Diverse and Distinct Bacterial Communities Induced Biofilm Fouling in

Oct 15, 2008 - AND LUDO DIELS †. Department of Environmental and Process Technology,. Flemish Institute for Technological Research. (VITO), B-2400 M...
7 downloads 0 Views 3MB Size
Environ. Sci. Technol. 2008, 42, 8360–8366

Diverse and Distinct Bacterial Communities Induced Biofilm Fouling in Membrane Bioreactors Operated under Different Conditions L I - N A N H U A N G , * ,†,‡ H E L E E N D E W E V E R , † AND LUDO DIELS† Department of Environmental and Process Technology, Flemish Institute for Technological Research (VITO), B-2400 Mol, Belgium, and School of Life Sciences, Sun Yat-Sen University, Guangzhou 510275, People’s Republic of China

Received May 9, 2008. Revised manuscript received August 12, 2008. Accepted August 25, 2008.

We conducted a laboratory-scale experiment using real municipal wastewater with identical submerged membrane bioreactors (MBR) operated under different conditions (sludge retention time (SRT) and membrane flux) for nearly 6 months. Membrane biofilm samples were periodically retrieved, and cultivation-independent molecular approaches were used to systematically elucidate the community composition and diversity of microorganisms responsible for biofilm formation in the MBRs. Membrane fouling occurred earlier and faster in the lowSRT reactors which had more active mixed liquor biomasses andhigherconcentrationsofdissolvedorganicmatters.Denaturing gradient gel electrophoresis (DGGE) analysis and comparative rRNA sequencing revealed that diverse and distinct bacterial communities significantly differing from those in the planktonic biomass developed on the microfiltration membrane surfaces, with phylotypes from the Proteobacteria (particularly the R and β subdivisions) and Bacteroidetes dominating the 16S rRNA gene libraries. This indicated that specific groups of bacteria were preferentially growing in the membrane habitats. Membrane flux had great impact on the predominant populations selected in the fouling biofilms. At lower fluxes, biofilm community composition was quite similar independent of sludge ages and biofilm formation seemed to be the result of a more natural process of colonization and biofilm development. In contrast, distinct biofilm communities developed on membrane surfaces at high fluxes. Despite the high convective forces, the biofilm composition was significantly different from the planktonic biomass, and selective enrichments of certain species were observed. Our study suggests that the microbial communities responsible for membrane biofouling in MBRs are far more complex and variable than expected and thus could be challenging to control.

1. Introduction The submerged membrane bioreactor (MBR) combines the activated sludge process with an effective solid-liquid * Corresponding author phone/fax: (8620) 84112399; e-mail: [email protected]. † Flemish Institute for Technological Research. ‡ Sun Yat-Sen University. 8360

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 22, 2008

separation of mixed liquor by membrane filtration. By offering several important advantages over conventional treatment systems, particularly in the production of high-quality permeate, the reliable and efficient MBR technology has increasingly become an attractive option worldwide for municipal and industrial wastewater treatment (1, 2). However, the wider application of MBR is retarded by membrane fouling which results in flux decline and transmembrane pressure (TMP) increase and thus significantly deteriorates system performance and production. Although it has been under intensive study for more than a decade, currently there is no consensus on the exact phenomenon occurring on the membrane interface during activated sludge filtration (3). Soluble microbial products (SMP) and extracellular polymeric substances (EPS) have been suggested to have relationships with fouling evolution in MBRs (4, 5); the relative contribution of each of its fractions (carbohydrates, proteins, humic compounds, and nucleic acids) is yet to be clarified, however (3). In addition, contradictory results on the correlation between these mixed liquor indices and membrane fouling have been reported (6). Recent research has clearly shown that biofilm formation induces membrane fouling (4, 7). However, the nature of the microbial communities involved remains largely unexplored. In particular, the influence of operating conditions on the biofilm development and community structure is poorly understood. Characterization of these sessile communities and identification of the major fouling microorganisms are crucial for the development of effective tools and strategies for the monitoring and control of biofilm formation fouling in MBRs. In comparison to other environments, the membrane filtration system provides a unique habitat for biofilm formation. The convective flow generates a special force for transporting particulates (including bacterial cells) in the mixed liquor to the membrane surface, whereas fluid/air shear over the membrane module may cause bacterial detachment from the membrane, particularly during the relaxation phase where filtration is paused and convective flow stops. Additionally, although membrane surfaces are often modified chemically to increase their antifouling properties, the inevitable deposition of mixed liquor microbial products with time makes them vulnerable to cell adhesion, an important step in biofilm establishment. Further, the SMPs produced by the planktonic biomass could serve as readily available substrates for the sessile microbial community when the mixed liquor supernatant is diffused across the membrane. The overall goal of this study was to systematically characterize and compare the microbial communities involved in biofilm formation in MBRs operated under different conditions. We operated simultaneously two laboratory-scale submerged MBRs at two different sludge retention times (SRT). Both reactors were subjected to a low-flux operation followed by a high-flux operation. Molecular approaches were used to examine the community structure and diversity of membrane biofilms retrieved from the MBRs at different fouling time points and then, by comparing these sessile communities with those in the planktonic biomass of mixed liquor, to determine whether the special membrane environments selected unique biofilms.

