Mixed bacterial communities commonly occur in microbial infections in humans. Alterations in composition of species and interspecies effects contribute to the complexity and severity of infectious diseases. Accordingly, there is a need for further, more detailed investigation. The analysis of terminal restriction fragment length polymorphisms (T-RFLP) has been demonstrated in numerous studies to be an effective tool to analyze the composition of mixed bacterial communities (1–6). Using capillary gel electrophoresis, multiple species can be separated from each other based on the length of previously labeled species-specific ribosomal DNA (rDNA) fragments, which are usually detected by laser-induced fluorescence. According to Trotha et al. (7), T-RFLP analysis can be applied for quantitative characterization of mixed communities through introduction of an internal quantification standard, i.e., an rDNA fragment from a known species with a defined cell concentration. Schmidt et al. (8) adapted this quantitative T-RFLP (qT-RFLP) method for the species-specific enumeration of a three-species mixed model community relevant to cystic fibrosis (CF) comprising strains of Pseudomonas aeruginosa, Staphylococcus aureus, and Burkholderia cepacia. This was implemented to characterize the growth of a defined mixed culture in chemostat cultivations. Furthermore, these experimental data were used in combination with a mathematical chemostat model to reveal interspecies effects (9). Recently, the qT-RFLP method was additionally used to study the efficacy of a cephalosporin-like antibiotic onto the growth of the described CF-relevant three-species mixed culture in batch cultures (10).
While these studies enabled quantitative measurement of species-specific growth dynamics, information concerning species-specific viability of respective CF-relevant bacteria was not obtained. However, knowledge concerning the viability of each species within a mixed culture may allow for a more detailed description of the growth dynamics and a better understanding of interspecies effects. Moreover, viability measurements are of high interest regarding susceptibility testing of antibiotics. qT-RFLP analysis cannot provide such data since DNA fragments are derived from total cell population. In contrast, flow cytometry provides a powerful technique to assess the viability of bacteria on a single cell level using fluorescent probes, which target or indicate specific cell functions such as respiratory activity, enzyme activity, substrate uptake, efflux pump activity, membrane potential, or membrane integrity (11). The viability of a bacterial cell is determined by different cellular processes, which can be related according to Nebe-von-Caron et al. (12) to reproductive growth, metabolic activity, and membrane integrity. In this study, viability was determined by membrane integrity analysis using propidium iodide (PI) as dead cell marker in combination with SYBR® Green I, also known as nucleic acid double-staining (NADS) protocol (13). It is assumed that the loss of cytoplasmic membrane integrity generally reflects the absence of reproductive growth and metabolic activity and leads irreversibly to cell death (14). Nevertheless, membrane integrity can also partially be lost during growth due to a short perforation of the cell wall during cell division and cell wall synthesis (15). Cells with a permeabilized membrane are stained by PI that binds to DNA and RNA, whereas PI is excluded from cells with an intact membrane due to its positive charge (16). In contrast, SYBR Green I penetrates all cells and binds selectively to double-stranded (ds) DNA (17). Therefore, viable cells exhibiting an intact membrane can accordingly be discriminated from dead cells with a permeabilized membrane. Considering the fact that the viability criterion only relies on membrane integrity, the detected viable cells are assumed to cover metabolically active and inactive cells as well as culturable and nonculturable cells with intact cytoplasmic membranes.
The NADS protocol for flow cytometry is widely used for viability determination of mixed bacterial communities prevalent in environmental samples, i.e., in waste water and activated sludge from waste water treatment plants (18–20), in sea water (21, 22), and from fresh water (23, 24). Surprisingly, in none of these studies, the NADS protocol was used in combination with taxa- or species-specific fluorescent labeling techniques to resolve the viability of individual species in mixed communities. In general, only few reports of viability studies dealing with species discrimination are available (12, 25, 26).
The aim of this study was to adapt and evaluate the flow cytometric NADS protocol for viability assessment of growth of Staphylococcus aureus and Burkholderia cepacia in mixed culture. To discriminate between two species, the method was additionally combined with Gram-specific fluorescent staining using wheat germ agglutinin (WGA). To our knowledge, this is the first report presenting an efficient flow cytometric method, which simultaneously allows for Gram-differentiation and viability assessment of mixed bacterial cultures.
