Journal of Applied Microbiology 2004, 97, 1123–1131 doi:10.1111/j.1365-2672.2004.02396.

Bacterial population dynamics and community structure

in a pharmaceutical manufacturing water supply system
determined by real-time PCR and PCR-denaturing
gradient gel electrophoresis

M. Kawai1, J. Yamagishi1, N. Yamaguchi2, K. Tani2 and M. Nasu2

1
Pharmaceutical Research and Technology Center, Dainippon Pharmaceutical Co., Ltd., Osaka, Japan, and 2Graduate School of
Pharmaceutical Science, Osaka University, Osaka, Japan

2004/0139: received 9 February 2004, revised 18 May 2004 and accepted 18 May 2004

ABSTRACT
M . K A W A I , J . Y A M A G I S H I , N . Y A M A G U C H I , K . T A N I A N D M . N A S U . 2004.
Aims: To control bacteria in the pharmaceutical water supply system.
Methods and Results: Bacteria were enumerated by conventional culture method and fluorescent vital staining.
Activated carbon treatment and storage in a tank provided favourable environments for bacterial growth. The
bacterial population of the water in both the post-activated carbon treatment and the tank was analysed by
denaturing gradient gel electrophoresis (DGGE) with PCR-amplified 16S rDNA fragments including V6, -7, and -8
regions. The bacterial community structure in activated carbon treated water was stable throughout the year.
Several kinds of bacteria such as genus Aquaspirillum and Methylobacterium were found in the water after activated
carbon treatment. The bacterial community structure was changed and other bacteria such as mycobacteria were
detected after storage. Mycobacteria were quantified in water samples using real-time PCR targeting the 16S rDNA
gene. Mycobacteria were also detected in tap water and their number was increased 103–104-fold higher after
storage.
Conclusion: These data suggest the importance of culture-independent methods for quality control of water used
in pharmaceutical manufacturing.
Significance and Impact of the Study: Critical steps and specified bacteria that should be controlled in the
water supply system were recognized by culture-independent methods. These data will enable effective control of
water used in the pharmaceutical industry.

Keywords: quality control, pharmaceutical manufacturing water supply system, real-time PCR, DGGE, bacterial
population dynamics, bacterial community structure.

multistep process. A variety of treatments (e.g. filtration,

INTRODUCTION
UV light, and heat) and designs (e.g. piping slopes for
Several grades of water are used both as components of drainage, smooth uniform internal surface) are employed in
pharmaceutical products and for washing equipment. The order to remove, destroy, and control bacteria. Despite these
grade of water is selected according to its role in the process. precautions, piping, filters, tanks, and other surfaces within
The production of pharmaceutical water is a complex the water supply system provide favourable places for
bacterial adhesion and cell growth. If contaminated water is
Correspondence to: Masao Nasu, Graduate School of Pharmaceutical Science,
Osaka University, 1-6 Yamada-oka, Suita, Osaka 565-0871, Japan used in the final pharmaceutical product, these microorgan-
(e-mail: nasu@phs.osaka-u.ac.jp). isms or their metabolic products may eventually cause
ª 2004 The Society for Applied Microbiology
1124 M . K A W A I ET AL.

