Upload
l-mathieu
View
214
Download
0
Embed Size (px)
Citation preview
w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 3 3 7 5 – 3 3 8 6
Avai lab le a t www.sc iencedi rec t .com
journa l homepage : www.e lsev ie r . com/ loca te /wat res
Reversible shift in the a-, b- and g-proteobacteria populationsof drinking water biofilms during discontinuous chlorination
L. Mathieua,*, C. Bouteleuxa,1, S. Fassb, E. Angela, J.C. Blockc
aLaboratoire d’hydroclimatologie medicale Environnement et Sante, Ecole Pratique des Hautes Etudes, UMR 7564, CNRS-Nancy Universite,
15 avenue du Charmois, F-54500 Vandoeuvre-les-Nancy, FrancebCellule Europe-Service Relations Internationales, 24-30 rue Lionnois, BP 60120, F-54003 Nancy Cedex, FrancecLCPME, Laboratoire de Chimie Physique et Microbiologie pour l’Environnement, UMR 7564, CNRS-Nancy Universite, 405 rue de Vandoeuvre,
F-54600 Villers-les-Nancy, France
a r t i c l e i n f o
Article history:
Received 22 January 2009
Received in revised form
29 April 2009
Accepted 5 May 2009
Published online 14 May 2009
Keywords:
Drinking water biofilm
Proteobacteria
Discontinuous chlorination
Bacterial population resilience
* Corresponding author. Tel.: þ33 (0) 3 83 68E-mail address: laurence.mathieu@mede
1 Present address: EDF R&D/LNHE, Groupe0043-1354/$ – see front matter ª 2009 Elsevidoi:10.1016/j.watres.2009.05.005
a b s t r a c t
As disinfection strategies could support a shift of some bacterial populations, the biodi-
versity of drinking water biofilms depending on the disinfectant concentrations was
explored. The effect of different chlorine sequences applied for several weeks (0.1–0.4–
0.1 mg Cl2 L�1 or vice versa) was tested on the abundance of the a-, b- and g-proteobacteria
populations, used as indicators of changes in bacterial populations within drinking water
biofilms. Using dynamic (industrial pilot) and batch (bench scale) conditions, our work
demonstrated the ability of the 3 proteobacteria subclasses to re-organize following
discontinuous chlorinations. The b- and g-proteobacteria subclasses were favoured by high
free residual chlorine concentrations (0.4 mg Cl2 L�1) while a-proteobacteria population
was sensitive to this oxidant level. The proteobacteria population shifts within the biofilm
exposed to discontinuous chlorination were reversible. The resilience of the biofilm pro-
teobacteria populations exposed to oxidant stress questioned the emergence of bacterial
population less sensitive to chlorine.
ª 2009 Elsevier Ltd. All rights reserved.
1. Introduction Tokajian et al., 2005). However such data are definitively
The composition and dynamics of bacterial communities in
drinking water distribution systems, especially in biofilms, are
far from being thoroughly assessed and understood today. As
shown by pioneering and recent studies, many culturable
bacteria in drinking water distribution systems belong to the
phylum Proteobacteria. Some of the commonly detected genera
included Pseudomonas, Sphingomonas, Caulobacter, Aeromonas,
Acinetobacter, Rhodobium, Aquabacterium and Acidovorax (Olson
and Nagy, 1984; LeChevallier et al., 1987, 1980; Norton and
LeChevallier, 2000; Martiny et al., 2005; Lee and Kim, 2003;
22 36; fax: þ33 (0) 3 83 68cine.uhp-nancy.fr (L. MatQualite de l’Eau et Enviroer Ltd. All rights reserved
biased as (1) the majority of bacterial cells in natural envi-
ronments and chlorinated waters are nonculturable by
current methods (Szewzyk et al., 2000; Colwell and Grimes,
2000; McFeters, 1990), and (2) culture methods do select some
phyla as proteobacteria (Martiny et al., 2005). Application of
nucleic acid-based approaches in drinking water research has
the ability to more thoroughly describe the presence, relative
abundances and the dynamics of different genera of bacteria
present in samples (Liu and Stahl, 2002). Under such modern
investigations, the bacterial communities in biofilms appear
much more diversified than expected and not only dominated
22 33.hieu).nnement, 6 quai Watier, F-78401 Chatou Cedex, France..
w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 3 3 7 5 – 3 3 8 63376
by the Proteobacteria but also well represented by the phyla
Actinobacteria, Firmicutes, Verrucomicrobia, Nitrospirae and Bac-
teriodetes (Martiny et al., 2005).
Even when they are not a dominant population, proteobac-
teria, particularly the a-, b- and g-proteobacteria, are system-
atically found indrinkingwater supply systems(Santo Domingo
et al., 2003; Schmeisser et al., 2003; Schwartz et al., 1998; Kalm-
bach et al., 1997; Eichler et al., 2006). Then, their proportion or
their variation could be used as an indicator of changes in the
bacterial populationexposedtoenvironmentalstresses. Indeed,
among the proteobacteria, the a, b and g subclasses were
reported to vary widely depending on the pipe material (Kalm-
bach et al., 1997; Schwartz et al., 1998; Norton and LeChevallier,
2000), the biofilm age (Martiny et al., 2003) and the disinfection
practice (Batte et al., 2003; Williams et al., 2004). As a result,
bacterial diversity differed depending on the drinking water
distribution system. For instance, the drinking water biofilms
from the Berlin, Mainz or Montreal distribution systems were
characterized by a high number of b-proteobacteria (Schwartz
et al., 1998; Kalmbach et al., 1997; Batte et al., 2003), while others
were characterized by a-proteobacteria (Williams et al., 2004;
Schmeisser etal., 2003).Finally,acommon thread between most
of these studies was the low abundance of g-proteobacteria,
which includes most pathogens and opportunistic pathogens,
and could account for approximately 30% of Eubacteria in some
studies (Batte et al., 2003).
As changes in environmental pressures may balance the
structure of the biofilm populations, the impact of different
disinfection strategies commonly used in distribution
systems (type and dose of disinfectant) on the bacterial
diversity within drinking water biofilms should be explored.
Therefore, the question about the emergence of undesirable
flora as a result of some disinfection practices is still open.
Few studies reported the microbial effects of changing disin-
fection regimes from chlorine to monochloramine in drinking
water biofilms, with population shifts and the emergence of
Legionella species in chlorinated biofilms and mycobacteria in
chloraminated ones (Pryor et al., 2004; Santo Domingo et al.,
2003; Williams et al., 2005). Codony et al. (2005) also demon-
strated that successive absence/presence of chlorine episodes
increased the number of culturable heterotrophic bacteria
(HPC) in drinking water, owing to a greater resistance of the
bacterial populations to disinfectants. The authors showed
that after each event of chlorine depletion, the reduction of
HPC due to chlorination was lower than in the previous event
leading to a lower microbial quality in the supply network.
In this context, the main objective of this study was to
monitor the effect of discontinuous chlorination on the abun-
dance of a-, b- and g-proteobacteria in drinking water biofilms,
using a culture-independent method: Fluorescent In Situ
Hybridization (FISH). To meet this objective, discontinuous
chlorinations were applied on an experimental distribution
system (i) to assess the variation of the a-, b- and g-proteo-
bacteria populations within the Eubacteria biofilm communi-
ties and the possible shifts in their proportion,(ii) to evaluate
the chlorine sensitivity of these populations exposed to 0.1 and
0.4 mg Cl2 L�1 residual free chlorine or vice versa, and (iii)
finally to determine the resilience of the microbial group of a-,
b- and g-proteobacteria within drinking water biofilms when
exposed to discontinuous chlorination. The assays were
conducted on a largely colonized experimental distribution
system composed of two loops supplied with drinking water
from the city of Nancy, which was chlorinated on the experi-
mental site in order to obtain a free chlorine residual of either
0.1 or 0.4 mg Cl2 L�1 in the experimental drinking water system.
