Nitric Oxide Treatment for the Control of Reverse OsmosisMembrane Biofouling

Robert J. Barnes,a,* Jiun Hui Low,a,b,c,d Ratnaharika R. Bandi,a Martin Tay,c Felicia Chua,a Theingi Aung,a Anthony G. Fane,b

Staffan Kjelleberg,d,e,f Scott A. Riced,e,f

Advanced Environmental Biotechnology Centre, Nanyang Environment and Water Research Institute, Nanyang Technological University, Singaporea; SingaporeMembrane Technology Centre, Nanyang Environment and Water Research Institute, Nanyang Technological University, Singaporeb; Interdisciplinary Graduate School,Nanyang Technological University, Singaporec; Singapore Centre on Environmental Life Sciences Engineering, Nanyang Technological University, Singapored; Centre forMarine Bio-Innovation and School of Biotechnology and Biomolecular Sciences, The University of New South Wales, Sydney, Australiae; School of Biological Sciences,Nanyang Technological University, Singaporef

Biofouling remains a key challenge for membrane-based water treatment systems. This study investigated the dispersal potentialof the nitric oxide (NO) donor compound, PROLI NONOate, on single- and mixed-species biofilms formed by bacteria isolatedfrom industrial membrane bioreactor and reverse osmosis (RO) membranes. The potential of PROLI NONOate to control ROmembrane biofouling was also examined. Confocal microscopy revealed that PROLI NONOate exposure induced biofilm dispersalin all but two of the bacteria tested and successfully dispersed mixed-species biofilms. The addition of 40 �M PROLI NONOate at 24-hintervals to a laboratory-scale RO system led to a 92% reduction in the rate of biofouling (pressure rise over a given period) by abacterial community cultured from an industrial RO membrane. Confocal microscopy and extracellular polymeric substances(EPS) extraction revealed that PROLI NONOate treatment led to a 48% reduction in polysaccharides, a 66% reduction in pro-teins, and a 29% reduction in microbial cells compared to the untreated control. A reduction in biofilm surface coverage (59%compared to 98%, treated compared to control) and average thickness (20 �m compared to 26 �m, treated compared to control)was also observed. The addition of PROLI NONOate led to a 22% increase in the time required for the RO module to reach itsmaximum transmembrane pressure (TMP), further indicating that NO treatment delayed fouling. Pyrosequencing analysis re-vealed that the NO treatment did not significantly alter the microbial community composition of the membrane biofilm. Theseresults present strong evidence for the application of PROLI NONOate for prevention of RO biofouling.

Membrane technology, for the conversion of seawater andwastewater through desalination and reclamation processes

into potable water, is vital for sustainable water management.However, membrane fouling by bacterial biofilms remains a keychallenge for these technologies (1–3). Initially, the adsorption oforganic species and suspended particles on the wetted membranesurface form a conditioning film. This enables attachment ofplanktonic cells to the membrane surface, followed by the forma-tion of microcolonies and biofilm maturation, where bacterialcells are embedded in a self-produced matrix of extracellular poly-meric substances (EPS) (1, 4, 5). The EPS is typically composed ofpolysaccharides, proteins, and nucleic acids. The attachment ofmicroorganisms to the membrane surface is affected by factorssuch as the membrane material, the roughness of the membranesurface, hydrophobicity, and membrane surface charge (6). Bio-film bacteria have several advantages over single planktonic cells,including optimization of growth and survival, improved acqui-sition of nutrients, and increased protection against environmen-tal stresses, including shear forces (2, 5, 7).

Biofilm formation on membrane surfaces results in a severedecline in flux, or an increase in transmembrane pressure (TMP;defined as the pressure gradient of the membrane, or the averagefeed pressure minus the permeate pressure) to maintain flux,higher energy consumption, and a deterioration of system perfor-mance and product water production (3, 8, 9). As the adhesive andcohesive matrix of biofilms, EPS has been suggested to be thepredominant culprit for biofouling of water treatment mem-branes (1, 10, 11). The EPS is composed mainly of polysaccharidesand proteins, which form hydrogel matrices (12). Common tech-

niques to reduce membrane fouling include membrane cleaningand pretreatment of the feed water. However, microorganismsmay survive pretreatment processes such as coagulation, floccula-tion, sand filtration, ultrafiltration, and cartridge filtration to sub-sequently colonize and foul the system (3). Membrane cleaning byphysical or chemical methods is used to regenerate the functionof fouled membranes, and the methods used and the frequency ofcleaning depend on the type of foulant as well as the resistance ofthe membrane to chemical cleaning agents (13). However, thesecleaning methods frequently shorten membrane life, further in-creasing operational costs (11, 14). Despite the widespread use ofsuch chemicals, they are ineffective in removing or killing themembrane biofilms (1), and regrowth quickly occurs, resulting in

Received 17 October 2014 Accepted 20 January 2015

Accepted manuscript posted online 30 January 2015

Citation Barnes RJ, Low JH, Bandi RR, Tay M, Chua F, Aung T, Fane AG, Kjelleberg S,Rice SA. 2015. Nitric oxide treatment for the control of reverse osmosis membranebiofouling. Appl Environ Microbiol 81:2515–2524. doi:10.1128/AEM.03404-14.

Editor: H. Nojiri

Address correspondence to Scott A. Rice, rscott@ntu.edu.sg.

* Present address: Robert J. Barnes, Centre for Marine Bio-Innovation and Schoolof Biotechnology and Biomolecular Sciences, The University of New South Wales,Sydney, Australia.

Supplemental material for this article may be found at http://dx.doi.org/10.1128/AEM.03404-14.

Copyright © 2015, American Society for Microbiology. All Rights Reserved.

doi:10.1128/AEM.03404-14

April 2015 Volume 81 Number 7 aem.asm.org 2515Applied and Environmental Microbiology

on July 30, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

poor system performance (15). Therefore, new strategies of bio-fouling prevention are required to reduce such impacts.

Recent research has demonstrated that the gas molecule andimportant biological messenger nitric oxide (NO) is a signal forbiofilm dispersal, inducing the transition from the biofilm modeof growth to the free swimming planktonic state (16, 17). NOinduces biofilm dispersal by stimulating phosphodiesterase activ-ity, resulting in the degradation of cyclic di-guanylate monophos-phate (c-di-GMP), culminating in changes to gene expression thatfavor the planktonic mode of growth (17). NO, which has a shorthalf-life in aqueous environments, can be delivered to biofilms byusing chemical compounds that generate NO in solution. The NOdonor compound, sodium nitroprusside (SNP), was shown todisperse biofilms of Pseudomonas aeruginosa, as well as other bac-terial species, including mixed-species biofilms at low, nontoxicconcentrations (16, 18). SNP is a well-characterized NO donorand is applied medically to control vasodilation. It has a half-life ofapproximately 2 min in aqueous solutions at neutral pH. Solu-tions of SNP are stable, that is, they do not release NO, at high pH,e.g., pH 10. A recent study investigated the biofilm dispersal po-tential of three NO donor compounds (MAHMA NONOate,SNP, and PROLI NONOate) using P. aeruginosa PAO1 (19). Theresults showed that MAHMA NONOate could reduce bacterialbiofilms by up to 40% over a 2-h exposure period but that thisreduction was partially due to growth inhibition. The addition ofSNP reduced biofilm biovolume by 40% over a 24-h period butled to enhanced growth over shorter periods. Due to the shorthalf-life of the PROLI NONOate (2 s at pH 7.4 and 37°C) (20), thisNO donor compound quickly dispersed PAO1 biofilms, reducingthe biovolume by 30% after 1 h of exposure (19), with no inhibi-tory or growth effects observed.

