Rapid enumeration of Oenococcus oeni during malolactic fermentation by flow cytometry

Authors


Correspondence

Marielle Bouix, AgroParisTech, INRA, UMR 782 Génie et Microbiologie des Procédés Alimentaires, Thiverval-Grignon, France. E-mail: marielle.bouix@grignon.inra.fr

Abstract

Aims

The aim of this study was to provide a method to rapidly enumerate Oenococcus oeni cells during malolactic fermentation (MLF). To keep MLF under control, it is important to monitor the growth of the bacteria O. oeni. However, the enumeration of O. oeni using the plate count technique requires a very long incubation time of about 10 days or more, which is not adapted to monitoring MLF in real time.

Methods and Results

Flow cytometry (FCM), in combination with several fluorescent probes, is a rapid method for counting large numbers of bacterial cells. However, probes based on fluorescein [FDA, carboxyfluorescein diacetate (cFDA)] did not give good results for O. oeni. For the first time, we propose using the BacLight™ kit for enumeration of O. oeni, and we compare the results with three methods: plate count and FCM, in combination with either fluorescein probes or the BacLight™kit. The last method provides a perfect correlation with the plate count method.

Conclusions

FCM coupled with the Baclight™ kit makes it possible to count O. oeni cells during MLF with a perfect correlation with the plate count method.

Significance and Impact of the Study

The result is obtained in 20 min vs 10 days with the reference method which will be very useful for wine microbiologists. Moreover, it should be emphasized that FDA/cFDA staining is not recommended because it can lead to an erroneous count during the latency period or at the end of growth due to the variation of intracellular pH in the O. oeni cells during growth.

Introduction

Malolactic fermentation (MLF) takes place after alcoholic fermentation in wines, converting malic acid to lactic acid and CO2 through decarboxylation. As a consequence, the organoleptic quality and the microbiological stability of the wine are improved. MLF is mainly driven by the lactic acid bacteria Oenococcus oeni (O. oeni), which is able to decarboxylate malic acid under unfavourable conditions such as low pH (Drici-Cachon et al. 1996; Lonvaud-Funel 1999). In practice, it is sometimes difficult to keep MLF in wines under control because of our lack of knowledge about the cell physiology of the bacteria O. Oeni. One requirement of growth physiology is viability. In concept, bacterial viability is defined by membrane integrity, protein synthesis and the energy production necessary to maintain cells (Breeuwer and Abee 2000b). In practice, bacterial viability is measured using the plate count technique and is assimilated to culturability. However, in the case of O. oeni, the plate count technique requires a very long incubation time of about 10 days or more, which is not adapted to monitoring MLF in real time.

Flow cytometry (FCM) is a rapid method for cell-by-cell analysis, in combination with several fluorescent probes. It is used to count viable bacteria (Jepras et al. 1995) and yeasts (Bouix et al. 1999). Esterase substrate such as fluorescein diacetate (FDA) or carboxyfluorescein diacetate (cFDA) are widely used for viability assessment of various yeasts and bacteria (Diaper et al. 1992; Diaper and Edwards 1994; Breeuwer and Abee 2000b; Veal et al. 2000; Bunthof et al. 2001; Ben Amor et al. 2002). These probes are nonfluorescent precursors that are taken up by the cell and converted by nonspecific esterase into a fluorescent compound. The labelling indicates enzymatic activity and membrane integrity (Bunthof et al. 2000). Cells that have taken up fluorescein appear with green fluorescence after excitation at 488 nm. Propidium iodide (PI) is a nucleic acid dye that is excluded from viable cells by intact membranes but that can enter dead cells with altered membranes and bind to the nucleic acids (Ben Amor et al. 2002). The dual-staining method with these two probes makes it possible to assess physiological states because both dyes are detected with fluorescence at different wavelengths after excitation at 488 nm: living cells are stained with cFDA (green fluorescence), whereas dead ones are labelled with PI (red fluorescence). A third possibility also exists: the detection of injured or damaged cells that are stained with both of the dyes. This method has been used to assess the viability of a great number of lactic acid bacteria including acid-stressed Streptococcus macedonicus (Papadimitriou et al. 2006), freeze-stressed Lactobacillus delbruekii (Rault et al. 2007), Lactococcus lactis (El Arbi et al. 2011), as well as O. oeni. However, (Malacrinò et al. 2001) reported that the fluorescence intensity varied considerably for this species, depending on the dye and the strain. Four strains of 10 of O. oeni were stained after 10 min, but 60 min were necessary for three of the 10 strains to stain viable cells, whereas three of the strains were never labelled.

Graça da Silveira et al. (2002) used the combination of the two fluorescent probes, cFDA and PI, in combination with FCM, to discriminate O. oeni ethanol-stressed cells, but not for enumeration.