The resilience of these proteobacteria communities within
drinking water biofilms is then discussed.
2. Materials and methods
2.1. Experimental drinking water distribution systemand sampling
The pilot distribution system used in this study was
comprised of two identical independent loops made of
industrial pipes (loops A and B). The 20-year-old pilot system
had been continuously fed with the drinking water from
Nancy city (France) and, as a result, was largely colonized by
its heterotrophic bacterial biomass. Each loop (31 m long,
100 mm in diameter, cement-lined cast iron) had a water flow
velocity of approximately 1 m s�1, a volume of 240 L, a theo-
retical hydraulic residence time of 24 h (flow rate: 10 L h�1) and
may be looked at as a perfectly mixed reactor.
Experiments were carried out between March 2005 and
May 2006, each loop was continuously fed with tap water from
Nancy city’s distribution system which was chlorinated on
site in two 50 L tanks (average hydraulic residence time of 3 h
before pumping in the loops A or B) to reach the residual free
chlorine concentrations of 0.1 or 0.4 mg Cl2 L�1 in the experi-
mental drinking water distribution network.
Special sampling devices (wettable area: 2 cm2; 21 devices
per loop) allowed PVC (unplasticized polyvinyl chloride
approved for the contact with cold water intended for human
consumption, material still largely used in distribution
systems in Europe) coupons to be placed on the inner wall of
the pipe for biofilm colonization. These coupons were placed
in the loops at the beginning of the experiments, about three
to seven months before chlorine discontinuities were applied.
Consequently, the biofilms analysed after the second discon-
tinuity sequence had also been exposed to the first one.
Water were sampled at the inlet of each loop, collected in 1 L
sterile bottles containing sodium thiosulfate (20 mg) and stored
at 6 �C for less than 2 h before bacterial analysis. PVC coupons
were placed in sterile flasks containing 25 mL of bacteria-free
distilled water. Less than 2 h later, the biofilm was dispersed by
gentle sonication (20 kHz; Vibra cell) for 2 min, using a micro-
probe placed 1 cm above the coupon (power output 10 W).
2.2. Chemical and physical water analysis
Dissolved organic carbon (DOC), pH, chlorine and temperature
were measured according to methods previously reported
elsewhere (Fass et al., 2003; Grandjean et al., 2005).
2.3. Counting of total bacteria and bacteria withmembrane integrity
The total number of bacteria and the number of bacteria with
membrane integrity (impermeant to iodide propidium) were
w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 3 3 7 5 – 3 3 8 6 3377
determined with Live/Dead BacLight kit (Molecular Probes, no.
L7012) (Boulos et al., 1999). Briefly, the stains (0.17 mM syto 9
and 1 mM propidium iodide) were added to a 1 ml aliquot of
the water sample, followed by incubation in the dark for
15 min. The sample was then filtered through a 25 mm
diameter, 0.2 mm pore-size black polycarbonate membrane
(Millipore), and the filter was mounted in BacLight mounting
oil. Observation was performed with an epifluorescence
microscope (BX40; Olympus) equipped with a�100 immersion
objective lens, a 470–490 nm excitation filter and a 520 nm
barrier filter Depending on the sample volume analysed, 30–
100 randomly chosen microscopic fields were counted to
reach a count of at least 100 cells for each sample. The total
number of stained cells (greenþ red fluorescent cells) and the
number of membrane undamaged cells (green fluorescent
cells) were expressed in cells per mL or cells per cm2 for water
and biofilm, respectively.
2.4. Fluorescent in situ hybridization procedure
The universal bacterial probe EUB338 (50-GCTGCCTCC
CGTAGGAGT-30) was used to estimate the density of FISH-
detectable cells (Amann et al., 1990; Loy et al., 2003). To assess
the abundance of the three proteobacteria populations,
biofilm bacteria were also targeted by the FISH probes: ALF1b
(50-CGTTCGYTCTGAGCCAG-30) for a-proteobacteria; BET42a
(50-GCCTTCCCACTTCGTTT-30) for b-proteobacteria; and
GAM42a (50-GCCTTCCCACATCGTTT-30) for g-proteobacteria
(Loy et al., 2003; Manz et al., 1992). Details on oligonucleotide
probes are available at probeBase. All probes were labelled
with Cy3 at the 50 end. Since probes BET42a and GAM42a
differed only by one base pair, hybridization was performed
with an unlabelled competitor as described by Manz et al.
(1992): an unlabelled BET42a probe (5 ng mL�1) was added for
hybridization with the labelled GAM42a probe, and an unla-
belled GAM42a probe (5 ng mL�1) was added for hybridization
with the labelled BET42a probe. The FISH protocol was adap-
ted from Batte et al. (2003). Each water sample was filtered
through a 25 mm diameter, 0.2 mm pore-size white poly-
carbonate membrane (Millipore) and was fixed with 3.7% (vol/
vol) formaldehyde for 30 min. The samples were washed twice
with phosphate-buffered saline (pH 7.4), air dried and dehy-
drated with 2 mL of increasing concentrations of ethanol (50%,
80% and 95%, 3 min each). Fifty microlitres of hybridization
solution were applied to the filter. The hybridization solution
contained 5 ng mL�1 of probe (and also 5 ng mL of unlabelled
probe used as a competitor when necessary), 0.9 M NaCl, 0.1%
SDS, 20 mM Tris-HCl [pH 7.2], and 35% formamide (for EUB338
probe) or 40% formamide (for the other probes). Hybridization
was performed for 2 h at 46 �C� 1 �C in a moisture chamber.
The filter was then washed twice for 15 min in 30 mL of 46 �C
preheated wash solution. This solution contained 20 mM
Tris-HCl [pH 7.2], 0.1% SDS, and 88 mM or 62.4 mM NaCl
(corresponding to the hybridization performed with 35% or
40% formamide, respectively). The filter was then counter-
stained with 1 mL of DAPI (0.05 mg mL�1), rinsed with ultrapure
water, air dried and then mounted on a slide with AF87
antifading reagent (Citifluor, Ltd., London, United Kingdom).
Hybridized cells were visualized by epifluorescence micro-
scopy (BX40, Olympus) with green light for Cy3 staining
(ref. U-MWG2, Olympus; dichroic mirror 570 nm, excitation
filter 510–550 nm and barrier filter 590 nm) or UV light for DAPI
staining (ref. U-MNU2, Olympus; dichroic mirror 400 nm,
excitation filter 360–370 nm and barrier filter 420 nm). Fifty to
one hundred microscopic fields were counted, depending on
bacteria concentration. For each of the analyses, the FISH-
hybridized cells were counted and on the same microscopic
field, these FISH cells were also checked for positive staining
by DAPI. This procedure prevents the counting of false-posi-
tive events under the microscope. Systematic checks of the
quality of the hybridization procedures were performed by
using species known to hybridize (or not) with the three
probes. The results were expressed in cells per mL or cells per
cm2 for water and biofilm, respectively. The percentage of
bacteria hybridized by the EUB338 probe among the total
population was calculated by comparison with the number of
bacteria counted by DAPI staining on the same slide. The
proportions of a-, b- and g-proteobacteria enumerated by FISH
were calculated by reference to the Eubacteria concentration.
2.5. Heterotrophic plate counts
Heterotrophic bacteria were cultured according to French
standard methods (AFNOR, 2003): a 1 mL aliquot of the sample
or of its decimal dilutions was mixed with melted, tempered
glucose-free nutrient agar and incubated for 15 days at 22 �C.
Colonies were expressed as CFU mL�1 and CFU cm�2 after 15
days of incubation for water and biofilm samples, respectively.
2.6. Experimental set-up for chlorine discontinuities onthe experimental water distribution system
Chlorine discontinuities were carried out by rapid change
(increasing or decreasing) in the chlorine treatment at the
inlet of the pilot network in order to get 0.1 or 0.4 mg Cl2 L�1 in
the distribution system. Relatively high concentrations of
chlorine were added as commercial bleach (up to 3 mg Cl2 L�1)
in the two 50 L tanks connected to the loops A and B, in order
to fulfil the chlorine demand of the system and to obtain the
expected chlorine residuals.
