Investigation of Natural Biofilms Formed During the Production of Drinking

ARTICLE IN PRESS

Water Research 38 (2004) 1197–1206

Investigation of natural bioﬁlms formed during the production of drinking water from surface water embankment ﬁltration Farahnaz Emtiazi, Thomas Schwartz*, Silke Mareike Marten, Peter Krolla-Sidenstein, Ursula Obst

Department of Environmental Microbiology, Forschungszentrum Karlsruhe GmbH, Institute for Technical Chemistry–Water Technology and Geotechnology Division, P.O. Box 3640, Karlsruhe D-76021, Germany

Received 5 June 2003; received in revised form 8 October 2003; accepted 13 October 2003

Abstract

Populations of bacteria in biofilms from different sites of a drinking water production system were analysed. Polymerase chain reaction (PCR) and denaturing gradient gel electrophoresis (DGGE) analyses revealed changing DNA band patterns, suggesting a population shift during bank filtration and processing at the waterworks. In addition, common DNA bands that were attributed to ubiquitous bacteria were found. Biofilms even developed directly after UV disinfection (1–2 m distance). Their DNA band patterns only partly agreed with those of the biofilms from the downstream distribution system. Opportunistic pathogenic bacteria in biofilms were analysed using PCR and Southern blot hybridisation (SBH). Surface water appeared to have a direct influence on the composition of biofilms in the drinking water distribution system. In spite of preceding filtration and UV disinfection, opportunistic pathogens such as atypical mycobacteria and Legionella spp. were found in biofilms of drinking water, and Pseudomonas aeruginosa was detected sporadically. Enterococci were not found in any biofilm. Bacterial cell counts in the biofilms from surface water to drinking water dropped significantly, and esterase and alanine-aminopeptidase activity decreased. b-glucosidase activity was not found in the biofilms. Contrary to the results for planktonic bacteria, inhibitory effects were not observed in biofilms. This suggested an increased tolerance of biofilm bacteria against toxic compounds. r 2003 Elsevier Ltd. All rights reserved.

Keywords: Bioﬁlms; Population shifts; Opportunistic pathogenic bacteria; Enzyme activities; Molecular-biological techniques

1. Introduction reduced and eliminated. Research results obtained with respect to the hydraulic, physico-chemical, and chemi- Bank filtration and artificial groundwater enrichment cal/biological processes revealed a good and stable long- are frequently employed for the production of drinking term cleaning efficiency of bank filtration. This also water. When surface water enters the aquifer via an applies to the removal of particles, pathogens, a number underground passage due to potential gradients gener- of organic, and most trace substances [1] except for some ated by wells, this is referred to as bank filtration. individual organic polar compounds which are persis- During underground passage, a variety of chemical and tent in a nearly unrestricted manner [2]. Microorganisms biological processes occur, by which compounds are significantly contribute to cleaning during the underground passage by enzymatic degradation or partial *Corresponding author. Tel.: +49-7247-826-802; fax: +49- metabolism of water impurities and by physico-chemical 7247-826-858. processes, e.g. adsorption. However, specific manipula- E-mail address: [email protected] tion and use of these microbial elimination processes (T. Schwartz). have failed, and knowledge of the processes is too

