Physiological Significance of the Cytometric Distribution of Fluorescent Yeasts After Viability Staining J. C. Bouchez, 1 M. Cornu, 2 M. Danzart, 1 J. Y. Leveau, 1 F. Duchiron, 3 M. Bouix 1 1 Ecole Nationale Supe ´rieure des Industries Agricoles et Alimentaires, 1 avenue des Olympiades, 91744 Massy Cedex, France; telephone: 69-93-51-43; fax: 69-93-50-84; e-mail: bouchez @ensia.fr 2 Agence FranB aise de Se ´curite ´ Sanitaire des Aliments, 22 rue Pierre Curie, BP332-94709 Maisons Alfort Cedex, France 3 Laboratoire de Microbiologie Industrielle, Universite ´ de Reims, BP 1039 Moulin de la Housse, 51687 Reims Cedex 2, France Received 15 April 2003; accepted 15 December 2003 Published online 8 April 2004 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/bit.20054 Abstract: This article describes a new method for the early detection of alcoholic fermentation arrest. This methodo- logy is based on the flow cytometric assessment of Saccharomyces cerevisiae yeasts stained with a carboxy- fluorescein diacetate fluorescent viability probe. Multi- component analysis of viable cell distribution constitutes a promising new tool to describe physiological and dynamic changes to heterogeneous viable populations during alcoholic fermentation, through its ability to discriminate between successful processes and those ending prema- turely. This framework, which is based on the comparison of cytometric histogram descriptors’ combinations that can be related to simple physiological significance comparison, quickly and simply, allows testing yeasts for their fermen- tation ability and can be used to detect any kind of viability loss so that fermentation arrest can be avoided. B 2004 Wiley Periodicals, Inc. Keywords: Saccharomyces cerevisiae; alcoholic fermenta- tion; yeast physiology; fluorescent viability probe; flow cytometric histogram modelisation INTRODUCTION Viable preparations of Saccharomyces yeasts are widely employed in the wine industry, particularly for the pro- duction of Champagne. In addition to the specificities of yeast strains, the consistency and quality of Champagne may be directly influenced by their physiological state or fermentation operating conditions. Because the process is often well controlled, producers are frequently faced with the problem of must composition, which is certainly the most variable parameter responsible for fermentation arrest prior to consumption of the entire carbon source. This is usually caused by a nitrogen deficiency or the persistence of pesti- cide residues in the must which hinder satisfactory fermen- tation because they affect the viability or vitality of yeasts. Yeast performance in alcoholic fermentation depends directly on yeast activity, which can be seen as a function of cell viability as well as the physiological state of viable cells. If it were possible to measure yeast viability and/or vitality rapidly and accurately, producers would then be able to take corrective action at the beginning of fermentation. Over the years, several methods have been developed to measure viability and vitality (Attfield et al., 2000; Boulton, 1996; Edwards et al., 1997; Lentini, 1993; Lloyd and Hayes, 1995; McFeters et al., 1995; Porter et al., 1996): plating, slide culture, vital stains, metabolic activity or other methods to assess viability, and metabolic activity, cell components, fermentation capacity, acidification potential or oxygen uptake ability to assess vitality. Among vital stains, fluorescent dyes are amenable to extremely rapid analysis, particularly when used in con- junction with a flow cytometer. Staining with carboxy- fluorescein diacetate is based on the assumption that only cells with an intact membrane and esterase activity are able to accumulate the fluorescent agent. This concept has been applied to the counting of viable yeasts and bacteria in a variety of food products (Laplace-Builhe ´ et al., 1993). Flow cytometric alternatives have been compared with standard methods for the assessment of yeast cultures in baking, wine-making, cider manufacture, and brewing. In all cases, the results of flow cytometry closely correlated with the standards. However, although knowledge of whether cells are alive or dead is important, it is inadequate because cells may be viable but weakly active and incapable of achieving fermentation. Recently, Bouix and Leveau (2001) used the carboxyfluorescein diacetate efflux phenomenon described by Breeuwer et al. (1994) to determine yeast vitality by calculating the relative decrease in the median fluorescent index of loaded viable cells after probe excretion for 15 min. The evolution of flow cytometric distribution during fermentation reveals dynamic changes in heterogeneous B 2004 Wiley Periodicals, Inc. Correspondence to: J. C. Bouchez
11
Embed
Physiological significance of the cytometric distribution of fluorescent yeasts after viability staining
