These authors contributed equally to this work.
Limits of propidium iodide as a cell viability indicator for environmental bacteria†
Article first published online: 9 APR 2007
Copyright © 2007 International Society for Analytical Cytology
Cytometry Part A
Volume 71A, Issue 8, pages 592–598, August 2007
How to Cite
Shi, L., Günther, S., Hübschmann, T., Wick, L. Y., Harms, H. and Müller, S. (2007), Limits of propidium iodide as a cell viability indicator for environmental bacteria. Cytometry, 71A: 592–598. doi: 10.1002/cyto.a.20402
Part of this manuscript was presented at the 11th Leipziger Workshop, April 27–29, 2006, BioCity Leipzig, Germany. Available at www.leipziger-workshop.de.
- Issue published online: 18 JUL 2007
- Article first published online: 9 APR 2007
- Manuscript Accepted: 15 NOV 2006
- Manuscript Revised: 12 NOV 2006
- Manuscript Received: 25 JUL 2006
- multiparametric flow cytometry;
- bacterial viability;
- population dynamics;
- live/dead staining;
- propidium iodide
Viability measurements of individual bacteria are applied in various scopes of research and industry using approaches where propidium iodide (PI) serves as dead cell indicator. The reliability of PI uptake as a cell viability indicator for dead (PI permeable) and viable (PI impermeable) bacteria was tested using two soil bacteria, the gram−Sphingomonas sp. LB126 and the gram+Mycobacterium frederiksbergense LB501T.
Bacterial proliferation activities observed viaDAPI and Hoechst 33342 staining were linked to the energy charge and the proportion of dead cells as obtained by diOC6 (3)-staining and PI-uptake, respectively. Calibration and verification experiments were performed using batch cultures grown on different substrates.
PI uptake depended on the physiological state of the bacterial cells. Unexpectedly, up to 40% of both strains were stained by PI during early exponential growth on glucose when compared to 2–5% of cells in the early stationary phase of growth.
The results question the utility of PI as a universal indicator for the viability of (environmental) bacteria. It rather appears that in addition to nonviable cells, PI also stains growing cells of Sphingomonas sp. and M. frederiksbergense during a short period of their life cycle. © 2007 International Society for Analytical Cytology
Studies on the activity of environmental bacterial species often lack detailed information on the physiological states and viability of individual bacteria. By applying cell fluorescence-based methods (for overview see Ref.1), flow cytometry allows to gain better information on the activity of single cells. Viability tests have been found to be relevant in medicine, biotechnology, and food industry, because of their power to assess the susceptibility of bacteria against antibiotics (2–4), their analysis of the physiological and metabolic states during product formation processes (5, 6), and their ability to recognize the cell death of production strains or infection by living pathogens (7–9). Although vast literature is available about organisms of biotechnological and medical relevance, only little is known about the single cell viability of environmental communities with most of the existing literature focusing on waste water treatment plants (10). This is a serious gap of knowledge as (oligotrophic) environmental bacteria are supposed to be key players in biogeochemical and pollutant transformation processes. The viability of environmental bacteria is often assessed with commercially available kits that rely on the propidium iodide (PI)-based assessment of membrane integrity. However, many users are unaware of the facts that such tests have been validated for a very limited number of bacterial species, only (11). Nevertheless, the use of PI-based kits for determining the viability of unknown environmental species is seductive because of their easy and quick application. Here, the reliability of PI as an indicator for the viability of two polycyclic aromatic hydrocarbon (PAH)-degrading soil bacteria, the gram−Sphingomonas sp. LB126 and gram+Mycobacterium frederiksbergense LB501T (12) was assessed in detail. The two strains were grown either in the presence of PAH (fluorene for strain LB126 and anthracene for strain LB501T) or glucose as sole carbon and energy sources. The numbers of PI stained cells from different growth phases were determined by flow cytometry. Additional physiological information was obtained by applying DAPI, Hoechst 33342, and diOC6 (3) fluorescent dyes and flow sorting to link the ability of individuals to form colonies on agar plates to the results obtained from the staining procedures.