Material and Methods
S. aureus ATCC 29213 and B. cepacia DSM 7288 were used in this study.
For cultivation of bacterial cells, Gibco® cell culture basal medium powder M199 (without NaCO3) (Life Technologies, Carlsbad, CA) was used as a chemically defined but rich medium without addition of complex components buffered with phosphate. To prevent precipitation, nitrilotriacetic acid was added additionally. Therefore, 800 mL of ultrapure water, 16 mL of 0.25 mM nitrilotriacetic acid solution (Sigma-Aldrich, Steinheim, Germany) in 0.6 M NaOH, and 25 mL of sodium potassium phosphate buffer (1.5 M NaH2PO4/K2HPO4, pH 7.0; Carl Roth, Karlsruhe, Germany) were mixed with the amount of powder indicated by the supplier and made up to 1 L with ultrapure water (8). Bacterial cells were grown in shake flask cultures in 20 mL culture medium in 250-mL wide-neck Erlenmeyer flasks in a humidified orbital shake incubator (Kuehner, Birsfelden, Switzerland) at 37°C, rotation speed 200 rpm, eccentric radius 1.25 cm, and relative humidity of 85%.
For optimization of viability staining, pure cultures were prepared. Briefly, each species was grown overnight. Cells were harvested, centrifuged at 3,522 g for 10 min at 4°C (Heraeus® Multifuge 1S-R; Thermo Scientific, Waltham, WA), and washed with PBS (8 g/L NaCl, 0.2 g/L KCl, 1.15 g/L NaH2PO4, 0.2 g/L K2HPO4; Carl Roth). Subsequently, the suspension was centrifuged again (3,522 g, 10 min, 4°C), and the cells were resuspended in 20 mL of fresh culture medium and cultivated for 5 h. Samples were taken from exponential and stationary growth phases after 2 or 5 h, respectively.
For viability assessment in mixed cultures, cultures were prepared using pure cultures of each species as inoculums as described above. In contrast, pure cultures were grown for 1.5 h to exponential growth phase. Cells were harvested and mixed in 50 mL prewarmed fresh culture medium for mixed culture cultivations. The inoculation volume was adjusted for each species to an initial concentration of 0.5 × 107 cells/mL based on optical density (OD650) measurements (OD at 650 nm; Photometer Ultraspec 3000; Amersham Biosciences, Otelfingen, Switzerland). Cultures were then cultivated for 32 h. Additionally, pure culture cultivations of each species were prepared to serve as controls in the same manner, except for inoculations with one species, in which a twofold higher inoculation volume was used. For both pure and mixed cultures, the initial total cell concentration was adjusted to 1×107 cells/mL. All cultivations were performed in three biological replicates.
Sample Preparation Before Staining
Harvested cells were centrifuged at 4,700 g for 10 min at 4°C (Heraeus Fresco; Thermo Scientific) and resuspended in sodium chloride (NaCl) phosphate buffer. This buffer was prepared according to conditions described by Müller et al. (27) (0.4 M Na2HPO4/NaH2PO4, 0.15 M NaCl, pH 7.2; Carl Roth). For optimization of viability staining, the OD650 of the samples was adjusted to 0.03 prior to staining by dilution with NaCl phosphate buffer to obtain defined cell concentrations for each species in the range of 107-108 cells/mL. Samples from 32 h cultures were diluted if necessary to an OD650 in the range of 0.01 to 0.04.
NaCl phosphate buffer was supplemented with 0.05 mg/mL glutaraldehyde (GTA) (50% grade I; Sigma-Aldrich), except for buffer in optimization of staining of S. aureus in pure culture.
As a positive control for membrane permeabilization, cell suspensions were treated with 70% (vol/vol) isopropanol (Merck, Darmstadt, Germany) for 1 h at room temperature (RT) and subsequently washed with NaCl phosphate buffer prior to adjustment of OD650.
Viability Staining Procedure
Cell suspensions were stained with SYBR Green I (Life Technologies, Carlsbad, CA) and PI (Sigma-Aldrich). Both dyes were simultaneously added to suspensions and incubated for 20 min in the dark at RT. For staining, a 1-mg/mL working solution of PI was prepared in ultrapure water and stored at 4°C prior to use. SYBR Green I was applied either directly from commercial stock solution after thawing (10,000 × concentrate in DMSO, stored at −20°C) or after dilution in ultrapure water. B. cepacia cells had to be treated with 0.05 mg/mL GTA for satisfying SYBR Green I staining. The applied GTA concentration did not impair membrane integrity, which was tested beforehand by PI staining experiments (Fig. 1).
To optimize staining procedure of each species, different SYBR Green I dilutions and PI concentrations were tested in single- or double-staining experiments. Positive controls were stained with SYBR Green I and PI according to tested and applied concentrations.
