APPLIED AND ENVIRONMENTAL MICROBIOLOGY, July 1998, p. 2681–2685 Vol. 64, No.

7
0099-2240/98/$04.0010
Copyright © 1998, American Society for Microbiology. All Rights Reserved.

A Fluorescent Gram Stain for Flow Cytometry and

Epifluorescence Microscopy
DAVID J. MASON,1 SUBO SHANMUGANATHAN,1 FIONA C. MORTIMER,2
1
AND VANYA A. GANT *

Infection and Immunity Laboratory, United Medical and Dental Schools of Guy’s and
St. Thomas’s Hospitals, London SE1 7EH,1 and Department
of Pharmacy, King’s College London, London SW3 6LX,2 United Kingdom
Received 6 August 1997/Accepted 19 April 1998

The fluorescent nucleic acid binding dyes hexidium iodide (HI) and SYTO 13 were used in combination as
a Gram stain for unfixed organisms in suspension. HI penetrated gram-positive but not gram-negative
organisms, whereas SYTO 13 penetrated both. When the dyes were used together, gram-negative organisms
were rendered green fluorescent by SYTO 13; conversely, gram-positive organisms were rendered red-orange
fluorescent by HI, which simultaneously quenched SYTO 13 green fluorescence. The technique correctly
predicted the Gram status of 45 strains of clinically relevant organisms, including several known to be gram
variable. In addition, representative strains of gram-positive anaerobic organisms, normally decolorized
during the traditional Gram stain procedure, were classified correctly by this method.

Gram’s staining method is considered fundamental in bac-
terial taxonomy. The outcome of the Gram reaction reflects
major differences in the chemical composition and ultrastruc-
ture of bacterial cell walls. The Gram stain involves staining a
heat-fixed smear of cells with a rosaniline dye such as crystal or
methyl violet in the presence of iodine, with subsequent expo-
sure to alcohol or acetone. Organisms that are decolorized by
the alcohol or acetone are designated gram negative.
Alternative Gram staining techniques have recently been
proposed. Sizemore et al. (19) reported on the use of fluores-
cently labeled wheat germ agglutinin. This lectin binds specif-
ically to N-acetylglucosamine in the peptidoglycan layer of
gram-positive bacteria, whereas gram-negative organisms con-
tain an outer membrane that prevents lectin binding. Although
simpler and faster than the traditional Gram stain, this method
requires heat fixation of organisms.
Other Gram stain techniques suitable for live bacteria in
suspension have been described. Allman et al. (1) demonstrat-
ed that rhodamine 123 (a lipophilic cationic dye) rendered
gram-positive bacteria fluorescent, but its uptake by gram-neg-
ative organisms was poor. This reduced uptake by gram-neg-
ative bacteria was attributed to their outer membranes. The
outer membrane can be made more permeable to lipophilic
cations by exposure to the chelator EDTA (4). Shapiro (18)
took advantage of this fact to form the basis of another Gram
stain, one which involved comparing the uptake of a carbo-
cyanine dye before and after permeabilizing organisms with
EDTA. All of these methods, however, rely on one-color flu-
orescence, making analysis of mixed bacterial populations dif-
ficult.
An alternative to the use of stains is the potassium hydroxide