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Physiological Significance of the CytometricDistribution of Fluorescent Yeasts AfterViability Staining
J.C. Bouchez,1 M. Cornu,2 M. Danzart,1 J.Y. Leveau,1 F. Duchiron,3 M. Bouix1
1Ecole Nationale Superieure des Industries Agricoles et Alimentaires, 1 avenuedes Olympiades, 91744 Massy Cedex, France; telephone: 69-93-51-43;fax: 69-93-50-84; e-mail: [email protected] FranBaise de Securite Sanitaire des Aliments, 22 rue Pierre Curie,BP332-94709 Maisons Alfort Cedex, France3Laboratoire de Microbiologie Industrielle, Universite de Reims, BP 1039 Moulinde la Housse, 51687 Reims Cedex 2, France
Received 15 April 2003; accepted 15 December 2003
Published online 8 April 2004 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/bit.20054
Abstract: This article describes a new method for the earlydetection of alcoholic fermentation arrest. This methodo-logy is based on the flow cytometric assessment ofSaccharomyces cerevisiae yeasts stained with a carboxy-fluorescein diacetate fluorescent viability probe. Multi-component analysis of viable cell distribution constitutesa promising new tool to describe physiological and dynamicchanges to heterogeneous viable populations duringalcoholic fermentation, through its ability to discriminatebetween successful processes and those ending prema-turely. This framework, which is based on the comparisonof cytometric histogram descriptors’ combinations that canbe related to simple physiological significance comparison,quickly and simply, allows testing yeasts for their fermen-tation ability and can be used to detect any kind of viabilityloss so that fermentation arrest can be avoided. B 2004Wiley Periodicals, Inc.
with cellular compounds, etc.). In practice, if percentage
fluorescence could not be evaluated and no assumption
could be made concerning the ethanol produced, the
median index alone did not allow determination of the
physiological state of cells. The introduction of additional
descriptors would assist this diagnosis, because it could
then be assumed that each physiological state has its own
distribution signature.
Indeed, width and dissymmetry indices, which have
never been described before, demonstrated reproducible
profiles throughout the process, showing that their values
were probably related to the physiological state of yeasts.
For these two descriptors, index values rose slightly
between 15 and 25 g.L�1 ethanol, beyond the values ex-
pected under a gaussian distribution (reference = 1) in
terms of width and reaching about 2.5 regarding dissym-
metry, demonstrating that the distributions were clearly
skewed to the left.
It is clear that histograms became much thinner and taller
as yeast homogeneity increased (which means that different
Figure 4. Discrimination between normal and aborted fermentations.
Mean and confidence interval (95%) plots for the three descriptors: (a)
median indice (normal –n– and aborted –5– fermentations); (b) width
index (normal –x – and aborted – w – fermentations); (c) dissymmetry
index (normal –E– and aborted –4– fermentations).
BOUCHEZ ET AL.: CYTOMETRIC DISTRIBUTION OF FLUORESCENT YEASTS 525
staining properties among the total viable population were
not markedly different). At the initiation of fermentation,
high width values could be explained by heterogeneity of
the inoculum. Thereafter, width values rapidly fell, reach-
ing a minimum at about 10 g.L�1 ethanol. At this stage of
fermentation the specifically high ethanol production rate
resulted from the contribution of an increasing proportion
of viable cells in the same healthy physiological state
(marked homogeneity). During the process, differences
among viable cells were revealed by flattening of the
histogram and greater cell heterogeneity, because of the
combined effect of yeast aging (disparity in budding states)
and increased ethanol production (disparity of stress sus-
ceptibility related to yeast aging and its effect on the
evolution of membrane composition).
As from 20 g.L�1 ethanol, cell homogeneity was likely
to reappear gradually, for two main reasons:
1) The oldest yeasts, which had previously contributed to a
fraction of less fluorescent cells and were incapable of
overcoming the combined stress of the production of
ethanol and other toxic substances, gradually disap-
peared because of lysis.
2) Those cells capable of recovery contributed to a more
homogeneous fraction of cells (viable, surviving cells),
although their number only represented a small
percentage of total cells.
As for the dissymmetry index profile, it is suggested that
their peak heterogeneity was due to an increase in the
number of less viable cells in the total viable population
due to cell death and loss of vitality, resulting in a drop in
the proportion of viable cells together with a decline in the
specific rate of ethanol production.
