Flow-cytometric study of vital cellular functions in
Escherichia coli during solar disinfection (SODIS)
Michael Berney, Hans-Ulrich Weilenmann and Thomas Egli
Swiss Federal Institute of Aquatic Science and Technology (EAWAG), PO 611, CH-8600
Du ¨bendorf, Switzerland
Received 21 October 2005
8 February 2006
Accepted 20 February 2006
The effectiveness of solar disinfection (SODIS), a low-cost household water treatment method for
developing countries, was investigated with flow cytometry and viability stains for the enteric
bacterium Escherichia coli. A better understanding of theprocess of injury or death of E. coli during
SODIS could be gained by investigating six different cellular functions, namely: efflux pump activity
(Syto 9 plus ethidium bromide), membrane potential [bis-(1,3-dibutylbarbituric acid)trimethine
oxonol; DiBAC4(3)], membrane integrity (LIVE/DEAD BacLight), glucose uptake activity
(2-[N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino]-2-deoxy-D-glucose; 2-NBDG), total ATP
in E. coli K-12 MG1655 cells that were exposed to either sunlight or artificial UVA light. The
inactivation pattern of cellular functions was very similar for both light sources. A UVA light dose
(fluence) of <500 kJ m”2was enough to lower the proton motive force, such that efflux pump
activity and ATP synthesis decreased significantly. The loss of membrane potential, glucose uptake
activity and culturability of >80% of thecells was observed at a fluence of ~1500 kJ m”2, and the
cytoplasmic membrane of bacterial cells became permeable at a fluence of >2500 kJ m”2.
Culturable counts of stressed bacteria after anaerobic incubation on sodium pyruvate-
supplemented tryptic soy agar closely correlated with the loss of membrane potential. The results
strongly suggest that cells exposed to >1500 kJ m”2solar UVA (corresponding to 530 W m”2
global sunlight intensity for 6 h) were no longer able to repair the damage and recover. Our
picture of the ‘agony’ of E. coli when it is stressed with sunlight.
Waterborne diarrhoeal diseases are prevalent in many
countries where sewage and drinking water treatment
are inadequate. Over 1?2 billion people are at risk because
they lack access to safe drinking water (World Health
Organization, 1996). Every day, diarrhoeal diseases such
as cholera (Fewtrell et al., 2005) claim the lives of ~5000
young children throughout the world and most of the
cases could be easily prevented (Hinrichsen et al., 1997).
The World Health Organization (WHO) and the United
Nations Children’s Fund (UNICEF) have recently claimed
that improvement of drinking water quality and basic
sanitation can cut this toll, and simple, low-cost, house-
hold water treatment has the potential to save further
lives because it cuts the primary transmission route for
diarrhoeal diseases (World Health Organization/United
Nations Children’s Fund, 2005).
Solar disinfection (SODIS) is such a water treatment
method. Through exposure of drinking water in poly(ethy-
bacteria in the water are inactivated (Acra et al., 1984;
Wegelin et al., 1994). Both UVA light and mild heat have
been shown to have inimical potential and, if the water
(Wegelin et al., 1994). Field trials in different geographic
regions, carried out by the Swiss Federal Institute of Aquatic
Science (EAWAG) Department of Water and Sanitation
in Developing Countries (SANDEC), have shown that
temperatures >50uC are rarely reached (R. Meierhofer,
personal communication). Acra et al. (1984) have pro-
posed that solar UVA irradiation accounts for >70% of
the negative effects of sunlight. Today, SODIS is one of
the recommended methods for household drinking water
disinfection (World Health Organization/United Nations
Children’s Fund, 2005). Nevertheless, the mechanism(s)
of disinfection are not yet known precisely. The two pri-
mary sources for bacterial inactivation in this method are
Abbreviations: 2-NBDG, 2-[N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino]-
2-deoxy-D-glucose; DiBAC4(3), bis-(1,3-dibutylbarbituric acid)trimethine
oxonol; EB, ethidium bromide; PEP, phosphoenolpyruvate; PEP-PTS,
phosphoenolpyruvate-phosphotransferase system; PI, propidium iodide;
SODIS, solar disinfection.
0002-8617 G 2006 SGM Printed in Great Britain1719
Microbiology (2006), 152, 1719–1729
believed to be mild heat and UVA light (Wegelin et al.,
1994). The first report on the inimical effect of sunlight on
bacteria was published by Downes (1886) and, 100 years
later, Acra et al. (1984) proposed the use of sunlight to
disinfect oral rehydration solutions. Subsequently, it was
demonstrated to be effective against bacteria and higher
organisms (Joyce et al., 1996; Lonnen et al., 2005; Wegelin
et al., 1994), and its applicability has been shown in a recent
health impact study (Hobbins, 2004). The effectiveness of
SODIS has been shown using classical cultivation of
micro-organisms on solid agar substrates, which repre-
sents the reference procedure for estimating the number of
viable bacteria. This method, however, only enumerates
culturable bacteria that are able to initiate cell division and
replication on agar plates. By using conventional cultur-
ability assays, there is a risk of overestimation of effec-
tiveness of disinfection processes. For example, bacteria
that are injured can fail to proliferate on plates and may
appear dead (Aldsworth et al., 1999). It has been shown
that injured bacteria do not grow due to the stress of high
nutrient concentrations, which causes a free radical burst
(Dodd et al., 1997). This behaviour has been referred to as
a suicide response (Aldsworth et al., 1999). Postgate had
already observed this behaviour much earlier and called it
substrate-accelerated death (Postgate & Hunter, 1964), but
did not link it to oxidative stress. Furthermore, in view of
the discussion about the viable but non-culturable (VBNC)
state, a careful analysis of bacterial viability after SODIS
treatment is of great importance (Barer & Harwood, 1999;
such a state exists at all and how this can be proven.
Multi-parameter flow cytometry has enjoyed increasing
popularity in microbiology, particularly in biotechnolo-
gical processing, food preservation and chemical disin-
fection (Hewitt et al., 1999; Nebe-von-Caron et al., 2000;
Porter et al., 1997). Hewitt & Nebe-Von-Caron (2004) have
shown that multi-parameter flow cytometry allows a func-
tional classification of the physiological state of single-celled
micro-organisms beyond that of culturability alone.