0043-1354/$ - see front matter r 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2003.10.056 ARTICLE IN PRESS 1198 F. Emtiazi et al. / Water Research 38 (2004) 1197–1206 incomplete. Still these biological processes during 2. Material and methods natural underground passage are suited as environmen- tally compatible and low-cost stage for sustainable 2.1. Biofilm sampling method drinking water processing. This is not only true for Central Europe with its large river basins, but also for For biofilm formation in drinking water distribution many Third World and threshold countries that have to systems, special devices were used [6,7]. The core of these produce drinking water from highly contaminated devices consisted of a hollow stainless steel cylindrical surface waters. Yet few data are available for the element, where stainless steel bolts holding steel platelets comparative description of function and population for biofilm growth were screwed into place. The platelet changes of adhesive bacteria (biofilms) during the (15 mm 30 mm 2.5 mm) was attached to the end of a underground passage and in the downstream drinking bolt with a small screw. To study natural biofilms, these water production facilities. devices were installed at different sampling points in the Among the major ecological units of aquatic systems, waterworks, i.e. downstream of the granular activated which affect water quality, are biofilms that cover carbon filtration (GAC) and downstream of UV practically all accessible wet surfaces. Biofilms may be disinfection (DIS). Two devices were installed in house composed of algae, bacteria, fungi and other eukaryotic branch connections within the municipal drinking water microorganisms, and cover surfaces, e.g. in storage distribution system (DW1 and DW2) 1–2 km down- basins, filter systems, and drinking water distribution stream of the waterworks. Additionally, platelets lines. As biofilms represent the predominating biological (170 mm 20 mm 2 mm) were incubated in surface form of life in habitats of water and soil, it is urgently water used for embankment filtration (SW, Rhine river) required to improve the understanding of the structure and raw water (RW) from a well downstream of bank and function of these biocoenoses. There is a lack of filtration. To examine the biofilm populations, the methods for the detection of adhesive bacteria in platelets were removed after 3–4 weeks, the biofilms classical drinking water analysis. Cultivation processes were scraped from the surface using a sterile cell scraper cover a small part of the natural planktonic population (Nunc, Wiesbaden, Germany) and suspended in 10 ml only [3]. The limitation of nutrients and environmental water from the sampling site. The biofilm suspensions stress situations may induce physiological and morpho- were centrifuged for 10 min at 10 000g. The pellet was logical changes in many aquatic bacteria. A dormant resuspended in 1 ml sterile water. DNA was extracted status was described for pathogenic bacteria, such from biofilms from the surface water and horizontal well as Campylobacter spp., E. coli, and Legionella pneumo- (QIAamps DNA Mini Kit-50, Qiagen). These DNA phila [4,5] which makes their isolation and identi- preparations were used as template in PCR. The biofilm fication problematic. Therefore, cultivation-independent suspensions from GAC, DIS, and DW were used molecular-biological methods targeting nucleic acids are without extraction for subsequent molecular-biological required in addition to biochemico-metabolic analyses analysis. [5]. Our investigations covered biofilms from surface 2.2. PCR-DGGE analysis of different aquatic biofilms water, raw water after bank filtration, processed drinking water prior to and after UV disinfection as PCR primers targeting the regions (V1-3) of 16S well as from the downstream municipal distribution rDNA of bacteria were used for biofilm analysis system. As an alternative to classical cultivation (Table 1) [8,9]. For DGGE analysis, a sequence of methods, molecular-biological methods were applied multiple guanines (G) and cytosines (C) was attached to with the DNA as target molecule, which exists in each the 50end of the forward primer [10]. A ‘‘touch-down’’ cell irrespective of its physiological state. Oligonucleo- PCR profile published by Kilb et al. [9] was applied. The tide probes developed for a number of environmental final 100 ml reaction mixture contained 2.5 U of HotStar bacteria and pathogens allowed to comprehensively Taq-DNA polymerase (Qiagen, Germany), 30 pmol of describe bacterial populations or identify pathogens each primer, 1 PCR buffer, 1.5 mM MgCl2, 200 mM without a cultivation pre-enrichment. By means of the dNTPs, and 10 ml biofilm suspension or template DNA. polymerase chain reaction (PCR), denaturing gradient A GeneAmp PCR System 9700 (Applied Biosystems) gel electrophoresis (DGGE), and southern blot hybridi- was used for PCR. Aliquots of 5 ml were analysed by sation (SBH), population shifts of the bacterial biofilms electrophoresis in a 1% agarose gel containing ethidium were investigated and the occurrence of facultatively bromide to check the sizes and amounts of the pathogenic bacteria, such as legionellae, mycobacteria, amplicons. PCR products were purified using phenol- and enterococci, was studied. Enzyme activities were chloroform-isoamyl alcohol (25:24:1 vol.), precipitated measured in biofilm samples to determine the metabolic with isopropanol, washed with 70% ethanol, dried, and performance and physiological states of the bacterial resuspended in 20 ml sterile water. DGGE analysis of biocoenoses. PCR products (B526 bp and 349 bp) was performed ARTICLE IN PRESS F. Emtiazi et al. / Water Research 38 (2004) 1197–1206 1199

Table 1 PCR primers and SBH probes used

Primer/probes Organism Sequence (50-30) Target site Reference

Primer GC27Fa Bacteria AGAGTTTGATCMTGGCTCAGb 8–27 (16S rDNA) [9] Primer 517R Bacteria ATTACCGCGGCTGCTGG 534–517 (16S rDNA) [9] Primer 342R Bacteria CTGCTGCCTCCCGTAG 357–342 (16S rDNA) [8] Primer-Eub338F Bacteria ACTCCTACGGGAGGCAGC 355–338 (16S rDNA) [8] Primer-Myc1 Mycobacterium (Genus) AAGGAAGGAAACCCACAC 829–847 (16S rDNA) [12] Probe-Myc2 Facultative pathogenic mycobacteria CCACCTACCGTCAATC 476–492 (16S rDNA) [12] Probe-Myc3 Mycobacterium (Genus) TTTCACGAACAACGCGACAA 609–590 (16S rDNA) [12] Primer-616V Bacteria AGAGTTTGATYMTGGCTCAGb 8–27 (16S rDNA) [6] Primer-630R Bacteria CAKAAAGGAGGTGATCCb 1542–1529 (16S rDNA) [6] Probe-Leg-705 Legionella sp. CTGGTGTTCCTTCCGATC 705–722 (16S rDNA) [13] Probe-Legpneu Legionella pneumophila ATCTGACCGTCCCAGGTT 16S rDNA [14] Primer-1019V Bacteria TAGCTGGTTCTCYBCGAAb 807–824 (23S rDNA) [15] Primer-1028R Bacteria CCTTCTCCCGAAGTTACGG 1691–1709 (23S rDNA) [15] Primer-118V Bacteria TCYGAATGGGGNAAC 121–136 (23S rDNA) [15] Primer-367R Bacteria CACGTGTYCCGCCGTACTCb 374–394 (23S rDNA) [15] Probe-DB6 Enterococcus faecium CACACAATCGTAACATCCTA 140–158 (23S rDNA) [15] Probe-DB8 Enterococcus faecalis TAGGTGTTGTTAGCATTTCG 342–361 (23S rDNA) [15] Probe-Pae1500 Pseudomonas aeruginosa AATCCGGGGTTTCAAGGC 1500–1517 (23S rDNA) [15]

a GC clamp: 50-CGC CCG CCG CGC CCC GCG CCC GTC CCG CCG CCG CCC CCG CCC C-30. b Wobbles according to IUPAC: M means A or C, Y means C or T, K means G or T, B means C, G or T, N means A, C, G or T.