Staining of S. aureus Using WGA
Prior to labeling, a working solution of 1.5 mg/mL WGA (Vector Laboratories, Burlingame, CA) was prepared in ultrapure water and stored at 4°C. WGA was fluorescently labeled using Mix-n-Stain™ CF405S antibody labeling kit (Biotium, Hayward, CA). Briefly, 75 μg of WGA were labeled according to the labeling procedure recommended by the supplier. After labeling, the WGA solution was transferred to the provided storage buffer resulting in a working solution of WGA-CF405S with a concentration of 0.18 mg/mL. After sample preparation (see above), cell suspensions of 100 μL were incubated with 20 μg/mL WGA-CF405S for 15 min in the dark at RT. Afterward, cells were centrifuged at 4,700 g for 10 min at 4°C (Heraeus Fresco; Thermo Scientific) and washed in NaCl phosphate buffer before analysis.
Three-Color Staining for Viability Assessment in Mixed Cultures
Cell suspensions were primarily stained with WGA-CF 405S as described above and subsequently incubated simultaneously with SYBR Green I (dilution 5 × 103) and PI (5 μg/mL) for 20 min in the dark at RT.
Flow cytometric analyses were carried out using a CyFlow® space flow cytometer (Partec, Münster, Germany) equipped with a 16-mW 375-nm UV diode laser and a 20-mW 488-nm argon solid state laser. Forward (FSC) and side scatter (SSC) was collected from 488 nm excitation. SSC was set as discriminator to reduce electronic background noise during analysis. WGA-CF405S fluorescence was excited by a 375-nm laser and collected through a 455/50-nm bandpass filter. SYBR Green I and PI fluorescence were excited by a 488-nm laser and collected through a 527/30-nm bandpass or a 630-nm long pass filter, respectively. All signals were amplified logarithmically (four decades). Sampling rate was adjusted to less than 1,000 particles/s. The total number of particles was set to 20,000 events. As sheath fluid, degassed ultrapure water was applied. Data were acquired with FloMax software (Version 2.70; Partec). Stained cell suspensions were immediately analyzed after dye incubation. If necessary, samples were diluted with sheath fluid prior to staining.
Flow Cytometric Data Analysis
Data analysis, gating, and compensation were performed using FlowJo software (Version 7.6.4; Tree Star, Ashland, OR). The geometric mean value of fluorescence intensities (MFI) was expressed as relative fluorescence units (RFU). SYBR Green I and PI fluorescence signals were compensated due to spectral overlapping. Therefore, single-stained controls of SYBR Green I and PI (isopropanol-treated cells) were processed in each experiment for each species in line with the applied dye concentration. Compensation was determined and applied for data analysis on a day-to-day basis for the respective species and the applied concentration.
All gates applied for population discrimination were set manually based on control samples. For viability determination, manually set gates were applied for each species based on SYBR Green I/PI fluorescence of positive controls in pure culture. Here, only fluorescence signals of dead cells were detected. Viability gates were defined separately and applied for each exponential and stationary growth phases.
Gates for species discrimination in three-color stained mixed culture samples were defined according to WGA-CF405S fluorescence signals of stained S. aureus cells and background signal of unstained B. cepacia cells each in pure culture.
Since WGA staining caused aggregation of S. aureus cells, a gate based on scatter signals was applied to consider only single cells for viability determination of S. aureus. This gate was defined by the FSC and SSC signal of SYBR Green I/PI stained S. aureus cells in pure culture, which predominately showed single cells.
Validation of Mixed Culture Staining Protocol
The mixed culture staining protocol using SYBR Green I (5 × 103) and PI (5 μg/mL) was validated for viability determination of each species in pure culture by flow cytometry. Linearity and homogeneity of variances (F-test; f1, f2 = 7; 95%) at the upper and lower working range were determined. Mixtures of viable and dead (isopropanol-treated) cells from stationary growth phase were prepared with defined ratios. Therefore, cell suspensions both from viable and dead cells were adjusted to a fixed cell concentration (OD650 = 0.03). Each ratio was prepared eightfold independent of each other. Subsequently, individual samples at each ratio were pooled to compensate for pipetting errors and stained with SYBR Green I and PI. Single replicates of each ratio and eight replicates from both upper and lower working ranges were analyzed. Frequencies of viable cells were determined taking into account all positive fluorescence events using the gates described earlier. Theoretical values of frequencies of viable cells were determined based on the defined ratio of viable to dead cells and the adjusted cell concentration.