In the work presented here, the SODIS method was assessed
with flow cytometry and viability stains. A set of essential
bacterial functions of E. coli, namely, membrane integrity,
membrane potential, efflux pump activity, glucose uptake
activity and culturability were measured.
Bacterial strains. Wild-type E. coli K-12 MG1655 (ATCC 700926)
was used for all experiments.
Growth media and cultivation conditions. Luria–Bertani (LB)
broth (10 g tryptone l21, 5 g yeast extract l21, 10 g NaCl l21) that
was filter-sterilized with membrane filters (Millex GP, 0?22 mm pore
size; Millipore), and diluted to 33% (v/v) of its original strength
(unless indicated otherwise) with ultrapure water (deionized and
activated carbon-treated), was used for batch cultivation (Miller,
1972). Precultures were prepared for each individual batch experi-
ment from the same cryo-vial stored at 280uC by streaking the
stock solution onto LB agar plates. After 15–18 h of incubation at
37uC, one colony was picked, loop-inoculated into a 125 ml
Erlenmeyer flask containing 20 ml diluted LB broth, and incubated
at 37uC on a rotary shaker at 200 r.p.m. At an OD546between 0?1
and 0?2 (measured spectrophotometrically at 546 nm in glass
cuvettes with a 1 cm light path using a Jasco V550 UV/VIS spectro-
photometer), an aliquot of the culture was transferred into 500 ml
Erlenmeyer flasks containing 50 ml prewarmed LB broth to obtain
an OD546of 0?002. With this procedure, no lag phase was observed.
These flasks were then shaken at 200 r.p.m. in a temperature-
controlled water bath (SBK 25D; Salvis AG) at 37uC for ~18 h
until stationary phase (specific growth rate, m=0 h21) was reached.
The specific growth rate m was calculated from five consecutive
OD546measurements. In drinking water, cells grow very slowly or
not at all; therefore, we used stationary-phase cells, which were
shown to be more resistant to SODIS than cells in the growth phase
(Berney et al., 2006; Reed, 1997).
Sample preparation and plating. Cells were harvested by centri-
fugation from batch culture (at 13000 g, Biofuge Fresco; Kendro),
washed three times with filter-sterilized (Nuclepore Track-Etch
Membrane, 0?22 mm pore size; Sterico) commercially available
bottled water (Evian) and diluted to an OD546of ~0?01 (corre-
sponding to 1–56107cells ml21). During exposure, aliquots were
withdrawn at different time points and diluted in decimal steps
(1021–1025) with sterile-filtered (0?22 mm), bottled mineral water
(Evian). Volumes of 1 ml of appropriate dilutions were withdrawn
and mixed with 7 ml liquid tryptic soy agar (TSA) (Biolife) at 40uC
(pour-plate method). After 20 min, the solidified agar was covered
with another 4 ml liquid TSA (40uC). Plates were incubated for
48 h at 37uC until further analysis. For cultivation under reduced
oxidative stress conditions, a 1 ml aliquot of the appropriate test
solution was mixed with 7 ml TSA supplemented with 0?05%
sodium pyruvate, covered after 5 min with the same agar, and then
placed in an anaerobic jar (Oxoid; Khaengraeng & Reed, 2005). Just
before closure of the jars, an AnaeroGen sachet (Oxoid) was placed
in the container to generate an anaerobic gas phase. The jars were
incubated for up to 72 h at 37uC. Plate counts were determined
with an automatic plate reader (Acolyte; Synbiosys).
Sunlight exposure. Samples of 10 ml bacterial suspension (see
above) were exposed to solar light in 30 ml quartz tubes, which
were placed into a temperature-controlled, acrylic glass container
with a quartz front glass, holding 25 tubes in total. A water bath was
used to control the temperature of the sample tubes in the con-
tainer. The container was adjusted regularly so that sunlight met the
tubes at an angle of 90±2u. At each time point, one tube was with-
drawn and its content immediately processed as described above.
Irradiation intensity data were obtained from a weather station,
which was located 300 m away from the exposure site (BUWAL/
NABEL, EMPA Du ¨bendorf, Switzerland). The fluence rates for sun-
light irradiation given in this work refer to the wavelength range
350–450 nm, which reflects that of the UVA lamp used (see below).
Conversion factors and calculations were used as described by
Wegelin et al. (1994). The light spectra were recorded with a cali-
brated LI-1800 portable spectroradiometer (LI-COR), 8 nm band-
width, fitted with a model 1800-10 detector head.
UVA exposure. Samples of 10 ml bacterial suspension (see above)
were exposed to UVA light in 30 ml quartz tubes placed in a carou-
sel reactor (holding 10 tubes) (adapted from Wegelin et al., 1994)
equipped with a medium-pressure mercury lamp (Hanau TQ718
Z4), which was operated at 500 W (for wavelength spectrum, see
Berney et al., 2006). The lamp was placed in a cooling jacket (Duran
50 borosilicate glass) in the centre of the carousel reactor. The light
emitted from the lamp passed through the glass jacket and through
35 mm of filter solution before reaching the cells in the quartz
tubes. The temperature of the filter solution was maintained at
1720 Microbiology 152
M. Berney, H.-U. Weilenmann and T. Egli
37uC, and the solution consisted of 12?75 g sodium nitrate l21with
a cut-off at 320 nm and a half maximum at 340 nm. The transmis-
sion property of the filter solution was measured before each experi-
ment. Chemical actinometry with p-nitroanisole/pyridine was used
to determine the fluence rate at the tube position (Wegelin et al.,
1994). Bacterial solutions were mixed intermittently on a magnetic
stirrer. At each time point, one tube was withdrawn and its contents
immediately processed as described above. UVA and light spectra
were recorded as outlined above.