with the D-Code-System (Bio-Rad Laboratories GmbH, 2.3. PCR and southern blot hybridisation (SBH) Germany) using polyacrylamide gels containing a gradient of 40–70% urea for primer combination All primers and probes used for the identification of GC27F/517R and 50–70% for primer combination hygienically relevant bacteria in different biofilms are GC27F/342R. 10 ml of the PCR products were loaded listed in Table 1. For the identification of mycobacteria on the gel. DGGE gels were run in 1 TAE buffer in different biofilms, PCR primer combinations were (40 mM Tris, 20 mM acetate, 1 mM EDTA, pH 7.4) at used according to Schwartz et al. [12]. The oligonucleo- 56C and 70 V for 16 h and stained with SYBRt Green I tides Myc1 and Myc3 were genus-specific for atypical (Sigma). Finally, the stained gels were analysed im- mycobacteria, whereas Myc2 was specific for faculta- mediately using the LumiImager Working Stationt tively pathogenic mycobacteria. For the detection of (Roche Diagnostics, Germany). DGGE fingerprints mycobacteria, semi-specific amplification of 16S rDNA were scored manually by the presence or absence of using the PCR primer Myc1 genus-specific for myco- bands, independently of intensity. The number of bacteria together with the primer Eub338R specific for bands is supposed to correspond roughly to the number most Eubacteria was accomplished. This primer set of the bacterial species in the microbial biofilm. amplified 500 bp amplicons. After amplification, PCR Community similarities were calculated in a pair-wise products were hybridised with a specific probe Myc3 for 1 manner using the Dice coefficient: Cs ¼ 2jða þ bÞ ; the detection of atypical mycobacteria and with probe where j is the number of bands common to samples A Myc2 for the detection of facultatively pathogenic and B, and a and b are the total numbers of bands in mycobacteria. The stringency of this detection was samples A and B, respectively. This index ranges from 0 adapted for Myc2 with 20% formamide content of the (no common bands) to 1 (100% similarity of band hybridisation buffer. In modification of the published patterns) [11]. PCR protocol, HotStar Polymerase from Qiagen (Hil- Intensively stained bands were excised, gel slices den, Germany) was used. crushed in 20 ml of sterile water, and equilibrated For identification of Legionella spp., DNA was overnight at 4C. 10 ml of the DNA extract were re- amplified using the universal primer combination amplified by PCR and re-run on DGGE gels to verify 616V/630R [6]. The PCR mix was heated up to 95C the purity of the PCR re-amplification product. PCR for 15 min to activate the polymerase. The profile of the products were purified with the QIAquick Spin PCR 35 cycles was as follows: Denaturing at 94C for 1 min, purification kit (Qiagen, Germany) and sequenced by annealing at 50C for 2 min, DNA extension at 72C for GENterprise (Mainz, Germany). 3 min.Probes Leg705 specific for Legionella spp. and ARTICLE IN PRESS 1200 F. Emtiazi et al. / Water Research 38 (2004) 1197–1206