Growth Monitoring by qT-RFLP Analysis
Growth of pure and mixed cultures was monitored over a period of 32 h by species-specific cell enumeration using qT-RFLP. Analysis was performed as described previously by Schmidt et al. (8). During cultivation, two samples were taken at each sampling point and analyzed in parallel. For quantification, aliquots of Campylobacter jejuni with a fixed cell concentration were used as the internal quantification standard. Applying a Genetic Analyzer ABI Prism 3100 Avant (Life Technologies, Carlsbad, CA), species-specific rDNA fragments were separated and quantified using laser-induced fluorescence. Absolute cell concentration for a species was determined by the ratio of detected peak area for the given species to the peak area of internal quantification standard (7, 8). Concentrations were declared as genome equivalents per milliliter. For better comparison of growth data between pure and mixed culture, absolute cell concentration was normalized according to the description of Riedele and Reichl (10) and expressed as the logarithm to the base two, which is equivalent to the number of cell doublings.
Viability Staining Optimization for Pure Culture Analysis
For the determination of viability, membrane integrity analysis using SYBR Green I and PI was applied. Since bacteria differ in cell wall structure and metabolic capabilities, it is most likely that the permeability of a fluorescent dye varies depending on the species of interest and its physiological state. Furthermore, the interaction of the dye with its intracellular biological target can also be influenced by species-specific mechanisms and the physiological state (28). Therefore, in this study, the fluorescence staining method was tested separately for each species in pure culture for different physiological states during growth, i.e. exponential and stationary phases. Different SYBR Green I dilutions were tested for staining to obtain optimal fluorescence signal intensities for discrimination of cells from electronic background noise. Fluorescence signal intensities were determined by MFI measurements. This correlates well with the quality of discrimination between stained cells and noise (data not shown). It was shown that the higher the MFI, the better is the discrimination between stained cells and noise. For S. aureus, all tested dilutions resulted in a clear discrimination of stained cells from noise (Fig. 2A). Maximum discrimination was observed for a series of SYBR Green I dilutions, ranging from 105 to 104. Results were valid and comparable for cells from both exponential and stationary growth phases. In contrast, B. cepacia cells had to be treated with 0.05 mg/mL GTA for satisfying SYBR Green I staining. The applied GTA concentration did not impair membrane integrity (Fig. 1). For SYBR Green I dilutions from 5 × 103 to 102, cells were clearly discriminated from noise (Fig. 2B). Maximum discrimination was obtained for cells from each growth phase for a SYBR Green I dilution of 2 × 102. Lower dilutions resulted in a significant decrease of the fluorescence signal. Interestingly, MFI for stained, exponentially growing B. cepacia cells was almost twofold higher than the MFI of stationary B. cepacia cells for dilutions ranging from 5 × 103 to 2 × 102. In general, appropriate staining of B. cepacia required much higher concentrations of SYBR Green I than was required for S. aureus. For further optimization experiments, a SYBR Green I dilution of 104 for S. aureus and 2 × 102 for B. cepacia was used.
Simultaneous staining of bacterial cells with SYBR Green I and PI resulted reproducibly in a characteristic viability pattern in plots of green fluorescence over red fluorescence for each species (Figs. 2C and 2D). For both species, dead cells with permeabilized membranes could clearly be discriminated from viable cells with an intact membrane. Viable cells showed intense green fluorescence and almost no red fluorescence, whereas dead cells demonstrated weak green fluorescence and intense red fluorescence. Cells treated with 70% (vol/vol) isopropanol served as positive controls and showed identical signals as compared to dead cells. For S. aureus, one additional subpopulation exhibiting intense SYBR Green I and PI fluorescence was detected. In the following, these events are referred to as cells with a slightly damaged membrane (damaged cells). To determine the optimal PI concentration for best discrimination between viable and dead cells, different PI concentrations were tested according to the approach of Barbesti et al. (29) in combination with a fixed SYBR Green I dilution (104 for S. aureus and 2 × 102 for B. cepacia). Therefore, cells from exponential and stationary growth phases were investigated. Here, the quality of discrimination was determined by calculating the ratio between MFI values of viable cells and dead cells for the SYBR Green I channel as well as the ratio between MFI values of dead and viable cells for the PI fluorescence channel. It was shown that a higher ratio led to better discrimination between the viable and dead cells in the corresponding fluorescence channel. For S. aureus, all tested PI concentrations in both channels resulted in a high discrimination between viable and dead cells for exponential (data not shown) and stationary growth phase (Fig. 2E). Additionally, damaged cells could clearly be discriminated from viable and dead cells. However, for B. cepacia, the discrimination was influenced strongly by increasing PI concentrations for cells from exponential (data not shown) and stationary growth phases (Fig. 2F), particularly in the PI channel. For both growth phases, the best discrimination between viable and dead cells was obtained for 5 μg/mL of PI. Considering the fact that the membrane of bacterial cells can become permeable by PI at higher concentrations (30), the viability of S. aureus and B. cepacia was determined. Here, the frequencies of viable, damaged, and dead S. aureus cells or frequencies of viable and dead B. cepacia cells were compared for all tested PI concentrations (data not shown). The viability of S. aureus remained relatively constant for all tested concentrations, whereas viability of B. cepacia, particularly of cells from the stationary growth phase, was sensitive to increasing PI concentrations. Here, the frequency of dead B. cepacia cells remained relatively low at about 2.7% at PI concentrations of 5 μg/mL and 10 μg/mL, but increased 1.5- to 3-fold at higher PI concentrations (data not shown). Consequently, for optimal viability staining, PI concentrations in the range of 5 to 20 μg/mL are recommended for S. aureus and a PI concentration of 5 μg/mL for B. cepacia.