adverse consequences. Periodical microbial monitoring of MATERIALS AND METHODS

selected spots throughout the water supply systems is
Collection of water samples
essential to ensure high-quality, uncontaminated final
products. Purified water prepared by softener, activated carbon,
It is necessary to provide the numbers and types of reverse osmosis, and ion-exchange from tap water was
organisms that have been deemed significant relative to evaluated. Water samples were collected every month
system control and product impact. Such microorganism between September 2002 and August 2003 from four sites
information may also be useful when identifying the source within the pharmaceutical water supply system (Fig. 1).
of microbial contamination in a product or process. These This water supply system has been running from May 2002.
results should be obtained rapidly and reliably. Some type of The holding tank after granular-activated carbon treatment
microbial isolate characterization should be a required can hold 200 l of water and retention time of water in the
element of water system monitoring. holding tank is about 1 h on average. All samples were
Enumeration or identification of bacteria based solely on collected in sterile bottles and stored on ice, for immediate
culture yields valuable information. However, the formation analysis. The sample water temperature was ca 25
C
of colonies by microorganisms in pharmaceutical water is throughout the year.
dependent on growth conditions such as nutrient media and
incubation temperature (Kawai et al. 2002). Additionally, it is
Culture method
well recognized that most aquatic microorganisms including
hazardous microorganisms cannot be cultivated under con- Colony-forming bacteria were detected by a filtration
ventional conditions, although they remain viable (Byrd et al. method. Bacteria in the sample water were trapped on
1991; Oliver 1995; Dailloux et al. 1999; Kawai et al. 2002). filters (mixed cellulose ester, pore size: 0Æ45 lm) and
Several culture-independent techniques to enumerate and incubated at ca 25
C for 5 days on R2A medium (Difco,
identify bacteria have been developed in order to study Detroit, MI, USA) (Reasoner and Geldreich 1985).
natural samples. Denaturing gradient gel electrophoresis
(DGGE) has emerged as a powerful tool (Muyzer et al.
Enumeration of esterase-active bacteria
1993; Nielsen et al. 1999; Casamayor et al. 2000; Iwamoto
et al. 2000) to reveal the genetic diversity of natural Samples for enumeration of esterase-active bacteria were
microbial communities. DGGE of PCR-amplified fragments prepared with reagents supplied by Chemunex (Paris,
coding for 16S rDNA can separate DNA fragments of the France) according to the manufacturer’s instructions.
same length but different base pair sequences. The utility of Bacterial cells in sample water were vacuum-filtered onto
these techniques in the assessment of purified water used in chemfilters (pore size: 0Æ2 lm). Counterstain E/2 (CSE/
pharmaceutical manufacturing has also been reported (Ka- 2), which reduces the incidence of background fluores-
wai et al. 1999, 2002). cence, was added to the surface of the filter and the
In this study, microbial population changes and identifi- vacuum applied. Each filter was carefully loaded on a
cation of the dominant bacterial groups were investigated droplet of the labelling solution and incubated for 30 min
throughout the water supply system by culture-independent at 30
C. After labelling, each membrane was then
techniques. Such investigations revealed microbial popula- placed onto the membrane holder and analysed in the
tion, which were contaminated, re-grown, and inhabited. ChemScan laser scanner following the instructions given
This leads to controlling and specifying bacteria that should by the manufacturer. The results were microscopically
be attracted for better effective control of water supplies. validated.

Storage in the Granular activated

Tap water Softener Filtration Filtration
holding tank carbon filtration

Sampling point A
Sampling point B

Next purification steps Storage in the Revers Osmosis/ Storage in the Fig. 1 Schematic diagram of the water sup-
(ultrafiltration, distillation) Purified water
holding tank Electric Deionization holding tank ply system used in pharmaceutical
Sampling point D manufacturing processes. The locations of the
Sampling point C sampling sites are indicated by dashed arrows

ª 2004 The Society for Applied Microbiology, Journal of Applied Microbiology, 97, 1123–1131, doi:10.1111/j.1365-2672.2004.02396.x
 BACTERIA IN PHARMACEUTICAL WATER SUPPLY 1125