A first chlorine exposure sequence was conducted within
the loop A: free chlorine was checked around 0.1 mg Cl2 L�1 in
the loop for 4.5 months (from March to July), then changed to
0.4 mg Cl2 L�1 (increasing discontinuity done in July) for 2
months and later decreased to 0.1 mg Cl2 L�1 (decreasing
discontinuity done in September) during 2.5 months (Fig. 1). A
second chlorine exposure sequence was conducted within the
loop B of the experimental distribution system: the residual
free chlorine concentration was controlled in the loop around
0.4 mg Cl2 L�1 for 5 months (from June to November), then
decreased to 0.1 mg Cl2 L�1 (decreasing discontinuity done
in November) for 2 months, and later again increased to
0.4 mg Cl2 L�1 (increasing discontinuity done in January of the
next year) over a 3-month period (Fig. 2). In other words,
the chlorine discontinuities of both 0.1 to 0.4 mg Cl2 L�1 and
0.4 to 0.1 mg Cl2 L�1 were performed twice in two independent
loops, at two different seasonal periods. During chlorine
exposure sequences, samplings were performed over a more
restricted zone, composed of five time periods (Figs. 1 and 2).
In all cases, water and biofilm samples (from 4 to 12 according
0
0.2
0.4
0.6
0.8
1
2005-03-01 2005-04-25 2005-06-22 2005-08-17 2005-10-10 2005-12-02 2006-02-01 2006-03-27 2006-05-230
5
10
15
20
25
30
35
Chlorine Loop A Temp. Loop A
A1
A2 A3
A4 A5
Free ch
lo
rin
e (m
g C
I2 L
-1)
A1
A2 A3
A4 A5
Fig. 1 – Water temperatures and residual free chlorine concentrations measured in the loop A of the experimental
distribution system with a chlorine sequence of 0.1–0.4–0.1 mg Cl2 LL1. Each point represents a single measure. The grey
area corresponds to the chlorine exposure period of the biofilm; A1 to A5 schematized the periods when regular samplings
were realized for biofilm analysis: A1 [ 3 weeks chlorination at 0.1 mg Cl2 LL1, A2 [ 3 weeks at 0.4 mg Cl2 LL1, A3 [ 10
weeks at 0.4 mg Cl2 LL1, A4 [ 3 weeks at 0.1 mg Cl2 LL1, A5 [ 10 weeks at 0.1 mg Cl2 LL1. The arrows below the x-axis
indicate the chlorine discontinuity’s events.
0
0.2
0.4
0.6
0.8
1
2005-03-01 2005-04-25 2005-06-22 2005-08-17 2005-10-10 2005-12-02 2006-02-01 2006-03-27 2006-05-230
5
10
15
20
25
30
35Chlorine Loop B Temp. Loop B
B1
B2
B5B4B1
B2
B5B4
Free ch
lo
rin
e (m
g C
I2 L
-1)
Fig. 2 – Water temperatures and residual free chlorine concentrations measured in the loop B of the experimental
distribution system with a chlorine sequence of 0.4–0.1–0.4 mg Cl2 LL1. Each point represents a single measure. The grey
area corresponds to the chlorine exposure period of the biofilm; B1 to B5 schematized the periods when regular samplings
were realized for biofilm analysis: B1 [ 3 weeks chlorination at 0.4 mg Cl2 LL1, B2 [ 3 weeks at 0.1 mg Cl2 LL1, B3 [ 10
weeks at 0.1 mg Cl2 LL1, B4 [ 3 weeks at 0.4 mg Cl2 LL1, B5 [ 10 weeks at 0.4 mg Cl2 LL1. The arrows below the x-axis
indicate the chlorine discontinuity’s events.
w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 3 3 7 5 – 3 3 8 63378
w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 3 3 7 5 – 3 3 8 6 3379
to the chlorination sequence) were taken out for analysis. All
samples were characterized in terms of total number of
bacteria, culturable bacteria and bacteria with undamaged
membrane (BacLight kit). In addition, the phylogenetic
composition was evaluated, using the FISH method, by tar-
geting the populations of Eubacteria and a-, b- and g-proteo-
bacteria as described above.
2.7. Experimental set-up for bacterial populations’sensitivity to chlorine in batch tests
In order to assess the chlorine effects on a-, b- and g-proteo-
bacteria, assays were performed in laboratory in batch
conditions. Drinking water was taken from the distribution
network of Nancy. Three flasks (A, B, C) were prepared con-
taining: (1) drinking water supplemented with sodium thio-
sulfate (dechlorinated water¼ control); (2) chlorinated water
at around 0.2 mg Cl2 L�1; and (3) chlorinated water at around
1.1 mg Cl2 L�1. All the flasks were incubated for 24 h at 20 �C in
the dark, without shaking. After 24 h incubations, analyses
were carried out on each flask to quantify the free chlorine, as
well as the total number of bacteria, the Eubacteria, and the a-,
b- and g-proteobacteria (before and after treatment). These
experiments were repeated twice.
2.8. Statistical analysis
Statistical analysis was performed on all the raw data of
bacterial counts for each loop separately. The Mann–Whitney
nonparametric test was used to assess the effect of chlorine
discontinuities on the microbiological quality of biofilm. To
this end, pre- and post-chlorine discontinuity data were
compared. All data were analysed with StatView 5.0 software
(SAS Institute Inc., Cary, NC, USA).
Table 1 – Average characteristics of chlorinated drinking wate
Periods of discontinuous chlorination
A1
Targeted chlorine concentrations (mg Cl2 L�1) 0.1
pH (n¼ 38 to 188)a 7.9 (0.1)
Residual free chlorine (mg Cl2 L�1) (n¼ 44 to 171)a 1.3 (0.4)
TOC (mg L�1) (n¼ 5 to 39)a 1.4 (0.2)
Culturable bacteria (CFU mL�1 after 15 days) (n¼ 7 to 19)a <1
Bacteria (cells mL�1) (n¼ 7 to 18)a 1.3� 105
(8.3� 104)
Membrane integrity (cells mL�1) (n¼ 7 to 18)a 3.0� 104
(2.0� 104)
Eubacteria (cells mL�1) (n¼ 7 to 19)a 2.9� 104
(2.4� 104)
a-proteobacteria (cells mL�1) (n¼ 7 to 18)a 8.7� 103
(7.2� 103)
b-proteobacteria (cells mL�1) (n¼ 7 to 19)a 1.1� 103
(9.4� 102)
g-proteobacteria (cells mL�1) (n¼ 7 to 19)a 4.8� 102
(4.0� 102)
nd¼ not determined.
a Depending on the loops and the time period of monitoring.
3. Results
3.1. Water characteristics at the inlet of the loops
The chlorine exposure sequences were performed in loop A
from March to December 2005 and in loop B during June 2005
to April 2006, with drinking water temperatures ranging from
15 �C to 32 �C (Figs. 1 and 2). These high temperatures of the
drinking water in the loops, especially in summer time, were
due to the fact that the pilot was located in a non-air-condi-
tioned technical hall.
The main characteristics of the chlorinated drinking
waters at the inlet of the loops turned out to be very similar
whatever the period of testing (Table 1). The slightly alkaline
pH (on average 7.9) was a result of the neutralization with
NaOH of the water corrosivity at the treatment plant of Nancy
(Grandjean et al., 2005) and the addition of bleach on the
experimental site at the inlet of the two loops. DOC was esti-
mated to be around 1.5 mg C L�1, of which biodegradable DOC
accounted for approximately 30% (Bouteleux et al., 2005).
