Journal of Applied Bacteriology 1995, 78. 3-15

The ability of membrane potential dyes and calcafluor white to distinguish between viable and non-viable bacteria

D.J. Mason, R. Lopez-Amoros, R. Allman, J.M. Stark and D. Lloyd School of Pure and Applied Biology, University of Wales College of Cardiff, Cardiff, UK

4934/05/94: received 25 May 1994, revised 2 November 1994 and accepted 4 November 1994

D.J. MASON, R . L O P E Z - A M O R ~ S , R . ALLMAN, J .M. S T A R K AND D. LLOYD. i w 5 . Various dyes were assessed for their ability to discriminate between viable and non-viable bacteria. Two methods of killing were employed : by heat treatment or by gramicidin treatment. Staining was carried out in two ways; by staining directly in the medium or by washing cells prior to staining in buffer. Carbocyanine and rhodamine 123 dyes only exhibited small changes in fluorescence between viable and non-viable populations of bacteria. Both oxonol dye (bis 1,3- dibutylbarbituric acid trimethine oxonol) and calcafluor white proved much more useful.

INTRODUCTION The use of fluorescent dyes as indicators of cell viability is widespread. Dyes such as ethidium bromide, acridine orange, propidium iodide and fluorescein diacetate have all been used successfully for this both in fluorescence micros- copy and flow cytometry.

Membrane potential sensitive dyes have also been used as indicators of bacterial cell viability, the fluorescent response of these dyes varies with the magnitude of the membrane potential (Mason et al. 1993). Rhodamine 123 (rh123) is one such dye; a cationic lipophilic dye (accumulated cytosolically by cells with an inside negative transmembrane electrochemical potential), it has been used extensively to study mitochondria in eukaryotic cells (Ronot et al. 1986; Farkas et al. 1989; Skowronek et al. 1990; Rhan et al. 1991) as well as bacterial viability (Matsuyama 1984; Kaprelyants and Kell 1992; Diaper et al. 1992; Davey et al. 1993).

The carbocyanine dyes are also a family of membrane potential sensitive lipophilic cations. These have been used to determine membrane potential in a diversity of cells and vesicles including mouse ascites tumour cells (Eddy 1989), cultured mammalian cells (Hargittai et al. 1991), lympho- cytes (Wilson et al. 1985), red blood cells (Sims et al. 1974), yeast (Pena et al. 1984) and bacteria (Zaritsky et al. 1984; Mason et al. 1993).

Another well used group of membrane potential sensitive dyes are those of the oxonol family; these are lipophilic

Correspondence to : Professor David Lloyd, School of Pure and Applied Biology, University of Wales College of CardtB; PO Box 915, Cardcff CFI 3TL, UK.

anions and thus unlike the carbocyanines and rh123 are not extensively accumulated cytosolically by organisms with an inside negative transmembrane electrochemical potential. Therefore, the fluorescent response is opposite to that of cationic dyes, i.e. the fluorescent response decreases with increase in potential. These dyes have been used in the study of mouse tumour cells (Brauner et al. 1984; Oyama 1991), lymphocytes (Wilson and Chused 1985), mouse thymus cells (Lakos et al. 1990) and to estimate the effects of antibiotics and antihngal agents on microbial popu- lations (Carter et al. 1993; Ordofiez and Wehman 1993).

Calcatluor white (CFW) is the disodium salt of 4,4’-bis(4 anilino-bis-diethyl amino-s-mazin-2-ylamino)-2,2’-stilbene disulphonic acid); it is one of a family of compounds used as ‘fluorescence brighteners’ in the dye industry. These compounds are highly fluorescent when excited by U.V.

light. Absorption and transport of these dyes by micro- organisms has been investigated (Darken 1962). CFW binds to the chitin of the fungal cell wall (Streiblova 1984). It has also been used to estimate cell viability in rodent cell lines (Berglund et al. 1987). Viable cells are able to exclude this dye whereas non-viable cells appear brightly fluores- cent although the mode of binding is unknown. Measure- ments of cell viability, staining with CFW have been found to correlate well with propidium iodide (PI) and fluorescein diacetate fluorescence, a correlation coefficient of 0.9886 was recorded with PI (Berglund et al. 1987).