electrophoresis, the gels were stained for 20 min with SYBR

DNA extraction for 16S rDNA analysis
Gold nucleic acid gel stain (Molecular Probes, Eugene, OR,
Bacterial cells in water samples were vacuum-filtered onto USA) as specified by the manufacturer. DGGE gels were
polycarbonate white filters (ADVANTEC, Tokyo, Japan; scanned with FluorImager (Molecular Dynamics, Sunny-
pore size: 0Æ2 lm). These filters were placed into a sterilized vale, CA, USA) using a 488 nm argon laser and digital
tube with ca 400 ll of sterile DNA-free water. The images were obtained by Image QuaNT (version 4-2-J) to
suspensions were mixed thoroughly and subsequently frozen generate a densitometric profile.
in liquid nitrogen then thawed at room temperature three
times (total) in succession. The suspensions were used for
Sequencing of denaturing gradient gel
PCR amplification of the extracted DNA.
electrophoresis fragments
Bands of 16S rDNA fragments in the DGGE gel were
Primers and PCR amplification
excised with a razor blade. The DNA was eluted overnight
16S rDNA fragments including the V6–V8 region, were at 4
C in a 1Æ5 ml tube containing 50 ll of sterile DNA-free
amplified with primers GC-clamp-EUB f933 (5¢-GC-clamp- water. The DNA was re-amplified with EUB f933 and EUB
GCACAAGCGGTGGAGCATGTGG-3¢) and EUB r1387 r1387 primers. The PCR products were purified with
(5¢-GCCCGGGAACGTATTCACCG-3¢), which are speci- MicroSpin Columns (Amersham Biosciences, Piscataway,
fic for universally conserved bacterial 16S rDNA sequences. NJ, USA) using the protocol recommended by the manu-
For DGGE analysis of the PCR product, a 40 bp GC-rich facturer and ligated into pGEM-T Easy Vector System
sequence (GC-clamp; 5¢-CGCCCGCCGCGCGCGGCGG (Promega, Madison, WI, USA). Ligated DNA was trans-
GCGGGGCGGGGGCACGGGGGG-3¢) was attached to formed into competent Escherichia coli JM109 cells. Plasmid
the 5¢ end of primer EUB f933. PCR was performed with were isolated and purified by alkaline lysis (Sambrook et al.
Ampli Taq GoldTM (Applied Biosystem, Foster City, CA, 1989). Plasmid inserts were PCR-amplified again with
USA). The PCR mixture, containing 0Æ25 ll of Ampli Taq primers GC-clamp-EUB f933 and EUB r1387, and analysed
Gold, 20 pmol of each primer, 6 ll of 25 mmol l)1 MgCl2 by DGGE. Plasmids inserts that produced a DGGE band at
solution, 5 ll of 2 mmol l)1 each deoxyribonucleoside tri- the same position as the excised band were selected. The
phosphate, 5 ll of 10· PCR buffer, and 0Æ5 ll of 2Æ5 mg ml)1 sequence analysis of selected plasmid DNA was performed
of 8-methoxypsoralen (Sigma, Munich, Germany) (Meier with M13F primer (5¢-CGCCAGGGTTTTCCCAGT-
et al. 1993) dissolved in dimethyl sulfoxide, was made up to CACGAC-3¢).
45 ll with DNA-free water. The tubes containing the PCR
mixture were irradiated from above at a distance of 1 cm with
Preparation of control DNA for real-time PCR
long-wave (366 nm) UV light at room temperature and the
DNA suspension was added last in a 5 ll volume. A hot-start The DNA of Mycobacterium tuberculosis H37 Rv was
PCR was performed at 95
C for 10 min followed by incubated for 30 min at 37
C with RNase (100 lg ml)1),
touchdown PCR performed as follows. The annealing tem- followed by incubation with Tris-EDTA buffer (pH 8) and
perature was initially set at 66
C and then decreased by sodium dodecyl sulfate (final concentration: 0Æ5%). The
0Æ5
C every cycle until it was 56
C; 20 additional cycles were DNA was purified with phenol–chloroform–isoamyl alco-
carried out at 56
C. Denaturing was carried out at 94
C for hol. After isopropanol precipitation, the DNA was resus-
1 min. Primer annealing was performed using the scheme pended in Tris-EDTA buffer (pH 8). DNA was quantified
described above for 1 min, and primer extension was using PicoGreen (Molecular Probes) as recommended by
performed at 72
C for 3 min. The final extension step the manufacturer. Serial dilutions of genomic DNA were
occurred for 7 min at 72
C. mixed with PicoGreen dye and fluorescence was analysed
photometrically. Results were compared to known kDNA
concentrations. The number of genomic copies per PCR
Denaturing gradient gel electrophoresis analysis
mixture was determined through calculation of molecular
PCR products were loaded onto a 6Æ5% (w/v) polyacryla- weight (genome size: 4043 bp) and subsequent serial
mide gel in 1· TAE [40 mmol l)1 Tris, 20 mmol l)1 acetic dilution.
acid and 1 mmol l)1 EDTA (pH 8Æ0)]. The polyacrylamide
gels (acrylamide/bisacrylamide [37Æ5 : 1]) were made with
Primers, probes, and PCR protocol for real-time
denaturing gradients ranging from 45 to 58% for 16S rDNA
PCR
fragments (100% denaturant contained 7 mol l)1 urea and
40% formamide). The electrophoresis was run at 60
C, for 16S rDNA fragments of mycobacteria were amplified with
10 min at 20 V and subsequently for 12 h at 100 V. After primers LC1 (5¢-GAGTTTGATCCTGGCTCAGGA-3¢)
ª 2004 The Society for Applied Microbiology, Journal of Applied Microbiology, 97, 1123–1131, doi:10.1111/j.1365-2672.2004.02396.x
1126 M . K A W A I ET AL.

and LC4 (5¢-TGCACACAGGCCACAAGGGA-3¢), which 7

are specific for universal and mycobacterial 16S rDNA
sequences, respectively. The FRET LC39 (5¢-GCAACGC- 6

Bacterial number (log cells 100 ml–1)