Water collected at the inlet of loops A and B was found to
contain 0.8 to 2.5� 105 bacterial cells mL�1 (Table 1), of which
less than 1& could be cultured, which was expected consid-
ering the high residual free chlorine concentration (0.8 to
1.7 mg Cl2 L�1) in water before entering the loop (Table 1).
Among the total number of bacterial cells, 12–15% were
detected as bacteria with intact membranes using the BacLight
kit. The Eubacteria averaged 18.8%� 9.3% and 15.5%� 4.8% in
the water collected at the inlet of loops A and B, respectively
(Table 1). Among this Eubacteria population, 46.6� 21% and
45.1� 17% were found to be proteobacteria in water of loops A
and B, respectively. There was a broad dominance of the
a-proteobacteria subclass with on average 45%� 20% (loop A)
rs at the inlet of the experimental distribution system.
Loop A Loop B
A2 and A3 A4 and A5 B1 B2 and B3 B4 and B5
0.4 0.1 0.4 0.1 0.4
7.8 (0.07) 7.9 (0.07) 7.9 (0.19) 7.8 (0.07) 7.9 (0.06)
1.7 (0.6) 1.1 (0.1) 1.3 (0.3) 0.8 (0.2) 1.4 (0.2)
1.5 (0.07) 1.4 (0.07) 1.5 (0.3) nd nd
<1 <1 <1 2 (2) <1
2.5� 105 1.5� 105 1.2� 105 8.0� 104 1.0� 105
(1.1� 105) (1.5� 105) (4.8� 104) (3.5� 104) (2.8� 104)
2.6� 104 1.5� 104 1.3� 104 7.2� 103 1.4� 104
(3.0� 104) (1.5� 104) (7.7� 103) (7.3� 103) (1.7� 104)
2.8� 104 4.9� 104 2.7� 104 1.2� 104 1.5� 104
(1.3� 104) (4.9� 104) (1.2� 104) (4.7� 103) (8.3� 103)
1.5� 104 1.6� 104 1.1� 104 5.0� 103 6.0� 103
(9.4� 103) (1.6� 104) (7.1� 103) (3.0� 103) (2.3� 103)
1.9� 102 1.4� 102 5.4� 102 2.0� 102 3.1� 102
(6.8� 101) (1.4� 102) (3.4� 102) (1.0� 102) (1.6� 102)
1.1� 102 1.1� 102 2.2� 102 1.4� 102 1.1� 102
(1.0� 102) (1.1� 102) (1.8� 102) (8.9� 101) (4.0� 101)
w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 3 3 7 5 – 3 3 8 63380
and 41%� 16.7% (loop B), against 2� 4% (loop A) and 1.9� 0.7%
(loop B) for the b subclass and 1� 1% (loop A) and 1� 0.5%
(loop B) for the g subclass (Table 1). The pH and DOC
concentrations of the drinking water entering the loops did
not appear to be altered when changing chlorine concentra-
tions at the inlet of the pilot distribution system. Moreover,
the free chlorine residuals contained in the water entering the
network did not seem to have any assessable impact on the
number of culturable bacteria ( p> 0.1 for both loops) which
was already very low before chlorination, nor on the number
of a-, b- and g-proteobacteria ( p> 0.1 for both loops) in chlo-
rinated drinking water (Table 1).
3.2. Effect of chlorine discontinuities 0.1 to0.4 mg Cl2 L�1 on drinking water biofilm
The chlorine discontinuity which aimed at increasing chlorine
concentration from 0.1 to 0.4 mg Cl2 L�1 was tested at two
periods: July 2005 and January 2006, on loop A and B, respec-
tively (Figs. 1 and 2).
Before increasing chlorine level in loops A and B, three
biofilm coupons were collected just before the chlorine
change in both loops in order to characterize the bacterial
biomass and served as a control before discontinuity. At this
stage, the biofilms had been exposed to chlorine concentra-
tions of around 0.1 mg Cl2 L�1 for 4.5 (loop A) to 2 (loop B)
months. In both loops, they were found to host bacteria within
the range of 2� 106 (loop A) to 9� 106 (loop B) cells cm�2 of
which 10% (loop A) and 5% (loop B) were culturable cells and
around 85% showed intact membranes (Fig. 3, sampling
periods A1 and B3). A twofold difference appeared in the
percentage of Eubacteria between the biofilms of the two loops
(loop A: 27.4%� 7.2%; loop B: 58.8%� 1.5%) (Fig. 3, sampling
periods A1 and B3). Despite a higher concentration of Eubac-
teria in loop B’s biofilm, the proportions of the proteobacteria
subclasses were very similar in both loops (on average 39%).
Among them, the a subclass prevailed, in the range of 37–38%
of Eubacteria, whereas b- and g-proteobacteria remained
1E+03
1E+04
1E+05
1E+06
1E+07
Bacteria (cells cm
-2) an
d cu
ltu
rab
le
bacteria (C
FU
15d
cm
-2)
Loop A
Period A1(0.1 mg Cl2 L
-1)Periods A2 + A3(0.4 mg Cl2 L
-1)
Fig. 3 – Characteristics of biofilms exposed to chlorine discontin
counts of total bacteria (cells cmL2) , bacteria with intact me
Eubacteria detected by in situ hybridization (probe EUB338) (cells
incubation) . Each point is an average value of 4–8 biofilm c
Figs. 1 and 2.
clearly low in this biofilm exposed continuously to
0.1 mg Cl2 L�1 free chlorine (<2% for b subclass and<0.3% for g
subclass). These biofilm bacterial dominances turned out to be
in accordance with those measured in the chlorinated
drinking water of both loops (data not shown).
Organic carbon transitory release in water was observed
when initiating chlorination from 0.1 to 0.4 mg Cl2 L�1 (data
not shown), in agreement with previous observations (Fass
et al., 2003). Increasing the residual of free chlorine from 0.1 to
0.4 mg Cl2 L�1 has only a slight effect on the biofilm bacterial
concentrations. Only an unexpected trend of increase in the
total bacterial count and in the number of bacteria with intact
membranes occurred within the biofilm (Fig. 3, sampling
periods A1 and A2þA3). These variations could be essentially
due to seasonal effect as the bacterial concentrations in the
drinking water at the inlet of the loops slightly increased at the
same time. When increasing chlorine concentration to
0.4 mg Cl2 L�1, the culturable fraction of biofilm bacteria in
loops A or B tended to decrease by a factor 2–4, but the biofilm
still contained around 1� 105 CFU cm�2 after 15 days of incu-
bation (Fig. 3).
The number of Eubacteria in biofilm did not change signif-
icantly as the free chlorine residual was changed from 0.1 to
0.4 mg Cl2 L�1. It remained in the range of 1.4� 106 cells cm�2
for loop A and 3–5� 106 cells cm�2 for loop B (Fig. 3). On the
contrary, the response of the biofilm proteobacteria exposed
to this chlorine discontinuity was different for each of the
three subpopulations considered. Even if the a-proteobacteria
remained the dominant subclass, a systematic decrease
occurred in their number (loop A: p¼ 0.04; loop B: p¼ 0.02) and
proportion (loop A: p¼ 0.1; loop B: p¼ 0.02), from 37–38% to 25–
29% whatever the loop, inversely with chlorine concentration.
