Recycled desalination membranes as a support material for biofilm development: A new approach for microcystin removal during water treatment

Science of the Total Environment 647 (2019) 785–793

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Recycled desalination membranes as a support material for biofilm development: A new approach for microcystin removal during water treatment Jesús Morón-López a,c, Lucía Nieto-Reyes a, Jorge Senán-Salinas a, Serena Molina a, Rehab El-Shehawy a,b,⁎ a b c

IMDEA Water Institute, Spain Department of Environmental Sciences and Analytical Chemistry, Stockholm University, Sweden Chemical Engineering Department, University of Alcalá, Madrid, Spain

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• Discarded RO membranes could be used as a support material for bacterial attachment. • The characteristics of discarded RO membranes promoted the biofilm development. • MC-degrading biofilm on discarded RO membranes removed 2 mg·L−1 of MC in 24 h. • The hereby developed R-MBfR proof-ofconcept proved its economic feasibility.

a r t i c l e

i n f o

Article history: Received 10 May 2018 Received in revised form 30 July 2018 Accepted 30 July 2018 Available online 01 August 2018 Editor: Zhen (Jason) He Keywords: Microcystin Degradation Biofilm Membrane Recycling Reactor

a b s t r a c t Increased harmful cyanobacterial blooms and drought are some negative impacts of global warming. To deal with cyanotoxin release during water treatment, and to manage the massive quantities of end-of-life membrane waste generated by desalination processes, we propose an innovative biological system developed from recycled reverse osmosis (RO) membranes to remove microcystins (MC). Our system, named the Recycled-Membrane Biofilm Reactor (R-MBfR), effectively removes microcystins, while reducing the pollution impact of RO membrane waste by prolonging their life span at the same time. This multidisciplinary work showed that the inherent flaw of RO membranes, i.e., fouling, can be considered an advantageous characteristic for biofilm attachment. Factors such as roughness, hydrophilic surfaces, and the role of calcium in cell-cell and cell-surface interactions, encouraged bacterial growth on discarded membranes. Biofilm development was stimulated by using a laboratoryscale membrane module simulator cell. The R-MBfR proved versatile and was capable of degrading 2 mg·L−1 of MC in 24 h. The economic feasibility of the scaling-up of the hypothetical R-MBfR was also validated. Therefore, this membrane recycling could be a future green cost-effective alternative technology for MC removal. © 2018 Elsevier B.V. All rights reserved.

1. Introduction ⁎ Corresponding author at: Department of Environmental Sciences and Analytical Chemistry, Stockholm University, Sweden. E-mail addresses: [email protected] (J. Morón-López), [email protected] (J. Senán-Salinas), [email protected] (S. Molina), [email protected] (R. El-Shehawy).

https://doi.org/10.1016/j.scitotenv.2018.07.435 0048-9697/© 2018 Elsevier B.V. All rights reserved.

Cyanobacterial mass proliferation as a harmful algal bloom (cyanoHABs) is an increasing phenomenon in eutrophic water bodies worldwide, due mainly to climate change (e.g. global warming and persistent drought) and eutrophication. To date, several types of toxins