Arrested Fermentations
During alcoholic fermentation in batches, the increasing
concentration of ethanol adversely affected the state and
the activities of the yeast population, including its specific
growth rate, specific rate of fermentation, and viability.
Under an increased process temperature, some of these
effects may have become more severe (Beney et al., 2000).
The underlying mechanisms are numerous and include
the irreversible denaturation and hyperbolic noncompeti-
tive inhibition of glycolytic enzymes, the exponential
noncompetitive inhibition of the glucose transport system,
depression of the optimum and maximum temperatures for
growth, and the enhancement of thermal death.
Biochemical Results
For this 40jC high-temperature fermentation, high concen-
trations of ethanol were formed and this occurred most
rapidly during the early part of the fermentation, during
active yeast growth and multiplication. At this stage the
intracellular concentration of ethanol may be high and par-
ticularly toxic, as it may act from within the cell to interact
with the cell membrane, leading to a loss of viability in the
overall population. These results are in line with previous
reports that yeast viability decreased as the temperature in-
creased. This decrease was thought to be due to a greater
accumulation of intracellular ethanol at higher temper-
atures, which would produce cell toxicity and alter the
structure of the membrane, thus reducing its functionality.
The usual growth curve, with a series of short-lag,
exponential, stationary, and decline phases, was observed,
but at this high, 40jC temperature, a large quantity of yeast
died earlier during the process. This high yeast mortality
may have induced fermentation to arrest with a high sug-
ar content.
Cytometric Results
The median fluorescence index (M) profile was quite
similar to that found during normal fermentations up to
about 15 g.L�1 ethanol, with a slight shift towards low
values which could be explained by one (or more) hy-
pothetical phenomena: a lower intracellular pH, a higher
probe efflux or fluorescence quenching resulting from a
higher intracellular probe concentration, because the
volume of stressed yeasts was lower (data not shown).
Because protection against extreme conditions such as
heat and high ethanol concentrations is dependent to a
large extent on a high membrane sterol content (because of
its protective function in modulating the fluidity of the
membrane phospholipid ‘‘bulk membrane function’’, and
also its role in the initiation of the cell cycle, or ‘‘sparking
function’’), it could be suggested that the median value
was low because of the adaptative increase of yeast sterols
inhibiting esterase activity (Jespersen and Jakobsen, 1994).
Once median index values had spiked, they were not
followed by the increase that would have been expected
because of a loss of viability in total cells (no cells capable
of recovery). Thus, in the case of arrested fermentations
due to a high temperature, the reasonable assumptions
which can be made are that more and more cells may have
severely damaged membranes, there is low esterase ac-
tivity, and low pHi. The pHi is very important, principally
because the fluorescence of fluorescein and cF is high-
ly pH-dependent, and also because pHi affects en-
zyme activity.
The width index (W) profile exhibited almost gaussian
homogeneity throughout fermentation. This was probably
due to the combined effect of yeast aging and increased
ethanol production, which appeared earlier in this case
because of the synergistic effect of high temperature
exposure (Piper, 1995) that led to more sustained hetero-
geneity than in the normal fermentations.
The dissymmetry index (D) also demonstrated an almost
normal profile up to an ethanol concentration of 15 g.L�1,
after which the histograms became skewed to the right.
Thus, fermentation arrest was consistent with loss of vi-
526 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 86, NO. 5, JUNE 5, 2004
ability: fewer and less viable cells with greater hetero-
geneity and a lower level of fluorescence.
The primary aim of this work was to demonstrate that
hitherto unused cytometric histogram descriptors related to
simple physiological parameters such as the median
fluorescence intensity, heterogeneity, and dissymmetry
of viable cells might characterize dynamic changes to
yeast vitality during normal fermentation. Differences
were observed throughout the process in all descriptors,
revealing that the fermentation stage, which was linked to a
particular physiological state of viable cells, could be
associated with range values for each descriptor. In some
cases, however, the separate consideration of descriptors
independently did not enable the differentiation of phy-
siological status, because descriptor values could be
common to quite different stages of fermentation.