GAAGAACCTTACCTGG-3¢-fluorescein) and LC40
(Light Cycler Red 640-5¢-TTTGACATGCACAGGACG- 5
3¢) probes were used for the detection of mycobacterium-
specific regions. Amplification PCR was performed with the
4
reagents supplied with LightCycler FastStart DNA Master
Hybridization Probes (Roche Molecular Biochemicals,
3
Mannhein, Germany). The commercially available ready-
to-use hot-start reaction mixture contains FastStart Taq
polymerase, reaction buffer, deoxynucleoside triphosphates, 2
and 3 mmol l)1 MgCl2 (final concentration). After supply-
ing primers at 18 pmol (final concentration: 0Æ9 lmol l)1) 1
per reaction mixture and DNA probes at 2 pmol (final
concentration: 100 nmol l)1) per reaction mixture, the 0
mixture was made up to 15 ll with sterile PCR-grade A B C D
water. The DNA suspension was added in a 5 ll volume. Sampling site
The amplification program was performed with a denatur- Fig. 2 Bacterial numbers in pharmaceutical water supply system.
ation step of 10 min at 95
C, followed by 50 cycles of PCR, Each bar shows the number of esterase-active bacteria (j), and colony-
with one cycle consisting of denaturation (3 s at 95
C), forming bacteria on R2A ((). Sampling sites A to D are shown in
touchdown annealing (2 s of a temperature ranging from 68 Fig. 1. These data were averages of 21 samples collected over the year
to 62
C), and extension (40 s at 72
C). For the first five
cycles, annealing was performed at 68
C (step delay) and
then reduced to 62
C with 1
C per cycle (step size). The 5–25-fold more often than colony-forming bacteria at every
temperature transition rate for all cycling steps was sampling point. In the water supply system, each result after
20
C s)1. The amplification program was followed by a activated carbon treatment yielded ca 100-fold higher
melting program of 95
C for 30 s (denaturation), 38
C for bacterial numbers than tap water. Furthermore, water
30 s (annealing), then 38 to 80
C at a transition rate of storage in a holding tank (sampling point C) resulted in
0Æ2
C s)1 with continuing monitoring of fluorescence 10-fold higher bacteria than found in water after activated
(Lachnik et al. 2002). Each LightCycler run included one carbon treatment (sampling point B). The microbial counts
capillary in which the template was replaced by water to in purified water by both methods (culture and vital
control for cross contamination, which might have occurred staining) were low, although the bacterial load in the water
at any time during preparation. supply systems obviously was increased in the earlier
treatment stages.
Nucleotide sequence accession number
Denaturing gradient gel electrophoresis analysis
The sequences obtained in this study have been deposited in
of bacterial community structure in water samples
the DDBJ database under accession nos. AB126884 to
AB126893. Water samples after activated carbon treatment (sampling
points B and C) were collected from September 2002 to
August 2003. Figure 3 shows the DGGE profiles of
RESULTS bacterial 16S rDNA fragments obtained from these samples.
At sampling point B (after activated carbon treatment), the
Enumeration of viable bacteria in purified water
banding pattern was relatively stable (Fig. 3a). Several bands
supply systems
of the same intensity appeared throughout the year.
Purified water was prepared by water softener, activated However, when the banding pattern at point B was
carbon, reverse osmosis, and ion-exchange from tap water. compared to that from point C (Fig. 3a,b, respectively),
The number of bacteria from four different sites in the distinct changes were observed. This indicates that other
purified water supply system (Fig. 2) was determined by the bacteria were found during filtration and storage. Although
conventional culture technique and fluorescent vital stain- the banding pattern was clearly unique at sampling point C,
ing. A total of 21 water samples were collected from August several bands of the same intensity appeared throughout the
2002 to August 2003. Esterase-active bacteria were detected year (e.g. Bands C5, C7).
ª 2004 The Society for Applied Microbiology, Journal of Applied Microbiology, 97, 1123–1131, doi:10.1111/j.1365-2672.2004.02396.x
 BACTERIA IN PHARMACEUTICAL WATER SUPPLY 1127

(a) (b)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 1 2 3 4* 5 6 7 8 9 10 11 12 13 14

Fig. 3 Denaturing gradient gel electrophoresis analysis of 16S rDNA fragments obtained from water samples in pharmaceutical water supply system.
(a) Water after activated carbon treatment (sampling point B); (b) Water after storage (sampling point C). Lanes: 1, September 3, 2002; 2, October 3,
2002; 3, November 9, 2002; 4, November 15, 2002; 4*, November 27, 2002; 5, December 19, 2002; 6, January 29, 2003; 7, February 14, 2003; 8,
March 14, 2003; 9, April 4, 2003; 10, April 30, 2003; 11, May 15, 2003; 12, June 10, 2003; 13, July 18, 2003; 14, August 6, 2003

Nucleotide sequence analysis and phylogenetic Table 1 Sequence similarities to closest relatives and phylogenetic
affiliation of bacteria detected in the water supply affiliation of DNA recovered from denaturing gradient gel electro-
system phoresis gel