2. Materials and Methods 2.1. MBRs and Operating Conditions. Two identical laboratory-scale submerged MBRs, designated MBR-8 and -30, were operated in parallel at two different SRTs (8 and 30 days) with the same feed wastewater delivered from a local municipal wastewater treatment plant. The reactors were 10.1021/es801283q CCC: $40.75

 2008 American Chemical Society

Published on Web 10/15/2008

FIGURE 1. Evolutions of TMP in MBR-8 (circles) and MBR-30 (crosses) during the two operation modes. Arrows indicate time points of biofilm sampling. Asterisks indicate major fouling time points for which clone libraries were constructed with membrane biofilms and the parallel activated sludge samples taken from the same MBRs. seeded with return sludge from the wastewater treatment plant and operated preliminarily for around 1 month for biomass acclimatization. Each MBR had a working volume of 20 L and was equipped with two new flat sheet microfiltration (MF) membranes (Kubota, surface area of 0.1 m2/ membrane) with pore size of 0.4 µm. Aeration was maintained with compressed air at 1.2 m3/h. Filtration was carried out with a constant flux (low-flux, 15 L/m2 · h) and followed an intermittent suction cycle of 8 min on and 2 min off. Sludge wasting was carried out daily to maintain the desired SRT for each MBR. The evolution of membrane fouling was indicated by the increase in TMP using a digital pressure transmitter. After running for nearly 5 months, both MBRs were switched to a high-flux mode (30 L/m2 · h) by using one new membrane for each reactor, and the operation was continued for approximately 1 month. Consequently, membrane flux was the only operating parameter that was changed between the two periods of operation. Each time when the maximal TMP was reached, the fouled membranes were removed for biofilm sampling and replaced with new ones. 2.2. Analytical Methods. Feed wastewater and membrane permeate were collected regularly and analyzed for a full set of conventional performance parameters (see the text in the Supporting Information). Mixed liquor was sampled weekly from the middle of the reactors. Then indices representing mixed liquor characteristics, including soluble EPS constituents, were determined (Supporting Information text). Specific oxygen uptake rate (SOUR) and specific nitrification rate (SNR) were also measured to indicate biological activity. 2.3. DNA Extraction and PCR Amplification. To monitor the planktonic bacterial population dynamics in the MBRs, weekly mixed liquor samples were taken and aliquots (1.5-3.0 mL) centrifuged for 10 min at 12 000g, 4 °C. The cell pellets were rinsed twice with a sodium phosphate buffer (120 mM, pH 8.0), and total community genomic DNA was extracted by a direct bacterial lysis which employs lysozyme treatment followed by a freeze-thaw procedure (8). Prior to precipitation with isopropyl alcohol, the DNA extracts were subjected to an additional purification step using polyvinylpyrrolidone to remove humic substances (9). Biofilm sampling was performed at different fouling time points (Figure 1). Membrane slices were taken from different locations on the fouled membrane module and formed a composite sample. After rinsing thoroughly with 120 mM sodium phosphate buffer (to remove loosely deposited sludge), the biofilms firmly attached on the membrane surface were removed with

