Environ. Sci. Technol. 2007, 41, 4728-4734
Community Structure Analysis of Reverse Osmosis Membrane Biofilms and the Significance of Rhizobiales Bacteria in Biofouling C H E E M E N G P A N G † A N D W E N - T S O L I U * ,‡ Department of Civil Engineering and Division of Environmental Science and Engineering, National University of Singapore, Block E2 04-07, 1 Engineering Drive 2, Singapore 117576, Singapore
The biofilm community structure of a biofouled reverse osmosis (RO) membrane was examined using a polyphasic approach, and the dominant phylotypes retrieved were related to the order Rhizobiales, a group of bacteria that is hitherto not implicated in membrane biofouling. A comparison with two other membrane biofilms using T-RFLP fingerprinting also revealed the dominance of Rhizobiales organisms. When pure culture RO biofilm isolates were cultivated aerobically in BIOLOG microplates, most Rhizobiales were metabolically versatile in their choice of carbon substrates. Nitrate reduction was observed in five RO isolates related to Castellaniella, Ochrobactrum, Stenotrophomonas, and Xanthobacter. Many of the key Rhizobiales genera including Bosea, Ochrobactrum, Shinella, and Rhodopseudomonas were detected by PCR to contain the nirK gene responsible for nitrite reductase activity. These findings suggest that Rhizobiales organisms are ecologically significant in membrane biofilm communities under both aerobic and anoxic conditions and may be responsible for biofouling in membrane separation systems.
1. Introduction Interest in wastewater reclamation have increased remarkably in the recent years. Given the superiority of membrane technologies, such as microfiltration (MF) and reverse osmosis (RO), over conventional water and wastewater treatment processes, membrane processes have established themselves as the preferred mode of treatment. However, a major challenge associated with membranes is biological fouling, where the formation of a surface-associated layer of microorganisms, or biofilms, can result in an unacceptable degree of membrane performance (1). To investigate the community structure of membrane biofilms, earlier studies usually employ culture-dependent methods (1). Acinetobacter, Flavobacterium, Pseudomonas, Bacillus, Serratia, and Micrococcus have been described based on m-SPC and R2A agar cultivation (2). Other bacteria including Actinomycetes, Aeromonas, Arthrobacter, Corynebacterium, and Mycobacterium have also been identified by culture-based techniques (3, 4). Among them, the Mycobacterium isolates recovered from full-scale RO installations were most extensively investigated. Their adhesion kinetics * Corresponding author phone: (+65) 65161315; fax: (+65) 67744202; e-mail:
[email protected]. † Department of Civil Engineering. ‡ Division of Environmental Science and Engineering. 4728
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 13, 2007
and correlation with membrane surface characteristics were elucidated (5, 6), and membrane cleaning strategies using surfactants have been optimized for these mycobacterial biofilms (7). However, culture-based approaches tend to select for the fittest and least fastidious of microorganisms, while counterselecting others by competitive exclusion. Further, cultivation conditions often lead to an enrichment bias, where the most abundant isolate retrieved is neither the most dominant nor the most ecologically relevant in the community (8). The more modern molecular methods circumvent these biases and provide a more accurate representation of microbial diversity and abundance (8). Nevertheless, the molecular characterization of biofilm communities on membrane surfaces remains limited in the literature. To date, the molecular analysis of biofilm communities in full-scale MF and RO membranes using 16S rRNA gene clone libraries and fluorescent in situ hybridization has suggested the dominance of Alphaproteobacteria in these biofilms (9), while another study (through 16S rRNA gene-based denaturing gradient gel electrophoresis and DNA sequencing) has indicated the presence of Flavobacterium in the biofilm community of a lab-scale nanofiltration (NF) membrane fed with synthetic wastewater (10). The observed difference in the two biofilm communities suggests a poorly understood relationship between selected biofilm populations and ambient environmental conditions in the membrane biofilm habitat. This paper describes the microbial community composition of a biofilm retrieved from a lab-scale RO membrane module using a polyphasic approach combining 16S rRNA gene-based molecular techniques and bacterial isolation. Comparisons between this biofilm and the two other previously described biofilms recovered from full-scale membrane installations (9) were performed to identify the dominant group of organisms commonly involved in membrane biofouling. Possible reasons for their ecological success in membrane biofilm environments are then investigated based on substrate utilization patterns and nitrate/nitrite respiration using pure culture representatives.