The partial 16S rDNA fragments originating from ten Accession Similarity
bacteria (numbered DGGE bands in Fig. 3) in the water No. no. Species (%)
supply systems were excised and re-amplified by PCR. B1 AB126884 Aquaspirillum sp. 99
Partial 16S rDNA sequences of ca 450 bases were obtained B2 AB126893 Unidentified
from the DNA derived from each dominant band. Com- B3 AB126885 Methylobacterium sp. 99
parison of 16S rDNA sequences with sequences available in C1 AB126886 Unidentified
the GenBank database revealed high similarity values for C2 AB126887 Cellvibrio sp. 98
some of these sequences. Table 1 shows related organisms C3 AB126888 Mycobacterium sp. 98
and their similarities. Analysis revealed that DGGE bands C4 AB126889 Mycobacterium sp. 97
B1 and B3 are closely related to the genus Aquaspirillum and C5 AB126892 Unidentified
the genus Methylobacterium, respectively. The closest relat- C6 AB126891 Unidentified
C7 AB126890 Mycobacterium sp. 98
ive of band B2 is genus Hydrogenophaga with 90%
similarity. Sequences obtained from DGGE bands C3, C4,
and C7 are related to those of species from the genus
Mycobacterium and the band C2 is related to the genus the organism represented by band C6 belongs to Alphapro-
Cellvibrio. Mycobacteria were remarkably detected in the teobacteria, and the organism represented by band C5 is
water after storage. most like those in genus Cytophaga. They exhibited 89%
The phylogenetic affiliation of 10 sequences, obtained similarity. The organism represented by band C1 was not
from the excised DGGE bands described previously, was closely related to any previously characterized bacteria. The
analysed with the sequences available in GenBank by Clustal bacterial species most closely related was an Acidobacterium
W 1Æ74. The phylogenetic tree shown in Fig. 4 indicates that sp. clone. They exhibited 92% similarity.
ª 2004 The Society for Applied Microbiology, Journal of Applied Microbiology, 97, 1123–1131, doi:10.1111/j.1365-2672.2004.02396.x
1128 M . K A W A I ET AL.

C3
Gammaproteobacteria
C7
Mycobacterium tuberculosis
Acinetobacter calcoaceticus
C4 C2
Flexibacter tractuosus Pseudomonas aeruginosa
Methylomonas methanica
Flavobacterium spiritivorum Enterobacter aerogenes
Escherichia coli
Salmonella enterica serovar
Cytophaga hutchinsonii Typhimurium
Enterobacter cloacae
Aeromonas hydrophila

B2
C5
Comamonas acidovorans
B1
Uncultured Acidobacterium sp.
clone KCM-C-25 Comamonas terrigena
Alcaligenes faecalis
Burkholderia cepacia
Betaproteobacteria
C1 Brevundimonas C6
diminuta Agrobacterium tumefaciens
B3
0·1 Sphigomonas Methylobacterium
paucimobilis extorquens Alphaproteobacteria

Fig. 4 Phylogenetic affiliations within the domain bacteria in the water supply system as revealed by comparative analysis of 16S rDNA sequences
from denaturing gradient gel electrophoresis bands and those stored in public nucleotide databases. Sequences determined in this study are shown in
bold. Compare Fig. 3a,b for location of each band

Quantitative analysis of genus Mycobacterium DISCUSSION

Mycobacteria were mainly detected in the water samples The strict bacterial control in water supply system is
after storage in the holding tank (sampling point C). achieved to supply high bacterial quality of pharmaceutical
There are some hazardous strains such as Myco. tubercu- water. Successful quality control in water supply systems
losis. Therefore, the mycobacterial population dynamic necessitates investigation of bacterial population at each step
was determined. The quantification of mycobacteria at in the system (McFeters et al. 1993; Gapp et al. 1999;
four sites in the water supply system (sampling points A Norton and LeChevallier 2000; Kulakov et al. 2002).
through D) was performed. Evidence for linearity was Bacterial number, species, and physiological activity as well
provided by analysis of seven different concentrations of as diversity of organisms must be determined. Such
genomic bacterial DNA. The minimum amount of DNA investigations will lead to the understanding of bacterial
which can reliably be detected was determined to be 1Æ6 populations and their characterization inhabiting water
genome copies per reaction. Therefore, threshold was supply system.
established at 16 genome copies per reaction. The number
of mycobacteria was calculated by using volume of the
sample water. These results are shown in Table 2. The Table 2 Abundance of mycobacteria in the water supply system
number of mycobacteria was 103–104 genome copies ml)1 determined by real-time PCR
in the water of the holding tank (sampling point C), and a
Number of mycobacteria (genome copies ml)1)
few in tap water (<1Æ3–11 genome copies ml)1). The
number of mycobacteria in purified water (sampling point Sampling data Point A Point B Point C Point D
D) was below the quantitation limit (<16 genome copies
June 10, 2003 <1Æ3 <1Æ3 1Æ2 · 104 <1Æ3
reaction)1). No inhibition of PCR amplification were
July 18, 2003 1Æ1 · 101 <1Æ4 5Æ8 · 103 <1Æ2
confirmed by the addition of genomic DNA in all August 6, 2003 2Æ3 1Æ4 2Æ6 · 104 <1Æ3
quantitative samples.