Another method using the live-dead bacterial viability kit (BacLight™; Invitrogen, Saint Aubin, France) was applied to estimate both the number of viable bacteria and the total bacterial count. BacLight™ is composed of two nucleic acid-binding stains, Syto9 that enters all of the bacterial cells and PI. With this kit, green fluorescent cells are live cells, and red and green dual-stained cells are dead cells. Initially developed to count viable bacteria as well as the total number of bacteria in drinking water by epifluorescence (Boulos et al. 1999; Defives et al. 1999) and in acetic fermentation (Mesa et al. 2003; Baena-Ruano et al. 2006; Fernandez-Perez et al. 2010), the BacLight™ kit is also used in combination with FCM on probiotic and other bacteria (Maukonen et al. 2006; Berney et al. 2007; Kramer et al. 2009), but its potential use in the oenological industry has not yet been tested. The aim of this study was to provide a rapid enumeration method for O. oeni during MLF by using fluorescent probes and FCM. We compare the potential of cFDA staining (or cFDA/PI dual-staining) and BacLight™ staining in combination with FCM to rapidly monitor the growth of O. oeni throughout MLF, with the plate count technique. We then discuss the results to propose a rapid and simple method for counting O. oeni during MLF.

Materials and methods

Strains and cultures

Three strains of O. oeni were used. MC1 was isolated from wine by Moet et Chandon, and was stored at −80°C. S1 and S2 were freeze-dried preparations marketed by Lallemand SA. All the cultures were performed in synthetic medium FT 80 broth (Cavin et al. 1989), adjusted to pH 3·2, and incubated at 30°C. Plate count was performed in FT 80 agar (FT 80 broth, 15 g l−1 agar). All the precultures were performed in the FT 80 broth at pH 5 for 96 h. The culture in synthetic medium was inoculated at a concentration corresponding approximately to 106 cell per ml (optical density at 600 nm of 0·025).

Bacterial growth was monitored using two methods. The enumeration of culturable cells was performed by plating appropriate serial dilutions on FT 80 agar. The count was obtained as the number of colony-forming units after incubation for 10 days at 30°C. The enumeration of viable and dead cells was performed by FCM on the basis of two types of fluorescence staining. All the measurements were performed once a day during MLF.

Viability staining

cFDA staining

The stain used was chemchrome V8 (AES-Chemunex, Ivry sur Seine, France), which contains carboxyfluorecein diacetate (cFDA). One millilitre of culture containing approx. 10cells was centrifuged in an MSE Micro Centaur SANYO centrifuge (13 000 g for 1 min). The pellet was resuspended in 1 ml of McIlvaine's buffer (citric acid: 0·1 mol l−1; disodium hydrogenophosphate: 0·2 mol l−1), at pH 7·3, and incubated for 10 min at 40°C in the presence of 10 μl of Chemchrome V8. The cFDA-stained sample was centrifuged, and the pellet was resuspended in 1 mol of McIlvaine's buffer at pH 3·2 (pH of the culture) before FCM analysis.

cFDA and PI double staining

One millilitre of culture containing approx. 106 cells was centrifuged (13 000 g, 1 min), and the pellet was suspended in 1 ml of McIlvaine's buffer, pH 7·3, and incubated for 10 min at 40°C in the presence of 10 μl of PI (1 mg ml−1 PI in water), and 10 μl of Chemchrome V8. The double-stained sample was centrifuged (13 000 g, 1 min), and the pellet was suspended in McIlvaine's buffer at pH 3·2 before FCM analysis.

Live-dead BacLight viability kit (molecular probe; Invitrogen) staining

One millilitre of culture containing approx. 106 cells was centrifuged (13 000 g, 1 min), and the pellet was suspended in 1 ml of McIlvaine's buffer, pH 7·3, and incubated for 15 min at room temperature in the dark in the presence of 1·5 μl of Syto9 and 1·5 μl of PI (solutions ready to use in the kit). The double-stained sample was centrifuged (13 000 g, 1 min), and the pellet was suspended in McIlvaine's buffer at pH 3·2 before FCM analysis.

FCM analysis

FCM analyses were performed with a CyFlow SL cytometer (PARTEC, Sainte Geneviève des Bois, France). The cytometer was equipped with a solid blue laser, emitting at 488 nm, and four band-pass filters: a forward-angle light scatter (FSC) combined with a diode collector, a side-angle light scatter (SSC) and two fluorescence signals collected with photomultiplier tubes, a 530-nm band-pass filter (515–545) to collect green fluorescence (FL1 channel), and a 630-nm long-pass filter to collect the red fluorescence (FL2 channel). The FCM analyses were performed using logarithmic gains and specific detector settings, adjusted on a sample of unstained cells, to eliminate cellular autofluorescence. Gating on FSC/SSC was used to discriminate bacteria from the background.