It has also been shown that NO stimulates biofilm formation inNitrosomonas europaea (21), Azospirillum brasilense (22), and She-wanella oneidensis (23) or that NO had no effect on biofilms (24,25). Thus, different bacteria may show individual responses toNO, and hence it is uncertain what the overall effect of NO mightbe on a natural, mixed-species biofilm community. Therefore, theaim of this study was to determine the potential of PROLI NONO-ate to disperse individual- and mixed-species biofilms formed bybacteria isolated from fouled industrial reverse osmosis (RO) andmembrane bioreactor (MBR) membranes. In addition, this studyexamined the application of PROLI NONOate as a novel strategyto control membrane biofouling in a laboratory-scale RO system.

MATERIALS AND METHODSBacterial isolates. Bacteria from industrial MBR and RO membraneswere isolated and identified as previously described (19). In addition,biofilm scrapings were taken from six 9-cm2 membrane segments from a5-year-old spiral wound RO module (Kranji NeWater plant, Singapore).Membrane scrapings were mixed together, and bulk genomic DNA wasextracted from two 0.1-g samples using sodium dodecyl sulfate hexade-cyltrimethyl ammonium bromide (26). A portion of the biofilm scrapingwas cultured overnight in R2A broth at 30°C with shaking at 200 rpm toprepare glycerol stocks of a mixed RO bacterial community. DNA wasextracted from the cultured cells as described above to determine thebacterial composition after culturing.

Pyrosequencing. DNA samples from both the raw membrane scrap-ing and the R2A culture were sequenced using the 454 pyrosequencingplatform (Research and Testing Laboratory, Texas, USA) using the eubac-terial primers Gray28F (5=-GAGTTTGATCNTGGCTCAG-3=) and

Gray519R (5=-GTNTTACNGCGGCKGCTG-3=) (27) for the initial PCRamplification.

Sequence analyses. The quality and adaptor trimming of the ampli-con sequencing reads was performed using QIIME (version 1.8.0) (28).The trimmed reads were then clustered into representative sequences ofoperational taxonomic units (OTUs), based on 97% genetic similaritythresholds for the species level, with the Uclust algorithm (29). Chimerascreening and depletion was performed on the representative sequenceusing UCHIME (30). After chimera depletion, the sequences were alignedagainst the Greengenes OTU database (31) using PyNAST (32). The rel-ative abundance for each OTU at the species level was then exported toPRIMER-E (Plymouth Routines in Multivariate Ecological Research) ver-sion 6 (33) to calculate the Bray-Curtis distances.

NO dispersal of batch (petri dish-grown) biofilms. The optimal con-centration and time exposure of the PROLI NONOate for biofilm disper-sal had been previously determined (19). A 10-ml liquid culture of eachbacterial isolate was grown aerobically overnight in its isolation medium(R2A broth, nutrient broth, or tryptone soya broth) at 30°C with shakingat 200 rpm. The bacterial culture was diluted to an optical density at 600nm (OD600) of 0.1 (approximately 1 � 108 CFU ml�1), and 1 ml was usedfor batch biofilm experiments. For mixed biofilms, 1 ml of each dilutedbacterial culture was mixed together, and 1 ml of this suspension wasused. The 12 RO and 10 MBR bacterial isolates used for mixed biofilmshave been previously identified (19). Biofilms were cultivated for 24 h ona glass slide in a petri dish immersed in 19 ml M9 (28.2 mM Na2HPO4, 22mM KH2PO4, 8.6 mM NaCl, 18.7 mM NH4Cl, 2 mM MgSO4, 0.09 mMCaCl2, and 22.2 mM glucose) with gentle shaking (50 rpm) at room tem-perature (24°C). After 24 h, the medium was replaced with 20% M9 me-dium and incubated for a further 1 h, at which time PROLI NONOate(Cayman Chemicals) solution in 10 mM NaOH was added to give a finalconcentration of 40 �M. For the untreated control, an equal amount of 10mM NaOH was added. For each experiment, treated and untreated bio-films formed by a mixture of isolates were run in triplicate. Each experi-ment was repeated three times to give an overall average. After 1 h ofexposure, slides were gently washed in 0.85% NaCl solution and stainedusing the LIVE/DEAD BacLight bacterial viability kit (Invitrogen) as perthe manufacturer’s instructions. The slides were then washed again in0.85% NaCl solution to remove excess stain. Live, SYTO 9-stained cellsand dead, propidium iodide (PI)-stained cells were visualized by confocallaser scanning microscopy (CLSM) (Nikon Eclipse 90i, part of the A1Rhybrid confocal spectral imaging system) at a magnification of �20, withan argon laser (488-nm wavelength for SYTO 9) and a diode laser(561-nm wavelength for PI). For each slide, z stack (three-dimensional[3D]) confocal images were obtained from 9 locations covering the glasssurface, and the average biovolume (�m3) was calculated using IMARISsoftware (Bitplane; version 7.3.1). Biofilms on glass slides were stored inpetri dishes containing water-soaked tissue paper to ensure humidity. Inaddition, slides were analyzed in a specific order so that any drying out ofbiofilm would not bias the results. No coverslips were used in order toprotect biofilm integrity. For each experiment, the percentage of biofilmdispersal was calculated by comparing the average biovolume of NO-treated slides to the untreated control.

Reverse osmosis system. The RO pressure reactor was assembled astwo independent cells in parallel, allowing one cell to be used as an un-treated control. Each stainless steel RO cell had a flat plate geometry andflow channel sizes of 150 by 30 by 0.8 mm (length by width by height) withan effective area of 0.0045 m2. The design (Fig. 1) and operation have beenpreviously described (34) and represent conditions that simulate typicallarge-scale RO processes. Each 10-liter feed tank was equipped with astirrer (IKA; model Eurostar) and was cooled using a chiller (PolyScience)to maintain the feed solution at 23 � 1°C (Fig. 1). A high-pressure cen-trifugal pump (Hydra-Cell) was used to deliver the feed solution andmaintain the cross-flow velocity (CFV) at 0.28 m s�1, while system pres-sure was set to 362.6 lb/in2 using a flow control valve (Swagelok; modelSS-4R3A). The feed flow rate was monitored with a turbine flow meter

Barnes et al.

2516 aem.asm.org April 2015 Volume 81 Number 7Applied and Environmental Microbiology

on July 30, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

(WF Waterflo). The pressure of feed and permeate streams was monitoredby pressure transducers (Ashcroft). RO membranes (DOW Filmtec; TW-30) were cut to size (3 cm by 15 cm) and soaked in Milli-Q water for atleast 12 h. The hydrated membranes were then sterilized in 70% ethanol(Merck) for 1.5 h followed by rinsing with Milli-Q water. The membraneswere compacted at a maximum flux, which was up to 60 liters/m2/h(LMH) overnight with Milli-Q water until a stable flux was achieved,which was monitored using a mass flow controller (Brooks Instrument;model 5882). Following compaction, the flux was set to 30 LMH, and anNaCl stock solution (3.4 M) was added to the feed tank to a final concen-tration of 0.034 M. The system was allowed to mix for 1.5 h before thenutrient source was added. For PAO1 experiments, a nutrient broth (NB)(Difco, BD) stock solution (8 g liter�1) was used to provide an averagenutrient concentration of 24 mg liter�1. This is a similar concentration toprevious biofouling studies in laboratory-scale RO systems (34–37). ForRO mixed-community experiments, an R2A broth (1.25 g liter�1 yeastextract, 1.25 g liter�1 proteose peptone, 1.25 g liter�1 Casamino Acids,1.25 g liter�1 starch, 6.9 mM glucose, 4.3 mM K2HPO4, 0.5 mM MgSO4,6.8 mM sodium pyruvate) stock solution (7.81 g liter�1) was used toprovide an average nutrient concentration of 25 mg liter�1. The systemwas allowed to mix for a further 1.5 h prior to the start of the experiment.The feed solution was replenished twice daily, and the total organic car-bon (TOC) was tested with a TOC analyzer (Shimadzu; model TOC-VWS) to ensure that the concentration of nutrients remained equal ineach parallel unit.