Flow-cytometric measurements. Flow-cytometric measurements
were made using a Partec PAS III flow cytometer with 488 nm excita-
tion from an argon ion laser at 20 mW. Five fluorescent dyes were
used alone or in different combinations: Syto 9 (Molecular Probes),
propidium iodide (PI; Molecular Probes), bis-(1,3-dibutylbarbituric
acid)trimethine oxonol [DiBAC4(3); Molecular Probes], ethidium
bromide (EB; Fluka Chemie) and 2-[N-(7-nitrobenz-2-oxa-1,3-diazol-
taken from irradiation experiments (sunlight and artificial UVA) were
divided into five subsamples and immediately stained with two mix-
tures of fluorescent dyes (Syto 9/PI and Syto 9/EB) and three single
fluorescent dyes [DiBAC4(3), Syto 9 and 2-NBDG]. Samples were
incubated in the dark at 37uC for 5 min (2-NBDG) or at 20uC
for 10 [DiBAC4(3)], 15 (Syto 9/EB), 20 (Syto 9/PI) and 25 min
(Syto 9), respectively, before analysis. Prior to flow-cytometric
analysis, cell samples were diluted with sterile-filtered bottled water
(Evian) to 1% (v/v) of the initial concentration (~16105cells ml21
final concentration). Stock solutions of the dyes were prepared as fol-
lows: PI and Syto 9 were used from the LIVE/DEAD BacLight kit
(Invitrogen), EB was prepared at 25 mM in distilled and filtered
water, DiBAC4(3) was prepared in dimethylsulfoxide at 10 mM, and
2-NBDG was dissolved in distilled and filtered water at 5 mM. All
stock solutions were stored at 220uC. The working concentrations of
Syto 9, PI, EB, DiBAC4(3) and 2-NBDG were 5, 30, 30, 10 and 5 mM,
respectively. 2-NBDG was added in combination with 2,4-dinitrophe-
nol (final concentration 2 mM) (Natarajan & Srienc, 2000). In the
flow cytometer, optical filters were set up such that PI and EB were
measured above 590 nm and Syto 9, DiBAC4(3) and 2-NBDG at
520 nm. The trigger was set for the green fluorescence (520 nm) chan-
The 2-NBDG uptake kinetics of E. coli K-12 MG1655 were measured
prior to the irradiation experiment by incubating a bacterial sample
with a mixture of 2-NBDG and 2,4-dinitrophenol at 37uC for 40 min
(Natarajan & Srienc, 1999). Every 5 min, a sample was withdrawn,
diluted with sterile-filtered mineral water (Evian) to 1% (v/v) of
of the bacterial population were gated (range-gate RN1) and the
geometric mean of green fluorescence intensity was used for data
Fluorescence stains and their function. The cellular properties
and physiological functions indicated by the fluorescent stains
used here are visualized in Fig. 1. Syto 9, a green fluorescent nucleic
Fig. 1. Viability indicators (fluorescence stains) applied in combination with flow cytometry and their function in E. coli
(simplified). All stains have to pass the cytoplasmic membrane to be detected by flow cytometry. Functions: Syto 9 (green
fluorescence) for total counts; propidium iodide (red) for membrane integrity; ethidium bromide (red) for efflux pump activity;
2-NBDG (green) for glucose uptake activity (PEP-PTglc, the PTS component for glucose transport); DiBAC4(3) for membrane
Flow cytometry of E. coli inactivated by sunlight
acid stain, has been shown to stain living and dead Gram-positive
and Gram-negative bacteria (Haugland, 2002). Therefore, it can be
used for total count measurements. EB binds to nucleic acid, and
can cross the intact cytoplasmic membrane, but is actively pumped
out of the cell via a non-specific, proton antiport transport system
in active cells (Midgley, 1987). It has been shown that the ethidium
efflux system is dependent on a transmembrane proton electroche-
mical gradient, but is ATP independent. PI is a red fluorescent dye
that intercalates with dsDNA and only enters permeabilized cyto-
plasmic membranes. If EB or PI is combined with Syto 9, a quench-
ing effect on the green fluorescence intensity is observed as soon as
the red fluorescent dye enters the cells. It is important to mention
that only the cation (ethidium or propidium) enters the cells; never-
theless, the stains are always referred to together with their anionic
constituent (bromide or iodide). DiBAC4(3) is a lipophilic and anio-
nic bis-oxonol. The uptake of this membrane potential-sensitive dye
is restricted to depolarized cells or cells with disrupted cytoplasmic
membranes. The fluorescent dye accumulates inside the cells by
binding to intracellular proteins and membranes. For further infor-
mation about the dyes described above, refer to Hewitt & Nebe-
Von-Caron (2004). 2-NBDG is a fluorescent glucose analogue,
which is taken up by the glucose-specific phosphoenolpyruvate-
phosphotransferase system (PEP-PTS) (Natarajan & Srienc, 1999;
Yoshioka et al., 1996). The system consists of five proteins and uses
PEP as an energy source.
Total ATP. For the determination of total ATP, the BacTiter-Glo
system (Promega) was used. The BacTiter-Glo buffer was mixed
with the lyophilized BacTiter-Glo substrate and equilibrated at room
temperature. The mixture was stored overnight at room temperature
to ensure that all ATP was hydrolysed (‘burned off’) and the back-
ground signal had decreased. A cell suspension of 100 ml was mixed
in a 2 ml Eppendorf tube with an equal volume of the previously
prepared BacTiter-Glo reagent (stored on ice). The sample was then
briefly vortexed and put into a water bath at 37uC for 30 s. The
luminescence of the sample was measured in a luminometer (model
TD-20/20; Turner BioSystems) immediately after incubation. A cali-
bration curve with dilutions of pure rATP (Promega, P1132) was
measured before each experiment. ATP concentration per cell
was then calculated using this calibration curve and the total count
measurements (Syto 9) from flow cytometry.
ent cellular functions in stationary-phase E. coli cells and
compared it to artificial UVA light. The application of
multi-parameter flow cytometry allows the measurement
of cellular functions at the single-cell level (as opposed to
bulk parameters), thus, one can acquire a more detailed
picture of the physiological state of bacterial cells than that
with plating alone.