Legpneu specific for Legionella pneumophila were used (final concentration: 2.4 108 mol/assay), l-alanin-4- for identification by Southern blot hybridisation [13,14]. methoxy-b-naphthylamide (1.2 107 mol/assay), and To confirm the presence or absence of Legionella spp. 4-nitrophenyl-b-d-glucopyranoside (3.32 108 mol/as- and L. pneumophila, the stringency of the subsequent say) as substrate, respectively [16]. Samples were hybridisation reaction was evaluated by testing forma- incubated at 25C for 120 min for the detection of mide contents ranging from 0%, 15%, 20% to 25% with esterase and at 30C for 19 h for the detection of cultures of Legionella pneumophila as reference strain. alanine–aminopeptidase and b-glucosidase. The concen- For high specificity the used formamide concentration tration of the released fluorescein and 4-methoxy-b- was 15% for probe Leg705 and 25% for probe Legpneu. naphthylamine was determined fluorimetrically at wa- Primer combination 1019V/1028R was used for the velengths of 485 nm and 355 nm (Labsystems Fluoros- amplification of bacterial 23S rDNA prior to the kan). The released 4-nitrophenol was measured detection of Pseudomonas aeruginosa. The PCR profile photometrically at 405 nm (Labsystems Multiskan MS). from Frahm et al. [15] was applied. For the identification of Enterococcus faecium and Enterococcus faecalis bacterial 23S rDNA was first amplified using primer pair 3. Results and discussion 118V and 367R. PCR was carried out as follows: activation of the Taq polymerase at 95C for 15 min, 3.1. Population analysis of biofilms by PCR and DGGE then 35 cycles consisting of denaturation at 94C for 1 min, annealing at 44C for 2 min, and extension at Molecular-biological methods allow to study the 72C for 3 min. For the specific detection of Enter- composition of the microflora in natural biofilms of ococcus faecium, Enterococcus faecalis, and Pseudomo- drinking water conditioning facilities without cultiva- nas aeruginosa, probes DB6, DB8, and Pae1500 were tion. In comparison to cultivation techniques, charac- used, respectively (Table. 1). Different formamide con- terisation of biofilms by means of PCR and DGGE centrations (0%, 20%, 25%, 30%) were tested in the analysis is rapid and may be more comprehensive hybridisation buffer to adjust stringent conditions. process. Within the framework of the present study, a Enterococcus faecium, Enterococcus faecalis and Pseu- touchdown technique was employed for PCR amplifica- domonas aeruginosa cultures were used as references. tion using two different primer combinations. Touch- For high specificity the used formamide concentration down techniques have been considered useful to amplify was 20% for DB6, 30% for DB8, and 30% for Pae1500. also rare DNA targets. Muyzer et al. [8] report that Southern blot hybridisation was performed according bacterial populations that make up at least 1% of the to Schwartz et al. [12]. Detection was accomplished by a total community can be detected by PCR-DGGE. LumiImager Working Stationt (Roche Diagnostics, In this study, a PCR-DGGE protocol was developed Mannheim, Germany). for the detection of population shifts of bacteria in different biofilms during the embankment purification 2.4. Total bacterial cell count and enzyme activities process from surface water to drinking water. The DGGE band patterns of the PCR product obtained with Total bacterial cells were counted using DAPI (40,6- the primer set GC27F and 517R are presented in Fig. 1. diamidine-20-phenyl-indole-dihydrochloride, Merck, The length of the PCR products was 526 bp. After Germany) staining [6]. separating the amplicons by DGGE, the number of Microbial esterase, alanine–aminopeptidase, and b- bands varied from 11 to 21, depending on the biofilm glucosidase activities and total cell counts were investi- sampling point analysed. DGGE profiles of a 349 bp gated in the different biofilms from all sampling points. PCR product obtained with the primer set GC27F and To determine concentration-effect correlations, bio- 342R, were different and showed fewer DNA bands films were scraped from the platelet substrates using (data not shown). Bands obtained with the primer set sterile cell scrapers (Nunc, Wiesbaden, Germany), GC27F/517R were more intense than those with suspended in 10 ml PBS (NaCl 8 g, KCl 0.2 g, KH2PO4 GC27F/342R. When DGGE profiles generated with 0.2 g, and Na2HPO4 2H2O 1.44 g per litre, pH 7.4) the two primer systems were compared, the similarities buffer, and enzyme activities were measured with an of biofilms decreased again continuously from surface aliquot of 200 ml biofilm suspension in microtiter plates. water to drinking water. In spite of PCR product The enzyme activity values were calculated as substrate purification, a substantial number of weak DNA bands turnover per hour and cm2 surface area due to the from less abundant bacteria was generally visible in all different sizes of used platelets. Biofilm suspensions were samples independently of the primer set used (Fig. 1, diluted with sodium chloride solution (0.14 mol l1)in lanes 2, 7). PCR efficiency may have been influenced by 10% steps from 0 to 100% for dose response experi- PCR inhibitors present particularly in the biofilm ments. Esterase, alanine–aminopeptidase, and b-glyco- samples, resulting in DGGE profiles with weaker sidase activities were detected using fluorescein diacetate DNA bands and reduced band numbers. It is hence ARTICLE IN PRESS F. Emtiazi et al. / Water Research 38 (2004) 1197–1206 1201

Fig. 1. DGGE band pattern of biofilm bacteria from different sampling sites using primer combination GC27F/517R after SYBR Green staining. Lane 1: Eschericha coli as positive control; lane 2: surface water SW1; lane 3: surface water SW2; lane 4: raw water RW; lane 5: drinking water after activated carbon filtration GAC; lane 6: drinking water after disinfection DIS1; lane 7: drinking water after disinfection DIS2; lane 8: distribution system DW1; lane 9: distribution system DW2. Arrows indicate bands excised for sequencing. pointed out that the different biofilm preparation ground passage. In some cases, the similarity values techniques (DNA extraction or direct analysis of biofilm generated with the two primer sets differed considerably. aliquots) may have had an effect on the abundance of This may be explained by differences in primer DNA band patterns. Generally, the PCR-DGGE mismatches between the two PCR systems. fingerprints of the biofilms did not exhibit any unique To obtain more detailed information about some of pattern from one sampling point to the next. Different the community members in these biofilms, a number of patterns were determined even within the distribution strong DGGE bands (see arrows in Fig. 1) were excised system, although the distribution area was supplied with from the gel, re-amplified, subjected to electrophoresis, the same drinking water (Fig. 1, lanes 8 and 9). and sequenced. Comparison of the 16S rDNA sequences Dice coefficients (Cs) for the pair-wise comparison of with sequences available in GenBank databases indi- biofilm community composition similarities are listed in cated that beta-Proteobacteria were most abundant Table 2. As expected, the Cs index for similar aquatic (similarity between 87% and 99%) (Table 3). Sequences habitats was higher than the similarity index between with high similarity to Dechloromonas (97–99%) were different aquatic compartments, with similarity Cs found in high abundance in the biofilms from raw water, values decreasing continuously from surface water to drinking water after UV disinfection, and the municipal drinking water. This indicates that surface water drinking water distribution system. Sequences with bacteria passing the different conditioning steps also similarity to Pseudomonas sp. and Acidovorax sp. were occur in the subsequent drinking water facilities. identified in biofilms from surface water only, and Oligotrophic drinking water systems may be preferred sequences similar to Azospirillum doebereinerae and by certain environmental groups of bacteria from Pseudomonas diminuta (96–99%) in biofilms from raw surface water. The lowest Cs values were measured water and disinfected drinking water. One DNA band of between raw water biofilms after embankment filtration a gamma-Proteobacterium (99% similarity to Pseudo- and biofilms within the distribution system. This may be monas marginalis) originated from a biofilm from the due to the more anaerobic conditions during under- municipal drinking water distribution system. The ARTICLE IN PRESS 1202 F. Emtiazi et al. / Water Research 38 (2004) 1197–1206