NO treatment for prevention of membrane biofouling. The mem-brane fouling rate of an untreated RO cell was compared to the parallel ROcell dosed with 40 �M PROLI NONOate every 24 h. The RO cells wereinoculated with either P. aeruginosa PAO1 or the mixed bacterial commu-nity cultured from the membrane scraping. The bacteria were grownovernight in either 750 ml nutrient broth (PAO1) or R2A broth (mixedcommunity) at 30°C with shaking at 200 rpm. The bacterial cells weresubsequently harvested by centrifugation at 18,514 � g for 10 min. Thepellet was washed and resuspended in 0.034 M NaCl to an OD600 of 0.1.The bacterial suspension was injected into the system before the feedsolution entered the RO cells by using an injection pump (ELDEX; model5979-Optos Pump 2HM). The RO system was operated in fully recycledmode, where the concentrate and permeate flows were returned to thefeed tank. Microfilters (KAREI; 5 and 0.2 �m) were installed downstreamof the pressure cell to prevent excess bacteria from entering the feed tank.Experiments were initiated by continuous injection of the bacterial sus-

pension into the flow line at a dilution rate of 1:800, giving an input loadof 105 CFU ml�1. The bacterial solution was replaced every 48 h. Experi-ments were conducted at a constant flux (30 LMH), and the difference inpressure between the feed and permeate stream (TMP) was monitoredcontinuously. For PROLI NONOate injection, the bacterial suspensionwas replaced with sterile 0.034 M NaCl, and the bacterial injection tubingwas flushed for 10 min at 2 ml min�1 to remove excess bacterial cells.PROLI NONOate (10,000 �M) in 10 mM NaOH was then injected intothe flow line for 30 min at a dilution rate of 1:250 (1.6 ml min�1), givinga final concentration of 40 �M. For the untreated cell, 10 mM NaOH,without PROLI NONOate, was used as a control. After dosing, the bacte-rial suspension was injected into the flow line as described before. PROLINONOate was injected into the treated cell every 24 h. This dosing intervalwas selected, as multiple dosing over shorter time periods did not enhancebiofilm dispersal in previous experimental work (19). Fouling was definedhere as the increase in TMP over time during the filtration process atconstant flux. Therefore, a higher TMP within the same time frame or thesame TMP within a shorter time frame represents more severe fouling.

Membrane autopsy. The fouled membranes were removed from theRO modules for examination at the conclusion of the experiment, and 1.5cm was removed from each end of the membrane while the remainder wasaseptically cut into 8 segments. Six segments (1 by 3 cm) of membranecovering the inlet, middle, and outlet of the RO cell were analyzed byfluorescence staining and CLSM observation to quantify the live and deadcells and the polysaccharide volume (representative of EPS). For the re-maining two segments (3 by 3 cm), viable bacterial counts were deter-mined, as were the concentrations of polysaccharide and protein ex-tracted from the membrane surface.

Fluorescent staining and CLSM. Segments of membrane were gentlywashed in 0.145 M NaCl solution and stained using SYTO 9 and PI (3 �gml�1 each) or 100 �g ml�1 ConA-fluorescein isothiocyanate (FITC) (Sig-ma-Aldrich). The segments were then washed again in NaCl solution toremove excess stain, mounted onto glass slides, and viewed using CLSM asdetailed above. For each section, z stack (3D) confocal images were ob-tained from 5 locations covering the membrane surface, and the averagebiovolume (�m3) and surface coverage (%) were calculated usingIMARIS (Bitplane; version 7.3.1).

Viable bacterial counts. Using a cell scraper, the biofilm was removedfrom each 3- by 3-cm segment of membrane and resuspended in individ-ual tubes containing 3 ml sterile phosphate-buffered saline (PBS) andvortexed for 30 s. For PAO1, viable heterotrophic counts were determined

FIG 1 Schematic of one of two independent cells of the laboratory-scale RO system.

Nitric Oxide for the Control of RO Membrane Biofouling

April 2015 Volume 81 Number 7 aem.asm.org 2517Applied and Environmental Microbiology

on July 30, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

using a modified method from Miles et al. (38). In brief, three 10-�l dropsof 10�1 to 10�6 dilutions were pipetted onto nutrient agar. For the mixedRO bacterial community, viable counts were determined by spread plat-ing, in triplicate, 100 �l of 10�4, 10�5, and 10�6 dilutions on R2A agar(Oxoid). Plates were incubated at room temperature for 24 h beforecounting. Viable cells were expressed as CFU/cm2 of membrane.

EPS extraction and quantification. The remaining bacterial suspen-sion was centrifuged at 18,514 � g for 10 min, and the supernatant wastransferred to a clean centrifuge tube for analysis of the soluble EPS frac-tion. The bound EPS was extracted by adding 2 ml 0.145 M NaCl and 12 �lof 37% formaldehyde (Sigma) to the bacterial pellet, and the suspensionwas vortexed and incubated at 4°C for 1 h. After incubation, 2 ml of 1 MNaOH and 0.5 ml Milli-Q water were added, and the solution was vor-texed and incubated at 4°C for 3.5 h. Subsequently, the bacterial suspen-sion was centrifuged at 18,514 � g at 4°C for 20 min, and the supernatantwas transferred to a clean centrifuge tube. A blank control consisted solelyof the EPS extraction reagents. The polysaccharide content of the EPS wasmeasured by the phenol-H2SO4 method (39). Briefly, 0.5 ml of 5% phenol(Sigma-Aldrich) and 2.5 ml of concentrated H2SO4 (Merck) were addedto 1 ml of sample and incubated at room temperature for 10 min. Theabsorbance of the solution was measured at 492 nm (Shimadzu; modelUV1800) and compared against a standard curve generated using glucose(Sigma-Aldrich). The protein content of the EPS was analyzed using theCoomassie (Bradford) protein assay kit (Thermo Scientific). A 1-ml vol-ume of the sample was mixed with 1 ml of working reagent and incubatedat room temperature for 10 min, and the absorbance was measured at 595nm (Shimadzu; model UV1800). Bovine serum albumin (Thermo Scien-tific) was used to generate a standard curve for quantification of proteins.

DNA extraction and pyrosequencing. For the mixed bacterial com-munity, the biofilm was removed from three 3-cm2 segments of mem-brane using a cell scraper for DNA extraction to compare the untreatedand PROLI NONOate-treated membrane bacterial communities. DNAextraction, pyrosequencing, and sequence analysis were performed as de-scribed above.

RESULTSNO induced dispersal of MBR and RO isolate batch (petri dish-grown) biofilms. Under batch conditions, the bacterial isolates

differed in their response to NO, with the percentage of biofilmdispersal ranging from 10% to 46% for the RO and MBR isolates(Fig. 2). A mixed biofilm consisting of all 12 RO isolates (19)showed 17% dispersal, while a mixed biofilm consisting of all 10MBR isolates (19) showed 23% dispersal (Fig. 2). An increase inbiofilm formation in the presence of PROLI NONOate was ob-served for one RO isolate, Duganella sp. (8%), and one MBR iso-late, Acinetobacter sp. (35%) (Fig. 2). In all batch experiments, theratio of dead to live cells did not increase after exposure to PROLINONOate (data not shown). Batch biofilms were established us-ing a complex, mixed-species community recovered from fouledRO membranes to determine the effect of NO on biofilm disper-sal. The NO-induced dispersal for the RO bacterial communitybiofilm was 25% (standard deviation [SD] � 7; n � 3 experi-ments) (data not shown), which was comparable to the artificiallyrecreated biofilms based on RO and MBR isolates (Fig. 2). Overall,the results suggest that NO can induce dispersal in RO and MBRisolates as well as mixed-species biofilms. Therefore, experimentswere designed to test the effect of NO on dispersing biofilms in anRO system.