a sterile stainless steel spoon, and total DNA was extracted (8) separately from the cell pellet and the membrane slice and then pooled. To more effectively release the bacterial cells sticking to the membrane, an additional bead-beating step (2 × 30 s) was included prior to DNA extraction. For clone library construction, nearly complete 16S rRNA gene fragments were amplified in triplicate with primers 27F and 1492R (10) as described previously (11). For denaturing gradient gel electrophoresis (DGGE) analysis, a 496 bp 16S rRNA gene fragment was amplified with primers GC-63F and 518R (Supporting Information text). 2.4. DGGE and Gel Pattern Analysis. DGGE analysis of GC-63F/518R PCR products was performed as described (9). Community fingerprint images were processed and analyzed with Bionumerics 4.0 software (Applied Maths). UPGMA (unweighted pair group method of arithmetic means) dendrograms relating band pattern similarities were calculated with the Dice coefficient. 2.5. Cloning, Sequencing, and Phylogenetic, Rarefaction, and Statistical Analysis. Triplicate PCR products were pooled (to minimize bias), purified with a QIAquick PCR purification kit (QIAGEN), and cloned using a TOPO TA cloning kit (Invitrogen). Eight clone libraries were prepared with DNA samples extracted from membrane biofilms and planktonic biomass of mixed liquors taken from the MBRs at major fouling time points (Figure 1). Plasmid DNAs were prepared (QIAprep Spin Miniprep Kit, QIAGEN) from more than 600 randomly selected clones containing correctly sized inserts and sequenced with primer 27F on an Applied Biosystems capillary 3730xl DNA sequencer. Chimeric sequences were identified as described (11) and excluded from subsequent analysis. Rarefaction and diversity statistics were calculated using DOTUR software (12) (Supporting Information text). Operational taxonomic units (OTU) were defined as groups in which sequences differed by e3%. Nearly complete 16S rRNA gene sequences were obtained for the representative biofilm OTUs by a second sequencing run starting from the opposite side of the vector with primer 1492R. Sequence data were processed and phylogenetic trees constructed using the ARB software package (13) (Supporting Information text). The 16S rRNA gene sequences from this study have been deposited in the EMBL/GenBank/DDBJ databases under accession numbers FM200855 to FM201247.

3. Results 3.1. Fouling Evolution and Mixed Liquor Characteristics. Despite substantial fluctuations in the influent concentration, operation of both MBRs was generally stable over the nearly 6 month period in terms of treated water quality (Supporting Information Tables S1 and S2). During the low-flux operation, severe and reproducible membrane fouling was observed in MBR-8 through TMP increase and the fouled membranes were removed from the reactor for biofilm sampling on days 70 and 135, respectively (Figure 1). Visual inspections revealed that the membrane surface was covered by a thick, brown homogeneous biofilm layer. By contrast, MBR-30 did not exhibit a significant TMP increase over the entire period of low-flux operation. In spite of this, it was found that the membrane surface was slightly covered by a brown homogeneous biofilm when the membranes were removed on day 135. The significantly thicker biofilm layers and faster biofilm development in MBR-8, which had a much lower concentration of mixed liquor suspended solids (MLSS, averaging at 1.7 g/L) (Supporting Information Table S3), correlated well with the higher fouling propensity in this reactor (Figure 1), a good indication of the link between biofilm formation and membrane fouling. The high-flux operation was initiated on day 147 by using one new membrane module for each reactor. MBR-8 experienced rapid and reproducible membrane fouling, and VOL. 42, NO. 22, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

8361

TABLE 1. TOC and Major Constituents of EPS in the Mixed Liquor Supernatanta

MBR-8 MBR-30 a

TOC (mg/L)

TOC (mg/g MLVSS)

carbohydrate (mg glucose/L)

protein (mg globulin/L)

UV254 (/m)

17.55 ( 3.60 13.89 ( 2.44

17.09 ( 6.21 4.13 ( 1.02

8.17 ( 2.82 6.57 ( 1.96

10.89 ( 2.53 9.27 ( 1.58

5.76 ( 1.37 5.29 ( 1.22

n ) 21, except for TOC where n ) 34.