ª 2004 The Society for Applied Microbiology, Journal of Applied Microbiology, 97, 1123–1131, doi:10.1111/j.1365-2672.2004.02396.x
 BACTERIA IN PHARMACEUTICAL WATER SUPPLY 1129

Conventional culture methods, although used for routine like river, lake, and pond. Some bacteria detected in water
monitoring over the last 50 years, have several problems. after activated carbon treatment may be populations typic-
The bacterial populations detectable by culture are affected ally found in activated carbon beds. Other bacteria were
by choice of media, incubation temperature, and culture found by storage. One bacterium was related to the genus
period. Recent studies describing the persistence of bacteria Cellvibrio, which belongs to Gammaproteobacteria. The
in aquatic environments have demonstrated that many of organisms detected in the holding tank were most closely
these organisms cannot be cultivated with conventional related to the genus Cytophaga and Acidobacterium clone
methods (Gapp et al. 1999; Colwell and Grimes 2000; which was uncultured. The genus Cytophaga, which aerobic
Kawai et al. 1999, 2002). Therefore, bacterial number and or microaerophilic, was generally present in freshwater and
the dominant bacteria were investigated by culture-inde- marine. The genus Acidobacterium was environmentally
pendent methods. widespread, especially in terrestrial environments (Barns
Ordinary tap water was purified by water softener, et al. 1999). Moreover, the genus Acidobacterium was
activated carbon, reverse osmosis, and ion-exchange to presently reported that most of them were detected by
produce purified water appropriate for use in pharmaceutical culture-independent methods (Hugenholtz et al. 1998).
manufacturing processes. Moreover, a final purification step Three main sequences obtained from the DGGE band
(ultra-filtration or distillation) was included in the water originating from the water in the holding tank were genus
supply system to produce highly purified water. The water Mycobacterium.
was evaluated by conventional culture method and fluores- The detected Mycobacterium may be considered an
cent vital staining (Mignon-Godefroy et al. 1997; Gapp et al. indigenous population typically found throughout the year
1999; Wallner et al. 1999). Bacteria were not detected in this in the holding tank. Mycobacteria, except the tubercle bacilli
highly purified water by these methods (data not shown). responsible for tuberculosis, are usually referred to as
Fluorescent vital staining techniques detected more physi- atypical or non-tuberculosis mycobacteria (NTM). NTM
ologically active bacteria than conventional culture methods are ubiquitous and have been recovered from a wide variety
in the water at all four sampling points located within the of environmental sources, including water, soil, dust, and
purified water supply system. Bacteria with esterase activity aerosols (Covert et al. 1999; Dailloux et al. 1999; Falkinham
may have major effects on pharmaceutical products such as et al. 2001). There are many reports that mycobacteria have
pyrogen contamination or alteration of active ingredients been isolated from public water distribution systems,
and, thus, should be accurately detected. The bacteria were including home distribution systems, hot and cold water
significantly increased in the activated carbon treatment and taps, ice machines, and showerheads (Dailloux et al. 1999;
holding tank water storage. The activated carbon bed adsorbs Le Dantec et al. 2002a). Mycobacteria are highly resistant to
low molecular weight organic materials and reduces oxidizing chlorine disinfection, especially in low-nutrient media such
additives, such as chlorine compounds, and removes them as that encountered in water (Taylor et al. 2000; Le Dantec
from the water. Bacteria then proliferate under these et al. 2002b). Moreover, there are some species of myco-
conditions such as on the activated carbon bed and in the bacteria, which rapidly form a biofilm (Hall-Stoodly and
storage tank. Therefore, these sites provide opportunity for Lappin-Scott 1998). Although the mechanisms responsible
critical steps in microbiological control. for the survival of mycobacteria in water are not well
In recent studies, several molecular techniques including understood, NMT in water distribution systems are able to
PCR-DGGE analysis helped to identify microorganisms colonize, survive, and grow in tap water.
without cultivation and revealed the enormous extent of Mycobacteria in the water samples were quantified
microbiological diversity. It has been shown previously that precisely using real-time PCR amplification and product
relatively short sequences, such as those obtained from detection by FRET probe. The number of mycobacteria in
DGGE analysis, are sufficient for an approximate phylo- tap water was <1Æ3–11 genome copies ml)1. The number of
genetic identification (Ward et al. 1990; Schmidt et al. 1991; mycobacteria was <1Æ3 genome copies ml)1 (under the
Nielsen et al. 1999). The bacterial diversity in the water quantification limit) after activated carbon treatment; how-
supply system was determined by PCR-DGGE and iden- ever, after storage, the number of mycobacteria increased to
tification of each dominant bacteria was performed. DGGE 5Æ8 · 103 up to 2Æ6 · 104. There was likely a mycobacterial
profiles showed that the diversity of the bacterial community biofilm in the holding tank or piping. Then, the number of
in the water after activated carbon treatment was relatively mycobacteria in purified water was decreased dramatically.
stable throughout the year. The genus Methylobacterium The removal of mycobacteria occurred efficiently during
belonging to Alphaproteobacteria was found in water after reverse osmosis filtration and electro-deionization treatment.
activated carbon treatment throughout the year. The genus This water system has next purification steps (ultrafiltra-
Aquaspirillum was also found populations. This organism tion or distillation) to supply water used in pharmaceutical
belonging to Betaproteobacteria was from aquatic habitats manufacturing process. Any bacterium was not detected in
ª 2004 The Society for Applied Microbiology, Journal of Applied Microbiology, 97, 1123–1131, doi:10.1111/j.1365-2672.2004.02396.x
1130 M . K A W A I ET AL.