PROLI NONOate treatment for the prevention of RO mem-brane biofouling by PAO1. Biofilm formation in the RO modulediffers from that in the batch systems described above, where theRO module includes continuous flow and flux (the passage ofwater through the membrane), which facilitate transport of bac-teria and nutrients to the surface. Additionally, the cross-flow in-creases aeration, which may result in faster inactivation of NOthrough binding to oxygen. Therefore, before testing the effect ofNO on a mixed-species biofilm, the effect of NO on dispersingbiofilms of the well-characterized P. aeruginosa was first tested.Results from multiple replicate experiments in the absence ofPROLI NONOate treatment showed that the TMP rise in bothmodules was very similar when inoculated with PAO1, with only a0.3% difference in the time required to achieve a maximum TMP

FIG 2 Average percentage reduction of 24-h batch biofilms formed by RO and MBR bacterial isolates when exposed to 40 �M PROLI NONOate for 1 h. Errorbars are standard deviations (n � 3 experiments).

Barnes et al.

2518 aem.asm.org April 2015 Volume 81 Number 7Applied and Environmental Microbiology

on July 30, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

increase of 120% (Fig. 3A). If the experiment was stopped afterone module had reached the maximum TMP percentage increase,the maximum TMP percentage difference between the two mod-ules was only 6%. Thus, differences of �6% between fouling rateswere considered to be significant.

The RO membrane without PROLI NONOate treatmentfouled in 140 h, a TMP increase of 120% (Fig. 3B). The TMPinitially showed a slow increase, �20% in the initial 100 h, fol-lowed by an abrupt rise in TMP over the next 40 h. Over the same140-h time period, the TMP of the RO injected with PROLINONOate increased by only 68%, a 52% difference between thetreated and untreated membranes fouled with PAO1. The biofilmbiovolume (total cells) of the treated membrane was 28% less thanthe untreated membrane (Fig. 3C). Additionally, the mean viablebacterial count of the treated membrane (3.89 � 106) was 18%lower than that of the untreated membrane (4.72 � 106 CFUcm2�1), reflecting the lower number of bacterial cells present onthe membrane. The ratio of dead to live cells was greater in theuntreated biofilm (0.34) than in the PROLI NONOate-treatedbiofilm (0.27), revealing that the PROLI NONOate treatment didnot have a bactericidal effect. The treated PAO1 biofilm was thin-ner than the untreated control (24 compared to 32 �m, treatedcompared to control) and had a lower surface coverage (83%compared to 99%, treated compared to control), with more areasof unfouled membrane (Fig. 3D). Staining with ConA-FITC re-vealed that the average polysaccharide biovolume was 20% less forthe treated membrane than for the control (Fig. 3C). This differ-

ence was also illustrated by EPS extraction, revealing a 52% reduc-tion in polysaccharides on the treated membrane. A 25% reduc-tion in protein was also observed for the treated membrane (Fig.3C). Collectively, the data clearly show that NO induced dispersalof the PAO1 biofilm, leading to reductions in CFU and biofilmmatrix components, and improved performance of the RO mod-ule. These results demonstrated that NO could be used in thecomplex RO module to induce dispersal at concentrations derivedfrom the batch biofilm experiments, and therefore the experimentwas repeated using an RO bacterial community.

PROLI NONOate treatment for the prevention of RO mem-brane biofouling by a mixed RO bacterial community. As de-scribed above for PAO1, the variability of fouling between the twoRO modules was determined in the absence of NO addition. TheTMP rise in both modules, when fouled with the R2A-culturedmixed RO bacterial community, was similar to that observed forthe PAO1, with a 0.4% difference in the time required for a max-imum TMP increase of 120% (see Fig. S1 in the supplementalmaterial). The TMP profile was similar to that of the PAO1, with aslow initial increase in TMP, �20% in the initial 92 h, followed bya more rapid increase where the maximum TMP was reached inthe next 25 h (see Fig. S1). If the experiment was stopped after onemodule had reached the maximum TMP percentage increase, amaximum TMP percentage difference between the two moduleswas 8%, and therefore differences of �8% were considered to besignificant.

The untreated RO membrane fouled in 117 h, a TMP increase

FIG 3 (A) TMP profiles of the RO system inoculated with PAO1, both cells untreated. TMP values were recorded every 1 min over the experimental period. (B)TMP profiles of the RO system inoculated with PAO1, untreated versus treated with 40 �M PROLI NONOate every 24 h. (C) RO membrane autopsy results foruntreated versus treated: biofilm biovolume as determined by staining with SYTO 9 and PI; biofilm biovolume as determined by staining with ConA-FITC;polysaccharide concentration as determined by EPS extraction; protein concentration as determined by EPS extraction. Error bars are SD, n � 3 experiments. (D)PAO1 biofilm on untreated versus treated membranes. Representative confocal images showing biofilm structure via live and dead bacterial cells stained withSYTO 9 and PI. Magnification, �20. Scale bar � 100 �m.

Nitric Oxide for the Control of RO Membrane Biofouling

April 2015 Volume 81 Number 7 aem.asm.org 2519Applied and Environmental Microbiology

on July 30, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

of 112% (Fig. 4A). Over the same 117-h time period, the TMP ofthe PROLI NONOate-treated RO module increased by only 20%,a 92% difference between the NO-treated and untreated mem-branes. The biofilm biovolume (total cells) of the treated mem-brane was 29% less than that of the untreated membrane (Fig. 4B).Additionally, the number of CFU of the treated (8.85 � 106 CFUcm2�1) membrane was 23% lower than that of the untreatedmembrane (1.15 � 107 CFU cm2�1). The ratio of live to dead cellswas lower in the untreated biofilm (0.83) than in the PROLINONOate-treated biofilm (0.92), indicating that the PROLINONOate was not bactericidal. The biofilm surface coverage(59% compared to 98%, treated compared to control) and thick-ness (20 �m compared to 26 �m, treated compared to control)was also significantly lower on the NO-treated membrane, whichalso showed more areas of unfouled membrane (Fig. 4C). Therewas 48% less polysaccharide and 66% less protein in in the EPSextract of the treated membrane biofilm relative to the control(Fig. 4B).

Due to the decreased biofouling rate in the PROLI NONOate-treated RO cell, the experiment was repeated and modified wherethe NO treatment was continued until the TMP reached its max-imum. The RO membrane without PROLI NONOate treatmentfouled in 115 h, a total TMP increase of 130% (Fig. 5). Over thesame 115-h time period, the TMP of the RO injected with PROLINONOate increased by only 33%, a difference of 97% between theNO-treated and untreated membrane, a similar result to the pre-vious experiment. However, biofouling could not be preventedindefinitely, and the treated RO membrane eventually became

fully fouled after 147 h, a 22% difference in time required toachieve the maximum TMP.