In this paper flow cytometry has shown the superiority of the oxonol dye bis-( 1,3dibutylbarbituric acid) tri- methine oxonol (DiBAC4(3)) and CFW as indicators of cell vitality over the more commonly used carbocyanine dyes and rh123.

310 D.J. MASON ET A L .

MATERIALS AND METHODS

Bacterial strains and growth medium

The bacterial strains used in this study were Escherichia coli NCTC 10418, E. coli K12, Salmonella fyphimurium STCC 14028 and Staphylococcus aureus NCTC 6571. Growth of all cultures was in nutrient broth (Oxoid, CM1) at 37°C in a reciprocating waterbath at 200 strokes min-'.

Heat treatment

Cultures of all strains were grown to early logarithmic phase of growth (0.D.600 0.3) and then heated to a tem- perature of 70°C for 30 min (laboratory pasteurization). Samples (1 ml) were removed and stained according to one of the two protocols described. A control culture for each strain was also grown to early logarithmic phase of growth before being sampled.

Gramicldln treatment

Cultures of each strain were grown to mid logarithmic phase (0.D.600 0.45), at which point samples (1 ml) were removed (one for each dye) and the remaining culture was further incubated with gramicidin S at 20 mg 1-'. One ml samples were then removed after 30 min.

Dyes used

The following dyes were added to separate samples to give the final concentrations shown: DiBaC4(3) 0.5 mg 1-', cal- cafluor white 2.5 mg I - ' , rhodamine 123 (rh123) 0.25 mg 1-' or 5 mg I-' , 3,3diethyloxacarbocyanine iodide (DiOC2(3)) 0.1 mg 1- ', 3,3-dihexyloxacarbocyanine iodide (DiOC6(3)) 0.1 mg 1- ', 3,3diheptyloxacarbocyanine iodide (DiOC,(3)) Omlmg 1-'.

Staining methods

Samples were stained using both the following protocols for each dye:

Method 1. Dyes were added directly to 1 ml of cells sus- pended in nutrient broth. There was no prior sample prep- aration for Gram-positive organisms. Samples of Gram- negative species samples were preincubated for 3 min at room temperature with EGTA 0.1 mmol I - ' (final concentration) except for those being stained with DiBAC4(3).

Method 2. Cells suspended in 1 ml of nutrient broth were sedimented for 30 s in an Eppendorf centrifuge at 13000 rev min-'. The supernatant fluid was removed and the

pellet gently resuspended using a vortex mixer in 10 mol 1 - ' Tris-HCl (pH 7.6). The samples were then centrifuged for 30 s, the supernatant fluid was discarded and the pellets resuspended in 1 ml of 10 mmol 1-' Tris-HC1 (pH 7.6) and then stained with one of the dyes.

These procedures may be regarded as minimally per- turbing, others have described the addition of 1 mmol 1- ' EDTA to permeabilize the outer membrane of Gram- negative bacteria for a better response from membrane potential dyes (Kaprelyants and Kell 1992; Diaper et al. 1992). However, EDTA was found to be extremely toxic to E. coli NCTC 10418 (results not shown). Nikaido and Varra (1984) reported that EDTA had been found to damage the outer membrane, of Gram-negative bacteria, by removing large portions of the structure. For this reason use of EDTA was avoided and EGTA used instead, which proved to be non-toxic at the concentation used (results not shown).

Viable counts

T o assess the effectiveness of the heat treatment on cul- tures, counts of colony-forming units were performed by the method of Miles and Misra (1938).

Flow cytometry

Flow cytometric analysis was carried out on a Skatron Argus 100, a mercury arc lamp based dual parameter flow cytometer. The excitation filter blocks used were: B1 filter (395-440 nm) for the carbocyanine dyes, FITC filter (47G 490 nm) for rh123 and DiBaC4(3) and the U.V. filter (365 nm) for calcafluor white. Sample flow was set to 5 pl min-' with a sheath fluid pressure of 1 kPa Data acquisition using linear amplification was gated by forward (logarithmic amplification) and wide angle light scatter (linear amplification). All samples were allowed to run for 1 min before data acquisition; 5000 cells were counted from each sample.