Table 2 Comparison of PCR-DGGE fingerprints from different biofilms by the Dice coefficient

Sample SW RW GAC DIS DW1 DW2

SWa 1.0 SWb 1.0 RWa 0.50 1.0 RWb 0.48 1.0 GACa 0.50 0.29 1.0 GACb 0.46 0.59 1.0 DISa 0.37 0.28 0.47 1.0 DISb 0.33 0.40 0.54 1.0 DW1a 0.32 0.26 0.45 0.46 1.0 DW1b 0.30 0.33 0.33 0.36 1.0 DW2a 0.41 0.37 0.46 0.46 0.56 1.0 DW2b 0.36 0.24 0.38 0.25 0.45 1.0

Two primer combinations GC27F/517R (a) and GC27F/342R (b) were used; for each primer combination 2-3 experiments were performed per sampling point. SW: surface water; RW: raw water; GAC: downstream of granular activated carbon ﬁlters; DIS: downstream of UV disinfection; DW1 and DW2: within the distribution system.

Table 3 Similarity of sequences of selected DGGE bands, as determined by BLAST nucleotide search

Selected DGGE bandsa Related sequence Similarity (%) Subclass Habitatb

P1 Nitrosomonas oligotropha 97 Beta-Proteobacteria RW P2 Beta Proteobacterium CRE-PA84 91 Beta-Proteobacteria DW1 P3 Beta Proteobacterium Spb298 99 Beta-Proteobacteria DIS P4 Pseudomonas marginalis, NZCX27 99 Gamma-Proteobacteria DW2 P5 Azospirillum doebereineri 99 Alpha-Proteobacteria DIS P6 Sludge bacterium S21 99 Beta-Proteobacteria DIS P7 Beta Proteobacterium A0640 97 Beta-Proteobacteria RW P8 Beta Proteobacterium UCT N117 99 Beta-Proteobacteria GAC P10 Pseudomonas sp. C96E 99 Gamma-Proteobacteria SW P11 Acidovorax sp. G8B1 97 Beta-Proteobacteria SW P12 Pseudomonas diminuta 98 Alpha-Proteobacteria RW P13 Dechloromonas spp. 98 Beta-Proteobacteria DW2 P14 Eubacterium F13.40 94 Beta-Proteobacteria GAC B1 Bacterium GKS2-174 87 Beta-Proteobacteria DIS B4b Azospirillum sp. Mat2-1a 95 Alpha-Proteobacteria DIS B6 Dechloromonas spp. 99 Beta-Proteobacteria DIS B8 Dechloromonas spp. 98 Beta-Proteobacteria DW2 B9 Bacterium clone IAFDn47 84 Beta-Proteobacteria DW1 B11 Pseudomonas spinosa, ATCC 93 Beta-Proteobacteria SW B12 Pseudomonas sp. clone Pseud3a 99 Gamma-Proteobacteria SW B12b Pseudomonas sp. clone Pseud3a 99 Gamma-Proteobacteria SW B13 Bacterium BVB72 98 Beta-Proteobacteria SW B15a Dechloromonas spp. 98 Beta-Proteobacteria RW B17a Dechloromonas spp. 96 Beta-Proteobacteria DW2 B17b Dechloromonas spp. 97 Beta-Proteobacteria DW2 B18a Brevundimonas sp. Dcm7A 99 Alpha-Proteobacteria DW2 B18b Alpha Proteobacterium FL14F11 96 Alpha-Proteobacteria DW2