(KOH) test. The method categorizes organisms on the basis of
differences in KOH solubility. After exposure to KOH, gram-
negative bacteria are more easily disrupted than gram-positive
organisms. This technique has been used to classify both aer-
* Corresponding author. Mailing address: Infection and Immunity
Laboratory, United Medical and Dental Schools of Guy’s and St.
Thomas’s Hospitals, Block 9, St. Thomas’s Campus, Lambeth Palace
Road, London SE1 7EH, United Kingdom. Phone: 44 171 928 9292,
ext. 1945. Fax: 44 0171 928 0730. E-mail: v.gant@umds.ac.uk.
obic and facultatively anaerobic bacteria, including gram-vari-
able organisms (8). In a study by Halebian et al. (9), however,
this technique incorrectly classified several anaerobic strains,
giving rise to the recommendation that the method should only
be used in conjunction with the traditional Gram stain.
In this study we demonstrate a Gram staining technique for
unfixed organisms in suspension, by using clinically relevant
bacterial strains and organisms notorious for their gram vari-
ability. The method uses two fluorescent nucleic acid binding
dyes, hexidium iodide (HI) and SYTO 13. Sales literature (11)
publishedbythemanufacturersofHI(MolecularProbes,Inc.,Eu-
gene, Oreg.), which displays a red fluorescence, suggests that
the dye selectively stains gram-positive bacteria. SYTO 13 is
one of a group of cell-permeating nucleic acid stains and flu-
oresces green (11). These dyes have been found to stain DNA
and RNA in live or dead eukaryotic cells (16). Both dyes are
excited at 490 nm, permitting their use in fluorescence instru-
ments equipped with the most commonly available light sources.
We reasoned that a combination of these two dyes applied to
mixed bacterial populations would result in all bacteria being
labeled, with differential labeling of gram-positive bacteria (HI
and SYTO 13) and gram-negative bacteria (SYTO 13 only).
The different fluorescence emission wavelengths of the two
dyes would ensure differentiation of gram-positive from gram-
negative bacteria by either epifluorescence microscopy or flow
cytometry when equipped with the appropriate excitation and
emission filters. While a commercial Gram stain kit produced
by Molecular Probes includes HI and an alternative SYTO
dye, SYTO 9, we are unaware of any peer-reviewed publica-
tions regarding either its use or its effectiveness with tradition-
ally gram-variable organisms.
MATERIALS AND METHODS
Bacterial strains and media. We examined a range of bacterial strains sup-
plied as fresh isolates from clinical specimens by the Department of Microbiol-
ogy, St. Thomas’s Hospital, London, United Kingdom, including Bacillus spp.,
Staphylococcus aureus (n 5 8), Staphylococcus saprophyticus (n 5 3), Enterococ-
cus spp. (n 5 6), Enterococcus faecalis (n 5 5), Acinetobacter spp., Escherichia coli
(n 5 10), Pseudomonas aeruginosa (n 5 4), Proteus spp. (n 5 4), and Klebsiella
spp. (n 5 5). We also studied several anaerobic organisms, including Clostridium
ramosum NCTC 11812 and Proprionibacterium acnes NCTC 0737, which were
obtained directly from the National Collection of Type Cultures, London, United
Kingdom. In addition, William Wade of the Department of Oral Medicine and