RESULTS

Results illustrated in this paper are typical of at least five repetitions of the experiments described.

Heat-treated cells

All strains after heat treatment consistently exhibited a sig- nificant increase in fluorescence, when stained with DiBAC4(3). A typical example is shown in Fig. 1. Figure 2 shows a typical fluorescent response obtained from heat- treated bacteria when stained with CFW, following either of the staining methods.

DYES TO ASSESS VIABILITY 311

a a

45 i

CFW fluorescence (channel number) DiBAC,+(J) fluorescence (channel number)

Flg. 1 A single parameter histogram of viable (- and heat-treated (- - - -) populations of Escherichia coli NCTC 10418; stained with DiBAC4(3) in nutrient broth

The type of fluorescent response obtained from rh123 varied according to the concentration used, method of staining and bacterial strain (Table 1).

Neither carbocyanine dye staining protocol produced consistent results. Figure 3 shows an example of two con- flicting responses from the same strain under identical con- ditions.

Viable counts of heat-treated organisms of all species were calculated as < 10' ml- '. Colony-forming units in all control cultures were calculated to be of the order of lo* mI-'.

Grrmlcldln-treated cells

Flg. 2 A single parameter histogram showing an example of the fluorescent response from viable (- ) and heat-treated bacterial cells (- - - -) stained with calcafluor white. Eschnichia coli NCTC 10418 cells stained in nutrient broth

A typical fluorescent response from gramicidin-treated bacteria stained either with CFW or rh123 is shown in Figs 5 and 6 respectively. Similar results were achieved with either staining method.

As with heat-treated cells the type of fluorescent response exhibited by rh 123 varied with the concentration used, staining method used and bacterial strain (Table 2).

The fluorescent response between gramicidin-treated and untreated bacteria stained with any of the carbocyanine dyes was inconsistent within all strains with either staining method. Figure 7 shows examples of conflicting fluorescent responses exhibited using a carbocyanine dye with one bac- terial strain.

Excellent discrimination between gramicidin-treated and untreated cells of all strains was achieved with DiBAC,(3). Similar results were obtained with both methods of stain- ing, a typical example is shown in Fig. 4.

D 1s c u s S l ON The fluorescent response of disrupted bacteria stained with DiBAC,(3) was similar with both methods of staining, also

Table 1 Patterns of fluorescent response for viable and non-viable bacteria stained with rh123

Staining conditions

0-25 mg 1- ' rh123 : cells 5 mg I - ' rh123 5 mg I - ' rh123 rh123: cells washed and cells suspended cells washed and suspended in suspended in in nutrient suspended in

0.25 mg I - '

Stnin nutrient broth Tris-HC1 broth Tris-HC1

E. coli NCTC L/D no FL D > L D > L L > D

E. coli K12 L/D no FL D > L D > L D > L

Salm. typhimurium LJD no FL D > L D > L D > L

Staph. aureus D > L D > L D > L L > D

10418 change

change

STCC 14028 change

NCTC 6571

FL, Fluorescence; L, viable cells; D, non-viable cells (heat-treated); D > L, greater fluorescence exhibited by non-viable cells; L > D, greater fluorescence exhibited by viable cells.

312 D.J. MASON ET AL

- '0 50.-

01 a"

DiOC,(3) fluorescence (channel number)

I I I I

DiOC6(3) fluorescence (channel number)

Fig. 3 Single parameter histograms showing the inconsistent fluorescent responses of the carbocyanine stained heat-treated bacteria (using DiOC,(3) in the illustration). Escherichia coli K12 cells were washed and stained in 10 mmol I-' Tris-HC1 pH 7.6 with 0.1 mmol I- ' EGTA. (a) Cells from the control culture (--- ) are showing marginally greater fluorescence than those from the heat-treated culture (- - - -). (b) On another occasion, under similar conditions, cells from the heat-treated culture (- - - -) had a greater fluorescence than those from the control culture (- 1

the discrimination between control and perturbed cells was excellent.