Data were collected and analysed with FlowMax software (PARTEC). The subpopulations were identified using dot plots. Gates were defined in the dot plots of FSC, SSC, green fluorescence and red fluorescence, thus allowing the software to separate the different events. Data were analysed using statistical tables that indicate numbers and percentages of stained cells determined by each detector, as well as the fluorescence intensity of each fluorescent signal.

Microscopic analysis

Labelled fluorescent cells were analysed with an Axiostar + epifluorescence microscope (CARL ZEISS, Le Pecq, France) equipped with a 50-watt mercury lamp. A JAI M70 camera (Imazys, Suresnes, France) linked to the ITEX software (Imazys) was used to take microphotographs (magnification ×1000).

Intracellular pH (pHin) measurement

To measure the pHin and establish the calibration curve, it is necessary to use two probes to cover the range of pH. Dicarboxy-dichloro fluorescein (CDCF; Invitrogen) was used for pH values ranging from 3 to 4·5 (Nedergaard et al. 1990), and carboxyfluorescein diacetate succimidyl ester (cFDA SE; Invitrogen) was used for pH values ranging from 4 to 7 (Breeuwer and Abee 2000a).

The staining protocol was the same as for viability staining, but Chemchrome V8 was replaced by 5 μl of CDCF (8·6 mmol l−1 in acetone) or 5 μl of (cFDA SE) (8·96 mmol l−1 in acetone) (molecular probes; Invitrogen). Cells containing CDCF or cFDA SE were centrifuged, and the pellet was suspended in 3 ml of McIlvaine's buffer at pH 3·2 before spectrofluorimetry analysis. Calibration curves were determined for each cell sample. After staining, cells were suspended in McIlvaine's buffer with pH values ranging from 3 to 7. Five microlitre of nigericin (0·2 mmol l−1 in ethanol) and 5 μl of valinomycin (0·2 mmol l−1 in ethanol) were added to equilibrate the pHin and the pH medium (pHex) before spectrofluorimetry analysis.

Spectrofluorimetry analysis

Fluorescence analyses were performed with a spectrofluorimeter (SHIMADZU RF 1501, Champs sur Marne, France). Fluorescence intensities were measured at excitation wavelengths of 490 and 440 nm, and the emission wavelength was 525 nm. The pHin was determined from the ratio of the 490/440 wavelength fluorescence intensities (Siegumfeldt et al. 1999).

Results

Growth curve, viability and culturability assessment

Growth curves established by plate count and FCM after cFDA staining are shown in Fig. 1.

Figure 1.

Comparison between Oenococcus oeni counts in FT 80 broth, pH 3·2, obtained by FCM after cFDA staining (● cFDA-stained cells) and plate count methods (Δ: CFU per ml). (a): MC1; (b) S1; (c) S2. For the three strains, the counts by culture and CMF were correlated only at the end of the exponential phase and the beginning of the stationary phase, but not at the beginning of the culture and in the stationary phase. cFDA, carboxyfluorescein diacetate; FCM, flow cytometry.

For the MC1 strain (Fig. 1a), the latency period was 50 h, and the stationary phase began after 168 h of culture. The final culturable cell concentration was 3 × 107 cells per ml. In the stationary phase, the FCM count dropped to 6 × 105 cells per ml after 240 h of culture, whereas the number of culturable cells remained constant. The number of cFDA-stained cells counted by FCM was 50-fold lower than the CFU on FT 80 agar. The counts by culture and CMF were then correlated only at the end of the exponential phase and the beginning of the stationary phase, but not at the beginning of the culture and in the stationary phase. For the S1 and S2 strains, we observed the same results (Fig. 1b,c). At the beginning of the culture and at the end of the stationary phase, O. oeni cells appeared to be culturable but nonviable cells. The observation of the dual parameter dot blots cF fluorescence vs PI fluorescence (double staining with cFDA and PI) (Fig. 2a) confirmed that the cells were alive and not dead (no population in the upper left quadrant Q1). Therefore, an increasing number of unstained cells (Q3 quadrant) appeared at the end of the stationary phase, whatever these cells were culturable. As a consequence, the count of viable cells of O. oeni by FCM was under-evaluated. The microscopic observations of the O. oeni cells confirmed the bright fluorescence of cells in the exponential phase and the lack of fluorescence of cells at the end of the stationary phase (Fig. 2b).

Figure 2.

Cytograms of Oenococcus oeni after cFDA-PI dual-staining (a) and microphotograph (b). Culture in FT 80 broth, pH 3·2. The dual parameter dot blots cF fluorescence (FL1) vs PI fluorescence (FL2) confirmed that the cells were alive and not dead (no population in the upper left quadrant Q1). Unstained cells (Q3 quadrant) appeared at the beginning and at the end of the MLF. cFDA, carboxyfluorescein diacetate; MLF, malolactic fermentation; PI, propidium iodide.