MATERIALS AND METHODS
The gram−Sphingomonas sp. LB126 and the gram+Mycobacterium frederiksbergense LB501T (12) were grown aerobically at 20°C and pH 7.5 in 500 ml shake flasks with 300 ml minimal medium at 150 rpm for batch experiments. The carbon and energy sources were either glucose (1 g l−1) or PAHs, such as crystalline anthracene (2 g l−1; ≥96.0% (GC), Fluka Chemie GmbH) for M. frederiksbergense and crystalline fluorene (2.7 g l−1; ≥99% (HPLC), Fluka Chemie GmbH) for Sphingomonas sp. Glucose was filtered through sterilized 0.2-μm filters (VWR International) before use and the carbon and energy sources were added separately at the beginning of the cultivation. The composition of the mineral medium and growth on crystalline PAH are described elsewhere (13). The strains were regularly cultivated on LB-agar (Lennox, Carl Roth GmbH; Agar from Difco) or on minimal-medium-agar. E. coli was grown aerobically at 30°C and pH 7.5 in 500 ml shake flasks at 150 rpm for batch experiments on 300 ml peptone media (l−1): 5 g peptone from meat (pancreatic), 3 g NaCl, 2 g K2HPO4, 10 g meat extract, 10 g yeast extract, 5 g glucose. Growth was measured spectrophotometrically by monitoring absorption at 578 nm (OD578 nm, 10 mm).
Cell Preparation and Staining Procedures
The harvested cells were centrifuged at 3,200g for 5 min, if necessary fixed with 10% NaN3, and stored at 4°C. This procedure was found to preserve the cells for at least 3 months. For flow cytometric measurements, either living or preserved cells were centrifuged again, washed in NaCl-phosphate buffer (0.4 M Na2HPO4/NaH2PO4, 150 mM NaCl, pH 7.2), and resuspended at a concentration of 3 × 108 cells ml−1.
DNA patterns of preserved cells were obtained in the following way: 2 ml of diluted cell suspension were treated with 1 ml solution A (2.1 g citric acid/0.5 g Tween 20 in 100 ml bidistilled water) for 10 min, washed, and resuspended in 2 ml solution B (0.24 μM 4′,6-diamidino-2′-phenylindole [DAPI, SIGMA], 400 mM Na2HPO4, pH 7.0) for at least 20 min in the dark at room temperature using a modification of a standard procedure (14). DNA contents of living cells were analyzed by using Hoechst 33342 (2,5′-Bi-1H-benzimidazole; Molecular Probes, OR) as described earlier, yet without the Tween 20 treatment (15).
To assess the membrane potential-related fluorescence intensity (MPRFI), living cells were resuspended in 20 mM imidazole buffer (pH 7.0), and immediately adjusted to 3 × 108 cells ml−1. The composition of the staining solution was taken from Shapiro (16). For optimal alignment, the MPRFI was defined by testing different dye concentrations, staining times, and the action of antibiotics (gramicidin and valinomycin) on exponentially growing cells. Accordingly, 15 μl (0.39 μM) and 5 μl (0.13 μM) of the dye stock solution were added to 2 ml of 3 × 108 cells ml−1 for 3 and 4 min to either Sphingomonas sp. or M. frederiksbergense. All measurements were carried out at 20°C. In addition, the measurements were daily equalized by using YG beads (0.5 μm, FluoSpheres 505/515, F-8813, Molecular Probes/Invitrogen, USA).
To determine membrane integrity, the living cells were washed and resuspended in 2 ml PBS (cell concentration of 3 × 108 cells ml−1) and stained by 20 μl PI (Sigma–Aldrich, Steinheim, Germany; final concentration: 1 μM, stock solution: 0.07 mg ml−1, in PBS, pH 7.2) for 10 min.
Combined double staining of living and dead cells was done by applying Hoechst 33342 and PI in the following way: 2 ml of a harvested and PBS washed cell suspension (3 × 108 cells ml−1) were incubated with 4 μl Hoechst 33342 for 35 min (final concentration: 3.24 μM). Then, 20 μl PI were added and the sample analyzed after additional 10 min.
Flow Cytometry and Cell Sorting
Multiparametric flow cytometry.