786

J. Morón-López et al. / Science of the Total Environment 647 (2019) 785–793

produced by cyanobacteria have been identified. Microcystins (hereafter MC) are among the most abundant and widely spread cyanotoxins worldwide. MC comprise a structurally diverse group of potent hepatotoxins with a cyclic heptapeptide structure that are produced by cyanobacteria as intracellular metabolites, released mainly by cell lysis. Under natural conditions, MC are relatively stable compounds when exposed to sunlight and a wide range of temperatures and pH (Wörmer et al., 2010). The half-life of MC-LR by photosensitized degradation in natural systems was estimated to be approximately 90–120 days per meter of water (Welker and Steinberg, 2000). As MC persist for long periods in the natural environment, developing mechanisms to avoid toxins reaching water supplies is mandatory. Nevertheless, most drinking water utilities still rely on conventional treatment processes that lack adequate optimized MC removal processes. To achieve drinking water quality standards, drinking water treatment plants (hereafter DWTPs) usually eliminate MC by a variety of physicochemical methods (Merel et al., 2010). However, these processes are costly and may produce harmful by-products. Biological filtration has also been used in drinking water treatment to remove cyanotoxins. Nonetheless, biological MC degradation is considered a slow process compared to chemical treatments and its costeffectiveness greatly depends on the type of support used (sand, activated carbon, anthracite, etc.) (Eleuterio and Batista, 2010; Ho et al., 2012a). Desalination of sea and brackish water by reverse osmosis (hereafter RO) membrane technology is broadly implemented worldwide given its low cost and easy handling. However, N840,000 end-of-life (hereafter EoL) membranes modules (N14,000 Tn/year) are discarded every year worldwide (Landaburu-Aguirre et al., 2016). There is global concern about managing these wastes because current management options for EoL-RO membrane modules are, unfortunately, landfilling and incineration, which pose relevant environmental risks. In general, EoL-RO membrane modules are formed by different polymers, such as epoxy resin, polypropylene, polyester, polyamide, polysulfone, etc. All these polymers are high-performance materials excellent high mechanical and chemical durability. Therefore, when these modules are disposed in landfills, they persist for prolonged periods of time, which aggravates the environmental problem. Of the different membrane types, polyamide thin film composite (PA-TFC) membranes currently account for over 95% of existing RO desalination plants (Geise et al., 2010). For this reason, PA-TFC RO membranes were studied herein. These membranes are formed by three layers: a non-woven polyester support, an asymmetric porous polysulfone (PSF) interlayer and a polyamide (PA) ultra-thin layer. Emerging membrane technologies, such as the membrane biofilm reactor (MBfR), have received attention in the last 20 years (Nerenberg, 2016). Unlike conventional filtration membranes, MBfRs do not separate solids from liquids. Instead biofilms form on the outer membrane surface and are able to remove pollutants, among several other reactions (Kinh et al., 2017; Martin and Nerenberg, 2012). The aim of the present research is to provide a new concept for biological MC removal during drinking water treatment. To this end, we reused the EoL RO membranes discarded by desalination plants as supports for biofilm immobilization that is capable of selectively removing MC. This approach is based on the most innovative biofilm reactor developments, and we provide the new development of reactors made from recycled materials (Halan et al., 2012). We call this pioneering concept the Recycled-Membrane Biofilm Reactor (hereafter R-MBfR) (Fig. 1). We studied the characteristics of two different membrane models and the influence of their previous fouling for bacterial association as an active biofilm. These biologically-active membranes were tested to remove MC by passing MC-polluted water. Finally, this hypothetical system was also economically detailed and analyzed in relation to current physico-chemical methods to provide a first understanding of its feasibility.

2. Materials and methods 2.1. MC-degrading bacteria The MC-degrading bacterium selected for this study was Sphingopyxis sp. strain IM-1, which possesses a specific cluster of genes for MC degradation (Lezcano et al., 2016).

2.2. Membranes Discarded RO membranes were used as the support material to grow MC-degrading strain IM-1. The membrane samples (coupons) studied herein came from EoL (PA-TCF) RO membranes (spirally wound modules) after treating; a) brackish water (hereafter BW discarded or BWd), TM720-400 (Toray), and b) seawater (hereafter SW discarded or SWd), HSWC3 (Hydranautics). In addition, new commercial membranes were used as a control: TM720-400 (Toray) for the BW type (hereafter BWc) and SW30HRLE-440i (DowFilmtec) for the SW type (hereafter SWc). It is noteworthy that we used the SW30HRLE-440i model as a control because the HSWC3 model is currently unlisted. Despite the different brands and models, according to the manufacturer's information both membranes have identical properties. All the membranes were conserved in Milli-Q water prior to use.

2.3. Extraction of microcystins MC were extracted from fresh cyanobacterial scum using methanol 100%, followed by a 15-minute sonication in an ultrasonication bath (P-Selecta Ultrasons), and were stored at 4 °C for 1 h to extract MC. The extract was then centrifuged (Heraeus Megafuge 16) at 4000 ×g for 15 min, and the supernatant was stored at −20 °C. When the pellet lost most of its chlorophyll (after repeating the above step 3 times), the extract was vacuum-dried at 40 °C in a rotavapor (Rotavapor-R, Büchi, Switzerland), resuspended in 10% methanol and bonded to 5 g C18 cartridges (Extrabond C18, 5 g, 20 mL, Sharlab) for partial MC purification. First, the cartridge was activated with 100% methanol, followed by MilliQ water and subsequent conditioning with 10% methanol. Afterward samples were added and washed with Milli-Q water and 30% methanol. MC were eluted in 20 mL of 90% methanol and dried in a SpeedVac concentrator (Savant, SPD131DDA) at 45 °C and 0.2 Torr for 2 h. Finally, MC were resuspended in Milli-Q water, passed through a sterile syringe with 0.22 μm filters (25 mm, Pall Corporation) for sterilization and stored at −20 °C. The MC solution was composed of 84.5% MC-LR, 9.86% MC-RR and 5.64% MC-YR.