Bacterial community composition of the membrane biofilmcommunities. To determine the effect of NO treatment on themixed-species biofilm community on the RO membranes, thecommunity present in the initial scraping, the R2A cultured com-munity, as well as the community that ultimately formed biofilmson the membrane surface during fouling were first determined.The raw biofilm scraping taken from the industrial RO membranewas quite diverse, containing at least 9 bacterial phyla (data notshown). The community was represented mainly by Proteobacte-ria (91%), followed by Actinobacteria (5.3%) and Bacteroidetes(2.75%) at lower abundances (data not shown). The Proteobacte-ria classes were dominated by Alphaproteobacteria (77.35%), fol-lowed by Betaproteobacteria (12.1%) and Gammaproteobacteria(1.4%) (data not shown). The scraping samples contained at least53 bacterial families from 29 orders (see Fig. S2 in the supplemen-tal material). The most abundant families were Bradyrhizobiaceae(33% � 1%), an unknown family of the order Rhizobiales (18% �2%), Sphingomonadaceae (12% � 2%), Nitrosomonadaceae (8%),Rhodospirillaceae (8% � 1%), Hyphomicrobiaceae (5% � 1%),and Mycobacteriaceae (3%). The Bray-Curtis index at the specieslevel revealed an 87% similarity between duplicate samples. TheProteobacteria (85.3%) remained as the dominant phylum in thecultured samples, followed by Bacteroidetes (11.9%) (data notshown). The Gammaproteobacteria class dominated the scrapingsamples (62.9%), followed by Betaproteobacteria (22.1%) (datanot shown). The dominant families after culturing were Aeromon-

FIG 4 (A) TMP profiles of the RO system inoculated with the mixed RO bacterial community, untreated versus treated with 40 �M PROLI NONOate every 24h. TMP values were recorded every 1 min over the experimental period. (B) RO membrane autopsy results for untreated versus treated: biofilm biovolumedetermined by staining with SYTO 9 and PI; polysaccharide concentration as determined by EPS extraction; protein concentration as determined by EPSextraction. Error bars are SD, n � 3 experiments. (C) RO bacterial community biofilm on untreated versus treated membranes. Representative confocal imagesshowing biofilm structure via live and dead bacterial cells stained with SYTO 9 and PI. Magnification, �20. Scale bar � 100 �m.

Barnes et al.

2520 aem.asm.org April 2015 Volume 81 Number 7Applied and Environmental Microbiology

on July 30, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

adaceae (43% � 8%), Oxalobacteraceae (17% � 3%), Pseudomon-adaceae (12% � 2%), and Flavobacteriaceae (6% � 2%). Thecommunities of the duplicate planktonic samples were very sim-ilar, with an 85% similarity at the species level as determined bythe Bray-Curtis index.

When the bacterial communities of two untreated membranebiofilms from the two experiments described above were exam-ined at maximum TMP, the Bray-Curtis index at the species levelrevealed a 92% similarity between samples (see Fig. S3 in the sup-plemental material), showing good reproducibility of the biofilmcommunity between the parallel cells of the RO system. The dom-inant phylum of the membrane biofilm was Proteobacteria(59.2%), followed closely by Bacteroidetes (40.6%). Of the Proteo-bacteria, the Betaproteobacteria class dominated the membranebiofilm samples (35.4%), followed by Gammaproteobacteria(22.6%) and Alphaproteobacteria (1.3%) (data not shown). Thedominant class of the Bacteroidetes phylum was Sphingobacteriia(34.3%). Despite bacteria of the family Aeromonadaceae being themost abundant in the R2A cultures (see Fig. S2), the dominantbacterial family of the membrane biofilms was Chitinophagaceae(23%) (see Fig. S3), suggesting that this organism is a strong bio-film former and thus plays a significant role in the biofouling ofRO system membranes. Other dominant bacterial families withinthe membrane biofilms were Oxalobacteraceae (23%), Enterobac-teriaceae (14%), Comamonadaceae (13%), Sphingobacteriaceae(11%), Aeromonadaceae (7%), and Flavobacteriaceae (6%) (seeFig. S3).

Bacterial community composition of the membrane bio-films and the effect of PROLI NONOate. The untreated andtreated membrane biofilms were examined after 115 and 147 h,respectively, once they had reached maximum TMP. In compar-ison to the untreated control, dosing with PROLI NONOate led toa decrease in the biofilm percentages of Chitinophagaceae (de-crease from 43% to 20%), Sphingobacteriaceae (decrease from

13% to 4%), and Comamonadaceae (decrease from 12% to 7%)but an increase in the percentages of Oxalobacteraceae (increasefrom 9% to 21%), Enterobacteriaceae (increase from 11% to 29%),and an unknown family of the order Rickettsiales (increase from6% to 9%) (see Fig. S3). However, the same six families of bacteriadominated the biofilms of both the treated and untreated mem-branes. The Bray-Curtis index at the species taxonomic level re-vealed a 76% similarity between samples, indicating that thePROLI NONOate treatment did not significantly alter the micro-bial composition of the membrane biofilm (see Fig. S3). The re-sults from the Bray-Curtis similarity index dendrogram indicatedthat the variation between the RO runs was higher than the vari-ation between the untreated and PROLI NONOate-treated mem-brane biofilms, further suggesting that the NO treatment did notsignificantly alter the biofilm community.

DISCUSSION

The growing need to improve water production for developed anddeveloping nations has focused attention on membrane-basedpurification methods. However, biofilm formation and biofoul-ing remains a significant issue for the technology, and thus there isa real need for novel approaches to control biological fouling. Onerecent discovery has been that bacteria produce and respond toNO and use this as a natural dispersal signal (16–18, 40, 41). Here,we have investigated the potential of PROLI NONOate to dispersebacteria isolated from fouled industrial reverse osmosis (RO) andmembrane bioreactor (MBR) membranes and explored the po-tential application of PROLI NONOate as a novel strategy to con-trol membrane biofouling in a laboratory-scale RO system.

It has been demonstrated in this study that micromolar con-centrations of PROLI NONOate induce biofilm dispersal in arange of bacteria isolated from industrial membrane bioreactorand RO membranes. Interestingly, results showed that differentbacteria responded differently to NO exposure, with biofilm dis-

FIG 5 TMP profiles of the RO system inoculated with mixed RO bacterial community, untreated versus treated with 40 �M PROLI NONOate every 24 h. TheRO system was run using feed solution with 31 mg liter�1 R2A broth and 0.034 M NaCl, at 30 LMH flux and a CFV of 0.17 m s�1. TMP values were recorded every1 min over the experimental period.

Nitric Oxide for the Control of RO Membrane Biofouling

April 2015 Volume 81 Number 7 aem.asm.org 2521Applied and Environmental Microbiology

on July 30, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

persal percentages ranging from 10% to 46%. These results corre-spond with the literature where NO induces biofilm dispersal insome bacteria (17) but may stimulate biofilm formation in others(21–23), where it was observed here that NO induced biofilmformation for Acinetobacter spp. and Duganella spp. The responseof these bacteria was different from that observed in a formerstudy, when the NO donor compound MAHMA NONOate wasemployed (19). This difference is likely a consequence of thegrowth inhibitory nature of the MAHMA NONOate, which wasshown to not be the case for the PROLI-NONOate (19). In addi-tion, each NO donor compound has a different half-life/releaserate of NO, which may have impacted the bacteria differently (20,42, 43). The concentration and time of NO gas exposure maytrigger different dispersal behavior in certain species of bacteriaand would be worth investigating in future studies. Overall, themixed biofilm effects are similar for both NO donor compoundstested, and hence the data reveal that it is not easy to predict effectson mixed communities under realistic conditions by using singlespecies. Importantly, the results showed that PROLI NONOate isnot bactericidal at the concentration tested in this study. The bio-film ratio of dead to live cells did not increase for any bacterialgenera tested after exposure to this compound.