Flow cytometric measurements were carried out using a MoFlo cell sorter (DakoCytomation, Fort Collins, CO) equipped with two water-cooled argon-ion lasers (Innova 90C and Innova 70C from Coherent, Santa Clara, CA). Excitation of 580 mW at 488 nm was used to analyze the forward scatter and side scatter as trigger signal at the first observation point. DAPI was excited by 180 mW of ML-UV (333–365 nm) atthe second observation point. The orthogonal signal was first reflected by a beam-splitter and then recorded after reflection by a 555 nm long-pass dichroic mirror, passage by a 505 nm short-pass dichroic mirror and a BP 488/10. DAPI fluorescence was passed through a 450/65 band pass filter, green fluorescence through BP 520/15 and, red fluorescence through BP 620/45. Photomultiplier tubes were obtained from Hamamatsu Photonics (models R 928 and R 3896; Hamamatsu City). Amplification was carried out at linear or logarithmic scales, depending on the application. Data were acquired and analyzed using Summit software (DakoCytomation, Fort Collins, CO). Fluorescent beads (Polybead Microspheres: diameter, 0.483 μm; flow check BB/Green compensation Kit, Polyscience, USA) were used to align the MoFlo. Also, an internal DAPI-stained bacterial cell standard was introduced for tuning the device up to a CV value not higher than 6%.
Cells from three independent cultures of Sphingomonas sp., M. frederiksbergense, and E. coli were treated with PI and sorted using the most accurate sort mode (single and one drop mode: highest purity 99%) channeling PI fluorescent or PI nonfluorescent cells onto LB agar plates in triplicate. Finally, each agar plate contained a grid of 96 (12 × 8) spots harboring 1 and 2 cells, respectively. After incubation at room temperature (Sphingomonas sp., M. frederiksbergense) or 37°C (E. coli), the numbers of colonies were counted. The culturability of PI fluorescent and PI nonfluorescent cells was calculated as the ratio of the numbers of colony forming units (CFUs) to the numbers of cells spotted. Bacterial survival of the sort procedure was tested as described elsewhere (15).
To verify reliable staining, the cells were subjected to image analysis (camera: DXC-9100P; software: Openlab 3.1.4., Improvision, Lexington, MA) using ultraviolet light from a 100 W mercury arc lamp. Fluorescence filters used were as follows: Zeiss filter set 02 for blue fluorescence (excitation G 365, BS 395, emission LP 420), Zeiss filter set 09 for green fluorescence (excitation BP 450-490, BS 510, emission LP 515), and Zeiss filter set 15 for red fluorescence (excitation 546/12, BS 580, emission LP 590).
Calibration of the Staining Procedures
All calibrations of the PI staining procedure were done with cells grown on glucose and harvested at the early stationary phase of growth (150 h of cultivation for M. frederiksbergense and 24 h for Sphingomonas sp.). Time and concentration dependencies were tested for both strains (Fig. 1). A stable staining equilibrium was obtained under conditions described in Materials and Methods, labeling up to 2.8% Sphingomonas sp. and 20.5% M. frederiksbergense cells.
The same conditions were used for the double staining of Hoechst 33342/PI. Calibration was necessary to prevent toxic effects of the dye combinations on cell viability. The commonly applied surfactant Tween 20 (81.4 nM) affected the membranes of Sphingomonas sp. and led to an increase of PI uptake to nearly 100% for high-exponential and to 38% for cells in the mid-exponential and early stationary growth phase, an effect that has been described already for other cells (17). Therefore, Tween 20 was omitted during further labeling procedures of both species.
MPRFI was calibrated by using diOC6 (3) for viable cells harvested at the early exponential growth phase. Stable staining conditions were found after application of 15 μl (0.39 μM) per 2 ml diluted Sphingomonas sp. cells within a time range of 3 min, as after longer exposure two apparent subpopulations of clearly differing fluorescence intensities were observed. For M. frederiksbergense 0.13 μM diOC6 (3) were necessary for reliable and constant staining within 4 min. The response of the fluorescence intensity of the cell-bound dye to changes in the membrane potential was verified by analyzing the effects of the ionophores valinomycin and gramicidin. Cells were treated with different concentrations of the two ionophores immediately after the staining procedure. Low concentrations of gramicidin (2–3 μM) decreased the MPRFI of both strains and led to an increase of cellular debris as analyzed by flow cytometry, whereas higher concentrations of valinomycin (20–30 μM) were necessary to depolarize the cells.