Culturability and total counts
Three independent experiments were conducted to measure
all cellular functions during artificial UVA irradiation. Cul-
turability on TSA was measured in all three experiments
to ensure comparability of results. The inactivation based
on this parameter was reproducible (Fig. 2). Exposure
to sunlight was repeated twice (on two different days)
while, in the experiment presented here, triplicate measure-
ments (three independent cultures were grown, diluted
and exposed to the same sunlight) of culturability were
made (Fig. 3). To prevent oxidative bursts in the cells
during aerobic incubation, samples were also incubated
on pyruvate-supplemented agar in anaerobic jars, which
resulted in higher c.f.u. In all experiments, total counts
remained constant over the whole irradiation period
(Fig. 4a). UVA light appeared to be slightly more effective
in inactivating E. coli cells than sunlight when assessed with
plating (Fig. 4b).
ATP and efflux pump activity
Soon after the start of exposure (400–600 kJ m22) to either
UVA light or sunlight, the ATP concentration per cell
dropped significantly and levelled out to ~5% of the initial
value (Fig. 4c). The uptake of EB (non-pumping cells)
closely followed the decline in ATP concentration per cell
(Fig. 4d). EB-positive cells (non-pumping cells) showed a
decrease in green fluorescence intensity (quenching effect)
and an increase in red fluorescence intensity on the flow
cytometer (Fig. 5a–c). The loss of culturability of bacterial
cells appeared to be time-delayed when compared to the
decrease in ATP.
Membrane potential and membrane integrity
The loss of membrane potential, which was measured with
the green fluorescent bis-oxonol DiBAC4(3), was very simi-
lar for both light sources (Fig. 4e). In the flow cytometer,
E. coli cells with polarized cytoplasmic membranes did not
appear, because the signal trigger was set on the green
fluorescence channel (FL1,520 nm;Fig. 5d),therefore, only
depolarized (green fluorescent) cells were counted (Fig. 5e,
f). The percentage of depolarized cells was calculated
using the total count measurements. In our experiment,
depolarization of the bacterial cells occurred later than
Fig. 2. Inactivation curves of E. coli K-12 MG1655 exposed to
artificial UVA light in three independent experiments (m, &, X).
Bacterial cells were harvested from a stationary-phase LB batch
culture, washed and diluted in mineral water (Evian). Initial cell
numbers in these experiments were 1?7–1?96107cells ml”1
Culturability of bacterial cells was measured with the pour-plate
method using TSA.
1722 Microbiology 152
M. Berney, H.-U. Weilenmann and T. Egli
the cessation of efflux pump activity but before mem-
brane integrity was lost (Fig. 4f). PI, a dye which only enters
cells with permeabilized cytoplasmic membranes, was
only detected after a dose of 2000–2500 kJ m22had been
applied. The permeabilized cell population showed a typical
increase in red fluorescence and decrease in green fluor-
escence intensity because of quenching (Fig. 5a–i). Inter-
estingly, green fluorescence intensity increased by a factor
of 4?3 before PI was able to enter the cells significantly.
At the end of the sunlight experiment, after a fluence
of 2088 kJ m22, ~50% of the cell population had lost
membrane integrity. From the UVA light experiment, we
learned that, after ~2500 kJ m22of UVA light, most of
the E. coli cell population (>95%) had lost membrane
integrity. Artificial UVA light seemed to be slightly more
effective in permeabilizing E. coli cells than sunlight.
Glucose uptake activity
The 2-NBDG uptake kinetics of healthy E. coli cells followed
a typical pattern (Fig. 6a). The mean green fluorescence
intensity of the cells increased linearly with time during
the first 10 min of incubation with 2-NBDG. Therefore,
irradiated cells were always incubated for 5 min to ensure
that the mean fluorescence intensity correlated with the
uptake rate. The mean 2-NBDG uptake rate of the cells
remained constant over the first 2 h of sunlight exposure
(800 kJ m22), thereafter, it dropped along with the num-
ber of fluorescent cells, indicating that the PEP-PTS sys-
tem was compromised (Fig. 6b). Bacterial cell size did not
change significantly during exposure, as was indicated by
the forward angle light scatter (data not shown). This loss
of the ability to take up glucose correlated well with the
loss of membrane potential.
The data acquired during artificial UVA irradiation for the
different cellular functions strongly resembled those of the
sunlight experiments. Although the light intensity applied
in the laboratory system was much higher than that during
sunlight exposure, the time-dependent pattern of the differ-
ent viability stains was similar. With the UVA lamp it was
possible to apply a higher fluence, which eventually led
to the loss of membrane integrity (PI staining). This was
the most apparent and essential difference between the two
different light sources.
Syto 9 staining characteristics
The analysis of membrane integrity with the Syto 9/PI mix-
ture during UVA irradiation revealed a unique fluorescence
pattern for the bacterial cell population (Fig. 5g–i). On a
two-dimensional dot plot, the whole population moved in
a circle from low green and red fluorescence intensity
(Fig. 5g) to an ~4?3-fold increased green fluorescence
intensity (Fig. 5h), and then to a state of increased red
and decreased green fluorescence intensity (Fig. 5i). The
increase of green fluorescence intensity coincided with the
loss of culturability of >97% of the cell population and,
only after prolonged irradiation, was PI able to enter the
cells and quench Syto 9 fluorescence. Hence, membrane
integrity was lost long after the culturability of the popu-
lation dropped below the detection limit. An increase in
green fluorescence intensity was also observed when E. coli
was exposed to sunlight. In these experiments, the geo-
metricalmeanofgreen fluorescenceintensityofstained cells
increased ~3?7-fold, while culturability was lost by >98%
of the population. This observation is probably related to
an increased uptake of Syto 9. This was tested by using the
Syto 9/PI mixture and Syto 9 alone in the same irradia-
tion experiment (Fig. 7). Using Syto 9 alone also led to a
significant 3?7-fold increase in green fluorescence intensity.