2. Experimental Procedures 2.1. Membrane Biofilm Samples. A lab-scale integrated membrane process developed for wastewater reclamation was previously described (11) (Supporting Information, Figure S1). Primary settled wastewater of 60% industrial and 40% municipal composition was treated in a bench-scale membrane bioreactor (MBR). At steady-state operation, the MBR effluent was blended with the RO concentrate to serve as feedwater (Table 1) for the RO membrane element (GE Osmonics) (Table S1). Chemical cleaning was performed as described (11) for organic fouling control and scale removal. A model VG1210 UV disinfection lamp (Aqua Nautica Aps) pretreated the RO feedwater as it entered the membrane module, but no specific control measure for the removal of biological foulants was undertaken. Biofilms on the RO membrane surface (denoted as MBR-RO) were recovered for microbial community analysis after 193 days of continuous operation and compared with two other previously described membrane biofilms (9). The first biofilm (SE-MF) was retrieved from a full-scale, hollow fiber microfiltration membrane module treating secondary effluent from a domestic wastewater treatment plant, and the other biofilm (PW-RO) was from a full-scale, spiral-wound RO membrane, which was used in the purification of potable water. 2.2. Sample Collection and Total Community DNA Extraction. The RO membrane containing biofilms and other 10.1021/es0701614 CCC: $37.00
2007 American Chemical Society Published on Web 05/24/2007
TABLE 1. Characteristics of Water Streams Associated with Three Membrane Biofilms MBR-ROa
a
SE-MFb
PW-ROb
parameter (mg/L)
influent
concentrate
permeate
influent
influent
TOC COD TDS TN NO3--N polysaccharides (as TOC)
51.0 171 3295 98.6 83.5 c
56.4 188 3650 109 91.1 15-25
0.8 5.8 140 11.7 3.3 c
9-10 c 645 c 3.1 c
∼1 c 200-350 c 0.11-1.6 c
Mean value over 147 days of operation (11, 21).
b
From ref 9. c No data available.
foulants were collected using sterile scalpel knives and transferred into sterile 1X phosphate buffer saline (1X PBS) solution. Three gentle changes of 1X PBS were performed to remove loosely adhered cells. The biofilm material was then physically removed using sterile cotton buds and resuspended in fresh 1X PBS by vortexing. Cell pellets were then collected by centrifugation at 16 440g for 10 min and stored at -80 °C. The total community DNA was extracted as described (12). 2.3. Terminal Restriction Fragment Length Polymorphism (T-RFLP), 16S rRNA Gene Clone Library Construction, and Phylogeny Analysis. T-RFLP of 16S rRNA genes was performed as described previously (12, 13). T-RFLP of nitrite reductase nirK genes was performed with primers nirK1F and nirK5R using a touchdown thermal cycling procedure (14). For the 16S rRNA gene clone library, total community DNA was amplified using Bacteria-specific primers 11F and 1512R as described (15). Near complete 16S rRNA gene sequences (>1300 bp) were obtained for representative clones and compared to those deposited in GenBank using the NCBI BLAST program. Chimeric sequences detected using the Pintail software (16) were discarded. A bootstrapped neighbor-joining tree was prepared in MEGA3 (17) based on the Jukes-Cantor distances computed for aligned sequences. 2.4. Bacterial Isolation and Culture Conditions. Bacterial biomass from the MBR-RO biofilm was recovered by manually scraping the membrane surface using sterile cotton buds, and bacterial isolates were retrieved using R2A agar as described (9). Pure culture bacteria strains were then preserved in aliquots at -80 °C. From separate digestions of RsaI and HhaI, 16S rRNA gene sequences (>1300 bp) of those isolates with distinctive restriction patterns were obtained for phylogenetic identification. For subsequent analyses, preserved cultures were revived by streaking on R2A agar and incubated at 30 °C for 2-7 days. Starter cultures were prepared by introducing a single colony into R2A broth and incubated at 30 °C for 24-48 h with vigorous shaking. The starter cultures were then transferred to fresh R2A medium and grown overnight before use. 2.5. BIOLOG GN2 MicroPlate Assay. Overnight cultures of bacterial cells were collected by centrifugation (8000g for 10 min), washed twice, and resuspended in 0.9% NaCl to an A600 value of 0.3. A total of 150 µL of this mixture was then inoculated into each well of a BIOLOG GN2 MicroPlate. The microplate was sealed with Parafilm and incubated at 30 °C. Respiratory activity in the wells reduced the tetrazolium dye, and the appearance of purple formazan precipitates was monitored by visual inspection and A596 for up to 48 h after inoculation. Data for principal component analysis (PCA) were transformed as described (18), and PCA was performed using MINITAB Statistical Software, Release 14. 2.6. Nitrate Reduction. The batch assay for dissimilatory nitrate reduction was performed in anaerobically prepared 150 mL serum bottles containing 5.5 mM sodium pyruvate (200 mg/L TOC) as the sole carbon source, 7.2 mM NaNO3 (100 mg/L NO3--N) as the sole nitrogen source, 3.4 mM Na2-
HPO4, 2.2 mM KH2PO4, and trace elements. One milliliter of overnight culture was inoculated by syringe injection into 99 mL of medium and incubated at 30 °C for 8 days without shaking. Prior to chemical analyses, the medium was filtered through 0.45 µm membrane filters and diluted using ultrapure water. Total dissolved organic carbon (DOC) and total nitrogen (TN) content were determined in triplicates using a Model 101 wet oxidation total carbon analyzer (O.I. Analytical), while nitrite and nitrate concentrations were measured by ion chromatography using a DX500 chromatography system (Dionex) as described in the ref 19. 2.7. Nucleotide Sequence Accession Numbers. DNA sequences for 16S rRNA genes and nirK genes were submitted to GenBank under the accession numbers EF219020 to EF219053 and EF363541 to EF363546, respectively.