In parallel, the number and proportion of b- and g-proteo-
bacteria ( p¼ 0.04 and p¼ 0.02 for b and g proportions
respectively, for both loops) were found to increase substan-
tially following the rise in chlorine residual levels (Fig. 5,
sampling periods A1 to A3 and B3 to B5). While the b-proteo-
bacteria accounted for 1–2% of the Eubacteria population in the
1E+03
1E+04
1E+05
1E+06
1E+07
Bacteria (cells cm
-2) an
d cu
ltu
rab
le
bacteria (C
FU
15d
cm
-2)
Loop B
Period B3(0.1 mg Cl2 L
-1)Periods B4 + B5(0.4 mg Cl2 L
-1)
uities from 0.1 to 0.4 mg Cl2 LL1 in loops A and B. Average
mbranes detected after BacLight staining (cells cmL2) ,
cmL2) and culturable bacteria (CFU cmL2 after 15 days
oupons. For a description of the periods of chlorination, see
w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 3 3 7 5 – 3 3 8 6 3381
biofilm exposed to 0.1 mg Cl2 L�1, they represented 3–4% in the
biofilm exposed for 2 months to 0.4 mg Cl2 L�1. In the same
way, 2–14 times more g-proteobacteria were detected in the
biofilm chlorinated at 0.4 mg Cl2 L�1 compared to the ones
chlorinated at 0.1 mg Cl2 L�1. Thus, while this bacterial
subclass in biofilm amounted to less than 0.1% of the Eubac-
teria population at 0.1 mg Cl2 L�1, it reached 1% and 0.6% ten
weeks after the change in residual chlorine in loops A and B,
respectively (Fig. 5, sampling periods A1 to A3 and B3 to B5).
It is also noteworthy that the proportions of the proteo-
bacteria subclasses measured in the biofilm changed gradu-
ally in the time course of the experiments (chlorine exposure
from 2 to 5 months). For instance, the effects of the chlorine
discontinuity on the number and percentages of g-proteo-
bacteria were already observed after 3 weeks of continuous
chlorination at 0.4 mg Cl2 L�1 and were confirmed and became
significant after 10 weeks (Loop A: p¼ 0.021) (Fig. 5). Similar
observations could be made for the other two proteobacteria
subclasses. A third assay, carried out one year later on another
loop of the same experimental distribution system (loop C),
showed the same shift in the proteobacteria populations
within the drinking water biofilm subjected to free chlorine
concentration changes (see supplementary data Fig. S1).
3.3. Effect of chlorine discontinuities 0.4 to0.1 mg Cl2 L�1 on drinking water biofilm
Two other independent discontinuity’s scenario with
decreasing chlorine concentrations were tested at two periods:
September 2005 and November 2005, on loops A and B,
respectively (Figs. 1 and 2). Before decreasing the concentration
of chlorine, three biofilm coupons were collected just before
the chlorine change in both loops A and B in order to charac-
terize the bacterial biomass and served as a control before
discontinuity (Figs. 1 and 2, sampling periods A3 to A5 and B1 to
B3). At this stage, the biofilms had been exposed to chlorine
concentrations of around 0.4 mg Cl2 L�1 for 2 (loop A) and 5
(loop B) months. Prior to discontinuity, the biofilm contained
on average 6–8� 106 cells cm�2, of which 82.9%� 3.6% (loop A)
and 63.5%� 1% (loop B) had intact membranes (BacLight kit),
1E+03
1E+04
1E+05
1E+06
1E+07
Ba
cte
ria
(c
ells
c
m-2) a
nd
c
ultu
ra
ble
ba
cte
ria
(C
FU
1
5d
c
m-2)
Loop A
Period A3(0.4 mg Cl2 L-1)
Periods A4 + A5(0.1 mg Cl2 L-1)
Fig. 4 – Characteristics of biofilms exposed to chlorine discontin
counts of total bacteria (cells cmL2) , bacteria with intact me
Eubacteria detected by in situ hybridization (probe EUB338) (cells
incubation) . Each point is an average value of 4–8 biofilm c
Figs. 1 and 2.
while the culturable fraction of bacteria within the biofilm was
found to be low: 1.2% in loop A and 5.7% in loop B. The chlori-
nated biofilm was characterized by around 40% of Eubacteria
(loop A: 41.7%� 10.3%; loop B: 43.4%� 2.0%) (Fig. 4). Proteobac-
teria phylum was slightly represented: 29.8%� 8.8% and
24.5%� 2.8% in loops A and B, respectively. Among the three
proteobacteria classes analysed, the a subclass was dominant
(on average 25–23%), whereas the b and g subclasses were
poorly represented (<3.2% and <1.1%, respectively) in the
drinking water biofilms exposed continuously to 0.4 mg Cl2 L�1
free chlorine (Fig. 5, sampling periods A3 to A5 and B1 to B3).
Chlorine discontinuities from 0.4 to 0.1 mg Cl2 L�1 were
found not to significantly modify the total number of bacteria
in the biofilms in loops A and B ( p¼ 0.23 and p¼ 0.12, respec-
tively), nor the number of bacteria with undamaged
membranes (loop A: p¼ 0.23; loop B: p¼ 0.3) or the number of
culturable bacteria ( p> 0.1 for both loops). Only a trend of
increase in the bacterial cultivability within the biofilm was
observed for loop A (Fig. 4). No significant change in Eubacteria
concentrations was observed following the chlorine disconti-
nuities 0.4 to 0.1 mg Cl2 L�1. Only a weak tendency to increase
appeared in loop A when reducing the chlorine dose from 0.4 to
0.1 mg Cl2 L�1 (1.4� 106 to 2.9� 106 cells cm�2) (Fig. 4). On the
contrary, all 3 proteobacteria classes found in the biofilm had
their levels considerably altered as the residual free chlorine
concentrations decreased, even if the dominance ranks within
the bacterial population remained unchanged. Indeed, the a-
proteobacteria within the biofilm saw their number increase
significantly (loop A: p¼ 0.006; loop B: p¼ 0.02) when chlorine
concentration reached 0.1 mg Cl2 L�1, accounting for almost
38% (loop A) and 35% (loop B) of the Eubacteria. On the contrary,
the b- and g-proteobacteria saw their number and proportion
decreased when chlorine levels dropped from 0.4 to
0.1 mg Cl2 L�1 (Fig. 5, sampling periods A3 to A5 and B1 to B3).
As observed for the 0.1–0.4 mg Cl2 L�1 chlorine sequence,
the shift in the 3 proteobacteria populations took place grad-
ually during the 3 months of experiments (Fig. 5). As an
example, the concentration of g-proteobacteria within the
biofilm was found to decline from 1.4� 104 cm�2 (sampling
period A3) at 0.4 mg Cl2 L�1 to 1.1� 104 cells cm�2 (sampling
1E+03
1E+04
1E+05
1E+06
1E+07
Ba
cte
ria
(c
ells
c
m-2) a
nd
c
ultu
ra
ble
ba
cte
ria
(C
FU
1
5d
c
m-2)
Loop B
Period B1(0.4 mg Cl2 L
-1)Periods B2 + B3(0.1 mg Cl2 L-1)
uities from 0.4 to 0.1 mg Cl2 LL1 in loops A and B Average
mbranes detected after BacLight staining (cells cmL2) ,
cmL2) and culturable bacteria (CFU cmL2 after 15 days
oupons. For a description of the periods of chlorination, see
mg Cl2 L-1
A
αα-p
ro
teo
bacteria
(%
E
UB
338)
0%
10%
20%
30%
40%
50%
60%Loop A
0%
10%
20%
30%
40%
50%
60%
Loop Bβ-
pro
teo
bacteria
(%
EU
B338)
0%
2%
4%
6%
8%
10%
0%
2%
4%
6%
8%
10%
γ-p
ro
teo
bacteria
(%
EU
B338)
0.0%
0.5%
1.0%
1.5%
2.0%
0.0%
0.5%
1.0%
1.5%
2.0%
0.10.4
0.1
0.4
0.10.4
A1 A2 A3 A4 A5 B1 B2 B3 B4 B5
B
Fig. 5 – Average evolutions of the proportions of a-, b- and g-proteobacteria among the Eubacteria in the drinking water
biofilms exposed to chlorine discontinuity sequences of 0.1–0.4–0.1 mg Cl2 LL1 in loop A (A) and 0.4–0.1–0.4 mg Cl2 LL1 in
loop B (B) (for each point n [ 3–8). For a description of the periods of chlorination, see Figs. 1 and 2.
w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 3 3 7 5 – 3 3 8 63382
period A4) and 7.8� 103 cells cm�2 (sampling period A5)
respectively 4 and 9 weeks after the chlorine change.