Establishment of Mixed Culture Staining Protocol
To determine viability of S. aureus and B. cepacia in mixed culture, a staining protocol using SYBR Green I and PI was established as a compromise between optimum conditions determined before for pure cultures. As a starting point, a SYBR Green I dilution of 5 × 103 and a PI concentration of 5 μg/mL were chosen. Additionally, 0.05 mg/mL GTA was applied, since the outer membrane of B. cepacia had to be permeabilized for SYBR Green I staining. Feasibility of these conditions was tested for each species in pure culture for cells from exponential and stationary growth phases. For each species, staining resulted reproducibly in detection of characteristic viability subpopulations based on green and red fluorescence (Figs. 3A and 3B). For S. aureus, fluorescence signals showed no difference in signal intensity in comparison with the optimum staining protocol described for pure culture. In contrast, SYBR Green I fluorescence intensity of B. cepacia was much lower compared to optimum conditions for pure culture. However, discrimination between viable and dead cells of B. cepacia was better using the mixed culture staining protocol. Finally, the viability of S. aureus and B. cepacia was determined and compared to results from optimal staining (Figs. 3C and 3D). Neither the viability of S. aureus nor the viability of B. cepacia was significantly affected by the chosen staining conditions.
The mixed culture staining protocol was validated for each species to evaluate the capability of the method to discriminate accurately between viable and dead cells. Therefore, mixtures of viable and dead (isopropanol-treated) cells from stationary growth phase with defined ratios were analyzed. For each mixed sample, the frequency of viable cells was determined. Although variances of the eight replicates at the lower working range and the upper working range were rather low for each species, variances were inhomogeneous for both (F-test; f1, f2 = 7; 95%). Consequently, weighted linear regression was applied and showed a very good correlation with the frequency of theoretically viable cells for both species (Fig. 4). The relative residual standard deviation of the method was determined for S. aureus as 17.50% and for B. cepacia as 4.17%. Limit of detection and limit of quantification were 0.054 and 0.163 for S. aureus and 0.019 and 0.056 for B. cepacia, respectively. For all further analyses, this staining protocol was applied for viability determination of S. aureus and B. cepacia in pure and mixed cultures.
Gram-Specific Labeling of S. aureus
In order to discriminate S. aureus from B. cepacia in mixed culture, fluorescently labeled WGA WGA-CF405S was applied. Due to its specific binding capacity to the peptidoglycan layer of Gram-positive bacteria, it labels S. aureus.
A final concentration of 20 μg/mL WGA-CF405S was chosen to discriminate S. aureus cells from noise and unstained B. cepacia cells (Fig. 5I-A or 5II-A). The frequency of stained S. aureus cells was always higher than 97%, irrespective of the growth phase and the state of membrane integrity. For B. cepacia, no unspecific binding to WGA-CF405S was observed, as exemplary shown for cells of the exponential growth phase (Fig. 5 III-A). Finally, results of three-color staining using WGA-CF405S were in good agreement with results of SYBR Green I/PI staining alone (data not shown).