With CFW a greater fluorescent response was obtained from gramicidin-treated bacteria compared to heat-treated bacteria. Possibly the heat treatment destroyed some of the

- 300\ 350

n 5 150

8 100

250

; 200

- -

- 501 k 0

d 0 50 100 150 200 250

DiBAC,(3) fluorescence (channel number)

Fig. 4 A single parameter histogram showing an example of the fluorescent response from gramicidin-treated (- - - -) and untreated bacteria (- ) to DiBAC,(3). Salmonella typhimurium STCC 14028 cells stained in nutrient broth

5 3001

I

250-

CFW fluorescence (channel number)

Fig. 5 A single parameter histogram showing an example of the fluorescent response from gramicidin-treated (- - - -) and untreated bacteria (- ) to calcafluor white. Escherichiu coli K12, stained in nutrient broth

CFW binding sites, or the heat-treated cells were viable but non-culturable. Cells in this state have been found to retain a membrane potential (Kaprelyants and Kell 1992; Davey et al. 1993) and thus may have the ability to exclude the dye due to its negative charge in solution. I f these cells are able to retain a membrane potential, then it is reasonable to believe that the integrity of their membranes is essentially intact and the dye could be excluded from the cells in a similar way to propidium iodide. Alternatively gramicidin- treated cells would have suffered disruption of membrane integrity resulting in an altered membrane potential and thus appeared more brightly stained.

Perturbed bacteria stained with one of the carbocyanine dyes or rh123 were expected to show a reduced fluorescent response. Instead, the carbocyanine dyes exhibited signifi- cant fluorescence with heat-treated and gramicidin-treated cells of all bacterial strains used, suggesting the Occurrence of non-specific energy-independent binding. It is possible that the dyes were binding to alternative cellular binding sites, possibly within the hydrophobic regions of the mem-

Rh123 fluorescence (channel number)

Fig. 6 A single parameter histogram showing an example of the fluorescent response from gramicidin-treated (- - - -) and untreated (- ) bacteria to rhodamine 123 ( 5 mg I-'). Escherichia coli NCTC 10418 cells washed and stained in 10 mmol I- ' Tris-HC1 (pH 7.6) with 0.1 mmol 1-' EGTA

DYES TO ASSESS VIABILITY 313

Table 2 Patterns of response for untreated and gramicidin-treated bacteria stained with rh123

Strain

Staining conditions

0.25 mg I - ' 0.25 mg I - ' rh123: cells 5 mg I - ' rh123 5 mg I - ' rh123 rh123: cells washed and cells suspended cells washed and suspended in suspended in in nutrient suspended in nutrient broth Tris-HC1 broth Tris-HCl*

E. coli G > U G > U U > G U > G

E . coli K12 G > U G > U G > U U > G Salm. typhimurium G > U G > U G > U U > G

Staph. aureus G > U G > U U > G U > G

NCTC 10418

STCC 14028

NCTC 6571

U, Untreated cells; G, gramicidin-treated cells; G > U, greater fluorescence exhibited by gramicidin-treated cells; U > G, greater fluor- escence exhibited by untreated cells. * A typical example of results in this column is shown in Fig. 6.

0 I50 200 250 DiOC,(3) fluorescence (channel number)

DiOC,(3) fluorescence (channel number)

Fig. 7 Single parameter histograms showing the inconsistent fluorescent responses of the carbocyanine stained gramicidin-treated bacteria (using DiOC,(3) in the illustration). Escherichia coli K12 cells were washed and stained in 10 mmol I - ' Tris-HC1 pH 7.6 with 0.1 mmol I - ' EGTA. (a) Cells from the control culture (- ) are more fluorescent than those from the gramicidin-treated culture (- - - -). (b) On another occasion, under similar conditions, cells from the gramicidin-treated culture (- - - -) were more fluorescent than those from the control culture (- 1

brane (Diaper et al. 1992), rather than reacting solely to membrane potential. Carbocyanine dyes have also been known to bind to proteins and have been found to be toxic to bacteria in some cases (Shapiro 1990; Diaper et al. 1992). They can perturb the membrane potential in some types of cell by changing the membrane conductivity and also inhibit energy metabolism (Shapiro 1990). The con- centration of the dyes used may have been too high. A high dye to cell ratio has been found to decrease the partitioning of carbocyanine dyes in mitochondria (Wilson et al. 1985). At a lower concentration (0.05 mg 1-') DiOC,(3) fluores- cence was found to vary in proportion to the membrane potential of E. cola C600 (Mason et al. 1993). However, attempts to use DiOC6(3) and the other carbocyanines at this concentration with the bacterial strains described in this paper resulted in a low fluorescence yield with viable cultures; thus, the higher concentration of 0-1 mg 1-' was used. The increased lipid solubility of DiOC,(3) (conferred by its longer alkyl side chains) could prevent it from being a good indicator of membrane potential.