3.4. Chlorine sensitivity of the a-, b- andg-proteobacteria populations in drinking water(laboratory experiments)
In an attempt to understand previous observations, laboratory
tests were performed with drinking water chlorinated in the
laboratory (batch assays) in order to determine the a-, b- and
g-proteobacteria sensitivity to chlorine. Quantification of the
three proteobacteria subgroups was conducted on drinking
water dechlorinated with sodium thiosulfate (control: flask A)
and chlorinated water (at 0.2 and 1.2 mg Cl2 L�1 for flask B and
C, respectively). After 24 h incubation at 20 �C, only
0.03 mg Cl2 L�1 of residual free chlorine was detected in flask
B, but flask C still contained 0.6 mg Cl2 L�1 free chlorine. In the
control flask A, the total number of bacteria remained stable
during the 24 h incubation (8.4� 104 cells mL�1 at T¼ 0 and
7.9� 104 cells mL�1 at T¼ 24 h), whereas a decrease in the
total number of bacteria in chlorinated drinking waters was
shown between T¼ 0 and T¼ 24 h in flasks B and C (Fig. 6).
The results showed that among the proteobacteria, the a,
b and g subclasses did not behave the same way. First, the a-
proteobacteria concentration was found to remain stable in
control flask A (dechlorinated drinking water) but to drop in
the presence of residual chlorine: the higher the chlorine
concentration, the more a-proteobacteria lost (Fig. 6). Second,
the b-proteobacteria concentration was shown to increase in
24 h in the dechlorinated drinking water (flask A) (þ0.4 log
after 24 h). While it proved to remain stable in slightly chlo-
rinated water (only unsignificant �0.02 log after 24 h), it
showed a more marked drop (�0.4 log after 24 h) in the
1.1 mg Cl2 L�1 chlorinated water (Fig. 6). Third, the population
of g-proteobacteria was the only one to be seen stable during
the 24 h incubation even for the higher chlorine dose tested
(þ0.1 log after 24 h). By way of consequence their proportion
among the Eubacteria increased from 2.1% to 3.8% and 9.4% on
average in chlorinated drinking waters chlorinated at
0.2 mg Cl2 L�1 and 1.1 mg Cl2 L�1, respectively.
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
Total number ofbacteria
Eubacteria Alpha-proteobacteria
Beta-proteobacteria
Gamma-proteobacteria
lo
g cells (T
24h
) - lo
g cells (T
0)
Dechlorinated water (control) (residual free chlorine <0.02 mg Cl2 L-1)Free chlorine applied at T0: 0.2 mg Cl2 L
-1 (residual free chlorine at T24h: 0.03 mg Cl2 L-1)
Free chlorine applied at T0: 1.1 mg Cl2 L-1 (residual free chlorine at T24h: 0.6 mg Cl2 L-1)
Fig. 6 – Laboratory tests carried out on planktonic drinking water bacteria. Ratio (logT24h L logT0) of the total number of
bacteria (DAPI staining), Eubacteria (probe EUB338), a-proteobacteria (probe ALF1b), b-proteobacteria (probe BET42a) and g-
proteobacteria (probe GAM42a) depending on the chlorine concentrations applied for 24 h at 22 8C (n [ 2).
w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 3 3 7 5 – 3 3 8 6 3383
4. Discussion
The characterization of the bacterial diversity of drinking
water is a pioneering field as only few works have docu-
mented the composition of biofilm communities using
nucleic-based approaches (Kalmbach et al., 1997; Martiny
et al., 2005; Schmeisser et al., 2003; Williams et al., 2005;
Schwartz et al., 1998). We used the in situ hybridization (FISH)
method in spite of the inability of the so-called universal
probes EUB338 to target all Eubacteria (Moter and Gobel, 2000).
Our results showed that, on average, Eubacteria represented 15
to 19% (in water) and 27 to 58% (in biofilm) of the total cells
stained with DAPI. The differences in the Eubacteria percent-
ages within our drinking water biofilm between the two loops
could be explained by the seasonal variations of the inlet
water and the age of the biofilm (4.5 and 7 months old in loops
A and B, respectively). These values were lower than those
recorded in other studies carried out on non-disinfected
drinking systems. For instance, Kalmbach et al. (1997) and
Manz et al. (1993) detected 23% and 37% of planktonic Eubac-
teria and 50% and 68% of biofilm Eubacteria, respectively.
Eubacteria may be not detected by FISH because: (1) some
phyla are not targeted by the EUB338 probe (Daims et al., 1999);
(2) hybridization efficiency may be reduced due to chlorine
reactivity with nucleic acids (Saby et al., 1997; Phe et al., 2005);
and (3) reduced signal intensity or false negative results have
been reported, due to the rRNA content of bacterial cells which
may significantly vary between species, especially under
adverse conditions which altered the physiological status of
bacteria (Delong et al., 1989; Poulsen et al., 1993).
Within our drinking water biofilm developed on PVC and
exposed to discontinuous chlorination, the Eubacteria contained
only 30 to 40% of a-, b- and g-proteobacteria, suggesting that
these three subclasses do not represent the biofilm predomi-
nant communities, as already reported by Martiny et al. (2005).
Besides, the proteobacteria within our chlorinated biofilm was
shown to be dominated by the a subclass, whereas the
b subclass prevailed in some other works (Batte et al., 2003). As
for the g subclass, it appeared to be little represented with less
than 2% on average within the chlorinated drinking water bio-
films which is consistent with other data (Williams et al., 2004;
Kalmbach et al., 1997; Manz et al., 1993. The abundance of
specific-group composition within the biofilm could be influ-
enced by the pipe material (Kalmbach et al., 1997). And as
demonstrated by Schwartz et al. (1998), plastic materials such as
PVC, widely used in domestic drinking water distribution
systems, appeared to be colonized very frequently by b- and
g-proteobacteria, whereas the b subclass was detected in higher
percentages on metallic materials (steel, copper).
The present study brings original data as it looked at biofilm
communities in response to discontinuous chlorinations
(from low to high levels and vice versa). First, it demonstrated
that the groups of b- and g-proteobacteria in biofilms were fav-
oured when chlorine concentrations increased, while the
number and proportion of a-proteobacteria declined. To our
knowledge, only the study from Batte et al. (2003) pointed out that
within a 113-day-old chlorinated drinking water biofilm the pro-
portions of b-proteobacteria decreased and those of
g-proteobacteria increased when applying chlorine (7 days at
1 mg Cl2 L�1). Additional chlorine discontinuity assays were
performed on the same experimental network in parallel with
a control loop exposed to continuous chlorine concentrations
of 0.4 mg Cl2 L�1 or 0.1 mg Cl2 L�1. The resulting data supported
our conclusions that the proteobacteria populations show
a shifting pattern within the drinking water biofilm subjected to
free chlorine concentration changes (see supplementary data
Figs. S1 and S2). Second, laboratory testing conducted with
chlorinated drinking water confirmed that the g subclass
behaves specifically and resists quite well to chlorine in spite of
high free chlorine concentrations applied (Fig. 6). Third, our study
demonstrated that, within a drinking water biofilm exposed to
discontinuous chlorination, a reversible shift in the proportions
of the three classes of proteobacteria was systematically
observed (at least for three reactors at two different seasons)
(Figs. 6 and S1, S2). Indeed, when testing chlorination sequences
of 0.1–0.4–0.1 mg Cl2 L�1 or 0.4–0.1–0.4 mg Cl2 L�1, as reported
on Fig. 5, the proportions of a- and g-proteobacteria within the
w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 3 3 7 5 – 3 3 8 63384
biofilm returned to their original values (i.e., those observed
before the first discontinuity when chlorine residual was 0.1 or
0.4 mg Cl2 L�1). Only the proportion of b-proteobacteria did not
systematically return to its initial level and was favoured and
increased in the biofilm exposed to high chlorine doses.