these water. However, purified water has been used in bulk Hugenholtz, P., Goebel, B.M. and Pace, N.R. (1998) Impact of
pharmaceutical chemicals and non-sterile products. It culture-independent studies on the emerging phylogenetic view of
should considered that bacteria in purified water have bacterial diversity. Journal of Bacteriology 180, 4765–4774.
potential to effect on pharmaceutical products from Iwamoto, T., Tani, K., Nakamura, K., Suzuki, Y., Kitagawa, M.,
Eguchi, M. and Nasu, M. (2000) Monitoring impact of in situ
pathogenecity or their metabolic product without our
biostimulation treatment on groundwater bacterial community by
observation. Therefore, water supply systems used in
DGGE. FEMS Microbiology Ecology 32, 129–141.
pharmaceutical manufacturing processes controlled and Kawai, M., Yamaguchi, N. and Nasu, M. (1999) Rapid enumeration of
monitored for these bacteria. physiologically active bacteria in purified water used in the
It is difficult for routine monitoring based on conventional pharmaceutical manufacturing process. Journal of Applied Microbio-
culture methods to detect mycobacteria, due to the media logy 86, 496–504.
required and their slow growth. The PCR-DGGE approach Kawai, M., Matsutera, E., Kanda, H., Yamaguchi, N., Tani, K. and
will provide early warning of an impending problem. Rapid Nasu, M. (2002) 16S ribosomal DNA-based analysis of bacterial
and convenient techniques such as real-time PCR should be diversity in purified water used in pharmaceutical manufacturing
useful for routine monitoring of bacterial contamination in processes by PCR and denaturing gradient gel electrophoresis.
purified water used in the pharmaceutical industry. Applied and Environmental Microbiology 68, 699–704.
Kulakov, L.A., McAlister, M.B., Ogden, K.L., Larkin, M.J. and
O’Hanlon, J.F. (2002) Analysis of bacteria contaminating ultrapure
ACKNOWLEDGEMENTS water in industrial systems. Applied and Environmental Microbiology
68, 1548–1555.
We thank Dr Tomotada Iwamoto, Kobe Institute of Health Lachnik, J., Ackermann, B., Bohrssen, A., Maass, S., Dephaus, C.,
Department of Bacteriology, for providing mycobacterial Puncken, A., Sterman, M. and Bange, F. (2002) Rapid-cycle PCR
DNA. and fluorimetry for detection of Mycobacteria. Journal of Clinical
Microbiology 40, 3364–3373.
Le Dantec, C.L., Duguet, J.-P., Montirl, A., Dumoutier, N., Dubrou,
REFERENCES S. and Vincent, V. (2002a) Occurrence of mycobacteria in water
Barns, S.M., Takara, S.L. and Kuske, C.R. (1999) Wide distribution treatment lines and in water distribution systems. Applied and
and diversity of members of the bacterial kingdom Acidobacterium in Environmental Microbiology 68, 5318–5325.
the environment. Applied and Environmental Microbiology 65, 1731– Le Dantec, C., Duguet, J.-P., Montiel, A., Dumoutier, N., Dubrou, S.
1737. and Vincent, V. (2002b) Chlorine disinfection of atypical Mycobac-
Byrd, J.J., Xu, H.-S. and Colwell, R.R. (1991) Viable but noncultur- teria isolated from a water distribution system. Applied and
able bacteria in drinking water. Applied and Environmental Micro- Environmental Microbiology 68, 1025–1032.
biology 57, 875–878. McFeters, G.A., Broadaway, S.C., Pyle, B.H. and Egozy, Y. (1993).
Casamayor, E.O., Schäfer, H., Bañeras, L., Pedrós-alió, C. and Distribution of bacteria within operating laboratory water purifi-
Muyzer, G. (2000) Identification of and spatio-temporal differences cation systems. Applied and Environmental Microbiology 59, 1410–
between microbial assemblages from two neighboring sulfurous 1415.
lakes: comparison by microscopy and denaturing gradient gel Meier, A., Persing, D.H., Finken, M. and Bottger, E.C. (1993)
electrophoresis. Applied and Environmental Microbiology 66, 499–508. Elimination of contaminating DNA within polymerase chain
Colwell, R.R. and Grimes, D.J. (2000) Nonculturable Microorganisms in reaction reagents: implications for a general approach to detection
the Environment. Washington, DC, USA: ASM press. of uncultured pathogens. Journal of Clinical Microbiology 31, 646–
Covert, T.C., Rodgers, M.R., Reyes, A.L. and Stelma, G.N. Jr. (1999) 652.
Occurrence of nontuberculous mycobacteria in environmental Mignon-Godefroy, K., Guillet, J-G. and Butor, C. (1997) Solid
samples. Applied and Environmental Microbiology 65, 2492–2496. phase cytometry for detection of rare events. Cytometry 27,
Dailloux, M., Laurain, C., Weber, M. and Hartemnn, P.H. (1999) 336–344.
Water and nontuberculous mycobacteria. Water Research 33, 2219– Muyzer, G., de Waal, E.C. and Uitterlinden, A.G. (1993) Profiling of
2228. complex microbial populations by denaturing gradient gel electro-
Gapp, G., Guyomard, S., Nabet, P. and Scouvart, J. (1999) Evaluation phoresis analysis of polymerase chain reaction-amplified genes
of the applications of a system for real-time microbial analysis of coding for 16S rRNA. Applied and Environmental Microbiology 59,
pharmaceutical water systems. European Journal of Parenteral Science 695–700.
4, 131–136. Nielsen, A.T., Liu, W.-T., Filipe, C., Grady, L. Jr., Molin, S. and
Falkinham, J.O. III, Norton, C.D. and LeChevallier, M.W. (2001) Stahl, D.A. (1999) Identification of a novel group of bacteria in
Factors influencing numbers of Mycobacterium avium, Mycobacterium sludge from a deteriorated biological phosphorus removal reactor.
intracellulare, and other mycobacteria in drinking water distribution Applied and Environmental Microbiology 65, 1251–1258.
systems. Applied and Environmental Microbiology 67, 1225–1231. Norton, C.D. and LeChevallier, M.W. (2000). A pilot study of
Hall-Stoodly, L. and Lappin-Scott, H. (1998) Biofilm formation by the bacteriological population changes through potable water treatment
rapidly growing mycobacterial species Mycobacterium fortuitum. and distribution. Applied and Environmental Microbiology 66, 268–
FEMS Microbiology Letters 168, 77–84. 276.