The second aim of this work was to use cytometric
histogram descriptors to detect any loss of viability and
thereby to distinguish, early in the process, between normal
and arrested fermentations. For most descriptors, differ-
ences in trends between normal and arrested fermentations
occurred throughout the process. In some cases, however,
the separate consideration of descriptors would not have
been sufficient because some descriptor values were
common to both arrested and normal fermentations, es-
pecially after ‘‘yeast pitching’’ with respect to median
fluorescence and dissymmetry indices and between 15 and
25 g.L�1 ethanol for the width index.
For the above reasons, a combination of descriptors was
the onlymeans of collating all data on viable cell distribution
and associating each histogram shape to a particular
physiological state, and thus anticipating the satisfactory or
poor progress of fermentation.
Early Discrimination Between Normaland Arrested Fermentations
Fermentations were divided into six phases based on
ethanol scale values. For each phase the means of des-
criptors were calculated taking into account all quantitative
values associated with ethanol concentrations within this
range. The means and their associated confidence
intervals (95%) are shown in Figure 5 for both normal
and arrested fermentations. For each histogram descrip-
tor, quantitative values were associated with one of three
possible qualitative indexes defined with respect to the
corresponding profile, with a view to permitting further
discrimination between normal and arrested fermentations
(Table I). The cutoffs were chosen visually after the
descriptor profiles had been established. Typical combi-
Figure 5. Discrimination between normal (n) and aborted (5)
fermentations for six ethanol concentration ranges. Mean and confidence
interval (95%) histograms for the three descriptors: (a) median index, (b)
width index, (c) dissymmetry index.
Table I. Classes of descriptors.
1 2 3
Median (M) M1 if < 1 M2 if between
1 and 1.5
M3 if > 1.5
Width (W) W1 if < 0.6 W2 if between
0.6 and 0.8
W3 if > 0.8
Dissymmetry (D) D1 if < 1 D2 if between
1 and 2
D3 if > 2
Quantitative values were associated with one of three possible
qualitative indexes. For instance, a width quantitative value of 0.85 would
be associated with qualitative index W3.
BOUCHEZ ET AL.: CYTOMETRIC DISTRIBUTION OF FLUORESCENT YEASTS 527
nations of normal and arrested fermentations are shown
in Table II.
Although within the 0–5 g.L�1 ethanol concentration
range typical combinations could be sketched for the two
fermentations, if the normal and arrested mean standard
deviations for each descriptor were taken independently,
they overlapped, not allowing accurate discrimination. As
from 5 g.L�1 ethanol, some combinations began to provide
very specific information on likely fermentation behavior.
If the width index (W) alone was considered at between
5–10 g.L�1 ethanol, it was possible to forecast the outcome
of fermentation: the thinner the histogram, the more likely it
was that fermentation would be completed. Inversely, viable
cell heterogeneity appeared as a result of cell aging and also
the combined effect of a rising ethanol concentration and a
high temperature. At lower temperatures (normal condi-
tions), cell division and ethanol effects appeared later
because of longer latency and lower ethanol toxicity.
It is likely that other, more subtle differences also
pertain, giving rise to differential staining, depending on
the cell cycle. All cells in a culture are not identical; in a
batch culture, each individual had its own characteristics
and in some sense may be unique. All stages of the growth
and division cycles are represented. It is therefore perhaps
readily understandable that the mean cell age of a yeast
population may be an important determinant of the
differences in the width index between fermentations
during the early stage of growth.
As from 10 g.L�1 ethanol, the width index was less and
less discriminating because the cell heterogeneities of
normal and arrested fermentations no longer differed
significantly. However, the median and dissymmetric
indices were of particular interest: during normal fermen-
tations, less fluorescent subpopulations began to merge
from overall viable cells (the mean D value increased),
whereas during arrested processes the histograms were
skewed to the right and the median value fell.
In addition, the combinations associated with arrested
fermentations were never those encountered at the
completion of normal fermentation. As from 30 g.L�1
ethanol (when the specific rate of ethanol production was
low or decreasing), viability losses in these two cases were
related to two distinct mechanisms, resulting in separate
physiological responses.
CONCLUSIONS
Two different approaches were applied to making an early
distinction between normal and arrested fermentations.
Under the first approach, cytometric histograms could be
recorded at random time points during the process. The
combinations of descriptors resulting from this experi-
mental distribution could be compared with a theoretical
database and suggest a probability of whether the fer-
mentation would be normal or not. Further, other histo-
grams could easily be recorded to confirm or invalidate the
first assessment.