With the live-dead bacterial viability BacLight™ kit, the FCM counts were well correlated with the plate counts throughout the culture for the three strains (Fig. 3). The R2 correlation was approx. 0·995.

Figure 3.

Comparison between Oenococcus oeni counts in FT 80 broth, pH 3·2, obtained by FCM after BacLight™ staining (■ BacLight™stained cells) and plate count methods (Δ: CFU per ml). (a) MC1; (b) S1; (c) S2. For the three strains, the counts by culture and CMF were well correlated; FCM, flow cytometry.

We can therefore conclude that the enumeration of O. oeni can be performed by FCM by using the live-dead bacterial viability BacLight™ kit. On the contrary, cFDA staining under-evaluated the bacterial concentration of culturable cells, except at the end of the exponential growth phase, and this phenomenon was observed for all of the strains tested.

Esterase activity, fluorescence intensity and pHin measurement

The fluorescence intensity of cells depends on two parameters: the esterase activity in cells, which cleaves cFDA in cF and the pHin of the cells because the fluorescence intensity of carboxyfluorescein depends on pH. Fluorescein has a pKa of 6·4, and its ionization equilibrium leads to pH-dependent absorption and emission over the range of 5–9. As a result, the lack of fluorescence in O. oeni cells may be due to the lack of esterase activity in the young or old cells, or to a low pHin.

We measured the esterase activity with a method adapted from (Bardi et al. 1993) and found that it remained constant (about 3 × 10−4 nmoles min−1 mg−1 of proteins) in the cells throughout MLF (data not shown).

We measured the pHin values of MC1 at two different times during culture. At the end of the exponential phase of the FML (144 h), the pHin of the MC1 cells was 6·1 and then decreased to 3·2 at the end of the stationary phase (240 h).

Discussion

We examined the usefulness of FCM for viability assessment of O. oeni with cFDA staining and dual Syto9/PI (BacLight) staining. To be useful, this FCM method must be as reliable as the plate count technique. The results showed that cFDA was successful in labelling the live cells only at the end of the exponential growth phase; at the beginning and during the stationary growth phase, the cells were culturable but not fluorescent enough to be taken into account by the flow cytometer. Bunthof et al. (2001) tested cFDA as a live-cell stain for lactic acid bacteria and concluded that the labelling with cF gave a clear discrimination between live and dead cells. Our results indicate that the assumption that live cells are fluorescent cells is not always true. In fact, cFDA is an esterase substrate that requires enzyme activity to be cleaved in fluorescein, and the fluorescence of the fluorescein is pH-dependent, implying that fluorescence intensity depends on bacterial cell pHin.

In our results, the lack of fluorescence could be explained by two hypotheses, the lack of cellular esterase to cleave the cFDA or a low intracellular pH. We verified that the esterase activity of O. oeni cells was equal in both the exponential phase and the stationary phase. This activity did not significantly vary during growth, regardless of the pH of the medium. The same value was found with Leuconostoc mesenteroides cells, used as a reference, and which were fluorescent after cFDA loading under the same conditions (data not shown). The lack of fluorescence of O. oeni cells in the stationary phase was not due to a loss of esterase activity. The low pHin (3·2) explains the lack of fluorescence in the O. oeni cells at the end of the culture because the fluorescence of the fluorescein was not detectable at pH 3·2, even though the same concentration of fluorescein was strongly fluorescent at pH 6. This difference in pHin during growth could perhaps account for the variability of staining in the study of Malacrinò et al. (2001).

These results showed that bacterial cells can be viable and culturable but not fluorescent with cFDA staining. This is the case of O. oeni, due to its low pHin at some phase of the culture. On the contrary, the staining principle is not the same with Syto9/PI, and the results were well correlated with the plate count method. All our results were obtained in FT80 synthetic medium for the settling of the method. We have verified that the same phenomenon of nonstained cells after cFDA staining was also observed in white wine at the end of the MLF. Now, the BacLight method must be validated in real conditions in wine and in presence of yeasts.

Conclusion

The FCM method coupled with Syto9/PI allows a rapid enumeration of O. oeni cells and represents a useful tool to more effectively monitor MLF because the result may be obtained in 20 min vs 10 days with the plate count method. FDA/cFDA staining is not recommended because it can provide an erroneous count during the latency period or at the end of growth due to the particular behaviour of the O. oeni species. The FCM syto9/PI count method must be validated in oenological conditions.

Acknowledgements

We thank Moet et Chandon and Lallemand SA for providing us with their O. oeni strains.

Ancillary