2681
2682 MASON ET AL. APPL. ENVIRON. MICROBIOL.

Pathology, Guy’s Hospital, London, United Kingdom, kindly supplied the fol-
lowing anaerobic strains: Eubacterium brachy ATCC 33089, Eubacterium infir-
mum NCTC 12940, Eubacterium timidum ATCC 33093, and Fusobacterium nu-
cleatum NCTC 11326.
Clinical isolates were maintained on Columbia blood agar or cysteine lactose
electrolyte-deficient medium (Unipath). Anaerobic strains were cultured on fas-
tidious anaerobe agar (LabM). Plates were incubated under anaerobic condi-
tions generated with an AnaeroGen (Oxoid) AN25 sachet in a 2.5-liter sealed gas
jar.
Experimental procedure. All isolates were cultured overnight at 37°C in Iso-
Sensitest broth (ISO-B) (Unipath) which had been passed through a 0.2-mm
(pore size) filter. Then 1 ml of each bacterial suspension was removed, pelleted
by spinning at 15,000 3 g for 1 min, washed once in ISO-B, and finally resus-
pended in fresh medium. Mixed bacterial populations were prepared by com-
bining equal volumes of E. coli and S. aureus suspensions. Alcohol-fixed organ-
isms were prepared from cultures of E. coli, Proteus spp., and Klebsiella spp. by
removing an additional 1-ml aliquot from each of these cultures and centrifuging
as before. The pellets were resuspended in 70% ethanol in distilled water for 10
min before being washed and resuspended in ISO-B. The effect of EDTA on HI
uptake by E. coli was investigated by adding 1 mM EDTA during the staining
procedure outlined below.
In the case of anaerobic organisms, single colonies were removed from culture
plates and resuspended in 1 ml of phosphate-buffered saline (pH 7.4) that had
been passed through a 0.2-mm (pore size) filter. These suspensions were incu-
bated with the dyes as described below.
Dyes and staining protocols. HI (5 mg) was dissolved in 1 ml of dimethyl
sulfoxide to give a stock solution of 5 mg/ml, which was further diluted 1:50
(vol/vol) in 10 mM Tris-HCl (pH 7.4) to give a working solution of 100 mg/ml.
SYTO 13 (Molecular Probes, Inc.) was supplied by the manufacturer as a 5 mM
solution in dimethyl sulfoxide, which was diluted 1:10 (vol/vol) in 10 mM Tris-
HCl to give a working solution of 0.5 mM. Bacterial suspensions were incubated
at room temperature with a 1:25 (vol/vol) addition of the SYTO 13 working
solution (to give a final concentration of 20 mM) or with a 1:10 (vol/vol) addition
FIG. 1. Photomicrograph of a mixed microbial population after staining with
of the HI working solution (to give a final concentration of 10 mg/ml) for 2 or 15
HI (10 mg/ml) and SYTO 13 (20 mM). Gram-negative bacteria (E. coli) and
min, respectively. Mixed bacterial suspensions of E. coli and S. aureus were
gram-positive bacteria (S. aureus) fluorescing green and orange-red, respectively,
incubated with both HI and SYTO 13 in the ratios indicated above for 15 min at
are shown.
room temperature.
Traditional Gram stains were carried out with a Gram stain kit (Sigma)
according to the manufacturer’s instructions.
Microscopy. Aliquots (10 ml) of stained bacterial suspensions were placed on
glass slides under cover slips and observed with a Nikon Diaphot (Kingston) taneously with SYTO 13 and HI, however, the green fluores-
inverted epifluorescence microscope fitted with a 100-W mercury arc lamp, cence emission of SYTO 13 was quenched by that of HI (Fig.
Nikon B-2A filter (excitation, 450 to 490 nm; emission, .520 nm) and a 390 oil 1). In contrast, the SYTO 13-associated fluorescence seen in
immersion objective lens. Traditional Gram stains were viewed by bright-field other gram-negative organisms persisted in the presence of HI,
transillumination.