FIGURE 2. Comparison of bacterial 16S rRNA gene community fingerprints (DGGE profile and respective UPGMA dendrogram) derived from membrane biofilms (BF) and activated sludge (AS) sampled at different fouling time points of the MBRs (see Figure 1). The scale bar indicates the percentage similarity at the nodes. LF, low-flux; HF, high-flux. biofilm sampling was carried out on days 151, 160, and 173, respectively. The TMP of MBR-30 also increased steadily, but the fouling rate was slower than that in MBR-8. In general, the high-flux biofilms were not as homogeneous and evenly spread on the membrane surface as those observed in the low-flux operation, and the difference in the thickness of membrane biofilms between the two reactors was not significant. Total organic carbon (TOC, contributed from SMP) and major constituents of EPS in the mixed liquor supernatant and effluent were monitored weekly (Table 1 and Supporting Information Table S4). Regardless of the membrane flux applied, the TOC and EPS components in MBR-8 were consistently higher than those in MBR-30 (P ) 0.000, paired t tests), although concentrations of these dissolved organic matters fluctuated considerably in response to changes in feed wastewater strength. Importantly, the difference in TOC concentration per unit biomass between the two reactors was even more profound, suggesting that the relative surplus of readily available substrates was >3 times higher for the microbial community in MBR-8 throughout the whole experiment. Consistent with this, both SOUR and SNR values were constantly higher in MBR-8 than in MBR-30 during the two operation modes (P ) 0.000, paired t tests) (Supporting Information Table S3), indicating a 100-130% more active biomass in the reactors with lower sludge concentration. Although operated at a relatively short SRT, MBR-8 was able to maintain a substantial nitrification as indicated by the SNR values and the ammonium levels in the effluent. 3.2. Fingerprints of Membrane Biofilm and Activated Sludge Communities. DGGE analysis was used to compare the composition of sessile and planktonic microbial populations from different fouling time points. Profoundly different 16S rRNA pattern types were distinguished between the membrane biofilm and activated sludge communities for all 8362

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 22, 2008

tested conditions. This was confirmed by the statistical analysis which formed two separate clades according to the environment (Figure 2). This finding indicated that specific groups of bacteria were preferentially selected by the unique habitats at the membrane interface. Interestingly, community fingerprints of the two low-flux biofilms from MBR-8 were highly similar, and they exhibited substantial similarities to that of the MBR-30 low-flux biofilm. This was further supported by the UPGMA dendrogram which showed that these sessile communities clustered together (Figure 2). In contrast, fingerprints of the high-flux biofilms exhibited the most different patterns visible by eye on the gel and were individually deeply branched in the dendrogram, suggesting the selection of highly divergent and distinct biofilm communities under the high membrane flux. Repetition of DNA extraction and PCR-DGGE analysis on replicate samples resulted in highly similar gels (data not shown). The planktonic bacterial communities in the mixed liquor were monitored weekly over the course of nearly 6 months. Community profiles were highly complex and indicated substantial fluctuations in community structure over time, in particular for MBR-8 (Supporting Information Figure S1). Although several bands were specific to the inocula and the startup phase and additional bands appeared in both community profiles, most DGGE bands were detected throughout the entire experimental period. Fluctuations were mainly due to shifts of the dominant bands and variations in the relative abundance of common bands among samples with operation time. 3.3. Phylogenetic Analysis. Sequencing and phylogenetic analysis of 310 randomly selected clones from the four biofilm libraries revealed 149 unique OTUs assigned to at least 13 distinct phyla (Supporting Information Table S5, Figure 3). Estimated library coverages ranged between 48% and 76%. The predicted minimum numbers of bacterial species present

FIGURE 3. Neighbor-joining tree showing the phylogenetic placement of the >600 bacterial 16S rRNA gene clone sequences retrieved from membrane biofilms (BF) and activated sludge (AS) sampled at major fouling time points of the MBRs. The numbers of biofilm and activated sludge OTUs are indicated in parentheses. Phylogenetic distribution of the OTUs in the eight clone libraries is given to the right (open columns, biofilms; closed columns, activated sludge). Some OTUs display very low levels of similarity to any known 16S rRNA gene sequences and thus could not be affiliated. Members of the Archaea domain were used as outgroups (not shown). LF, low-flux; HF, high-flux. in the analyzed biofilms ranged from 38 to 160 (Chao 1 estimator) and 40 to 185 (ACE estimator) (Supporting Information Table S5). The vast majority of the OTUs comprised a single clone, indicating high levels of bacterial diversity. This was further supported by the steep rarefaction curves (Supporting Information Figure S2) and the traditional Shannon diversity indices which ranged from 2.55 to 3.87 ( Supporting Information Table S5). Only a small fraction of the retrieved sequences represented by 14-27% of the total OTUs was affiliated with 16S rRNA gene sequences of known bacterial species with g97% identity (Supporting Information Table S6). The remaining 16S rRNA sequences represent putative phylotypes with no cultivated representatives or novel phylotypes never described before. The most frequently detected phyla were Proteobacteria and Bacteroidetes. Other minor lineages detected included Nitrospira, Verrucomicrobia, Acidobacteria, Actinobacteria, and even Planctomycetes, Spirochaetes, OP10 and TM7. Sequence analysis of 203 clones from the two low-flux biofilms revealed a total of 104 OTUs belonging to 12 different