Course of Cell Growth Followed by DAPI, Hoechst 33342, diOC6 (3), and PI
To determine the most active growth states, cells were analyzed with regard to their DNA contents (by DAPI) while cultivated on the following carbon and energy sources: glucose (both strains, up to 120 h), anthracene (M. frederiksbergense, up to 900 h), and fluorene (Sphingomonas sp., up to 300 h). The Sphingomonas sp. presented up to three subpopulations with different chromosome contents referred to as C1n, C2n, and C4n, all standing for multiples of chromosomes. The chromosome nomenclature was already explained by Müller and Babel (18). On glucose Sphingomonas sp. showed no C1n cells yet an up to 50% increase of C4n cells during the exponential growth phase though up to 90% C2n cells dominated the stationary phase of growth. On fluorene, the proliferation behavior was very similar with, however, only 20% of total cells in the C4n subpopulation during the exponential growth phase. Also, a minor content of C1n cells (up to 10%) was additionally observed for a short time period. Growth of M. frederiksbergense on glucose presented up to 50% C2n cells during exponential growth and an increase in C4n and even C8n cells during stationary growth probably due to their clotting behavior as well as the formation of small hyphae-like morphology under limiting growth conditions. During growth on anthracene C2n subpopulations remained constant at about 50% during the whole observation period with always only low numbers of C8n cells. Since the anthracene was given in crystalline form because of its low aqueous solubility, even after 900 h of cultivation, a constant dissolution-limited anthracene flux to the cells was provided and no classical stationary phase was observed within this time period of cultivation (13).
Using the same batch-cultures, the quantity of dead cellswas determined as the fraction of PI permeable cells. However, unexpectedly, the quantities of PI positive cells were always found to be very high during the early growth phase. These results were surprising and demanded a closer investigation. Therefore, cultivation duration was extended to up to 2,600 h for glucose and over 3,000 h for both fluorene and anthracene (Fig. 2). The data showed an obvious increase in PI stained Sphingomonas sp. cells up to 40% during exponential growth on glucose (see also the inset in Fig. 2A) and up to 8% on fluorene (Fig. 2B). The same behavior was observed for M. frederiksbergense on its two substrates (Figs. 2C and 2D). The temporarily high PI uptake leveled down to nearly 2–5% on all given substrates of the two species quickly afterwards, only to increase after about 400 h cultivation for Sphingomonas sp. and notedly after 2,600 h of cultivation for M. frederiksbergense.
To demonstrate the viability of PI-stained cells in the early exponential growth phase, the MPRFI related stain diOC6 (3) was applied to provide further information on viability of the cells. Changes in energy charge were followed over the same time ranges in coevally and independently harvested samples. As expected, the energy charge-related fluorescence intensities were homogeneously distributed and went along with high PI uptake (Fig. 2, insets b).
Hoechst 33342 was used to elucidate which cell states of the cell cycle might have captured PI maximally. The data for M. frederiksbergense are presented in Figure 3. A clear assignment was difficult to make due to dye pumping (Fig. 3A, 48 h, gate 1) and quenching of the Hoechst 33342 fluorescence intensities when high amounts of PI (Fig. 3a, 48 h, gate 2) entered the cells. Such a quenching was already described in double stained Hoechst 33342/PI yeast cells (19). However, referring to the part of unaffected glucose grown cells, those with the lowest DNA contents had the highest quantities of PI stained cells during the first 48 and 168 h of cultivation (arrows with one star), yet at the end of cultivation there was a higher number of PI cells with higher chromosome contents, even though the C2n was the dominant subpopulation (arrow with two stars). Growth on anthracene revealed a different behavior of the population, no pumping and nearly no quenching was observed, also the quantities of dead cells were nearly equally distributed between the subpopulations.
Viability of PI Stained Cells
The viability of PI stained cells of M. frederiksbergense, Sphingomonas sp., and E. coli (as control) was determined as their culturability on LB agar. The culturability was calculated as the ratio of the number of CFUs to the number of cells spotted; the culturability of PI treated but unstained cells was also analyzed for comparison.
Initial experiments revealed remarkable species-specific differences of culturabilities even if cells had never been treated with PI with 96% for E. coli, 84% for M. frederiksbergense, and 20% for Sphingomonas sp. (inset of Fig. 4). Similar degrees of culturability were observed with PI treated but unstained cells (Fig. 4). Hence, the mere presence of PI did not apparently affect cellular growth.
In contrast, the culturability of fluorescent cells was reduced. This was most obvious with E. coli. Only 4% of fluorescent E. coli cells were culturable, indicating a large majority of 96% dead bacteria among the PI stained cell population. Exponentially grown and PI stained cells of the two soil bacteria displayed a different behavior: almost 60% of M. frederiksbergense and 10% of Sphingomonas sp. cells grew to colonies. Taken into account the species-specific cultivabilities, 50–75% of all PI stained cells were culturable. In other words, the PI-stained population consisted of more culturable than dead cells.