Photomicrographs were obtained with a Nikon FE camera and Kodak Ekta- presumably reflecting the ability of these unfixed organisms to
chrome ISO 400 color slide film. exclude this dye. Finally, fixation of gram-negative organisms
Flow cytometry. Stained suspensions were also analyzed with an enhanced with ethanol rendered them permeable to and fluorescent with
Bryte HS (Bio-Rad) xenon arc lamp-based flow cytometer. The instrument is HI.
equipped with two light scatter detectors (one each for light scattered ,15° and
.15°) and three fluorescence detectors fitted with filter blocks that deliver Flow cytometry. Flow cytometric analysis of organisms
wavelengths with the following specifications, as designated: FL1, 515 to 565 nm; stained with HI or SYTO 13 confirmed microscopy results and
FL2, 565 to 605 nm; and FL3, .605 nm. Excitation wavelength was restricted to allowed quantitative assessment of staining intensity and per-
470 to 490 nm by using the standard fluorescein isothiocyanate filter block, which centage. The intensity of HI-associated fluorescence from
also allowed emission from 520 to 560 nm, with a beam splitter at 510 nm. SYTO
13 (maximum emission at 509 nm) green fluorescence was detected on FL1. HI gram-positive strains and F. nucleatum was at least one log
(maximum emission at 605 nm) orange-red fluorescence was detected primarily order greater than that from other unfixed gram-negative or-
on FL2, with some spectral overlap into FL3. Logarithmic amplification was used ganisms (Fig. 2). Table 1 summarizes flow cytometric results
throughout, and fluorescence acquisition was gated by light scatter parameters. obtained from all strains following incubation with HI. All
Electronic compensation for spectral overlap of the dyes was set in the instru-
ment software, the sample flow rate was set to 2 ml/min, and at least 5,000 strains tested were rendered fluorescent by SYTO 13, in con-
organisms were acquired for analysis. trast to HI, irrespective of Gram staining status (Fig. 3). Data
obtained from a mixed bacterial population (S. aureus and
E. coli in a ratio of 1:1) stained with both SYTO 13 and HI are
RESULTS
shown in Fig. 4. Figure 4 shows that while distinguishing be-
Microscopy. All clinical isolates (with the exception of Acin- tween these two bacterial species by light scatter parameters
etobacter spp.) were stained correctly as defined by cell wall alone is difficult if not impossible (Fig. 4A), the two microbial
structure by the traditional Gram stain technique. The Acin- populations can be clearly separated on the basis of differential
etobacter spp. gave a gram-positive result. In addition, C. ramo- fluorescence wavelength (Fig. 4B).
sum, E. infirmum, E. timidum, and P. acnes had a “washed out” Ethanol-fixed gram-negative bacteria fluoresced with HI to
appearance that precluded classification. an intensity equivalent to that of unfixed gram-positive organ-
Fluorescence microscopy revealed that all of the gram-pos- isms. Exposure of E. coli to EDTA (1 mM) induced the uptake
itive organisms and the traditionally gram-negative organism of HI in some (58%) organisms.
F. nucleatum fluoresced bright orange-red when stained with
HI, whereas no fluorescence was observed with other unfixed DISCUSSION
gram-negative organisms with this dye. In contrast, all strains
fluoresced bright green in the presence of SYTO 13. When We have found the HI-SYTO 13 dye combination to be
gram-positive organisms and F. nucleatum were stained simul- effective, allowing the rapid identification and enumeration of
VOL. 64, 1998 FLUORESCENT GRAM STAIN FOR FLOW CYTOMETRY 2683