phyla or candidate divisions, with Proteobacteria- (specifically R- and β-Proteobacteria) and Bacteroidetes-affiliated sequences accounting for 64-65% and 18-24% of the corresponding gene libraries, respectively (Supporting Information Table S5, Figure 3). Bacteria from the R-Proteobacteria and Nitrospira were clearly enriched in the membrane biofilms in comparison with the planktonic biomass. Abundantly detected phylotypes were phylogenetically associated with environmental sequences retrieved from conventional activated sludges (e.g., MBR-30_LF_BF41 and 90) or closely related to Sphingomonas or Nitrospira species isolated from water and wastewater treatment processes (e.g., MBR-8_LF_BF90 and MBR-30_LF_BF31) (Figure 4 and Supporting Information Figure S3). A total of eight OTUs were found in both low-flux biofilms where they comprised 18-24% of the total clones. Furthermore, two of the three most abundant MBR-8 biofilm OTUs were also identified in the MBR-30 biofilm library. These findings indicated that the two low-flux sessile communities are related, as was also supported by the DGGE analysis. VOL. 42, NO. 22, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

8363

FIGURE 4. Neighbor-joining tree showing the phylogenetic affiliation of nearly complete Proteobacteria and Bacteroidetes 16S rRNA gene sequences from membrane biofilms. Sequences are color-coded according to reactor/flux. Additional symbols (colored circles) show the relative frequency (see the legend of Supporting Information Figure S3 for the calculation) of a sequence in their respective clone libraries (red, MBR-8, low-flux; purple, MBR-8, high-flux; blue, MBR-30, low-flux; green, MBR-30, high-flux). With the exception of δ-Proteobacteria sequences, only OTUs comprising g2 clones among the 310 randomly analyzed biofilm clones are shown. Bootstrap values of >50% are indicated at branch points. Clostridium aminobutyricum and Clostridium formicaceticum were used as outgroups. Sequence analysis of 107 high-flux biofilm clones indicated seven distinct phyla (Supporting Information Table S5, Figure 3). Similar to the low-flux biofilms, Proteobacteriaand Bacteroidetes-affiliated phylotypes dominated the two high-flux biofilm libraries. In particular, Bacteroidetes species were highly abundant in the MBR-30 biofilm. Although 8364

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 22, 2008

R-Proteobacteria was also enriched in the high-flux biofilms, β- and γ-Proteobacteria were obviously less favored in the membrane habitats. The most abundant phylotypes were moderately related to Sphingomonas spp. (MBR-8_HF_BF43, 34% of total clones) and affiliated with Fluviicola taffensis (MBR-30_HF_BF11, 24% of the gene library) (Figure 4). Unlike

the low-flux biofilms, the two high-flux biofilm libraries only had one OTU in common, and it only comprised minor fractions of the corresponding sessile communities. Consistent with the community fingerprinting analysis, little overlap in the identified OTUs was found between the lowflux and high-flux membrane biofilms from the same SRT MBRs. The community compositions of sessile and planktonic bacteria differed significantly as reflected by the limited number of OTUs detected in both environments and the unproportionate distribution of the shared OTUs in the respective clone libraries. Of the OTUs identified in the membrane biofilms, only 3-20% were detected in the parallel planktonic populations. More importantly, most of the dominant biofilm OTUs were not identified or could only be occasionally found in the mixed liquors, indicating that these organisms were preferentially growing on the membrane surface. Likewise, the most frequently detected OTUs in the planktonic biomasses were rarely found in the parallel biofilm libraries. This separation of sessile and planktonic populations was in good agreement with the DGGE fingerprinting results. Rarefaction analysis was performed to compare the relative species richness between the two environments. All rarefaction curves of the biofilm libraries were lower than those of the corresponding activated sludge libraries (Supporting Information Figure S2). In particular, both the MBR-30 low-flux and the MBR-8 high-flux biofilms had a relative species richness significantly (P < 0.05) lower than that of parallel planktonic biomass samples. These provided further evidence that certain bacteria were preferentially selected in the unique membrane habitats.