Viability Assessment in Pure and Mixed Culture
Viability of both S. aureus and B. cepacia was assessed during growth in mixed culture over a cultivation period of 32 h using the established three-color staining protocol. Sample pretreatment, staining, and analysis of samples from one sampling point (three biological replicates) can be completed within 1 h. To compare the viability between pure and mixed culture, viability was investigated for each species in parallel in pure culture. Additionally, the growth of each species was monitored by qT-RFLP. In pure culture, S. aureus doubled its cell concentration about 7.5 times during the first 8 h until the stationary phase was reached (μmax = 0.57 h−1) (Fig. 6A). In contrast, B. cepacia doubled its cell concentration only about six times (μmax = 0.50 h−1). In mixed culture, both species grew faster and the number of cell doublings increased for S. aureus and B. cepacia to about 10 (μmax = 0.84 h−1) and 8 (μmax = 0.80 h−1), respectively, until stationary phase was reached (Fig. 6B).
The viability of S. aureus in pure culture was initially detected to be about 95% (Fig. 6A). During the first 1 h of growth, the frequency of viable cells decreased to about 92%. After 2 h of growth, the frequency of viable cells increased again to about 98% and remained relatively constant until the stationary phase was achieved (10 h). Afterward, the frequency of viable cells dropped continuously to about 49%. The frequency of damaged cells correlated well with the time course of viable cells during the first 16 h of growth, whereas the frequency of dead cells was constant. After 16 h of growth, the frequency of dead cells increased significantly and correlated well with the time course of viable cells toward the end of cultivation. In comparison to pure culture growth, the viability of S. aureus in mixed culture did not change significantly during the first 7 h (Fig. 6B). Afterward, the time courses differed significantly, i.e., viability in mixed culture was always lower than in pure culture with a difference of about 37% at the end of cultivation (32 h). Correspondingly, the frequencies of damaged and dead cells increased faster in mixed culture with final values of 26% and 62%, respectively. Interestingly, during the first 16 h of growth in both pure and mixed culture, the loss of viability of S. aureus resulted in an increase in the frequency of damaged cells, whereas later, only the frequency of dead cells considerably increased. In the end of cultivation, the frequencies of dead cells were even higher than those of damaged cells (32 h).
Over the first 2 h of cultivation, the viability of B. cepacia in pure culture was as high as for S. aureus (Fig. 6A). During the exponential growth phase, the frequency of viable cells decreased to about 88% (3 h to 6 h) and increased again to remain constant at about 98% toward the end of cultivation. Accordingly, the frequency of dead cells correlated well with this growth pattern. Interestingly, and in contrast to S. aureus, the viability of B. cepacia in mixed culture was comparable to pure culture throughout the entire cultivation.
Different Optima for Viability Staining of Each Species
SYBR Green I and PI were applied for flow cytometric determination of viability of S. aureus and B. cepacia for membrane integrity analysis. Fluorescence staining was optimized separately for each species in pure culture with respect to different physiological states during growth, thereby considering differences between the bacteria of interest in cell wall structure and metabolic capabilities. Cells from exponential and stationary growth phases were tested at different dye concentrations. The MFI of SYBR Green I stained bacteria increased with dye concentration either until the fluorescence signal was saturated (S. aureus) or a maximum level was achieved (B. cepacia), suggesting maximal binding of this dye to nucleic acids. Due to relatively high dye concentrations, it can be assumed that binding of SYBR Green I is highly selective to ds DNA (17). Hence, a contribution of SYBR Green I bound to RNA or single-stranded DNA to overall fluorescence can be largely neglected. The decrease of MFI observed at high concentrations of dye as shown for B. cepacia might reflect an efflux of SYBR Green I from cells through active extrusion pumps. Another reason might be the quenching of emitted fluorescence by bound SYBR Green I molecules due to their close proximity on DNA strands. The clear difference in uptake of SYBR Green I observed between S. aureus and B. cepacia may be caused by differences in cell wall composition (15), particularly due to the outer membrane of B. cepacia, which acts as a barrier for penetration of hydrophobic dyes (31). Only staining with GTA in buffer resulted in satisfying fluorescence staining of B. cepacia. This is most likely due to the fact that GTA increases the permeability of the outer membrane to hydrophobic SYBR Green I by crosslinking outer membrane proteins through hydrophilic free amino groups (32, 33). Similar findings were reported by Morono et al. (34), who showed that fluorescence staining with hydrophobic carboxyfluorescein diacetate can be improved considerably by using GTA.
