metabolites such as pyruvate, boiling ethanol solution was necessary for com-
NAD + and NADH were relatively unstable during plete release of metabolites, whereas cell lysis was
extraction and a destruction rate ranging from 1·5 complete from 30% (v/v final) of boiling ethanol
to 10% per min could be calculated. To circumvent (not shown). Interestingly, a complete release of
this serious problem, the ethanol solution was ADP was obtained with a relatively low concen-
buffered around pH 7·0 with different buffers. It tration of ethanol (25% v/v final), whereas 50% of
was found that the loss of metabolites was reduced glucose-6P and pyruvate, and only 10% of ATP
to less than 1% per min when Hepes buffer, pH 7·5, and NAD + were extracted at this concentration.

 . 13: 1347–1356 (1997) ? 1997 John Wiley & Sons, Ltd.
   
Figure 1. Effect of increasing the final concentration of the
boiling buffered ethanol solution upon the release of metabo-
lites from yeast cells. Samples (5 ml) from a yeast suspension
incubated for 5 min with 1% glucose were immediately dropped
into boiling ethanol buffered with Hepes, pH 7·5 at 70 m final
concentration and incubated for 5 min at 80)C. The results
shown are from one experiment that was repeated three times
with a standard error of less than 5%.
Figure 2. Effect of incubation time at 80)C on the complete release of
metabolites from yeast cells. Same procedure as described in Figure 1, except
that the final concentration of the boiling ethanol solution was 75% (v/v final)
buffered with 70 m-Hepes, pH 7·5.
? 1997 John Wiley & Sons, Ltd.
1351
The low extractability of these latter two molecules
at low ethanol concentration could be due either to
their different compartmentation or to their bind-
ing to macromolecules or both. Because metabo-
lites have been extracted from stationary yeast cells
resuspended in glucose, it could be thought that a
great proportion of ATP is sequestered in active
mitochondria whose membrane is perhaps more
resistant to ethanol than the plasma membrane. It
is remarkable that the profile of nucleotide extrac-
tion shown in Figure 1 is similar to that obtained
by Theobald et al. (1993), who investigated the
effect of increasing concentration of perchloric
acid on nucleotide release from continuous cul-
tures of yeast that had been subjected to glucose
pulses. These data support the notion that differ-
ential metabolite extraction may be related to
either compartmentation and/or tight association
to macromolecules rather than their stability in the
extracting agents.
The next step was to determine the incubation
time required for complete release of metabolites
from yeast when using a 75% (v/v final) boiling
buffered ethanol solution. As shown in Figure 2, a
3-min incubation at 80)C was optimal for complete
metabolite release. However, it should be stressed
 . 13: 1347–1356 (1997)
1352 .   .
Table 2. Extraction of metabolites from yeast by boiling buffered ethanol or by cold HClO4.

Metabolite concentrations (ìmol/g dry weight)

Glucose-6P ADP ADP Pyruvate NAD

Ethanol 5·93&0·4 7·22&0·3 6·41&0·2 11·97&1·1 9·65&0·7

HClO4 5·68&0·2 6·83&0·1 6·68&0·4 7·45&0·5 6·02&0·4

5 ml of yeast cultures at the exponential phase of growth on glucose were sprayed either into 75% (volume/final volume) of boiling
buffered ethanol or into 14% (final concentration) of ice-cold HClO4. Extraction and metabolite assays were performed as
described in Materials and Methods. Measurements are the average&standard deviation of three extracts from the same culture.

Table 3. Effect of HClO4 upon the stability of commercial metabolites.

Amount before acid Amount after acid Recovery

treatment (ì) treatment (ì) (%)

Glucose-6P 227·1 224·2&3·8 99

ADP 159·1 153·4&5·9 96
Pyruvate 199·2 177·5&8·1 89
NAD 235·4 95·9&4·7 41

Commercial metabolites were treated with 14% (final concentration) of HClO4 as described in Materials and Methods.
Measurements are the average&standard deviation of three experiments.