4. Discussion Operating conditions have significant influence on the microbial community structure and physiological state and other important properties of mixed liquor. These in turn will greatly affect the progress of biofilm fouling in the MBR and the predominant populations selected in the membrane biofilms. In particular, SRT (and consequently the food-tomicroorganism (F/M) ratio), which ultimately controls biomass characteristics, is probably the most important operating parameter impacting on fouling propensity (3). While full-scale MBRs are typically operated at relatively long SRTs to reduce reactor footprint and minimize excess sludge, some wastewater treatment plants could prefer a short SRT operation to maximize sludge production for biogas generation (14). Although it was originally believed that high MLSS concentration associated with long SRT operation will increase membrane filtration resistance (15-17), recent studies have revealed that MBRs operated at low SRT (or high F/M ratio) exhibited faster membrane fouling rates (4, 6). Likewise, biofilm formation and membrane fouling occurred earlier and faster in MBR-8 which was operated at a much higher F/M ratio. The high F/M ratio supported a more active biomass that had a high biofilm formation potential, and these microbial activities maintained constantly higher TOC and EPS concentrations in the mixed liquor supernatant in MBR-8. These SMPs could have served as readily available substrates for the fouling microorganisms and thus promoted faster biofilm growth on the membrane surface (4). Both DGGE and clone library analysis indicated that, in contrast to the highly distinct high-flux biofilms, the low-flux biofilm communities from the two different SRT MBRs were related. The selection of similar predominant populations in the membrane biofilms during the low-flux operations may reflect a more natural process of colonization and biofilm development since the convective flow plays a less important role in transporting to the membrane surface the bacterial cells that trigger biofilm formation. If this is true, then there might be a certain range of low membrane fluxes, where

biofilm fouling in MBRs operated under different SRTs is induced by the sessile growth of highly similar bacterial populations and thus could be controlled by the same strategy. Further research is required to investigate this point. Membrane flux represents another important operating factor that has great effect on membrane fouling (6), and selection of an optimal flux is critical for a submerged MBR to achieve long-term and sustainable operation. This was supported by the observation that both MBR-8 and -30 exhibited very rapid membrane fouling during the high-flux operation, despite the fact that the fouling-related indices of mixed liquors did not differ significantly from those of the low-flux operation (data not shown). In addition to transporting more fouling components toward the membranes, the strong convective force associated with the high membrane flux would provide a major mechanism for transporting bacterial cells, including nonmotile ones, to the membrane surface and thus trigger a faster biofilm development and lead to different predominant populations in the biofilms. This was evidenced by the significant differences between the biofilm communities from the same SRT MBRs operated at different fluxes. On the other hand, highly similar biofilms were found associated with the reproducible membrane foulings observed in MBR-8 during the low-flux operation (Figure 2), although the community fingerprints of high-flux biofilms from this reactor are highly variable. These results suggest that membrane flux has great impact on biofilm development and community structure with the low-flux communities seems easier to predict and control, an important implication for the mitigation of biofilm fouling in MBRs. Our comprehensive phylogenetic analyses have revealed that the membrane biofilms retrieved from MBRs operated under different conditions harbored diverse bacterial communities which are significantly different from those in the planktonic biomass. Although some of the biofilm phylotypes are phylogenetically associated with genera or taxa that have been identified by cultivation in membrane biofilms from water and wastewater treatment processes (e.g., Sphingomonas, Acinetobacter, Flavobacterium, Pseudomonas, Aeromonas, Mycobacterium, and Rhizobiales) (18-21), the vast majority of OTUs represent putative phylotypes never obtained in culture or not described before. Similarly, novel bacterial OTUs were abundantly recovered from the 16S rRNA gene library of mature biofilm developed on the hollowfiber MF membrane surface of a pilot-scale MBR treating municipal wastewater (7). These results indicate that the microbial communities responsible for biofilm fouling in MBRs are much more complex than expected and remain largely unexplored. Although the wide variety of different microorganisms would allow the biofilm communities to respond to varying environmental conditions in the MBRs, it could also render the control of sessile growth of these microorganisms more challenging, probably requiring more integrated strategies. On the other hand, our results support recent findings suggesting that specific groups of bacteria were preferentially selected in the membrane biofilms (7, 18). In particular, the significant enrichment of R-Proteobacteria species in the four different biofilms indicates that bacteria from this subphylum are specialized for the specific conditions at the membrane interface. Rhizobiales organisms, which have recently been found to be ecologically significant in membrane biofilm communities in membrane separation systems for secondary effluent purification (18, 22), were only occasionally found in our fouling biofilms. Instead, Sphingomonas-related OTUs were frequently (8-34%) detected, particularly in the MBR-8 high-flux biofilm library. The metabolically versatile Sphingomonas spp. have previously been identified as important members of biofilm communities in drinking water VOL. 42, NO. 22, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