Bacteria from the late stationary phase showed reduced viability upon PI staining. Only 34% of cells of M. frederiksbergense and 3% of Sphingomonas sp. formed colonies. This indicated a large proportion of dead bacteria, even so, the number of culturable cells was unexpectedly high. Hence, both with exponentially and stationary phase cells, PI staining did not reliably indicate death of M. frederiksbergense and Sphingomonas sp.
As bacteria are of outstanding relevance in medicine, biotechnology, and environmental microbiology, the identification of their activity at given conditions requires reliable and quick staining-based viability assessments. The application of such staining cocktails needs detailed knowledge of the dye's effects as well as the physiology and structure of the microorganisms studied. As PI has been described manifold in literature to penetrate injured membranes and cell walls, the dye is often applied to distinguish living from dead microorganisms. However, although PI uptake was described to occur in compartmented Bifidumbacterium species (20), producing segregate double stained cFDA/PI cells, large quantities of these cells were reproducibly culturable on agar plates, a phenomenon attributed to transiently injured or dead (compartmented, probably as part of a hyphae-like morphology), but recovering cells.
As many commercially available viability kits are developed for medically relevant strains, such as E. coli, PI-staining of exponentially growing E. coli was included in this study and compared to the staining behavior of environmental isolates. In contrast to the latter, growing E. coli cells were found to take up negligible amounts of PI (1–2%, not shown). However, up to 40% of the environmental isolates were stained by PI during the exponential growth phase and large quantities of PI-stained cells of both Sphingomonas sp. and M. frederiksbergense remained culturable on LB-agar plates after cell sorting. This applied also, yet to minor quantities, to PI stained and sorted cells from the stationary growth phase (Fig. 4). The presence of viable cells was corroborated by the observation that the exponentially growing cells showed cell cycle activities (analyzed by DAPI), performed pumping of Hoechst 33342 and also had a high and stable energy charge as analyzed by MPRFI.
Although these data clearly question the utility of PI as universal indicator for cell death, the question why the cell walls of some actively growing cells became permeable for PI remains unanswered. Generally, there are three main protein families responsible for the bacterial SEDS (shapes, elongation, division, and sporulation) functioning: PBPs (penicillin binding proteins, peptidoglycan synthesis complex), Fts-proteins and RodA (tubulin homologues, responsible for the cell division site), and MreB (actine homologue and probably responsible for cell elongation). The loss of peptidoglycan stability has been described to be connected to the functioning of these molecules. For gram− bacteria it was assumed that the membrane integrity is endangered if subunits of glycan strands are inserted into broken single cell wall glycan-layers during cell growth (see for review Ref.21). Additionally, the cell membrane integrity is also known to be temporarily affected during periods of fast cell size growth and to be particular sensitive to a variety of antibiotics due to physical cell wall reconstruction at new cell poles or sites of cell-division (22). The significant differences of PI-uptake between fast growing E. coli cells (μ = 0.347 h−1) and slowly proliferating Mycobacteria (glucose: μ = 0.027 h−1; anthracene: μ = 0.013 h−1) and Sphingomonas species (glucose: μ = 0.046 h−1; fluorene: μ = 0.01 h−1) may thus be explained by the inability of the latter to quickly close broken glycan strands by highly efficient enzymes.
Mycobacteria are further known to exhibit porin-mediated transport of hydrophilic compounds across the outer membrane. Although the porin MspA is described to be at least 1,000-fold less efficient than those of gram− bacteria (23), its expression was found both to increase the glucose uptake and to accelerate growth. Consequently, Mailaender et al. (24) discussed porins as the mediator for a tenfold increased uptake of the fluorescent dye Syto 9 due to presumed local membrane interruptions upon expression of mspA.
In conclusion, the data presented question the utility of PI as universal indicator for the viability and culturability of environmental bacteria. Careful testing of the reliability of PI-based viability assays thus is recommended for their application outside of standardized procedures.
The authors thank Christine Süring and Helga Engewald for skilled technical assistance.
- 6The use of multi-parameter flow cytometry to study the impact of limiting substrate, agitation intensity and dilution rate on Bacillus licheniformis CCMI 1034 aerobic continuous culture fermentations. Biotechnol Bioeng 2005; 92: 568–578., , , , , .
- 11The Handbook—A Guide to Fluorescent Probes and Labeling Technologies, 10th ed. Eugene, OR: Molecular Probes; 2005. Section 15.3; sheet mp07007..
- 16Practical Flow Cytometry. New York: Alan R. Liss Inc.; 1988; p 186–192 and p 296–298.