that incubation longer than 10 min resulted in a
significant loss (3–10%) of metabolites including
fructose-1,6-P2, pyruvate, triose phosphates and
NADH (not illustrated).
The reliability of this method was directly com-
pared with perchloric acid extraction, which is
commonly used to extract acid-stable metabolites
including NAD + (Table 2). Aliquots were taken
from the same yeast culture at the exponential
phase of growth on glucose and were treated with
either boiling buffered ethanol or HClO4 solution.
While similar values for glucose-6P, ATP and
ADP were obtained with the two extracting agents,
extraction with HClO4 led to a 30–40% loss of
pyruvate and NAD + . The dramatic loss of these
two metabolites could be mainly accounted for by
their high instability in the high acidic medium
prevailing at the early step of the extraction
(Table 3).
To test for the lack of enzymatic or chemical
conversions during extraction, a number of metab-
olites were added to the cells at a concentration
close to that expected to be found intracellularly.
As reported in Table 4, the recovery of the exo-
genous plus endogenous metabolites ranged from
91% to 110%, using the boiling buffered ethanol
solution, while it was inconsistent with the HClO4
method. These data further confirmed the low
reliability of the latter method and demonstrated
that yeast metabolism was instantaneously
stopped in contact with boiling ethanol mixture.
Because the neutral extraction using chloroform
at "35)C also appears doubtful as a complete
absence of enzymatic conversion cannot be guar-
anteed during extraction (de Koning and Van
Dam, 1992), we conclude that the use of a 75%
boiling buffered ethanol solution appears today
the more accurate method for the determination of
the true levels of intermediary metabolites in yeast.
Levels of metabolites in yeast before and after
quenching in cold methanol
The major drawback of this extraction method-
ology is that it cannot be readily applicable for
determining metabolites which freely diffuse out,
or that are excreted from cells, such as succinate,
acetate, pyruvate and other carboxylic acids. It
is also not appropriate with cultures made in
complex media containing yeast extract and/or
bactopeptone, because the yellow colour of these
compounds interferes with the photometric and
the fluorometric detection of NADH. Therefore,
we used the quenching method developed by de

 . 13: 1347–1356 (1997) ? 1997 John Wiley & Sons, Ltd.
    1353
Table 4. Metabolite recovery after extraction in buffered boiling ethanol or perchloric acid in the presence of
yeast cells.

Metabolite measured
in control cell extract
(ìmol/g dry weight)
Amount added
(ìmol/g dry weight)
Metabolite measured
in extract with
metabolites added
before extraction Recovery
(ìmol/g dry weight) (%)

Glucose-6P 5·2&0·3 (6·2&0·1) 13 (15·14) 17·7&1·8 (22·3&0·4) 97 (104)

ATP 6·7&0·3 (7·6&0·9) 5·6 (7·24) 11·2&0·5 (14·8&0·6) 91 (100)
ADP 5·9&0·2 (7·2&0·4) 9·1 (8·85) 16·1&1·1 (16·8&0·5) 107 (105)
Pyruvate 16·2&0·1 (6·7&0·1) 12·1 (10·31) 31·3&2·8 (12·9&1·1) 111 (76)
NAD 12·1&0·8 (7·0&0·1) 10·3 (10·56) 22·3&1·7 (8·5&0·3) 99 (48)
NADH 0·9&0·1 (ND) 1·4 (ND) 2·1&0·3 (ND) 91 (ND)

Results after extraction in perchloric acid are shown in parentheses. Exogenous metabolites were added to a suspension of yeast
incubated for 5 min with 1% glucose just before spraying the cells into 75% (volume/final volume) of boiling buffered ethanol or
into 14% (final volume) of ice-cold HClO4. Extraction of metabolites by the two methods was done from two different yeast
cultures (the absolute data are not comparable). Metabolite assays were performed as described in Materials and Methods.
Measurements are the average&standard deviation of four extracts.