8365

distribution systems (23). They were considered as the key organisms contributing to biofouling in a reverse osmosis membrane for freshwater purification due to their ability of twitching and swarming motility and to produce viscous exopolysaccharides which collectively facilitate bacterial adhesion and colonization of surfaces (19). Pang and Liu (24) recently reported that Sphingomonadaceae-related organisms served as pioneer colonizers and subsequently continued to thrive in biofilms developed in secondary effluents from a conventional activated sludge process. A substantial fraction of the membrane biofilm OTUs are affiliated with the β-Proteobacteria and Bacteroidetes. While β-Proteobacteria have recently been identified as the most numerically important group in mature biofilms grown on hollow-fiber membranes for wastewater treatment (7), the predominance of Bacteroidetes phylotypes sparticularly in the MBR-30 high-flux biofilms suggests that this phylogenetic group could have a competitive advantage over other colonizers and become a major player in the development of membrane biofilms under specific ecological conditions. In addition, many biofilm 16S rRNA sequences from this study cluster with environmental sequences previously retrieved from biofilms developed in a variety of different habitats (rhizosphere, lagoons, bioreactors, and oxygentransferring membranes for wastewater treatment), indicating a ubiquitous distribution of the corresponding organisms and, potentially, also their biofilm formation characteristics. The results reported here significantly expand our knowledge of the diversity of bacteria residing in the membrane biofilms. Cultivation and characterization of these as yet uncultured organisms, particularly the predominant phylotypes that are likely to play a substantial role in the biofilm formation process, will lead to a better understanding of the mechanisms and controlling factors of biofilm fouling in MBRs for wastewater treatment.

Acknowledgments We thank the European Commission for financial support of L.-N. H. through a Marie Curie Fellowship (MIF1-CT-2005021768), the Prodem team for reactor maintenance and chemical analyses, and X. Lin and M. Caluwaerts for their contributions in the molecular experiments.

Supporting Information Available Characteristics of influent wastewater, effluent and mixed liquor, diversity and predicted richness and novelty of 16S rRNA gene sequences; DGGE profiles of planktonic bacterial communities, rarefaction curves, phylogenetic tree of nonProteobacteria/Bacteroidetes 16S rRNA sequences from membrane biofilms. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Judd, S. The MBR Book: Principles and Applications of Membrane Bioreactors in Water and Wastewater Treatment; Elsevier: Amsterdam, The Netherlands, 2006. (2) Yang, W.-B.; Cicek, N.; Ilg, J. State-of-the-art of membrane bioreactors: worldwide research and commercial applications in North America. J. Membr. Sci. 2006, 270, 201–211. (3) Le-Clech, P.; Chen, V.; Fane, T. A. G. Fouling in membrane bioreactors used in wastewater treatment. J. Membr. Sci. 2006, 284, 17–53. (4) Ng, H. Y.; Tan, T. W.; Ong, S. L. Membrane fouling of submerged membrane bioreactors: impact of mean cell residence time and the contributing factors. Environ. Sci. Technol. 2006, 40, 2706– 2713. (5) Nagaoka, H.; Kono, S.; Yamanishi, S.; Miya, A. Influence of organic loading on membrane fouling in membrane separation activated sludge process. Water Sci. Technol. 2000, 41, 355–362.