Previous studies have investigated only NO-induced biofilmdispersal of P. aeruginosa and other bacterial species in batch ex-periments (16, 18, 19). However, in this study, it has been shownthat the regular addition of an NO donor compound to a contin-uously operating laboratory-scale RO system can significantly re-duce the rate of biofouling by both a single species of P. aeruginosaand also, more importantly, a mixed bacterial consortium isolatedfrom a fouled industrial RO membrane. In the mixed bacterialcommunity, dosing with PROLI NONOate led to a decrease in thebiofilm percentages of Chitinophagaceae, Sphingobacteriaceae, andComamonadaceae but an increase in the percentages of Oxalobac-teraceae, Enterobacteriaceae, and an unknown family of the orderRickettsiales. It therefore also seems apparent in this study thatwhile the dispersal of some bacteria is induced during exposure tolow levels of NO, biofilm formation may be encouraged in others.It is important to note that the pyrosequencing does not providean absolute quantification of the number of each organism pres-ent, only a relative abundance. Therefore, it is possible that all ofthese organisms are generally dispersed, but their individual re-sponses differ, as shown in Fig. 2, resulting in a change in theirrelative abundances. Despite these different responses, the samesix families of bacteria dominated the biofilms of both the treatedand untreated RO membranes, with a microbial composition sim-ilarity of 76%. Thus, the addition of PROLI NONOate did notsignificantly alter the microbial composition of the membranebiofilm. In addition, the increase in certain biofilm members didnot affect the success of the PROLI NONOate treatment to dra-matically reduce the rate of biofouling in the RO system.

Previous work has shown that while the aggregation of bacte-rial cells on the membrane surface may decrease salt rejection andpermeate flux by a biofilm-enhanced osmotic pressure mecha-nism, the EPS biofouling layer adversely impacts permeate flux byincreasing the hydraulic resistance to permeate flow (10, 11). Arecent study showed that the biofouling rate is dependent on boththe quantity of EPS components and the surface coverage of themembrane biofilm, rather than the number of bacterial cells pres-ent (34). A lack of biofilm polysaccharides combined with areas ofunfouled membrane dramatically reduced any hydraulic resis-

tance to permeate flow, thus delaying the rate of biofouling (34).This study showed similar results, where a reduction in EPS con-stituents (protein and polysaccharide) and a decreased biofilmsurface coverage after NO treatment led to significantly lower bio-fouling rates. A reduction in biofilm cells was also observed afterPROLI NONOate treatment. There is an increasing interest indeveloping alternative, nontoxic methods to control microbialfouling in membrane-based water purification, and in particularthere has been a focus on modification of the biofilm developmentprogram. For example, Yeon et al. recently demonstrated thatdisruption of microbial cell-cell signaling could similarly delay theTMP rise associated with biofouling of a membrane bioreactor(44). Thus, biologically derived strategies may represent an addi-tional approach for the control of fouling communities.

This study has contributed to existing knowledge on the struc-ture and diversity of industrial RO membrane biofilm communi-ties (3, 35, 45–49). The advancement of modern molecular meth-ods, such as pyrosequencing, combined with ever-decreasinganalysis costs enable current studies to portray a detailed charac-terization of sample microbial diversity and abundance. Identify-ing common bacteria that contribute to the biofouling of indus-trial RO membranes could enable the development of tailoredsolutions for its control. The industrial RO membrane biofilmsupported a bacterial community dominated by Proteobacteria,with a relatively low abundance of other phyla. The dominance ofProteobacteria on a fouled industrial RO membrane has been ob-served in other studies (3, 15, 35, 45–48), covering RO plants inthe Netherlands (3), Australia (48), Israel (47), Italy (49), andSingapore (35, 46). The dominant class within this phylum wasobserved to be Alphaproteobacteria, also similarly observed inother studies (3, 35, 46, 48), followed by Betaproteobacteria (3, 35,48). Two studies by Bereschenko et al. (15, 45) revealed that Beta-proteobacteria are dominant over Alphaproteobacteria within im-mature biofilms, and this changes to a dominance of Alphaproteo-bacteria over Betaproteobacteria as the biofilm matures (15, 45).This is consistent with the complex community used here to in-oculate the RO system, which was dominated by Alphaproteobac-teria, and that was collected from fouled membranes that had beenin use for more than 5 years. However, after inoculation and foul-ing for approximately 5 days, the community was observed to bedominated by Betaproteobacteria, consistent with previous obser-vations of early-stage biofouling communities.

A recent study has exposed the difficulty in determining themost predominant and problematic microbial species in a specificwater recycling plant, because species composition varies signifi-cantly from one plant to another depending on site-specific con-ditions such as feed water quality and seasons (48). However, bycomparing the bacterial community data from the raw KranjiNeWater membrane scraping in this study with that of other pub-lished work, some common bacteria can be observed to occur onfouled industrial RO membranes. These include bacteria of theorders Sphingomonadales (47, 48), Rhizobiales (46, 47), Pseu-domonadales (47), Xanthomonadales (47), and Burkholderiales(45, 47), bacteria in the families Comamonadaceae (3, 15, 45, 47),Nitrosomonadaceae (48), Flavobacteriaceae (48), Microbacteri-aceae (48), Pseudomonadaceae (48), Kineosporiaceae (48), andPlanctomycetaceae (3), and bacteria of the genera Bacillus (19, 35),Bradyrhizobium (15, 35, 48), Rhizobium (48), Sphingomonas (3,15, 35, 45, 46, 48, 49), Sphingopyxis (3, 15, 45, 48), Pseudomonas(3, 15, 19, 45, 48), Mycobacterium (3, 15, 48), Hyphomicrobium (3,

Barnes et al.

2522 aem.asm.org April 2015 Volume 81 Number 7Applied and Environmental Microbiology

on July 30, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

45), Acidovorax (3, 45), Janthinobacterium (3), Microbacterium (3,15, 35, 46), Geothrix (35), Brevundimonas (49), Novosphingobium(15), Sphingobium (15), Acidovorax (15), and Flavobacterium(15). These bacteria therefore seem to play a significant role in ROmembrane biofouling at a number of RO plants covering fivedifferent countries, spanning three continents.

Micromolar concentrations of PROLI NONOate have beenshown to disperse a wide diversity of single-species biofilmsformed by bacteria isolated from industrial membrane bioreactorand RO membranes, as well as isolates combined to generatemixed-species biofilms. The addition of PROLI NONOate every24 h to a laboratory-scale RO system reduced the rate of biofoul-ing by 92% for a bacterial community cultured from an industrialRO membrane. NO treatment led to a reduction in biofilm con-stituents (polysaccharides, proteins, and total cells) as well as bio-film surface coverage and thickness. The addition of PROLINONOate also led to a 22% increase in the time required for theRO module to reach its maximum TMP. Importantly, communitysequence analysis revealed that the NO treatment did not signifi-cantly alter the microbial composition of the membrane biofilm.Further work aimed at optimizing the dosing protocol is likely toimprove the efficacy observed here. Overall, the results presenteddemonstrate strong evidence for the application of PROLINONOate for delaying RO biofouling.

ACKNOWLEDGMENTS

This research was supported by a research grant (MEWRC651/06/177)from the Environment and Water Industry Programme Office of Singa-pore. This research was also partly supported by (i) the Singapore Na-tional Research Foundation under its Environmental & Water Technolo-gies Strategic Research Programme and administered by the Environment& Water Industry Programme office (EWI) of the PUB, (ii) PUB, NationalWater Agency of Singapore, and (iii) NTU. The Singapore MembraneTechnology Centre, Nanyang Environment and Water Research Institute,Nanyang Technological University, is supported by the Economic Devel-opment Board of Singapore.