The variations of the a-, b- and g-proteobacteria populations
appeared to be an interesting indicator of the impact of oxidant
stress because these three proteobacteria subclasses were
characterized by a relevant degree of flexibility as their relative
proportions returned to their original levels after being disturbed
by chlorine. This phenomenon is described in the literature
under the term ‘‘resilience’’ of bacterial communities, as
opposed to ‘‘resistance’’ defined asthedegreetowhich microbial
composition remained unchanged in the face of a disturbance
(Allison and Martiny, 2008). In the light of these two definitions,
withinourchlorinateddrinking waterbiofilm, thegroupsof a-, b-
and g-proteobacteria appeared to be resilient and able to quickly
(i.e., within few weeks) return to their original levels depending
on the residual free chlorine concentrations.
Lower chlorine sensitivity of g-proteobacteria compara-
tively to a-proteobacteria population is difficult to explain at
this stage of knowledge and only hypothesis can be settled,
referring to factors considered to be responsible for biofilm
tolerance (Lewis, 2001).
The first hypothesis is that the g subclass, described as
pioneer during the formation of drinking water biofilm
(Kalmbach, 1998), could be located deeper within the biofilm
and thus be less susceptible to chlorine which diffusion is
limited by reaction–diffusion interactions within the biofilm
matrix (De Beer et al., 1994; Stewart and Franklin, 2008).
However, this possible explanation is not consistent with our
batch assays carried out on planktonic drinking water bacteria
exposed to increasing doses of chlorine, which showed that
the abundance of the planktonic g-proteobacteria increased
as free residual chlorine increased while the number of
planktonic a-proteobacteria decreased.
The second hypothesis is that g-proteobacteria population
grew slowly may be because of limited access to nutrients and
oxygen, which may contribute to increase its tolerance to
chlorine. This low susceptibility of slow growing bacteria is
well-known for biofilm bacteria exposed to antimicrobial
agents (Drenkard, 2003; Lewis, 2001; Costerton et al., 1999).
Again, this assumption is not consistent with our short-term
batch test (24 h chlorination), because planktonic a-, b- and g-
proteobacteria showed rapid responses to chlorine stress
similar to those observed in the biofilms (i.e., an increase in
the number of g-proteobacteria).
The third and most probable hypothesis that could explain
the distinct response to chlorine between the three proteo-
bacteria subclasses is the differential expression of stress
oxidant resistance genes within the bacterial communities
(Storz and Imlay, 1999). There is however no information on
the existence of different and specific oxidant stress regula-
tory systems between the three proteobacteria subclasses.
Additional experiments are required to characterize the
expression of genes encoding the oxidative stress response in
the a-, b- and g-proteobacteria populations associated with
differential induction of resistance mechanisms.
A key question for the future is whether changes in chlo-
rine concentrations in the drinking water systems can lead to
a permanent occurrence of some bacterial populations, such
as g-proteobacteria, to assess the accumulation of undesirable
microorganisms as this has already been observed in hospital
hot-water distribution systems (colonization by Pseudomonas
or Legionella).
5. Conclusions
This work enabled us to study the effect of discontinuous
chlorination on the a-, b- and g-proteobacteria used as indi-
cators of changes in bacterial populations within drinking
water biofilms. This work demonstrated the ability of these
bacterial communities to re-organize following a chlorine
stress, and suggested the resilience of proteobacteria biofilm
communities. In particular, it demonstrated that:
- The g-proteobacteria population (subclass sparsely repre-
sented but including more than 40 different genus among
which are most of the pathogens and opportunistic patho-
gens carried by water) within drinking water biofilms is less
sensitive to chlorine than its a counterpart. The g-proteo-
bacteria increase in proportion and number when increasing
chlorine concentration. This was also confirmed by labora-
tory tests on chlorinated drinking water.
- The change in the numbers and proportions of the three pro-
teobacteria subclasses within the biofilm appeared gradually
in the time course of the experiments (from 2 to 5 months).
- The population shifts within a drinking water biofilm
exposed to discontinuous chlorination were reversible.
The application of chlorine in a network modifies the
bacterial communities of the biofilm and supports de facto
bacterial populations less sensitive to chlorine (in particular
the g-proteobacteria). These results questioned the emer-
gence of bacterial populations less susceptible to chlorine
according to the chlorination practices. A more complete
picture of microbial community diversity and interspecies
relationships should facilitate a better understanding of
disinfection resistance phenomena.
Acknowledgments
The results of this study were obtained within the scope of
a study coordinated by the Centre International de l’Eau de
Nancy (NANCIE) and supported by the following partners:
Anjou Recherche and VEOLIA EAU, the Syndicat des Eaux d’Ile
de France, the Agence de l’Eau Seine-Normandie, the Com-
munaute Urbaine du Grand Nancy (CUGN, France), and the
Centre International de l’Eau de Nancy.
Appendix.Supplementary data
Supplementary data associated with this article can be found
in the online version, at doi:10.1016/j.watres.2009.05.005.
w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 3 3 7 5 – 3 3 8 6 3385
r e f e r e n c e s
Allison, S.D., Martiny, J.B., 2008. Resistance, resilience andredundancy in microbial communities. Proc. Natl. Acad. Sci.105 (1), 11512–11519.
Amann, R.I., Krumholz, L., Stahl, D.A., 1990. Fluorescent-oligonucleotide probing of whole cells for determinative,phylogenetic, and environmental studies in microbiology.J. Microbiol. 172, 762–770.
Batte, M., Mathieu, L., Laurent, P., Prevost, M., 2003. Influence ofphosphate and disinfection on the composition of biofilmsproduced from drinking water, as measured by fluorescencein situ hybridization. Can. J. Microbiol. 49, 741–753.
Boulos, L., Prevost, M., Barbeau, B., Coallier, J., Desjardins, R., 1999.LIVE/DEAD BacLight: application of a new rapid stainingmethod for direct enumeration of viable and total bacteria indrinking water. J. Microbiol. Methods 37, 77–86.
Bouteleux, C., Saby, S., Tozza, D., Cavard, J., Lahoussine, V.,Hartemann, P., Mathieu, L., 2005. E. coli behavior in the presenceof organic matter released by algae after exposure to watertreatment chemicals. Appl. Environ. Microbiol. 71, 734–740.
Codony, F., Morato, J., Mas, J., 2005. Role of discontinuouschlorination on microbial production by drinking waterbiofilms. Water Res. 39, 1896–1906.
Colwell, R.R., Grimes, D.J., 2000. Nonculturable micro-organismsin the environment. In: Colwell, R.R., Grimes, D.J. (Eds.).American Society for Microbiology, Washington, DC, 354 pp.
Costerton, J.W., Stewart, P.S., Greenberg, E.P., 1999. Bacterialbiofilms: a common cause of persistent infections. Science284, 1318–1322.
Daims, H., Bruhl, A., Amann, R., Schleifer, K.H., Wagner, M., 1999.The domain-specific probe EUB338 is insufficient for thedetection of all bacteria: development and evaluation ofa more comprehensive probe set. Syst. Appl. Microbiol. 22,434–444.
De Beer, D., Srinivasan, R., Stewart, P.S., 1994. Directmeasurement of chlorine penetration into biofilms duringdisinfection. Appl. Environ. Microbiol. 60, 4339–4344.
Delong, E.F., Taylor, L.T., Marsh, T.L., Preston, C.M., 1989.Visualization and enumeration of marine planktonic archaeand bacteria by using polyribonucleotide probes andfluorescent in situ hybridization. Appl. Environ. Microbiol. 65(12), 5554–5563.
Drenkard, E., 2003. Antimicrobial resistance of Pseudomonasaeruginosa biofilms. Microb. Infect. 5, 1213–1219.