Second, cytometric histograms could also be recorded
at key stages of the process, within a known ethanol
concentration range. It was therefore possible to determine
within which minimum ethanol concentration range a
distinction could be made between normal and ‘‘heat-
arrested’’ fermentations. As soon as a level of 5 g.L�1
ethanol was attained, it was possible to forecast whether a
fermentation was likely to abort, because at this concen-
tration the viable population was more heterogeneous,
according to its highest width index value. As from
15 g.L�1 ethanol, the median and dissymmetry indices were
the only useful descriptors for discrimination.
These interesting, reproducible differences in descriptor
profiles and combinations highlight the complex nature of
yeast physiology and the changes which may occur during
normal anaerobic and arrested fermentations. However,
even though flow cytometry may reveal some subpopula-
tions which merge according to the shape of the histogram
and thus provide a more precise picture of the overall
fluorescence response of the yeasts under test, the questions
raised by the physiological significance of descriptors
cannot be entirely answered because some necessary, ac-
companying biological measurements were not made at the
time of the experiment.
The biological interpretation of the cytometric variability
observed, that mainly relied on esterase activity and
membrane integrity, should be verified for different strains
because their enzyme pool and ethanol tolerance may vary.
Taking account of the fact that our study was carried out in
a synthetic medium using a selected Champagne strain, the
median profiles for normal fermentations were quite similar
to those obtained by Jespersen and Jakobsen (1994) for a
brewing strain fermenting wort, and also quite similar in
must with the same strain and the same synthetic medium
for another, confidential Champagne strain (unpubl. study).
We were therefore encouraged to think that the only
differences may occur in terms of quantitative index values,
and that the combinations of descriptors may be the same
Table II. Combination of descriptors for six ethanol concentration ranges
that permitted discrimination between normal and aborted fermentations.
Combination of descriptorsEthanol
concentration range Median Width Dissymmetry Fermentation
0–5 M2 W2 D2 Normal
M1 W3 D1 Aborted
5–10 M2 W1 D2 Normal
M2 W3 D1 Aborted
10–15 M2 W2 D2 Normal
M1 W3 D1 Aborted
15–20 M2 W3 D3 Normal
M1 W3 D1 Aborted
20–30 M2 W1 D2 Normal
M1 W3 D2 Aborted
>30 M3 W1 D2 Normal
(low activity)
528 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 86, NO. 5, JUNE 5, 2004
with respect to the cutoffs chosen. Nevertheless, our
purpose of presenting the applications described here was
to illustrate how the theoretical framework could be
applied. We did not present these results as definitive
proof that all arrested or sluggish fermentations lead to the
same shapes and patterns.
Ideally, a study of this nature should teach us how to
proceed in the reverse direction, i.e., beginning with the
histogram and using it to infer a unique type of fermentation
behavior. That goal still eludes us in many situations.
In this article, we have demonstrated the efficiency of
using combinations of cytometric histogram descriptors to
discriminate at an early stage between normal and arrested
fermentations. Because the alcoholic fermentation process
for Champagne wine is often well controlled, exposure to
high temperature, which is the easiest way to induce
fermentation arrest, is probably a theoretical rather than a
realistic cause of arrest. On the other hand, as producers are
mainly faced with problems of must composition, it was
encouraging to think, therefore, that the technique presented
here, enabling characterization of the inherent variability of
viable cell distributions, could provide a basis for the study
of cytometric histogram signatures for viable yeasts in the
case of slow and sluggish fermentations due to nitrogen
deficiency or the persistence of pesticide residues in the
must, because these two factors are known to hinder
satisfactory fermentation.
NOMENCLATURE
Y Ethanol concentration (g.L�1)X Total biomass concentration (g.L�1)i and j Channel numbers of cytometer in log scalejV Channel number of cytometer in linear scaleAi Number of events on channel ia Exponential fit coefficientaj Calculated number of events on channel j after exponential fitr Ethanol production specific rate
The authors thank Moet & Chandon Research Laboratories, the
Champagne Wine Trade Committee, and Europol’Agro for their
contributions to this work.
References
Attfield PV, Kletsas S, Veal DA, Van Rooijen R, Bell PJL. 2000. Use of
flow cytometry to monitor cell damage and predict fermentation
activity of dried yeasts. J Appl Microbiol 89:207–214.
Bagwell CB. 1979. Theory and application of DNA histogram analysis.