ª 2004 The Society for Applied Microbiology, Journal of Applied Microbiology, 97, 1123–1131, doi:10.1111/j.1365-2672.2004.02396.x
 BACTERIA IN PHARMACEUTICAL WATER SUPPLY 1131

Oliver, D.J. (1995) The viable but non-culturable state in the human Taylor, R.H., Falkinham, J.O. III, Norton, C.D. and LeChevallier,
pathogen Vibrio vulnificus. FEMS Microbiology Letters 133, 203–208. M.W. (2000) Chlorine, chloramines, chlorine dioxide, and ozone
Reasoner, D.J. and Geldreich, E.E. (1985) A new medium for the susceptibility of Mycobacterium avium. Applied and Environmental
enumeration and subculture of bacteria from potable water. Applied Microbiology 66, 1702–1705.
and Environmental Microbiology 49, 1–7. Wallner, G., Tillmann, D. and Haberer, K. (1999) Evaluation of the
Sambrook, J., Fritsch, E.F. and Maniatis, T. (1989) Molecular Cloning chemscan system for rapid microbiological analysis of pharmaceu-
a Laboratory Manual, 2nd edn. New York: Cold Spring Harbor tical water. PDA Journal of Pharmaceutical Science and Technology
Laboratory Press. 53, 70–74.
Schmidt, T.M., DeLong, E.F. and Pace, N.R. (1991) Analysis of a Ward, D.M., Weller, R. and Bateson, M.M. (1990) 16S rRNA
marine picoplankton community by 16S RNA gene cloning and sequences reveal numerous uncultured microorganisms in a natural
sequencing. Journal of Bacteriology 173, 4371–4378. community. Nature 345, 63–65.

ª 2004 The Society for Applied Microbiology, Journal of Applied Microbiology, 97, 1123–1131, doi:10.1111/j.1365-2672.2004.02396.x