Eichler, S., Christen, R., Holtje, C., Westphal, P., Botel, J., Brettar, I.,Mehling, A., Hofle, M.G., 2006. Composition and dynamics ofbacterial communities of a drinking water supply system asassessed by RNA- and DNA-based 16S rRNA genefingerprinting. Appl. Environ. Microbiol. 72 (3), 1858–1872.
Fass, S., Block, J.-C., Boualam, M., Gauthier, V., Gatel, D., Cavard, J.,Benabdallah, S., Lahoussine, V., 2003. Release of organic matterin a discontinuously chlorinated drinking water network.Water Res. 37, 493–500.
French association for normalisation (AFNOR), 2003. WaterQuality – Enumeration of Cultivable Heterotrophic BacteriaUsing Solid Culture Media Incorporation Method. NF EN ISO6222 (in French).
Grandjean, D., Fass, S., Tozza, D., Cavard, J., Lahoussine, V., Saby, S.,Guilloteau, H., Block, J.-C., 2005. Coliform culturability in over-versus undersaturated drinkingwaters.WaterRes. 39, 1878–1886.
Kalmbach, S., 1998. Polyphasic Characterization of the MicrobialPopulation of Drinking Water Biofilms. PhD thesis, Universityof Berlin, Berlin, Germany, 125 pp.
Kalmbach, S., Manz, W., Szewzyk, U., 1997. Dynamics of biofilmformation in drinking water: phylogenetic affiliation andmetabolic potential of single cells assessed by formazan
reduction and in situ hybridization. FEMS Microbiol. Ecol. 22,265–279.
LeChevallier, M.W., Babcock, T.M., Lee, R.G., 1987. Examinationand characterization of distribution system biofilms. Appl.Environ. Microbiol. 53, 2714–2724.
LeChevallier, M.W., Seidler, R.J., Evans, T.M., 1980. Enumerationand characterization of standard plate count bacteria inchlorinated and raw water supplies. Appl. Environ. Microbiol.40, 922–930.
Lee, D.-G., Kim, S.-J., 2003. Bacterial species in biofilm cultivatedfrom the end of the Seoul water distribution system. J. Appl.Microbiol. 95, 317–324.
Lewis, K., 2001. Riddle of biofilm resistance. Antimicrob. AgentsChemother. 45 (4), 999–1007.
Liu, W.-T., Stahl, D.A., 2002. Molecular approaches for themeasurement of density, diversity and phylogeny. In: Hurst, C.J.,Crawford, R.L., Knudsen, G.R., McInerney, M.J., Stetzenbach, L.D.(Eds.), Manual of Environmental Microbiology, second ed. ASMPress, Washington, DC, pp. 114–134.
Loy, A., Horn, M., Wagner, M., 2003. ProbeBase: an online resourcefor rRNA-targeted oligonucleotide probes. Nucleic Acids Res.31, 514–516.
Manz, W., Amann, R., Ludwig, W., Wagner, M., Schleifer, K.H.,1992. Phylogenetic oligodeoxynucleotide probes for the majorsubclasses of proteobacteria: problems and solutions. Syst.Appl. Microbiol. 15 (44), 593–600.
Manz, W., Szewzyk, U., Ericsson, P., Amann, R., Schleifer, K.H.,Stenstrom, T.A., 1993. In situ identification of bacteria indrinking water and adjoining biofilms by hybridization with16S and 23S rRNA-directed fluorescent oligonucleotide probes.Appl. Environ. Microbiol. 59 (7), 2293–2298.
Martiny, A.C., Albrechtsen, H.J., Arvin, E., Molin, S., 2005.Identification of bacteria in biofilm and bulk water samples froma nonchlorinated model drinking water distribution system:detection of a large nitrite-oxidizing population associated withNitrospira spp. Appl. Environ. Microbiol. 71, 8611–8617.
Martiny, A.C., Jørgensen, T.M., Albrechtsen, H.J., Arvin, E.,Molin, S., 2003. Long-term succession of structure anddiversity of a biofilm formed in a model drinking waterdistribution system. Appl. Environ. Microbiol. 69, 6899–6907.
McFeters, G.A., 1990. Enumeration, occurrence, and significanceof injured indicator bacteria in drinking water. In: DrinkingWater Microbiology: Progress and Recent Developments.Springer-Verlag New York, Inc., New York, pp. 478–492.
Moter, A., Gobel, U.B., 2000. Fluorescence in situ hybridization(FISH) for direct visualization of microorganisms. J. Microbiol.Methods 41, 85–112.
Norton, C.D., LeChevallier, M.W., 2000. A pilot study ofbacteriological population changes through potable watertreatment and distribution. Appl. Environ. Microbiol. 66,268–276.
Olson, B.H., Nagy, L.A., 1984. Microbiology of potable water. Adv.Appl. Microbiol. 30, 73–132.
Phe, M.H., Dossot, M., Guilloteau, H., Block, J.-C., 2005. Nucleicacid fluorochromes and flow cytometry prove useful inassessing the effect of chlorination on drinking water bacteria.Water Res. 39, 3618–3628.
Poulsen, L.K., Ballard, G., Stahl, D.A., 1993. Use of rRNAfluorescence in situ hybridization for measuring the activity ofsingle cells in young and established biofilms. Appl. Environ.Microbiol. 59 (5), 1354–1360.
ProbeBase. http://www.microbial-ecology.net/probebase/default.asp (accessed 03/2009).
Pryor, M., Springthorpe, S., Riffard, S., Brooks, T., Huo, Y.,Davis, G., Sattar, S.A., 2004. Investigation of opportunisticpathogens in municipal drinking water under different supplyand treatment regimes. Water Sci. Technol. 50 (1), 83–90.
Saby, S., Sibille, I., Mathieu, L., Paquin, J.-L., Block, J.-C., 1997.Influence of water chlorination on the counting of bacteria
w a t e r r e s e a r c h 4 3 ( 2 0 0 9 ) 3 3 7 5 – 3 3 8 63386
with DAPI (40-6-diamidino-2-phenylindole). Appl. Environ.Microbiol. 63 (4), 1564–1569.
Santo Domingo, J.W., Meckes, M.C., Simpson, J.M., Sloss, B.,Reasoner, D.J., 2003. Molecular characterization of bacteriainhabiting a water distribution system simulator. Water Sci.Technol. 47 (5), 149–154.
Schmeisser, C., Stockigt, C., Raasch, C., Wingender, J., Timmis, K.N., Wenderoth, D.F., Flemming, H.C., Liesegang, H.,Schmitz, R.A., Jaeger, K.E., Streit, W.R., 2003. Metagenomesurvey of biofilms in drinking-water networks. Appl. Environ.Microbiol. 69, 7298–7309.
Schwartz, T., Hoffmann, S., Obst, U., 1998. Formation andbacterial composition of young, natural biofilms obtainedfrom public bank-filtered drinking water systems. Water Res.32, 2787–2797.
Stewart, P.S., Franklin, M.J., 2008. Physiological heterogeneity inbiofilm. Microbiology 6, 199–210.
Storz, G., Imlay, J.A., 1999. Oxidative stress. Curr. Opin. Microbiol.2, 188–194.
Szewzyk, U., Szewzyk, R., Manz, W., Schleifer, K.-H., 2000.Microbiological safety of drinking water. Annu. Rev. Microbiol.54, 81–127.
Tokajian, S.T., Hashwa, F.A., Hancock, I.C., Zalloua, P.A., 2005.Phylogenetic assessment of heterotrophic bacteria froma water distribution system using 16S rDNA sequencing. Can.J. Microbiol. 51, 325–335.
Williams, M.M., Santo Domingo, J.W., Meckes, M.C., 2005.Population diversity in model potable water biofilmsreceiving chlorine or chloramine residual. Biofouling21 (5–6), 279–288.
Williams, M.M., Santo Domingo, J.W., Meckes, M.C., Kelty, C.A.,Rochon, H.S., 2004. Phylogenetic diversity of drinking waterbacteria in a distribution system simulator. J. Appl. Microbiol.96, 954–964.