Furthermore,stainingwith DiBAC4(3)revealedthat the loss
of membrane potential coincided with the increase in green
fluorescence intensity in samples that were stained with
the Syto 9/PI mixture (Fig. 5e, h). This implies that Syto 9
is also subject to active dye exclusion and accumulates
intracellularly when efflux pump activity and membrane
potential are lost.
This work enables us to present an inactivation pattern of
the essential cellular functions in E. coli during exposure
to both sunlight and artificial UVA light. Our finding
that artificial UVA light produces an almost identical
Fig. 3. Culturability of E. coli K-12 MG1655
during sunlight exposure for 7 h. Bacterial
cells were harvested from a stationary-phase
LB batch culture, washed and diluted in
mineral water (Evian). Samples were plated
either on unsupplemented TSA (pour-plate
method) and incubated under aerobic condi-
tions (open bars), or on TSA with 0?05%
sodium pyruvate and incubated in anaerobic
jars (grey bars). Error bars represent SD from
Flow cytometry of E. coli inactivated by sunlight
inactivation pattern to that of sunlight indicates that
there is a similar inactivation mechanism for both light
sources. UVA light seems to affect the functioning of
the electron transport chain, because ATP synthesis and
efflux pump activity (both functions are dependent on a
trans-membrane proton gradient) are inactivated quickly.
The loss of culturability correlated well with the loss of
membrane potential and glucose uptake activity. We also
describe a yet-unreported feature of Syto 9 uptake, namely
its dependence on membrane potential and efflux pump
activity. Furthermore, our results stress the importance of
using several viability indicators, apart from culturability, to
characterize the physiological state of a stressed bacterial
Fig. 4. Analysis of cellular functions of E. coli K-12 MG1655 exposed to sunlight (circles) or artificial UVA light (triangles)
with five viability indicators. Bacterial cells were harvested from a stationary-phase LB batch culture, washed and diluted
in mineral water (Evian). Results were calculated as a percentage relative to the total cell count (Syto 9-stained cells) at a
given sampling point. Unstressed control samples (in the dark at 376C) are displayed as open symbols. (a) Total counts (Syto
9-stained cells, dashed lines) measured flow-cytometrically compared to c.f.u. (solid lines). (b) Culturability of UVA- (m)
and sunlight-irradiated cells [aerobic (&) and anaerobic ($) incubation] as a percentage of total counts. (c) Total ATP
concentration per cell. (d) Non-pumping cells; loss of efflux pump activity measured with Syto 9/EB staining. (e) Depolarized
cells; loss of membrane potential measured with DiBAC4(3). (f) Permeabilized cells; loss of membrane integrity measured with
Syto 9/PI staining by flow cytometry. All experiments were done at least twice. Data from one representative experiment are
1724 Microbiology 152
M. Berney, H.-U. Weilenmann and T. Egli
Inactivation pattern of vital cellular functions
The inactivation of cellular functions in E. coli measured
during exposure to light followed a typical inactivation
pattern. Shortly after the start of exposure, the total ATP
concentration decreased rapidly, indicating the cessation
of ATP synthesis by ATPases. At the same time, efflux
pump activity stopped as well. Both functions are depen-
dent on the proton motive force. Gradually, the mem-
brane potential was also lost and the glucose uptake rate
diminished accordingly. Finally, the cytoplasmic mem-
brane of bacterial cells became permeable. The loss of
culturability was dependent on the incubation method
and the growth medium. Anaerobic incubation on sodium
pyruvate-supplemented agar, a procedure known to also
recover injured cells (Bromberg et al., 1998; Czechowicz
et al., 1996), showed a decrease in culturability that closely
correlated with the loss of membrane potential (r2=0?96).
The rapid decrease of total ATP during irradiation implies a
direct or indirect inhibition of ATP synthesis and the
utilization of the remaining ATP by ATP-dependent
functions. Nevertheless, a certain level of ATP was still
measured after prolonged irradiation. This implies that cells
could either maintain alternative ways of ATP synthesis,
such as glycolysis, or that the remaining ATP could not be
used due to damage to systems which require ATP. It has
been shown that the activity of ATPase from Saccharomyces
Fig. 5. Flow-cytometric analysis of E. coli K-12 MG1655 irradiated with artificial UVA light. Bacterial cells were harvested
from a stationary-phase LB batch culture, washed and diluted in mineral water (Evian). Bacterial cell samples were stained
with Syto 9/EB, DiBAC4(3) or Syto 9/PI and analysed on a flow cytometer after being exposed to different fluences (irradiation
intensity6time). After 1000 kJ m”2, >95% of the cells were non-pumping (b), 40% were depolarized (e), and <1%
permeabilized (upper polygon-gate RN2) (h). After 2500 kJ m”2, 100% of the cells were non-pumping (c), 100% depolarized
(f), and >90% permeabilized (polygon-gate RN2) (i). SSC, side scatter.
Flow cytometry of E. coli inactivated by sunlight
cerevisiae solubilized from the plasma membrane and
exposed to UVA light remains constant irrespective of
dosage, indicating that the ATPase molecule itself is
not damaged by UVA irradiation (Arami et al., 1997a, b).
The authors proposed that the reduction of ATPase activity
in the membrane by UVA irradiation is attributable to
conformational changes resulting from an alteration in
the higher-order structure of the membrane due to photo-
chemical decomposition of ergosterol. For E. coli K-12
growing on succinate, it has been shown that ubiquinone
Q-8 is a chromophore for inhibition of ATP synthesis
(Lakchaura et al., 1976). Nevertheless, the same authors
concluded that the oxidative phosphorylation system is
not a primary factor in the induction of growth inhibition
in E. coli by UVA light, because the doses required for
inhibition of growth are only one-sixth of those required for
inhibition of ATP synthesis. This contradicts our findings
that the inhibition of ATP synthesis and efflux pump acti-
vity occurs earlier than the loss of culturability. However,
it has to be mentioned that, in their experiments, stressed
cells were plated on unsupplemented complex media. Such
media can enhance an oxidative burst in bacterial cells,
which eventually leads to cell death (Aldsworth et al., 1999;
Dodd et al., 1997; George et al., 1998). Furthermore, the
strain used in their study (AB2480) was recA-deficient
and, therefore, was probably more susceptible to UVA light.