Koning and Van Dam (1992) to separate cells
from the medium and we compared the values
obtained after direct extraction with those
obtained in cell pellets after quenching in 60%
methanol kept at "40)C in dry ice/ethanol and
extracted in a boiling buffered ethanol solution. As
can be seen in Figure 3, levels of all metabolites in
the extracts from whole yeast cultures were higher
than those found in the cell pellets. This raised
the question whether methanol treatment caused
a leak of metabolites out of the cells. Actually,
significant amounts of glucose-6P, ATP and
NAD + were measured in the supernatant from
methanol-quenched cells. However, the amounts
of these metabolites were roughly equal to those
found in the culture medium (Figure 3), which
therefore accounted for the higher values of
metabolites measured in the whole yeast culture.
These experiments demonstrated the validity of the
quenching method to efficiently stop yeast metab-
olism and to not alter cell permeability. Further-
more, the extraction of metabolites using boiling
buffered ethanol solution instead of chloroform at
"35)C will speed up the whole procedure. How-
ever, as emphasized by de Koning and Van Dam
(1992), the quenching procedure is reliable as long
as the temperature is kept well below "20)C
throughout the centrifugation step. In favour of
their claims, we have observed that the levels of
glucose-6P and ATP were two- to three-fold
lower when the temperature reached "10)C after
centrifugation.
The results of the experiment shown in Figure 3
also raise intriguing questions about the release
of glucose-6P, ATP and NAD + in the culture
medium, as these molecules are supposed not to be
excreted by yeast. This release is not likely to be
due to the centrifugation since, as stated above,
similar levels were found in the supernatants of
cells whose metabolism had been frozen by
quenching in 60% methanol at "40)C. It may be
due to permeabilization and/or lysis of cells during
growth, since we found that about 20% of the yeast
population in the exponential phase of growth on
a glucose synthetic medium turned to dark blue
upon incubation with 2% methylene blue (not
shown). Interestingly, this observation was made
with two different prototrophic strains, CBS8066
and CENPK133-7D. While it does not modify the
validity of the boiling ethanol method to extract
metabolites, the apparent leakage of intermediary
metabolites out of the cells needed to be checked
with any type of yeast strains and for any
experiments devoted to metabolic flux analysis.
To conclude, we have developed a simple, accu-
rate and reliable method for the extraction of
metabolites based on a 75% (v/v final) ethanol
solution buffered with 70 m-Hepes, pH 7·5. This
method does not require sophisticated equipment.
It can be performed with very high dilute cell
density culture as the concentration step is carried
out after the inactivation of metabolism and the
extraction of metabolites. An additional advantage
is that the extracts are ready for use on various