8366

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 22, 2008

(6) Kimura, K.; Yamato, N.; Yamamura, H.; Watanabe, Y. Membrane fouling in pilot-scale membrane bioreactors (MBRs) treating municipal wastewater. Environ. Sci. Technol. 2005, 39, 6293– 6299. (7) Miura, Y.; Watanbe, Y.; Okabe, S. Membrane biofouling in pilotscale membrane bioreactors (MBRs) treating municipal wastewater: impact of biofilm formation. Environ. Sci. Technol. 2007, 41, 632–638. (8) Tsai, Y. L.; Olson, B. H. Rapid method for direct extraction of DNA from soil and sediments. Appl. Environ. Microbiol. 1991, 57, 1070–1074. (9) Hendrickx, B.; Dejonghe, W.; Boe¨nne, W.; Brennerova, M.; Cernik, M.; Lederer, T.; Bucheli-Witschel, M.; Bastiaens, L.; Verstraete, W.; Top, E. M. Dynamics of an oligotrophic bacterial aquifer community during contact with a groundwater plume contaminated with benzene, toluene, ethylbenzene, and xylenes: an in situ mesocosm study. Appl. Environ. Microbiol. 2005, 71, 3815–3825. (10) Dojka, M. A.; Hugenholtz, P.; Haack, S. K.; Pace, N. R. Microbial diversity in a hydrocarbon- and chlorinated-solvent contaminated aquifer undergoing intrinsic bioremediation. Appl. Environ. Microbiol. 1998, 64, 3869–3877. (11) Huang, L. N.; Zhou, H.; Zhu, S.; Qu, L. H. Phylogenetic diversity of bacteria in the leachate of a full-scale recirculating landfill. FEMS Microbiol. Ecol. 2004, 50, 175–183. (12) Schloss, P. D.; Handelsman, J. Introducing DOTUR, a computer program for defining operational taxonomic units and estimating species richness. Appl. Environ. Microbiol. 2005, 71, 1501– 1506. (13) Ludwig, W.; Strunk, O.; Westram, R.; Richter, L.; Meier, H.; Yadhukumar Buchner, A.; Lai, T.; Steppi, S.; Jobb, G. ARB: a software environment for sequence data. Nucleic Acids Res. 2004, 32, 1363–1371. (14) Ng, H. Y.; Hermanowicz, S. W. Membrane bioreactor operation at short solids retention times: performance and biomass characteristics. Water Res. 2005, 39, 981–992. (15) Hwang, E. J.; Sun, D. D.; Tay, J. H. Operational factors of submerged inorganic membrane bioreactor for organic wastewater treatment: sludge concentration and aeration rate. Water Sci. Technol. 2003, 47, 121–126. (16) Lee, J.; Ahn, W.-Y.; Lee, C.-H. Comparison of the filtration characteristics between attached and suspended growth microorganisms in submerged membrane bioreactor. Water Res. 2001, 35, 2435–2445. (17) Yamammoto, K.; Hiasa, H.; Mahmood, T.; Matsuo, T. Direct solid liquid separation using hollow fiber membrane in an activated sludge aeration tank. Water Sci. Technol. 1989, 21, 43–54. (18) Pang, C. M.; Liu, W.-T. Community structure analysis of reverse osmosis membrane biofilms and the significance of Rhizobiales bacteria in biofouling. Environ. Sci. Technol. 2007, 41, 4728– 4734. (19) Pang, C. M.; Hong, P. Y.; Guo, H. L.; Liu, W.-T. Biofilm formation characteristics of bacterial isolates retrieved from a reverse osmosis membrane. Environ. Sci. Technol. 2005, 39, 7541–7550. (20) Dudley, L. Y.; Christopher, N. S. J. Practical experiences of biofouling in reverse osmosis systems. In Biofilms in the Aquatic Environment; Keevil, C. W., Godfree, A., Holt, D., Dow, C., Eds.; The Royal Society of Chemistry: Cambridge, U.K., 1999. (21) Ridgway, H. F.; Kelly, A.; Justice, C.; Olson, B. H. Microbial fouling of reverse osmosis membranes used in advanced wastewater treatment technology: chemical, bacteriological, and ultrastructural analyses. Appl. Environ. Microbiol. 1983, 45, 1066– 1084. (22) Chen, C.-L.; Liu, W.-T.; Chong, M.-L.; Wong, M.-T.; Ong, S. L.; Seah, H.; Ng, W. J. Community structure of microbial biofilms associated with membrane-based water purification processes as revealed using a polyphasic approach. Appl. Microbiol. Biotechnol. 2004, 63, 466–473. (23) Koskinen, R.; Ali-Vehmas, T.; Kampfer, P.; Laurikkala, M.; Tsitko, I.; Kostyal, E.; Atroshi, F.; Salkinoja-Salonen, M. Characterization of Sphingomonas isolates from Finnish and Swedish drinking water distribution systems. J. Appl. Microbiol. 2000, 89, 687– 696. (24) Pang, C. M.; Liu, W.-T. Biological filtration limits carbon availability and affects downstream biofilm formation and community structure. Appl. Environ. Microbiol. 2006, 72, 5702–5712.

ES801283Q