These circumstances can lead to an overestimation of the
negative effects of UVA light on the culturability of E. coli.
It has been hypothesized that when a cell is stressed energet-
ically, the active transport systems will cease first, then
the cytoplasmic membrane will depolarize, and finally per-
meabilization will occur, indicating cell death (Nebe-von-
Caron et al., 2000). These authors observed that Salmonella
typhimurium, which was stored for 25 days on nutrient agar
at 4uC and resuspended in growth medium, could be
differentially stained with a combination of EB, DiBAC4(3)
and PI. In their case, 35% of the depolarized cells could
still form colonies when sorted onto nutrient agar plates,
while all permeabilized cells were unable to grow. In our
experiments, depolarization of bacterial cells correlated well
with the loss of culturability. In contrast to their measure-
the outer membrane. Still, the uptake of DiBAC4(3) seemed
to be specific without the addition of EDTA, because per-
meabilization of the cytoplasmic membrane was observed
at a much higher fluence. Without disrupting the outer
membrane with EDTA, one probably overestimates the
fluence needed to depolarize and permeabilize the cyto-
plasmic membrane of E. coli. In the case of SODIS this is
rather favourable, because an underestimation could have
severe consequences for the user of SODIS.
The specific glucose uptake rate by the PEP-PTS decreased
simultaneously with the loss of membrane potential,
although the cause of inhibition remains unclear. Changes
in uptake rates due to cell size can be ruled out, because
bacterial cell size remained constant during exposure (data
not shown). Also, the loss of membrane potential cannot
be a direct cause, because 2-NBDG uptake is measured in
combination with an uncoupler (2,4-dinitrophenol) which
itself causes a collapse of the bacterial membrane potential
(by preventing ATP synthesis, hence the degradation of 2-
NBDG) (Natarajan & Srienc, 1999; White, 1995). One can
speculate that enzymes of the PEP-PTS are inactivated by
UVA light, and that glucose is taken up only until all PEP
is used up. It has been proposed that components of the
electron transport chain, possibly menaquinone or dehy-
drogenases, are damaged by UVA light and that this inhi-
bits permeases such as the lactose permease (Jagger, 1981).
Therefore, the high-affinity glucose permease could be an
Fig. 6. (a) 2-NBDG uptake by E. coli K-12 MG1655 harvested
from a stationary-phase LB batch culture, washed and diluted
in sterile-filtered, bottled mineral water (Evian). A bacterial cell
sample was incubated with 2-NBDG and 2,4-dinitrophenol at
376C. Green fluorescence intensity (520 nm) of sub-samples
was analysed every 5 min in a flow cytometer. On the graph,
the geometrical means (G-means) of green fluorescence inten-
sity (measured in range-gate RN1) are displayed in the inset.
(b) 2-NBDG uptake of E. coli K-12 MG1655 (pre-treatment as
above) exposed to sunlight. Bacterial cell samples were incu-
bated with 2-NBDG and 2,4-dinitrophenol for 5 min at 376C
and analysed immediately in a flow cytometer. ($) 2-NBDG-
positive cells as a percentage of total cell counts; (m) G-mean
of green fluorescence intensity (520 nm) of the cell population.
Unstressed control samples are displayed with open symbols.
1726 Microbiology 152
M. Berney, H.-U. Weilenmann and T. Egli
indirect target of UVA. Recently, it has been revealed by
and elongation factor G (EF-G) (FusA) of S. typhimurium
specifically increased in cells just after heat treatment and
recovery in tryptic soy broth (TSB) (Kobayashi et al., 2005).
PykF catalyses the synthesis of pyruvate from PEP in the
sugar phosphotransferase system. Kobayashi and coworkers
suggest that the increase of PykF in cells just after heating
and recovery of cells might facilitate the production of
ATP in the electron transport system, because pyruvic acid
is a key substrate for the citric acid cycle and subsequent
electron transport system (Kobayashi et al.,2005). The same
authors state that ATP seems to be particularly important,
because many reactions involved in recovery require ATP
as an energy source. If PykF were a direct or indirect tar-
get of UVA light, it would prevent recovery by the process
described above. This is in line with our finding that cells
which are inhibited in ATP synthesis can still be recovered
Fig. 7. Flow-cytometric analysis of E. coli
K-12 MG1655 irradiated with artificial UVA
light. Bacterial cells were harvested from a
stationary-phase LB batch culture, washed
and diluted in mineral water (Evian). Staining
with Syto 9/PI or Syto 9 alone was com-
pared to culturability on TSA (pour-plate
method). The loss of culturability coincided
with an increase in green fluorescence inten-
sity (range-gate RN2).
Flow cytometry of E. coli inactivated by sunlight
bition of PykF would not directly inhibit glucose uptake
by the PEP-PTS, because PEP is not converted to pyruvate
and is still available for the uptake of glucose molecules, as
All viability indicators were applied again to the irradiated
cells 5 days after irradiation. No regrowth or recovery of
over, a significant decrease in all measured cellular func-
tions, especially in cells which received >1500 kJ m22solar
UVA (corresponding to 530 W m22global sunlight inten-
sity for 6 h), was observed. This indicates that the damage
in E. coli caused by UVA light is irreversible. It has to be
mentioned, however, that resuscitation attempts with sterile
supernatant, as carried out in previous studies (Kaprelyants
& Kell, 1993; Mukamolova et al., 1998), were not performed
in this study. However, to our knowledge, the resuscitation
of permeabilized cells has never been shown and seems
Differences in viability stains
The death of a bacterial cell has long been defined as the
inability of a cell to grow to a visible colony on bacterio-
logical media. With culturability methods, one can only
observe bacterial death in retrospect (Postgate, 1989).
Today, several viability indicators can be assessed without
culturing cells, and each method is based on criteria that
reflect different levels of cellular integrity or functionality.