We also thank the Singapore Public Utilities Board (PUB) for provid-ing the RO membrane samples, Chong Tzyy Haur for designing the lab-oratory-scale RO system, and Miles Rzechowicz and Stanislaus Suwarnofor help with system troubleshooting.

REFERENCES1. Flemming HC. 2002. Biofouling in water systems— cases, causes and

countermeasures. Appl Microbiol Biotechnol 59:629 – 640. http://dx.doi.org/10.1007/s00253-002-1066-9.

2. Mayer C, Moritz R, Kirschner C, Borchard W, Maibaum R, WingenderJ, Flemming HC. 1999. The role of intermolecular interactions: studies onmodel systems for bacterial biofilms. Int J Biol Macromol 26:3–16. http://dx.doi.org/10.1016/S0141-8130(99)00057-4.

3. Bereschenko LA, Heilig GH, Nederlof MM, van Loosdrecht MC, StamsAJ, Euverink GJ. 2008. Molecular characterization of the bacterial com-munities in the different compartments of a full-scale reverse-osmosiswater purification plant. Appl Environ Microbiol 74:5297–5304. http://dx.doi.org/10.1128/AEM.00387-08.

4. O’Toole G, Kaplan HB, Kolter R. 2000. Biofilm formation as microbialdevelopment. Annu Rev Microbiol 54:49 –79. http://dx.doi.org/10.1146/annurev.micro.54.1.49.

5. Costerton JW, Lewandowski Z, Caldwell DE, Korber DR, Lappin-ScottHM. 1995. Microbial biofilms. Annu Rev Microbiol 49:711–745. http://dx.doi.org/10.1146/annurev.mi.49.100195.003431.

6. Park N, Kwon B, Kim IS, Cho J. 2005. Biofouling potential of various NFmembranes with respect to bacteria and their soluble microbial products(SMP): characterizations, flux decline, and transport parameters. J Mem-brane Sci 258:43–54. http://dx.doi.org/10.1016/j.memsci.2005.02.025.

7. Hall-Stoodley L, Costerton JW, Stoodley P. 2004. Bacterial biofilms:

from the natural environment to infectious diseases. Nat Rev Microbiol2:95–108. http://dx.doi.org/10.1038/nrmicro821.

8. Miura Y, Watanabe Y, Okabe S. 2007. Membrane biofouling in pilot-scale membrane bioreactors (MBRs) treating municipal wastewater: im-pact of biofilm formation. Environ Sci Technol 41:632– 638. http://dx.doi.org/10.1021/es0615371.

9. Huang LN, De Wever H, Diels L. 2008. Diverse and distinct bacterialcommunities induced biofilm fouling in membrane bioreactors operatedunder different conditions. Environ Sci Technol 42:8360 – 8366. http://dx.doi.org/10.1021/es801283q.

10. Herzberg M, Kang S, Elimelech M. 2009. Role of extracellular polymericsubstances (EPS) in biofouling of reverse osmosis membranes. EnvironSci Technol 43:4393– 4398. http://dx.doi.org/10.1021/es900087j.

11. Herzberg M, Elimelech M. 2007. Biofouling of reverse osmosis mem-branes: role of biofilm-enhanced osmotic pressure. J Membrane Sci 295:11–20. http://dx.doi.org/10.1016/j.memsci.2007.02.024.

12. Wingender J, Neu T, Flemming H-C. 1999. What are bacterial extracel-lular polymeric substances? p 1–19. In Wingender J, Neu T, FlemmingH-C (ed), Microbial extracellular polymeric substances. Springer, Berlin,Germany.

13. Ramesh A, Lee DJ, Wang ML, Hsu JP, Juang RS, Hwang KJ, Liu JC,Tseng SJ. 2006. Biofouling in membrane bioreactor. Sep Sci Technol41:1345–1370. http://dx.doi.org/10.1080/01496390600633782.

14. Ng HY, Tan TW, Ong SL. 2006. Membrane fouling of submerged mem-brane bioreactors: impact of mean cell residence time and the contribut-ing factors. Environ Sci Technol 40:2706 –2713. http://dx.doi.org/10.1021/es0516155.

15. Bereschenko LA, Prummel H, Euverink GJ, Stams AJ, van LoosdrechtMC. 2011. Effect of conventional chemical treatment on the microbialpopulation in a biofouling layer of reverse osmosis systems. Water Res45:405– 416. http://dx.doi.org/10.1016/j.watres.2010.07.058.

16. Barraud N, Hassett DJ, Hwang SH, Rice SA, Kjelleberg S, Webb JS.2006. Involvement of nitric oxide in biofilm dispersal of Pseudomonasaeruginosa. J Bacteriol 188:7344 –7353. http://dx.doi.org/10.1128/JB.00779-06.

17. Barraud N, Schleheck D, Klebensberger J, Webb JS, Hassett DJ, Rice SA,Kjelleberg S. 2009. Nitric oxide signaling in Pseudomonas aeruginosa bio-films mediates phosphodiesterase activity, decreased cyclic di-GMP levels,and enhanced dispersal. J Bacteriol 191:7333–7342. http://dx.doi.org/10.1128/JB.00975-09.

18. Barraud N, Storey MV, Moore ZP, Webb JS, Rice SA, Kjelleberg S.2009. Nitric oxide-mediated dispersal in single- and multi-species bio-films of clinically and industrially relevant microorganisms. Microb Bio-technol 2:370 –378. http://dx.doi.org/10.1111/j.1751-7915.2009.00098.x.

19. Barnes RJ, Bandi RR, Wong WS, Barraud N, McDougald D, Fane A,Kjelleberg S, Rice SA. 2013. Optimal dosing regimen of nitric oxide donorcompounds for the reduction of Pseudomonas aeruginosa biofilm and iso-lates from wastewater membranes. Biofouling 29:203–212. http://dx.doi.org/10.1080/08927014.2012.760069.

20. Saavedra JE, Southan GJ, Davies KM, Lundell A, Markou C, HansonSR, Adrie C, Hurford WE, Zapol WM, Keefer LK. 1996. Localizingantithrombotic and vasodilatory activity with a novel, ultrafast nitricoxide donor. J Med Chem 39:4361– 4365. http://dx.doi.org/10.1021/jm960616s.

21. Schmidt I, Steenbakkers PJ, op den Camp HJ, Schmidt K, Jetten MS.2004. Physiologic and proteomic evidence for a role of nitric oxide inbiofilm formation by Nitrosomonas europaea and other ammonia oxidiz-ers. J Bacteriol 186:2781–2788. http://dx.doi.org/10.1128/JB.186.9.2781-2788.2004.

22. Arruebarrena Di Palma A, Pereyra CM, Moreno Ramirez L, XiquiVazquez ML, Baca BE, Pereyra MA, Lamattina L, Creus CM. 2013.Denitrification-derived nitric oxide modulates biofilm formation in Azos-pirillum brasilense. FEMS Microbiol Lett 338:77– 85. http://dx.doi.org/10.1111/1574-6968.12030.

23. Plate L, Marletta MA. 2012. Nitric oxide modulates bacterial biofilmformation through a multicomponent cyclic-di-GMP signaling network.Mol Cell 46:449 – 460. http://dx.doi.org/10.1016/j.molcel.2012.03.023.

24. Zaitseva J, Granik V, Belik A, Koksharova O, Khmel I. 2009. Effect ofnitrofurans and NO generators on biofilm formation by Pseudomonasaeruginosa PAO1 and Burkholderia cenocepacia 370. Res Microbiol 160:353–357. http://dx.doi.org/10.1016/j.resmic.2009.04.007.