TABLE 1. Data obtained from flow cytometric analysis of

all bacterial strains after incubation with HI
Gram-positive strain % Gram-negative strain %
(sample no.) Staineda (sample no.) Staineda

Bacillus spp. 90.0 Acinetobacter spp. 1.2

C. ramosum NCTC 11812 93.6 E. coli (1) 9.7
Enterococcus spp. (1) 95.8 E. coli (2) 13.3
Enterococcus spp. (2) 87.0 E. coli (3) 14.6
Enterococcus spp. (3) 98.3 E. coli (4) 10.5
Enterococcus spp. (4) 95.1 E. coli (5) 6.9
Enterococcus spp. (5) 94.0 E. coli (6) 14.4
Enterococcus spp. (6) 92.7 E. coli (7) 4.9
E. faecalis (1) 99.1 E. coli (8) 8.3
E. faecalis (2) 98.4 E. coli (9) 14.5
E. faecalis (3) 99.7 E. coli (10) 13.0
E. faecalis (4) 86.4 F. nucleatum 89.0
E. faecalis (5) 99.8 Klebsiella spp. (1) 2.1
E. infirmum NCTC 12940 86.8 Klebsiella spp. (2) 8.8
E. brachy ATCC 33089 89.7 Klebsiella spp. (3) 6.1
E. timidum ATCC 33093 97.2 Klebsiella spp. (4) 3.0
P. acnes NCTC 0737 97.4 Klebsiella spp. (5) 4.8
S. aureus (1) 85.4 Proteus spp. (1) 7.4
S. aureus (2) 95.5 Proteus spp. (2) 1.0
S. aureus (3) 80.0 Proteus spp. (3) 2.1
S. aureus (4) 93.1 Proteus spp. (4) 2.7
FIG. 2. Comparison of a gram-negative culture and a gram-positive culture
stained with HI (10 mg/ml). Data are displayed as flow cytometric histograms of S. aureus (5) 87.9 P. aeruginosa (1) 7.2
5,000 bacterial events in which the axes represent the relative number of cells S. aureus (6) 97.7 P. aeruginosa (2) 13.0
(y axis) and the cell-associated fluorescence on a logarithmic scale (x axis). h, S. aureus (7) 93.8 P. aeruginosa (3) 15.2
Gram-negative culture (Proteus spp.); ■, gram-positive culture (Enterococcus S. aureus (8) 96.0 P. aeruginosa (4) 17.9
spp.). The marked region represents stained organisms, i.e., those with a fluo- S. saprophyticus (1) 90.4
rescence intensity above the first decade. These histograms are typical of all of S. saprophyticus (2) 95.4
the gram-negative and gram-positive strains tested. S. saprophyticus (3) 70.0
a
The percentage of stained organisms refers to the proportion of bacterial
cells with a fluorescence intensity above the first decade (see Fig. 2).
gram-negative and gram-positive bacteria in suspension. Fur-
thermore, a washing step following labeling is not necessary, also been noted for propidium iodide (6) and SYTOX green
and the technique is equally applicable to flow cytometry and (17). We believe that the HI staining of the traditionally gram-
epifluorescence microscopy. Our technique also correctly clas- negative organism F. nucleatum reflects its recent classifica-
sified those organisms that were either incorrectly or poorly tion to a cluster of low-GC gram-positive organisms (15). This
stained by the traditional Gram stain technique. Thus, Acineto-
bacter spp. were incorrectly characterized as gram-positive
organisms by the traditional method due to incomplete decol-
orization. This phenomenon has been related to the use of
“stabilized” polyvinylpyrrolidone iodine as a mordant rather
than iodine in potassium iodide solution (10). The mordant
used in our Gram stain kit, however, consisted of iodine
(0.33% [wt/vol]) in potassium iodide (0.66% [wt/vol]). False-
gram-positive staining of Acinetobacter spp. has been associ-
ated with delayed diagnosis and inappropriate antibiotic ther-
apy (7, 20). All of the anaerobic gram-positive organisms in
this study were characteristically decolorized (5, 12) dur-
ing the traditional Gram stain procedure. Using a modified
Gram stain to study the staining reaction by electron micros-
copy, Beveridge (2) demonstrated that cytoplasmic voids were
formed close to the cell wall and septation sites in P. acnes,
causing the cells to lyse and appear gram negative. Clostridium
spp., on the other hand, were found to alter their conformation
with time, with thinning of the peptidoglycan cell wall towards
late-exponential-growth phase, presumably explaining their in-
ability to retain the crystal violet.
Exposure of gram-negative bacteria to EDTA destabilizes
bacterial lipopolysaccharides (LPS) after removal of essential
cations (13–14). Such events may account for the HI uptake
observed in EDTA-treated E. coli, suggesting that the LPS-rich
outer membrane may play a part in the exclusion of the dye FIG. 3. Comparison of stained and unstained cultures of S. aureus. The dye
from these organisms. Alcohol fixation similarly abolishes the used was SYTO 13 (20 mM). The data are displayed as flow cytometric histo-
grams of 5,000 bacterial events in which the axes represent the relative number
ability of gram-negative organisms to exclude HI, suggest- of cells (y axis) and the cell-associated fluorescence on a logarithmic scale
ing that this dye provides an indication of outer membrane (x axis). h, Unstained culture; ■, stained culture. These histograms are typical of
integrity in gram-negative organisms. Such properties have all of the strains tested.
2684 MASON ET AL. APPL. ENVIRON. MICROBIOL.