Consequently, the interpretation of viability is often am-
biguous. Hewitt & Nebe-Von-Caron (2004) have shown
that multi-parameter flow cytometry allows a functional
classification of the physiological state of single-celled
micro-organisms. They claim that, with this technique,
it is possible to resolve a cell’s physiological state beyond
culturability and to determine population heterogeneity.
Our study confirms their findings and shows that the use
of only one viability indicator, as often applied, is not
sufficient to describe the physiological status of a bac-
terial cell under stress. Therefore, these viability indicators
or indirect methods do not provide short cuts by which
viability may be determined (Postgate, 1967). We pro-
pose that only the sum of all these methods, including the
detection of culturability, can give us some certainty about
the physiological state of a bacterium.
Implications for the SODIS method
Our results show that a UVA fluence of >1700 kJ m22is
needed to depolarize the cytoplasmic membrane of >95%
of the cell population, while 2500 kJ m22is needed for per-
meabilization. This corresponds to about 600–880 W m22
of sunlight intensity (global intensity) over a period of 6 h,
a threshold which can be achieved in most developing
countries (Martin-Dominguez et al., 2005). The same study
also confirms that temperature is not a predominant fac-
tor in the elimination of bacteria with sunlight, but that it
is mainly radiation which determines the efficiency of
the method. Furthermore, our culturability data fitted well
with those of Khaengraeng & Reed (2005), in which a solar
simulator at 410 W m22for a period of 6 h was used to
irradiate stationary-phase E. coli cells. A one log reduc-
tion was achieved with a UVA fluence of ~1200 kJ m22.
This corresponds well with the 1400 kJ m22measured in
Important for the user of SODIS is the loss of infectivity of
pathogenic bacteria in the treated water. It has been shown
that S. typhimurium C5Nxrexposed to simulated sun-
light (probably ~700 W m22) for 8 h and a temperature
regime temporarily reaching 55uC fail to produce detect-
able infections in BALB/c mice (Smith et al., 2000). Even
culturable cells that have been irradiated for 1?5 h are less
infective (virulent) than their non-irradiated counterparts.
Although infectivity of target organisms is the ultimate
parameter to measure in disinfection processes, it is not
a feasible alternative to viability measurements, because
appropriate models for certain pathogens are rare and
difficult to perform. Our results suggest that cells that
(i) have lost culturability under anaerobic conditions in
sodium pyruvate-supplemented agar, (ii) are not able to
take up glucose, and (iii) have lost membrane potential, will
not be able to regain viability in the human intestine. The
highest certainty about cell death, however, will be achieved
if fluences are applied that lead to membrane permeabi-
lity. It is likely that these results also apply to other enteric
bacteria, but this still has to be investigated. Also, the effect
of sunlight on eukaryotes, such as Cryptosporidium or
This project was financially supported by the Velux Foundation,
Switzerland (grant no. 119) and by EAWAG internal funding. We
thank Martin Wegelin, Silvio Canonica, Regula Meierhofer and
Frederik Hammes for valuable discussions.
Acra, A., Raffoul, Z. & Karahagopian, Y. (1984). Solar Disinfection of
Drinking Water and Oral Rehydration Solutions. New York: United
Nations Children’s Fund.
Aldsworth, T. G., Sharman, R. L. & Dodd, C. E. (1999). Bacterial
suicide through stress. Cell Mol Life Sci 56, 378–383.
Arami, S., Hada, M. & Tada, M. (1997a). Reduction of ATPase
activity accompanied by photodecomposition of ergosterol by near-
UV irradiation in plasma membranes prepared from Saccharomyces
cerevisiae. Microbiology 143, 2465–2471.
Arami, S., Hada, M. & Tada, M. (1997b). Near-UV-induced absor-
bance change and photochemical decomposition of ergosterol in the
plasma membrane of the yeast Saccharomyces cerevisiae. Microbiology
Barer, M. R. & Harwood, C. R. (1999). Bacterial viablility and
culturability. Adv Microb Physiol 41, 93–137.
M. Berney, H.-U. Weilenmann and T. Egli
Berney, M., Weilenmann, H., Ihssen, J., Bassin, C. & Egli, T. (2006).
Specific growth rate determines the sensitivity of E. coli to thermal,
UVA and solar disinfection. Appl Environ Microbiol 72, 2586–2593.
Bromberg, R., George, S. M. & Peck, W. (1998). Oxygen sensiti-
vity of heated cells of Escherichia coli O157:H7. J Appl Microbiol 85,
Colwell, R. R. (2000). Viable but nonculturable bacteria: a survival
strategy. J Infect Chemother 6, 121–125.
Czechowicz, S. M., Santos, O. & Zottola, E. A. (1996). Recovery of
thermally-stressed Escherichia coli O157:H7 by media supplemented
with pyruvate. Int J Food Microbiol 33, 275–284.
Dodd, C. E. R., Sharman, R. L., Bloomfield, S. F., Booth, I. R. &
Stewart, G. S. A. B. (1997). Inimical processes: bacterial self-
destruction and sub-lethal injury. Trends Food Sci Tech 8, 238–241.
Downes, A. (1886). On the action of sunlight on micro-organisms,
&c., with a demonstration of the influence of diffused light. Proc R
Soc 40, 14–22.
Fewtrell, L., Kaufmann, R. B., Kay, D., Enanoria, W., Haller, L. &
Colford, J. M., Jr (2005). Water, sanitation, and hygiene interventions
to reduce diarrhoea in less developed countries: a systematic review
and meta-analysis. Lancet Infect Dis 5, 42–52.
George, S. M., Richardson, L. C., Pol, I. E. & Peck, M. W. (1998).
Effect of oxygen concentration and redox potential on recovery
of sublethally heat-damaged cells of Escherichia coli O157:H7,
Salmonella enteritidis and Listeria monocytogenes. J Appl Microbiol
Haugland, R. P. (2002). Handbook of Fluorescent Probes and Research
Products, 9th edn. Edited by J. Gregory. Eugene, OR: Molecular Probes.