25. Falsetta ML, McEwan AG, Jennings MP, Apicella MA. 2010. Anaerobicmetabolism occurs in the substratum of gonococcal biofilms and may be

Nitric Oxide for the Control of RO Membrane Biofouling

April 2015 Volume 81 Number 7 aem.asm.org 2523Applied and Environmental Microbiology

on July 30, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

sustained in part by nitric oxide. Infect Immun 78:2320 –2328. http://dx.doi.org/10.1128/IAI.01312-09.

26. Ager D, Evans S, Li H, Lilley AK, van der Gast CJ. 2010. Anthropogenicdisturbance affects the structure of bacterial communities. Environ Microbiol12:670–678. http://dx.doi.org/10.1111/j.1462-2920.2009.02107.x.

27. Baker GC, Smith JJ, Cowan DA. 2003. Review and re-analysis of domain-specific 16S primers. J Microbiol Methods 55:541–555. http://dx.doi.org/10.1016/j.mimet.2003.08.009.

28. Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD,Costello EK, Fierer N, Pena AG, Goodrich JK, Gordon JI, Huttley GA,Kelley ST, Knights D, Koenig JE, Ley RE, Lozupone CA, McDonald D,Muegge BD, Pirrung M, Reeder J, Sevinsky JR, Turnbaugh PJ, WaltersWA, Widmann J, Yatsunenko T, Zaneveld J, Knight R. 2010. QIIMEallows analysis of high-throughput community sequencing data. NatMethods 7:335–336. http://dx.doi.org/10.1038/nmeth.f.303.

29. Edgar RC. 2010. Search and clustering orders of magnitude fasterthan BLAST. Bioinformatics 26:2460 –2461. http://dx.doi.org/10.1093/bioinformatics/btq461.

30. Edgar RC, Haas BJ, Clemente JC, Quince C, Knight R. 2011. UCHIMEimproves sensitivity and speed of chimera detection. Bioinformatics 27:2194 –2200. http://dx.doi.org/10.1093/bioinformatics/btr381.

31. DeSantis TZ, Hugenholtz P, Larsen N, Rojas M, Brodie EL, Keller K,Huber T, Dalevi D, Hu P, Andersen GL. 2006. Greengenes, a chimera-checked 16S rRNA gene database and workbench compatible with ARB.Appl Environ Microbiol 72:5069 –5072. http://dx.doi.org/10.1128/AEM.03006-05.

32. Caporaso JG, Bittinger K, Bushman FD, DeSantis TZ, Andersen GL,Knight R. 2010. PyNAST: a flexible tool for aligning sequences to a tem-plate alignment. Bioinformatics 26:266 –267. http://dx.doi.org/10.1093/bioinformatics/btp636.

33. Clarke KR, Gorley RN. 2006. PRIMER v6: user manual/tutorial.PRIMER-E, Plymouth, Ivybridge, United Kingdom.

34. Barnes RJ, Bandi RR, Chua F, Low JH, Aung T, Barraud N, Fane AG,Kjelleberg S, Rice SA. The roles of Pseudomonas aeruginosa extracellularpolysaccharides in biofouling of reverse osmosis membranes and nitricoxide induced dispersal. J Membrane Sci 466:161–172. http://dx.doi.org/10.1016/j.memsci.2014.04.046.

35. Chen CL, Liu WT, Chong ML, Wong MT, Ong SL, Seah H, Ng WJ.2004. Community structure of microbial biofilms associated with mem-brane-based water purification processes as revealed using a polyphasicapproach. Appl Microbiol Biotechnol 63:466 – 473. http://dx.doi.org/10.1007/s00253-003-1286-7.

36. Suwarno SR, Chen X, Chong TH, Puspitasari VL, McDougald D,Cohen Y, Rice SA, Fane AG. 2012. The impact of flux and spacers onbiofilm development on reverse osmosis membranes. J Membrane Sci405– 406:219 –232. http://dx.doi.org/10.1016/j.memsci.2012.03.012.

37. Chong TH, Wong FS, Fane AG. 2008. The effect of imposed flux onbiofouling in reverse osmosis: role of concentration polarisation and bio-film enhanced osmotic pressure phenomena. J Membrane Sci 325:840 –850. http://dx.doi.org/10.1016/j.memsci.2008.09.011.

38. Miles AA, Misra SS, Irwin JO. 1938. The estimation of the bactericidalpower of the blood. J Hyg 38:732–749. http://dx.doi.org/10.1017/S002217240001158X.

39. Daniels L, Hanson RS, Phillips JA. 1994. Chemical analysis, p 512–554.In Gerhardt P, Murray RGE, Wood WA, Krieg NR (ed), Methods forgeneral and molecular bacteriology. American Society for Microbiology,Washington, DC.

40. Roy AB, Petrova OE, Sauer K. 2012. The phosphodiesterase DipA(PA5017) is essential for Pseudomonas aeruginosa biofilm dispersion. JBacteriol 194:2904 –2915. http://dx.doi.org/10.1128/JB.05346-11.

41. Schlag S, Nerz C, Birkenstock TA, Altenberend F, Gotz F. 2007. Inhi-bition of staphylococcal biofilm formation by nitrite. J Bacteriol 189:7911–7919. http://dx.doi.org/10.1128/JB.00598-07.

42. Keefer LK, Nims RW, Davies KM, Wink DA. 1996. “NONOates” (1-substituted diazen-1-ium-1,2-diolates) as nitric oxide donors: convenientnitric oxide dosage forms, p 281–293. In Lester P (ed), Methods in enzy-mology, vol 268. Academic Press, Waltham, MA.

43. Hrabie JA, Klose JR, Wink DA, Keefer LK. 1993. New nitric oxide-releasingzwitterions derived from polyamines. J Org Chem 58:1472–1476.

44. Yeon KM, Cheong WS, Oh HS, Lee WN, Hwang BK, Lee CH, BeyenalH, Lewandowski Z. 2009. Quorum sensing: a new biofouling controlparadigm in a membrane bioreactor for advanced wastewater treatment.Environ Sci Technol 43:380 –385. http://dx.doi.org/10.1021/es8019275.

45. Bereschenko LA, Stams AJ, Euverink GJ, van Loosdrecht MC. 2010.Biofilm formation on reverse osmosis membranes is initiated and domi-nated by Sphingomonas spp. Appl Environ Microbiol 76:2623–2632. http://dx.doi.org/10.1128/AEM.01998-09.

46. Pang CM, Liu WT. 2007. Community structure analysis of reverse osmo-sis membrane biofilms and the significance of Rhizobiales bacteria in bio-fouling. Environ Sci Technol 41:4728 – 4734. http://dx.doi.org/10.1021/es0701614.

47. Al Ashhab A, Herzberg M, Gillor O. 2014. Biofouling of reverse-osmosismembranes during tertiary wastewater desalination: microbial commu-nity composition. Water Res 50:341–349.

48. Ayache C, Manes C, Pidou M, Croue JP, Gernjak W. 2013. Microbialcommunity analysis of fouled reverse osmosis membranes used in waterrecycling. Water Res 47:3291–3299. http://dx.doi.org/10.1016/j.watres.2013.03.006.

49. Chiellini C, Iannelli R, Modeo L, Bianchi V, Petroni G. 2012. Biofoulingof reverse osmosis membranes used in river water purification for drink-ing purposes: analysis of microbial populations. Biofouling 28:969 –984.http://dx.doi.org/10.1080/08927014.2012.724679.

Barnes et al.

2524 aem.asm.org April 2015 Volume 81 Number 7Applied and Environmental Microbiology

on July 30, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from