? 1997 John Wiley & Sons, Ltd.  . 13: 1347–1356 (1997)
1354
Figure 3. Levels of metabolites in whole culture, culture
medium and in cell pellets after quenching in methanol and
removal of methanol by centrifugation. Samples (5 ml) from
yeast at the exponential phase of growth on glucose (A660 =1·5
or about 2#107 cells/ml) were harvested and dropped either
directly into boiling buffered ethanol solution, or into
methanol-buffered solution as described in Materials and
Methods. After centrifugation at "20)C and careful draining
of the supernatant, which was saved for metabolite measure-
ment, the pellet was resuspended into 5 ml of boiling buffered
ethanol solution. Samples (5 ml) of the culture medium were
also collected by a 2-min centrifugation at 4)C and lyophilized.
The residue was resuspended in 0·5 ml of water. Metabolites
were determined in whole cell culture (black bar), in the culture
medium (open bar), in the cell pellet after methanol quenching
(horizontal hatched bar) and in the supernatant from the
methanol-quenched suspensions (vertical hatched bars). Note
that metabolites in the culture medium are expressed in mol per
g dry weight. Mean values and standard deviations of four
experiments are shown.
types of chromatography columns. Finally, this
method can be extended to the determination of
metabolites in yeast on a subsecond time scale and
could be applied to other micro-organisms includ-
ing filamentous fungi (H. Hajjaj, B. Gonzalez and
J. François, unpublished data).
ACKNOWLEDGEMENTS
This work was supported by grants from the ECC
(program Cell Factories BIO.CT95.132) to J.F.
and from the French Ministry of Education (pro-
gram ACC-SV 14 ‘Biotechnologies’) to M.R. B.G.
is supported by a fellowship from the Ministère
de la Recherche et de l’Enseignement Supérieur
(MENESR).
 . 13: 1347–1356 (1997)
.   .
REFERENCES
Bailey, J. E. (1991). Toward a science of metabolic
engineering. Science 252, 1668–1675.
Bergmeyer, H. U. (1985). Methods in Enzymatic Analy-
sis, vol. VI and VII. VCH Publishers.
Boles, E. and Zimmermann, F. (1993). Different signals
control activation of glycolysis in the yeast Saccharo-
myces cerevisiae. Yeast 9, 761–770.
Ciriacy, M. and Breitenbach, I. (1979). Physiological
effects of seven different blocks in glycolysis in
Saccharomyces cerevisiae. J. Bacteriol. 139, 152–160.
de Graaf, A. A., Wittig, R. M., Probst, U.,
Strohhaecker, J., Schobert, S. M. and Sahm, H.
(1992). Continuous-flow NMR bioreactor for in vivo
studies of microbial cell suspensions with low biomass
concentration. J. Magn. Res. 98, 654–659.
De Koning, W. and Van Dam, K. (1992). A method for
the determination of changes of glycolytic metabolites
in yeast on a subsecond time scale using extraction at
neutral pH. Anal. Biochem. 204, 118–123.
Entian, K. D., Zimmermann, F. K. and Scheel, I. (1977).
A partial defect in carbon catabolite repression mu-
tants of Saccharomyces cerevisiae with reduced hexose
phosphorylation. Mol. Gen. Genet. 156, 99–105.
Franco, C. M. M., Smith, J. E. and Berry, D. R. (1984).
Effect of nitrogen and phosphate on the levels of
intermediates in baker’s yeast grown in continuous
culture. J. Gen. Microbiol. 130, 2465–2472.
François, J., Van Schaftingen, E. and Hers, H. G.
(1984). The mechanism by which glucose increases
fructose-2,6-bisphosphate concentration in Saccharo-
myces cerevisiae. Eur. J. Biochem. 145, 187–193.
Gamo, J., Portillo, F. and Gancedo, C. (1993). Charac-
terization of mutations that overcome the toxic effect
of glucose on phosphoglucose isomerase less strains
of Saccharomyces cerevisiae. FEMS Microbiol. 106,
233–238.
Gancedo, J. M. and Gancedo, C. (1973). Concentrations
of intermediary metabolites in yeast. Biochimie 55,
205–211.
Hartbrich, A., Schmitz, G., Weuster-Botz, D., de Graaf,
A. A. and Wandrey, C. (1996). Development and
application of a membrane cyclone reactor for in vivo
NMR spectroscopy with high microbial cell densities.
Biotechnol. Bioeng. 51, 624–635.
Klingenberg, M. (1974). In Bergmeyer, H. U. (Ed.),
Methods of Enzymatic Analysis. Academic Press, New
York, pp. 2102–2110.
Ruijter, G. J. G. and Visser, J. (1996). Determination
of intermediary metabolites in Aspergillus niger.
J. Microbiol. Methods 25, 295–302.
Sahm, H. (1993). Metabolic design. In Rehm, H. J.
and Reed, G. (Eds), Biotechnology, vol. 1. VCH
Weinheim, pp. 190–221.
Theobald, U., Mailinger, W., Reuss, M. and Rizzi, M.
(1993). In vivo analysis of glucose-induced fast
glycolytic changes in yeast adenine nucleotide pool
? 1997 John Wiley & Sons, Ltd.
   
applying a rapid sampling technique. Anal. Biochem.
214, 31–37.
Verduyn, C., Postma, E., Scheffers, W. A. and Van
Dijken, J. P. (1992). Effect of benzoic acid on metabo-
lite fluxes in yeasts: A continuous-culture study on the
regulation of respiration and alcoholic fermentation.
Yeast 8, 501–517.
Weibel, K. E., Mor, J. R. and Fiechter, A. (1974). Rapid
sampling of yeast cells and automated assays of
? 1997 John Wiley & Sons, Ltd.
1355
adenylate, citrate, pyruvate and glucose-6-phosphate
pools. Anal. Biochem. 58, 208–216.
Westerhoff, H. V. and Kell, D. B. (1996). What Bio-
Technologists knew all along . . .? J. Theor. Biol. 182,
411–420.
Weuster-Botz, D. and de Graaf, A. A. (1996). Reaction
engineering methods to study intracellular metabolite
concentrations. Adv. Biochem. Eng. Biotech. 34, 75–
110.
 . 13: 1347–1356 (1997)