Hewitt, C. J. & Nebe-Von-Caron, G. (2004). The application of
multi-parameter flow cytometry to monitor individual microbial cell
physiological state. Adv Biochem Eng Biotechnol 89, 197–223.
Hewitt, C. J., Nebe-Von Caron, G., Nienow, A. W. & McFarlane, C. M.
(1999). Use of multi-staining flow cytometry to characterise the
physiological state of Escherichia coli W3110 in high cell density fed-
batch cultures. Biotechnol Bioeng 63, 705–711.
Hinrichsen, D., Bryant Robey, M. A. & Upadhyay, U. D. (1997).
Solutions for a Water-Short World. Population Reports. Baltimore:
Johns Hopkins School of Public Health. (http://www.infoforhealth.
Hobbins, M. (2004). Home-based water purification through sunlight:
from promotion to health effectiveness. Doctoral thesis. Swiss Tropical
Institute, University of Basel.
Jagger, J. (1981). Near-UV radiation effects on microorganisms.
Photochem Photobiol 34, 761–768.
Joyce, T. M., McGuigan, K. G., Elmore-Meegan, M. & Conroy, R. M.
(1996). Inactivation of fecal bacteria in drinking water by solar
heating. Appl Environ Microbiol 62, 399–402.
Kaprelyants, A. S. & Kell, D. B. (1993). Dormancy in stationary-
phase cultures of Micrococcus luteus: flow cytometric analysis of
starvation and resuscitation. Appl Environ Microbiol 59, 3187–3196.
Khaengraeng, R. & Reed, R. H. (2005). Oxygen and photoinac-
tivation of Escherichia coli in UVA and sunlight. J Appl Microbiol 99,
Kobayashi, H., Miyamoto, T., Hashimoto, Y., Kiriki, M., Motomatsu, A.,
Honjoh, K. & Iio, M. (2005). Identification of factors involved in
recovery of heat-injured Salmonella enteritidis. J Food Prot 68, 932–941.
Lakchaura, B. D., Fossum, T. & Jagger, J. (1976). Inactivation of
adenosine 59-triphosphate synthesis and reduced-form nicotinamide
adenine dinucleotide dehydrogenase activity in Escherichia coli by
near-ultraviolet and violet radiations. J Bacteriol 125, 111–118.
Lonnen, J., Kilvington, S., Kehoe, S. C., Al-Touati, F. & McGuigan,
K. G. (2005). Solar and photocatalytic disinfection of protozoan,
fungal and bacterial microbes in drinking water. Water Res 39,
Martin-Dominguez, A., Alarcon-Herrera, M. T., Martin-Dominguez,
I. R. & Gonzalez-Herrera, A. (2005). Efficiency in the disinfection of
water for human consumption in rural communities using solar
radiation. Solar Energy 78, 31–40.
Midgley, M. (1987). An efflux system for cationic dyes and related
compounds in Escherichia coli. Microbiol Sci 4, 125–127.
Miller, J. (1972). Experiments in Molecular Genetics. Cold Spring
Harbor, NY: Cold Spring Harbor Laboratory.
Mukamolova, G. V., Yanopolskaya, N. D., Kell, D. B. & Kaprelyants,
A. S. (1998). On resuscitation from the dormant state of Micrococcus
luteus. Antonie Van Leeuwenhoek 73, 237–243.
Natarajan, A. & Srienc, F. (1999). Dynamics of glucose uptake by
single Escherichia coli cells. Metab Eng 1, 320–333.
Natarajan, A. & Srienc, F. (2000). Glucose uptake rates of single
E. coli cells grown in glucose-limited chemostat cultures. J Microbiol
Methods 42, 87–96.
Nebe-von-Caron, G., Stephens, P. J., Hewitt, C. J., Powell, J. R. &
Badley, R. A. (2000). Analysis of bacterial function by multi-colour
fluorescence flow cytometry and single cell sorting. J Microbiol
Methods 42, 97–114.
Nystrom, T. (2001). Not quite dead enough: on bacterial life,
culturability, senescence, and death. Arch Microbiol 176, 159–164.
Porter, J., Deere, D., Hardman, M., Edwards, C. & Pickup, R. (1997).
Go with the flow cytometry in environmental microbiology. FEMS
Microbiol Ecol 24, 93–101.
Postgate, J. R. (1967). Viability measurements and the survival of
microbes under minimum stress. Adv Microb Physiol 1, 2–23.
Postgate, J. R. (1989). A microbial way of death. New Scientist 122,
Postgate, J. R. & Hunter, J. R. (1964). Accelerated death of Aerobacter
aerogenes starved in the presence of growth-limiting substrates. J Gen
Microbiol 34, 459–473.
Reed, R. H. (1997). Solar inactivation of faecal bacteria in water: the
critical role of oxygen. Lett Appl Microbiol 24, 276–280.
Smith, R. J., Kehoe, S. C., McGuigan, K. G. & Barer, M. R. (2000).
Effects of simulated solar disinfection of water on infectivity of
Salmonella typhimurium. Lett Appl Microbiol 31, 284–288.
Wegelin, M., Canonica, S., Mechsner, K., Fleischmann, T., Pesaro, F.
& Metzler, A. (1994). Solar water disinfection: scope of the process
and anlysis of radiation experiments. J Water SRT-Aqua 43, 154–169.
White, D. (1995). The Physiology and Biochemistry of Procaryotes. New
York: Oxford University Press.
World Health Organization (1996). Guidelines for Drinking-Water
Quality, vol. 2, Health Criteria and Other Supporting Information,
2nd edn. Geneva: World Health Organization.
World Health Organization/United World Nations Children’s Fund
(2005). Water for Life: Making it Happen. Geneva: World Health
Yoshioka, K., Saito, M., Oh, K. B., Nemoto, Y., Matsuoka, H.,
Natsume, M. & Abe, H. (1996). Intracellular fate of 2-NBDG, a
fluorescent probe for glucose uptake activity, in Escherichia coli cells.
Biosci Biotechnol Biochem 60, 1899–1901.
Flow cytometry of E. coli inactivated by sunlight