a Band numbers P1 to P14 were amplified using primer set GC27F/517R (arrows in Fig. 1) and band numbers B1 to B18b were amplified using primer set GC27F/342R (not shown). b SW, surface water; RW, raw water; GAC, drinking water after activated carbon filtration; DIS, drinking water after disinfection; DW, drinking water distribution system. ARTICLE IN PRESS F. Emtiazi et al. / Water Research 38 (2004) 1197–1206 1203 results confirm previous studies on biofilms grown at the facultatively pathogenic mycobacteria in the biofilm same drinking water sampling points within the from drinking water directly after UV disinfection and distribution system. In situ hybridisation experiments in 1 sample from the distribution system (Fig. 2b2). using subclass-specific labelled probes for Proteobacteria Previously, Schwartz et al. [12] demonstrated the revealed that beta-Proteobacteria were most frequently presence of non-tuberculous and non-pathogenic myco- found, but also alpha- and gamma-Proteobacteria could bacteria in native biofilm samples from the same be detected in significant minor percentages in biofilms drinking water distribution systems. Atypical mycobac- from embankment filtered drinking water [6].In teria occurred more frequently in biofilms from bank- drinking water biofilm communities from distribution filtered drinking water than in biofilms from drinking systems supplied with conditioned groundwater in water conditioned from groundwater. Hall-Stoodley Hamburg, Berlin, Mainz, and drinking water in Stock- et al. [19] found the rapidly growing facultative holm, which is gained from surface water, a higher pathogens Mycobacterium fortuitum and Mycobacterium number of beta-Proteobacteria, particularly Aquabacter- chelonae in soil, freshwater, seawater, wastewater and ium commune, was detected [17]. In our analyses none of potable water. The environment and even drinking the dominant PCR amplicons showed any significant water must hence be considered as a potential source of homology with A. commune. Kawai et al. [18] reported infection with facultatively pathogenic mycobacteria. constrastingly that alpha- and gamma-Proteobacteria Bacteria of the genus Legionella and Legionella were dominant in purified water which had been pneumophila, the etiologic agent of the Legionnaires’ prepared by ion exchange from tap water. disease, normally inhabit fresh water or wet soil and can live as intracellular parasites of amoebae and ciliates or 3.2. Identification of hygienically relevant bacteria in cyanobacteria. They are found both in natural and man- different biofilms made environments, such as cooling towers [20,21]. Lye et al. [22] detected significant amounts of Legionella spp. Bacterial re-growth in biofilms can become a problem in groundwater and potable water. Legionella spp. and in water distribution systems. In the present study, the especially Legionella pneumophila were described to be occurrence of hygienically relevant bacteria, such as adapted to warm water systems, where they multiply facultatively pathogenic mycobacteria, Legionella spp., most effectively [20,21]. In larger water distribution enterococci and Pseudomonas aeruginosa was analysed systems Legionella spp. find good survival conditions, by PCR and Southern blot hybridisation (SBH) with increased temperatures, and nutrients (such as sediments specific primers and probes in biofilms from surface and biofilms). In the present study, Legionella spp. were water to drinking water (Table 1). Saprophytic myco- detected using genus-specific probes in biofilms from bacteria were detected in all samples (Fig. 2b1), and surface water to drinking water, but there was no

Fig. 2. Identification of saprophytic and facultatively pathogenic atypical mycobacteria by PCR and SBH: (2a) PCR with primer set Eub338R/Myc1; (2b1) SBH with the probe Myc3 for identification of saprophytic mycobacteria; (2b2) SBH with the probe Myc2 for identification of facultatively pathogenic atypical mycobacteria; lane 1: surface water SW1; lane 2: surface water SW2; lane 3: raw water RW; lane 4: drinking water after activated carbon filtration GAC; lane 5: drinking water after disinfection DIS; lane 6: distribution system DW1; lane 7: distribution system DW2; lanes 8 and 9: Mycobacterium avium and Mycobacterium kansasii as positive controls. ARTICLE IN PRESS 1204 F. Emtiazi et al. / Water Research 38 (2004) 1197–1206

Table 4 Enzyme activities and total cell counts in different bioﬁlms

Bioﬁlm samples from Enzyme activities (mol h1 cm2) Total cell countsa

Esterase Alanine-aminopeptidase (DAPI counts cm2)

Surface water 8.8 106 (74.3 107) 4.0 106 (73.4 108) 7.2 106 Raw water 8.4 106 (71.0 107) 1.9 106 (72.6 107) 5.6 106 Drinking water after activated carbon ﬁltration 5.7 106 (71.8 107) 3.4 107 (71.7 108) 7.4 105 Drinking water after UV disinfection 2.7 106 (72.9 107) 2.3 107 (72.0 109) 3.0 105 Drinking water from the distribution system 4.9 106 (71.1 107) 6.6 107 (72.6 108) 5.7 105

a Standard error, 20–30%; n ¼ 223:

evidence for Legionella pneumophila. Using PCR and was much lower than of esterase, and glucosidase SBH techniques, Schwartz et al. [6] also detected activity was very low, indicating a high general cell Legionella spp. in different biofilms of a cold drinking and protein but a low carbohydrate metabolism. In water distribution system. parallel with the DAPI counts, enzymatic substrate Pseudomonas aeruginosa is an important pathogen in utilisation was decreased after filtration and disinfection nosocomial infections and its frequent presence in but still detectable (Table 4). For the downstream recreational and drinking water is a significant threat distribution system esterase and alanine-amino-pepti- to public health [23]. We detected P. aeruginosa in one dase activities slightly increased. The decrease of enzyme biofilm sample from drinking water, but not in the activity was proportional to the decrease of bacterial whole distribution system. The indicators of fecal numbers. Hence, the biofilms did not differ in specific contamination, Enterococcus faecium and Entercoccus enzyme activities. faecalis, were not detected in the investigated biofilms. To identify effects of soluble water components on In previous molecular biological and conventional biofilm enzyme activities, serial dilutions of the sus- microbiological studies, enterococci were detected only pended biofilms were tested. There was a linear sporadically after pipe bursts in biofilms from these correlation between enzyme activities and biofilm drinking water distribution systems [6]. biomass (not shown) indicating non-toxic conditions. These results indicate that saprophytic as well as A non-linear relationship was expected, if inhibitors facultatively pathogenic atypical mycobacteria, Legio- were present. nella spp., and Pseudomonas aeruginosa may have been Lehman et al. [24] reported increased amino-peptidase transferred from their natural habitat of surface water to activity of attached organisms compared to unattached the drinking water distribution system, although water organisms in artificial systems supplied with ground- treatment involved bank filtration and disinfection. water. Enzyme activities depend on bacterial numbers, water quality, natural substrates, and other environ- 3.3. Microbial cell counts and enzyme activities mental factors (e.g. slow sand filtration, bank filtration of groundwater) [24–26]. Many of these studies focussed Survival and physiology of bacteria depend on on free-living bacteria without considering attached nutrient supply, and this affects the purification biocoenoses. In contrast to previous studies with performance during raw water conditioning. Bacterial planktonic systems, which described dose-dependent counts per biofilm surface area measured by DAPI inhibition effects in bulk water [27], no enzyme staining decreased by 1–2 orders of magnitude from inhibition was observed in our biofilm samples. This surface water to raw water and further from non- suggests that biofilms are more resistant to toxic disinfected drinking water after filtration to disinfected compounds than free living bacteria. Miettinen et al. drinking water directly after UV disinfection (Table 4) [25] reported that many xenobiotics inhibited exoenzyme probably due to filtration and disinfection. In the activities. This inhibition has been used as a biochemical downstream water distribution system, bacterial counts toxicity test for the pollution of surface waters [16]. per biofilm surface area were slightly increased. Hendel et al. [26] reported that the microbial community Enzyme activities of esterases for general metabolic in the sand filter probably acts as a biological buffer activities, b-glucosidase for carbohydrate metabolism against increases in the loads of organic matter and and alanine-amino-peptidase for protein metabolism nutrients in the recharge plant. were measured to physiologically characterise biofilms High extracellular enzyme activities reflect high in surface water and drinking water treatment steps. In substrate availability and typically are characteristic of all biofilm samples, activity of alanine-amino-peptidase untreated water. Reductions in cell counts and enzyme ARTICLE IN PRESS F. Emtiazi et al. / Water Research 38 (2004) 1197–1206 1205 activities of biofilms during the purification process from Acknowledgements surface water to drinking water (Table 4) reflected an improvement in its trophic status, indicating that the We thank the EU for funding the study (EVK1-CT- purification process principally fulfilled its function. 1999-00001). Special thanks are given to Dr. Birgit However, the presence of (facultative) pathogens in Kuhlmann, Dr. Beate Kilb, Dr. Holger Volkmann, and biofilms after UV treatment and in the distribution Erik Ziemann for their helpful support. We also thank system is a potential threat to human health. Therefore, Silke Kirchen for her technical assistance during microbiological, molecular biological and physiological sampling and laboratory analyses. We thank the controls are useful tools for monitoring the quality of technical staff of the municipal drinking water supplier drinking water and the efficiency of the conditioning for installing the modified devices. processes.

References 4. Conclusions [1] Brauch H, Sacher F, Denecker E, Tacke T. The effective- ness of bank filtration for the removal of polar organic * The characterisation of biofilms by means of PCR, trace elements. Wasser Abwasser 2000;141(4):226–34. DGGE, and sequence analysis is a rapid cultivation- [2] Heberer T, Schmidt-Baumler. K, Stan H-J. Occurrence independent method to gain insight into the micro- and distribution of organic contaminants in the aquatic bial population composition of aquatic habitats. system in Berlin. Acta Hydrochim Hydrobiol 1998;26(5): 272–8. * The biofilms of the different drinking water con- [3] Byrd J, Xu H, Colwell R. Viable but nonculturable ditioning sampling sites showed different band bacteria in drinking water. Appl Environ Microbiol patterns, and were apparently made up by different 1991;57(3):875–8. populations. Even within one compartment, the [4] Buswell C, Herlihy Y, Lawrence L, McGuiggan J, Marshc biofilms were not unique. P, Keevil C, Leach S. Extended survival and persistence of * The similarity index in similar aquatic habitats was Campylobacter spp. in water and aquatic biofilms and their higher than the similarity index between different detection by immunofluorescent-antibody and rRNA aquatic compartments, indicating that similar ecolo- staining. Appl Environ Microbiol 1998;64(2):733–41. gical niches are inhabited by similar bacterial [5] Theron J, Cloete T. Emerging waterborne infections: populations. contributing factors, agents and detection tools. Crit Rev Microbiol 2002;28:1–26. * Most of the bacteria identified belonged to beta- [6] Schwartz T, Hoffmann S, Obst U. Formation and Proteobacteria. bacterial composition of young, natural biofilms obtained * Saprophytic mycobacteria and legionellae were de- from public bank-filtered drinking water systems. Water tected in all biofilms studied from surface to drinking Res 1998;32(9):2787–97. water within the distribution system, indicating a [7] Kalmbach S, Manz W, Szewzyk U. Dynamics of biofilm possible passage of bacteria from surface to drinking formation in drinking water: phylogenetic affiliation and water. Positive Southern Blots for Pseudomonas metabolic potential of single cells assessed by formazan aeruginosa and facultatively pathogenic mycobacter- reduction and in situ hybridization. FEMS Microbiol Ecol ia were obtained from some biofilms of drinking 1997;22:265–79. water directly after UV disinfection and from the [8] Muyzer G, Waal EC, Uitterlinden AG. Profiling of complex microbial population by denaturing gradient gel downstream distribution system. This potential electrophoresis analysis of polymerase chain reaction— source of infection is hygienically of concern, amplified genes coding for 16S rRNA. Appl Environ drinking water may need improvement. Microbiol 1993;59(3):695–700. * Legionella pneumophila and the indicators of faecal [9] Kilb B, Kuhlmann B, Eschweiler B, PreuX G, Ziemann E, contamination, Enterococcus faecium and E. faecalis, Schottler. U. Community structures of different ground- were not detected in the biofilms by specific probing. water habitats investigated using methods of molecular Data for fecal indicators hence do not allow to draw biology. Acta Hydrochim Hydrobiol 1998;26(6):349–54. conclusions on the presence of pathogenic bacteria in [10] Sheffield VC, Cox DR, Myers RM. Attachment of a 40-bp biofilms. G+C rich sequence (GC-clamp) to genomic DNA * Effects of toxic compounds, as typically observed for fragments by polymerase chain reaction results in im- proved detection of single-base changes. Proc Natl Acad bulk water samples, were not detected for our biofilm Sci USA 1989;86:232–6. samples, indicating an increased tolerance of biofilm [11] Murray AE, Hollibaugh JT, Orrego C. Phylogenetic communities against toxic compounds. composition of bacterioplankton from two California * The reduction in enzyme activities during the estuaries compared by denaturing gradient gel electro- purification process from surface water to drinking phoresis of 16S rDNA fragments. Appl Environ Microbiol water reflect an improvement in its trophic status. 1996;62:2676–80. ARTICLE IN PRESS 1206 F. Emtiazi et al. / Water Research 38 (2004) 1197–1206