FIG. 4. Dual-parameter histograms of 5,000 bacterial events acquired from a mixed population of gram-negative (E. coli) and gram-positive (S. aureus) bacteria
stained with HI (10 mg/ml) and SYTO 13 (20 mM). (A) Data accumulated from light scatter parameters, where the axes represent side-angle scatter (y axis) and
forward-angle scatter (x axis). (B) Data accumulated from fluorescence parameters, where the axes represent cell-associated HI fluorescence (y axis) and cell-associated
SYTO 13 fluorescence (x axis). The percentages relate to the proportion of particles found in each quadrant.

discrepant finding between Gram and HI staining suggests that
quite different mechanisms are responsible for the interaction
of either crystal violet or HI with bacteria. Unlike Gram stain-
ing, staining with HI may be reflecting the “true” Gram status
as determined by sequence relatedness studies, and this occurs
irrespective of the close structural relationship of the lipid A
fragment of F. nucleatum LPS to that of the Enterobacteriaceae
(3).
This staining technique has potential for widespread appli-
cation in clinical and environmental microbiology. Its useful-
ness will depend on whether this dye combination performs
equally well with other organisms of clinical or environmental
relevance, which still needs to be determined before its unre-
stricted application. Once this reservation has been addressed
and overcome, the method could conceivably be adapted for
the rapid and possibly automated detection and assessment of
VOL. 64, 1998 FLUORESCENT GRAM STAIN FOR FLOW CYTOMETRY
gram reactivity of bacteria present in liquid samples from in-
dustrial, environmental, and clinical sources.
ACKNOWLEDGMENTS
We thank Gary French for his financial support and without whom
we would have had no ultrapure water and UMDS for the laboratory
space. This work was not supported by any charitable organization.
REFERENCES
1. Allman, R., R. Manchee, and D. Lloyd. 1993. Flow cytometric analysis of
heterogeneous bacterial populations, p. 27–47. In D. Lloyd (ed.), Flow cy-
tometry in microbiology. Springer-Verlag, London, England.
2. Beveridge, T. J. 1990. Mechanism of Gram variability in select bacteria.
J. Bacteriol. 172:1609–1620.
3. Dahlen, G., and I. Mattsby-Baltzer. 1983. Lipid A in anaerobic bacteria.
Infect. Immun. 39:466–468.
4. Davey, H. M., and D. B. Kell. 1996. Flow cytometry and cell sorting of
heterogeneous microbial populations: the importance of single cell analyses.
Microbiol. Rev. 60:641–696.
5. Esteban, J., G. Garcı́a-Calvo, P. Jiménez-Castillo, and F. Soriano. 1996.
Failure of Gram stain to detect Propionibacterium acnes in specimens from
clinically significant infections. J. Clin. Microbiol. 34:2051.
6. Gant, V. A., G. Warnes, I. Phillips, and G. F. Savidge. 1993. The application
of flow cytometry to the study of bacterial responses to antibiotics. J. Med.
Microbiol. 39:147–154.
7. Goodhart, G. L., E. Abrutyn, R. Watson, R. K. Root, and J. Egert. 1977.
Community-acquired Acinetobacter calcoaceticus var. anitratus pneumonia.
JAMA 238:1516–1518.
8. Gregerson, T. 1978. Rapid method for distinction of gram-negative from
2685
gram-positive bacteria. Eur. J. Appl. Microbiol. Biotechnol. 5:123–127.
9. Halebian, S., B. Harris, S. M. Finegold, and R. D. Rolfe. 1981. Rapid method
that aids in distinguishing gram-positive from gram-negative anaerobic bac-
teria. J. Clin. Microbiol. 13:444–448.
10. Harrington, B. J. 1997. Letter. Clin. Microbiol. Newsl. 19:191.
11. Haughland, R. P. 1996. Nucleic acid detection, p. 143–168. In M. T. Z.
Spence (ed.), Handbook of fluorescent probes and research chemicals. Mo-
lecular Probes, Inc., Eugene, Oreg.
12. Johnson, M. J., E. Thatcher, and M. E. Cox. 1995. Techniques for controlling
variability in Gram staining of obligate anaerobes. J. Clin. Microbiol. 33:
755–758.
13. Nikaido, H., and M. Vaara. 1985. Molecular basis of outer membrane per-
meability. Microbiol. Rev. 49:1–32.
14. Nikaido, H., and M. Vaara. 1987. Outer membrane, p. 7–22. In F. Neidhardt,
J. Ingraham, K. B. Low, B. Magasanik, M. Schaechter, and H. E. Umbarger
(ed.), Escherichia coli and Salmonella typhimurium: cellular and molecular
biology. American Society for Microbiology, Washington, D.C.
15. Olsen, G. J., C. R. Woese, and R. Overbeek. 1994. The winds of (evolution-
ary) change: breathing new life into microbiology. J. Bacteriol. 176:1–6.
16. Poot, M., L. L. Gibson, and V. L. Singer. 1997. Detection of apoptosis in live
cells by MitoTracker red CMXRos and SYTO dye flow cytometry. Cytom-
etry 27:358–364.
17. Roth, B. L., M. Poot, S. T. Yue, and P. J. Millard. 1997. Bacterial viability
and antibiotic susceptibility testing with SYTOX green nucleic acid stain.
Appl. Environ. Microbiol. 63:2421–2431.
18. Shapiro, H. M. 1995. Practical flow cytometry, 3rd ed., p. 367–425. Alan R.
Liss, Inc., New York, N.Y.
19. Sizemore, R. K., J. J. Caldwell, and A. S. Kendrick. 1990. Alternate Gram
staining technique using a fluorescent lectin. Appl. Environ. Microbiol. 56:
2245–2247.
20. Verghese, A., and S. L. Berk. 1983. Bacterial pneumonia in the elderly.
Medicine 62:271–285.