Although maximum MFIs were comparable for both species, the concentration of dye required for optimal staining of B. cepacia was much higher than that needed for S. aureus. This phenomenon may be explained by an efflux of SYBR Green I from B. cepacia. Another possible explanation would be differences in total DNA content and in base composition of DNA of both species (35) since SYBR Green I is known to preferentially bind adenine-thymine (AT) rich regions (17). Genomic data for each species obtained through an Integrated Microbial Genome (IMG) database search supports this hypothesis. B. cepacia (IMG Project ID Gc00309) exhibit a larger genome than S. aureus (IMG Project ID Gc01218). Yet, its DNA contains a comparatively low number of AT bases.
Remarkably, for B. cepacia, SYBR Green I staining was dependent on the physiological state of cells. Exponentially growing cells showed higher MFIs than stationary cells. This might be due to the high proliferation activity of exponentially growing cells exhibiting more than one genome in average, whereas stationary cells contain only the minimum number of genomes because of starvation (36).
For each species, simultaneous staining of tested bacteria with SYBR Green I and PI resulted reproducibly in a characteristic viability pattern. For both species, dead and viable cells were detected with permeabilized and intact cytoplasmic membranes, respectively. These subpopulations could clearly be discriminated from each other due to a fluorescence resonance energy transfer (FRET) from SYBR Green I to PI: If cells are double-stained, green fluorescence is quenched and red fluorescence is increased (29). Concomitantly, displacement of SYBR Green I by PI may play an additional role in the discrimination of viable and dead cells due to a higher binding affinity of PI to nucleic acids as reported by Stocks (37) for a similar staining protocol using the commercially available Baclight™ kit. PI concentration testing revealed optimal PI concentrations, which had no effect on membrane integrity and enabled complete FRET from SYBR Green I to PI for discrimination of dead from viable cells. For S. aureus, an additional subpopulation exhibiting intense SYBR Green I and PI fluorescence was detected. Taking into account the findings of other authors (13, 20), these events were referred to as damaged cells, i.e., cells with slightly damaged membranes. Here, a lower amount of PI entered cells resulting in an incomplete FRET from SYBR Green I to PI (29). Even though a loss of membrane integrity is generally associated with absence of reproductive growth and metabolic activity, the membrane of growing cells can also be perforated during cell division and cell wall synthesis (15). Staining would then result in false PI positive cells.
Staining Protocol for Viability Determination in Mixed Culture
In order to determine the viability of S. aureus and B.cepacia in mixed cultures, the staining protocol described above was modified and extended to enable species discrimination.
For detection of species-specific viability, Gram-specific staining of S. aureus using fluorescently labeled WGA was successfully employed in combination with SYBR Green I and PI using the mixed culture staining protocol. WGA is known for its specific binding to the peptidoglycan layer of Gram-positive bacteria, specifically to N-acetylglucosamine and N-acetylneuroaminic acid residues (38). In this study, WGA did not bind to B. cepacia, which is in agreement with results reported by Sizemore et al. (39), who explained this finding by the presence of the outer membrane of Gram-negative bacteria, which covers the peptidoglycan layer. In contrast to this, reports from other authors showed WGA binding to Gram-negative bacteria, in particular for Escherichia coli, where the enterobacterial common antigen was labeled (40, 41). On the other hand, flow cytometric studies of Holm and Jespersen (42) demonstrated no WGA binding to E. coli among other Gram-negative bacteria. Therefore, it was important to test the specificity of WGA for the species and the physiological state of interest. WGA staining of S. aureus was highly efficient and reproducible. More than 97% of S. aureus could be stained irrespective of growth phase and membrane integrity. In addition, the binding capacity of the lectin did not decrease significantly over the time period covered in our cultivations as observed by Heine et al. (43) for labeling of yeast cells with concanavalin A during brewing processes.
In contrast to widely used fluorescence in situ hybridization-based techniques used for specific detection of bacterial species or taxa in mixed communities involving fluorescently labeled single stranded nucleic acid probes, no permeabilization of cytoplasmic membrane is needed for specific labeling by WGA. Furthermore, the specificity of WGA is superior compared to commercially available fluorescence-based Gram-staining kits involving hexidium iodide. These kits stain not only Gram-positive bacteria (44) but also Gram-negative bacteria if lipopolysaccharides are destabilized (45). Therefore, WGA staining can generally be employed for flow cytometry in combination with different viability staining techniques.