Non-specific energy-independent binding may also explain why, under all but one, of the staining conditions, perturbed cells stained with rh123 consistently exhibited a higher fluorescence. The lower concentration (0.25 mg 1- ') was chosen to avoid the effects of non-specific binding and of self-quenching. A concentration similar to this was found to give good results for Micrococcus luteus (Kaprelyants and Kell 1992). The higher concentration (5 mg 1-') has been found to give a good response with a number of different species (Matsuyama 1984; Diaper et al. 1992). However, in this study, using rh123 at 0.25 mg 1-' only served to

314 D.J . MASON ET A L .

reduce the background fluorescence ; there was no improve- ment in fluorescent response compared with the higher concentration. The results also indicate that to obtain the expected reduction in fluorescent response from perturbed bacterial cells, stained with rh123, samples must be washed first, which risks possible further cell disruption. It must be emphasized that involved methods of sample preparation which risk possible disturbance of the cells are to be avoided at all costs, especially when studies are undertaken of conditions that produce small changes in cell vitality and viability.

Normally there was cell shrinkage with non-viable cells causing a decrease in forward angle light scatter. However, any changes in forward angle light scatter from heat-treated or gramicidin-treated cells were not as marked as the changes in fluorescence seen with some of the dyes. For this reason only the fluorescence changes were shown in the results.

There is now an increasing awareness that fluorescent responses from dyes of this kind depend not only on the dye but also on the bacterial species employed (Matsuyama 1984). Even within a single bacterial species the response can vary between strains, as illustrated by rhl23-stained strains of E. coli. T h e fluorescent response can also depend on the method of cell perturbation as illustrated by CFW- stained bacteria.

T h e overall conclusion from this study is that DiBAC,(3) is more universally successful a t detecting per- turbation in bacterial cells than other dyes studied in this paper.

ACKNOWLEDGEMENT

The authors wish to thank the Welsh Scheme for the Development of Health and Social Research (Welsh Office) for the financial support of this work.

REFERENCES

Berglund, D.L., Taffs, R.E. and Robertson, N.P. (1987) A rapid analytical technique for flow cytometric analysis of cell viability using calcafluor white M2R. Cytometry 8,421-426.

Briuner, T., Hulser, D.F. and Strasser, R.J. (1984) Comparative measurements of membrane potentials with microelectrodes and voltage-sensitive dyes. Biochimica et Biophysica Acta 771,

Carter, E.A., Paul, F.E. and Hunter, P.A. (1993) The cytometric evaluation of antifungal agents. In Flow Cytometry in Micro- biology ed. Lloyd, D. pp. 11 1-120. London : Springer-Verlag.

Darken, M.A. (1962) Absorption and transport of fluorescent brighteners by microorganisms. Applied Biology 10, 387-393.

Davey, H.M., Kaprelyants, A.S. and Kell, D.B. (1993) Flow cyto-

208-2 I 6.

metric analysis, using rhodamine 123, of Micrococcus luteus at low growth rate in chemostat culture. In Flow Cytometry in Microbiology ed. Lloyd, D. pp. 83-93. London: Springer- Verlag.

Diaper, J.P., Tither, K. and Edwards, C. (1992) Rapid assessment of bacterial viability by flow cytometry. Applied Microbiology and Biotechnology 22, 1-5.

Eddy A.A. (1989) Use of carbocyanine dyes to assay membrane potential of mouse ascites tumor cells. Methods in Enzymology 172,95-101.

Farkas, D.L., Wei, M., Febbroriello, P., Carson, J.H. and Loew, L.M. (1989) Simultaneous imaging of cell and mitochondrial membane potentials. Biophysical Journal 56, 1053-1069.