[12] Schwartz T, Kalmbach S, Hoffmann S, Szewzyk U, Obst formation under high and low nutrient conditions. J Appl U. PCR-based detection of mycobacteria in biofilms from Microbiol Symp Suppl 1999;85:60–9. a drinking water distribution system. J Microbiol Meth [20] Atlas RM. Legionella: from environmental habitats to 1998;34:113–23. disease pathology, detection, and control. Environ Micro- [13] Manz W, Amann R, Szewzyk R, Szewzyk U, Stenstrom. biol 1999;1:283–95. TA, Hutzler P, Schleifer KH. In situ identification of [21] States SJ, Wadowsky RM, Kuchta JM, Wolford RS, Legionellaceae using 16S rRNA-targeted oligonucleotide Colnley LF, Yee RB. Legionella in drinking water. In: probes and confocal laser scanning microscopy. Micro- McFeters GA, editor. Drinking water microbiology. New biology 1995;141:29–39. York: Springer; 1990. p. 340–67. [14] Grimm D, Merkert H, Ludwig W, Schleifer KH, Hacker J, [22] Lye D, Fout S, Crout SR, Danielson R, Thio CL, Paszko- Brand BC. Specific detection of Legionella pneumophila: Kolva CM. Survey of ground, surface, and potable waters construction of a new 16S rRNA targeted oligonucleotide for the presence of Legionella species by EnviroampR PCR probe. Appl Environ Microbiol 1998;64(7):2686–90. Legionella Kit, culture, and immunofluorescent staining. [15] Frahm E, Heiber I, Ludwig W, Obst U. Rapid parallel Water Res 1997;31:287–93. detection of hygienically relevant microorganisms in water [23] Trautmann M, Michalsky T, Wiedeck H, Radosavljevic V, samples by PCR and specific hybridisation in microtiter Ruhnke N. Tap water colonization with Pseudomonas plates. Syst Appl Microbiol 2001;24:423–9. aeruginosa in a surgical intensive care unit (ICU) and [16] Obst U, WeXler A,Wiegand-Rosinus M. Enzyme inhibition relation to Pseudomonas infection of ICU patients. Infect for examination of toxic effect in aquatic systems. In: Wells Control Hosp Epidemiol 2001;22(1):49–52. P, Lee K, Blaise C, editors. Microscale testing in aquatic [24] Lehman RM, O’Connell SP. Comparison of extracellular toxicology. Boca Raton: CRC Press; 1997. p. 77–96. enzyme activities and community composition of attached [17] Kalmbach S, Manz W, Bendinger B, Szewzky U. In situ and free-living bacteria in porous medium columns. Appl probing reveals Aquabacterium commune as a widespread Environ Microbiol 2002;68(4):1569–75. and highly abundant bacterial species in drinking water [25] Miettinen IT, Vartiainen T, Martikainen PJ. Bacterial biofilms. Water Res 2000;34(2):575–81. enzyme activities in ground water during bank filtration of [18] Kawai M, Matsutera E, Kanda H, Yamaguchi N, Tani K, lake water. Water Res 1996;30:2495–501. Nasu M. 16S Ribosomal DNA-based analysis of bacterial [26] Hendel B, Marxsen J, Fiebig D, PreuX G. Extracellular diversity in purified water used in pharmaceutical manu- enzyme activities during slow sand filtration in a water facturing processes by PCR and denaturing gradient gel recharge plant. Water Res 2001;35(10):2484–8. electrophoresis. Appl Environ Microbial 2002;68(2): [27] Schmitt-Biegel B, Obst U. Inhibition of the microbial 699–704. purification in the river Rhine and in the groundwater [19] Hall-Stoodley L, Keevil CM, Lappin-Scott HM. Myco- influenced by the river Rhine. Vom Wasser 1989;73: bacterium fortuitum and Mycobacterium chelonae biofilm 315–22.