Viability During Growth in Pure and Mixed Culture
The optimized three-color staining protocol was successfully applied for the assessment of viability of S. aureus and B.cepacia during growth in mixed culture over a cultivation period of 32 h. To compare pure and mixed culture cultivations, viability was determined additionally in pure culture. Furthermore, for each culture, qT-RFLP analysis was performed to analyze growth. In mixed culture, the growth of both species was not affected significantly by the presence of the other species. Even though it was expected, neither growth inhibition of S. aureus due to substrate competition nor an growth advantage of B. cepacia due to an existing food chain could be observed as had previously been reported for CF-relevant mixed communities (10, 46). This may be due to differences in species composition as Riedele and Reichl (10) characterized three-species communities or differences in mixed culture conditions as Hesseler et al. (46) described competition in chemostat cultivations. However, growth of both species was improved in mixed culture in this study. However, this is due to batch-to-batch variations or an interspecies effect remained to be clarified.
Viability of S. aureus was initially high for pure as well as for mixed culture but decreased within the first 1 h or 2 h of growth, respectively. Concomitantly, the frequency of damaged S. aureus cells increased. This may be caused by PI entry mediated by cell wall perforation encountered during cell division and cell wall synthesis (15, 47), especially for exponentially growing cells. Accordingly, PI uptake in highly reproductive bacteria during exponential growth was reported in several studies (47, 48). Furthermore, some of the events here classified as damaged might be associated with aggregation of viable and dead S. aureus cells, since it had previously been shown that Staphylococci clump during growth (49). This was also observed in our experiments by increased FSC and SSC signals during growth (data not shown). After 3 h, the frequency of viable S. aureus increased again to high levels (above the initial values) and remained relatively constant until stationary phase was reached. This may be explained by increased cell concentrations during this phase as indicated by cell doublings observed during qT-RFLP analysis. Furthermore, cells grew slower after exponential growth phase due to glucose limitation (data not shown) and, therefore, reduced sensitivity to PI staining (48). The decrease of viability of S. aureus during stationary growth phase in pure and mixed cultures could probably be attributed to the exhaustion of nutrients. As a result, metabolic activity was reduced and impairment of active transports led eventually to membrane permeabilization (14). The tremendous increase in frequency of dead cells toward the end of cultivation was in agreement with results reported by other authors (50, 51), which suggested the loss of membrane integrity as the last detectable cellular event after damage. Furthermore, this may explain the existence of damaged S. aureus cells, which reflects a transient state during loss of membrane integrity. Furthermore, this suggests a slower damage process of the cytoplasmic membrane in comparison to B. cepacia.
The time course of viability of B. cepacia during exponential growth phase in pure and mixed culture was comparable to S. aureus. The correlated increase and decrease in the frequency of dead cells is probably due to the same mechanisms as discussed for S. aureus. In contrast, the viability of B. cepacia remained relatively constant during stationary growth phase until the end of cultivation. This may imply the ability of B. cepacia to maintain membrane integrity during stationary phase, despite of nutrient limitations. For S. aureus, during stationary growth phase clear differences in the time courses of viability in mixed and pure culture were found. In mixed culture, viability dropped earlier and decreased faster compared to pure culture, especially toward the end of cultivation. These observations suggest interspecies effects with B. cepacia in mixed culture. This could probably be attributed to lytic activity of B. cepacia against S. aureus inducing membrane damage, e.g., by peptidoglycan hydrolases (52). However, obtained qT-RFLP data did not indicate any cell lysis. Furthermore, proteomic analyses of a three-species mixed community of relevance did not reveal staphylolytic activity of B. cepacia (53). The mechanism of the observed interspecies effect remained to be clarified. In contrast to S. aureus, viability dynamics of B. cepacia were comparable in mixed and pure culture, which suggest that no negative interspecies effects were induced by the presence of S. aureus. This is in line with results reported by Riedele and Reichl (10).
In conclusion, a novel flow cytometric three-color staining method using SYBR Green I, PI, and fluorescently labeled WGA was established, validated and successfully tested for assessment of viability of S. aureus and B. cepacia in pure and mixed cultures. The method allowed rapid simultaneous Gram-differentiation and viability monitoring of bacterial mixed cultures and is therefore recommended for quantitative analysis of dynamics of mixed cultures, not only for medical research but also for biotechnological applications, food processing, and environmental studies. In combination with methods for species-specific enumeration, in particular qT-RFLP assays, a rapid and reproducible analysis of mixed culture dynamics is achieved.
The authors thank Corina Siewert for excellent technical assistance. Jan Lenke from Institute of Bioprocess Engineering, Flensburg University of Applied Sciences, Germany, is acknowledged for helpful discussion on viability analysis of Gram-negative bacteria using flow cytometry.