Hargittai, P.T., Youmans, S.J. and Lieberman, E.M. (1991) Determination of the membrane potential of cultured mamma- lian Schwann cells and its sensitivity to potassium using a thio- carbocyanine fluorescent dye. GIia 4,611-616.

Kaprelyants, AS. and Kell, D.B. (1992) Rapid assessment of bac- terial viability and vitality using rhodamine 123 and flow cyto- metry. Journal of Applied Bacteriology 72,410-422.

Lakos, Z. , Somogyi, B., Balazs, M., Matko, J. and Damjanovich, S. (1990) The effect of transmembrane potential on the dynamic behaviour of cell membranes. Biochimica et Biophysica Acta 1023,4146,

Mason, D.J., Allman, R. and Lloyd, D. (1993) Uses of membrane potential dyes with bacteria. In Flow Cytometry in Microbiology ed. Lloyd, D. pp. 69-81. London: Springer-Verlag.

Matsuyama, T. (1984) Staining of living bacteria with rhodamine 123. FEMS Microbiology Letters 21, 153-157.

Miles, A.A. and Misra, S.S. (1938) The estimation of the bacteri- cidal power of the blood. Journal of Hygiene 38,732-737.

Nikaido, H. and Vaara, M. (1985) Molecular basis of bacterial outer membrane permeability. Microbiological Reviews 49, 1-32.

Ordbiiez, J.V. and Wehman, N.M. (1993) Rapid flow cytometric antibiotic susceptibility assay for Staphylococcus aureus. Cyto- metry 14, 811-818.

Oyama, Y., Chikahisa, L., Tomiyoshi, F. and Hayashi, H. (1991) Cytotoxic action of triphenyltin on mouse thymocytes: a flow cytometric study using fluorescent dyes for membrane potential and intracellular Ca2 + . Japanese Journal of Pharmacology 57, 419-424.

Pena, A., Uribe, S., Pardo, J.P. and Borbolla, M. (1984) The use of a cyanine dye in measuring membrane potential in yeast. Archives of Biochemistry and Biophysics 231, 217-225.

Rhan, C.A., Bombick, D.W. and Doolittle, D.J. (1991) Assess- ment of mitochondria1 membrane potential as an indicator of cytotoxicity. Fundamental and Applied Toxicology 16,435-448.

Ronot, X., Benel, L., Adolph, M. and Mounolou, J. (1986) Mito- chondrial analysis in living cells: the use of rhodamine 123 and flow cytometry. Biology of the Cell 57, 1-8.

Shapiro, H. M. (1990) Cell membrane potential analysis. In Methods in Cell Biology ed. Darzynkiewicz, Z. and Crissman, H. A. Vol. 33, pp. 41 1-426. London, Academic Press Inc.

Sims, J., Waggoner, AS. , Wang, C.H. and Hoffman, J.F. (1974) Studies on the mechanism by which cyanine dyes measure membrane potential in red blood cells and phosphatidylcholine vesicles. Biochemistry 33 15-3327.

Skowronek, P., Krummeck, G., Hoferkamp, 0. and Rodd, G.

DYES T O ASSESS VIABILITY 315

(1990) Flow cytometry as a tool to discriminate respiratory- competent and respiratory-deficient yeast cells. Current Genetics 18,265-267.

Streiblova, E. (1984) The yeast cell w a l l a marker for cell cycle controls. In The Microbial Cell Cycle ed. Nurse, P. and Strei- blova, E. pp. 141-172. Florida: CRC Press.

Wilson, H. A. and Chused, T. M. (1985) Lymphocyte membrane potential and Ca2 + sensitive potassium channels described by oxonol fluorescence measurements. Journal of Cellular Physiol- 00 125,7241.

Wilson, H.A., Seligmann, B.E. and Chused, T.M. (1985) Voltage- sensitive cyanine dye fluorescence signals in lymphocytes : plasma membrane and mitochondria1 components. Journal of Cellular Physiology 125, 61-71.

Zaritsky, A., Kihara, M. and MacNab, R.M. (1981) Measurement of membrane potential in Bacillus subtilis: a comparison of lipo- philic cations, rubidium ion, and a cyanine dye as probes. Journal of Membrane Biology 63, 215-231.