3. Results 3.1. Comparison of Influent Water Quality. Depending on the source of wastewater, the influent characteristics supporting microbial growth in the three membrane biofilm communities were very different (Table 1). In the MBR-RO membrane, the influent contained 51 mg/L TOC, at least 5-fold higher than what was encountered in the SE-MF system and about 50-fold higher than the influent TOC to the PWRO membrane. In addition, at total dissolved solids (TDS) levels reaching 3300 mg/L, the MBR-RO influent was considerably richer in inorganic substances like metallic and nonmetallic ions. Total nitrogen (TN) in the MBR-RO influent was present mainly as NO3--N, and its concentration at 84 mg/L was at least 25 times greater than that found in the influents to the SE-MF and PW-RO membranes. Dissolved oxygen (DO) was maintained at 5.0 mg/L in the MBR but was not measured in the downstream RO feed tank (Figure S1). However, the retention time of the MBR effluent in the feed tank was at least 14 h and was possibly long enough to cause a drop or even depletion in the DO of the influent to the RO element. 3.2. Biofilm Community Structure as Revealed by 16S rRNA Gene-Based Clone Library and Bacterial Isolation. The 80 clones retrieved from the MBR-RO biofilm were assigned to 18 distinct phylotypes from five recognized bacterial lineages (Table S2). Among them, Alphaproteobacteria phylotypes were most dominant at 54%, followed by phylotypes related to Gammaproteobacteria (20%), Betaproteobacteria (11%), and Bacteroides (7.5%). From isolation analyses, 12 of the 13 phylotypes recovered were also distributed in these lineages with the remaining one being related to Actinobacteria. Gammaproteobacteria and alphaproteobacteria isolates were most abundant at 46 and 32%, respectively. On the basis of phylogeny analysis (Figure 1), the majority of alphaproteobacteria clones and isolates were distributed in a single cluster belonging to the order Rhizobiales. Among these, the most abundant was Ochrobactrum-related phylotypes represented by clones RO233 and RO238 (20% of total clones) and isolates ROi15 and ROi52 (24% of total VOL. 41, NO. 13, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4729
FIGURE 1. Phylogenetic relationships of 16S rRNA gene sequences retrieved from clone library and isolation analyses. The phylogenetic tree was constructed using a neighbor-joining algorithm with the Jukes-Cantor distance in MEGA3. The 16S rRNA gene sequence of A. pyrophilus (M83548) was selected as the outgroup. Bootstrap (number ) 1000) values greater than 50% are shown at the nodes, and the bar represents one substitution per 20 nucleotides. The abundance of each clone and isolate is shown in parenthesis. Theoretical T-RF lengths based on in silico MspI and HhaI digestions are also provided, and those in bold can be assigned to an actual peak in the community T-RFLP electrophoregram. N.A.: not assigned as the sequence cannot be amplified using the Cy5-modified forward primer 47F. isolates). Other Rhizobiales phylotypes such as those related to Bosea, Oligotropha, Rhodopseudomonas, and Methylocella were only retrieved in the clone library, while Xanthobacterand Shinella-related phylotypes were recovered exclusively by isolation. Gammaproteobacteria phylotypes were also numerically abundant in the MBR-RO community. Stenotrophomonas and Thermomonas phylotypes each accounted for about 10% the biofilm community. Isolation also revealed the presence of phylotype ROi44 affiliated to Pseudoxanth4730
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 13, 2007
omonas. For Betaproteobacteria, clones RO118 and RO219 belonged to Hydrogenophaga, while the isolate ROi28 was associated with Castellaniella. In the case of Bacteroidetes, the dominant phylotypes were related to Sphingobacteria. Finally, the minor members of the MBR-RO biofilm community included those related to Planctomycetes and candidate divisions TM6 and TM7. 3.3. Biofilm Community Structure as Revealed by 16S rRNA Gene-Based T-RFLP. The MspI-digested MBR-RO
FIGURE 2. T-RFLP fingerprints obtained from the three membrane biofilm samples. 16S rRNA gene-based T-RFLP profiles are produced by digestion with MspI, while nirK T-RFLP patterns are generated using HaeIII. T-RFs shown in italics are consistently retrieved in all three biofilms. Sphingo.: Sphingobacteria. T-RFLP (Figure 2) was characterized by the presence of more than 20 T-RFs, most of which were present in abundances