2.4. Initial fouling characterization To determine the percentage of organic and inorganic fouling, a thermogravimetric analysis (TGA) of membrane fouling from the RO discarded membranes was carried out. The membrane fouling TGA data were recorded in a TGA Q500 analyzer in an oxidative (air) atmosphere at a heating rate of 5 °C/min from 45 °C to 800 °C. To complement this analysis, SEM images were taken in the discarded membranes, according to Section 2.9.

2.5. Membrane surfaces characterization 2.5.1. Zeta potential measurements The surface Zeta potential was measured via electrophoretic light scattering in a Zetasizer Nano ZS (Zen 1020). Measurements were taken at 25 °C in 10 mM KCl and pH 7.0 using an aqueous solution with of 0.5% (w/w) poly (acrylic acid) (450 kDa) as a tracer. pH was adjusted using 1 M KOH and 1 M HCl (Santiago-Morales et al., 2016).

J. Morón-López et al. / Science of the Total Environment 647 (2019) 785–793

787

Fig. 1. Diagrammatic representation of the Recycled-Membrane Biofilm Reactor (R-MBfR) concept for MC removal. Letters indicate different membrane layers: A, B and D represent (PATFC) RO membranes; C represents the feed spacers between (PA-TFC) membrane layers. The MC-degrading biofilm grow over PA active layers (A, B and D), while gap A–B correspond to the inactive intermembrane cavity, which will not be used in the current proposal. Arrows mean the flux of water polluted by MC.

2.5.2. Contact angle measurement The surface wettability tests were carried out using a static contact angle meter by the sessile drop technique with a KSV CAM200 instrument (KSV Instruments, USA). After fixing a membrane sample on a glass support, 4.5 mL of Milli-Q water were placed on the membrane surface with a Hamilton syringe at room temperature. Ten measurements from two different pieces of the same dried membrane were taken to know the average value (Molina et al., 2015). 2.5.3. Atomic-force microscopy (AFM) The atomic force microscopy (AFM) experiments were performed in the tapping mode with a Multimode AFM (Vecco Instruments, Santa Barbara, CA, USA) equipped with a Nanoscope Iva control system (software version 6.14r1). Silicon tapping probes (RTESP, Veeco) were used with a resonance frequency of ∼300 kHz and at a scan rate of 0.5 Hz, 5 × 5 and 2 × 2 μm2. The AFM images were taken for each sample. Topography was examined by topographical AFM (Molina et al., 2018). 2.6. Analysis of the biofilm generating capacity Biofilm formation on membranes followed three main steps: a) enhancing bacterial aggregation (Section 2.6.1); b) bacterial association on different membranes and initial biofilm formation (Section 2.6.2); c) biofilm development on the selected membrane by the R-MBfR simulator cell (Section 2.6.3). 2.6.1. Induction of bacterial aggregation In order to promote the cell-cell interaction of strain IM-1, quorumsensing molecules and divalent cations were added separately to the R2A liquid medium, and biofilm formation was encouraged according

to Merritt et al. (2011). OD was measured at 595 nm in a 96-well plate reader (Multiskan FC, Thermo Scientific, Shanghai, China). The experiment was carried out in triplicate. 2.6.2. Bacterial association and initial biofilm formation The bacterial association with membrane surfaces and the initial biofilm formation were studied on membranes BWd, SWd, BWc and SWc. All the membranes were previously washed with 50% ethanol and taped to the pre-autoclaved methacrylate carriers, with the PA layer outwardly exposed. Sterile membranes were held perpendicularly inside a continuous culture of Sphingopyxis sp. strain IM-1. The continuous culture was incubated at room temperature, bubbled by an aquarium air pump and constantly fed the R2A medium enriched with 0.01 M CaCl2 at a rate of 1.4 mL·min−1. Culture density was maintained at an OD600 value of 0.4. Samples were collected at 0, 24, 48, 72, 96, 120, 144 h. The highest coverage was achieved after 96 h (data not shown). Hence all the subsequent experiments were incubated until this time. Negative controls were also included. 2.6.3. Biofilm development in the simulator cell (R-MBfR simulator) Once the time needed for bacterial association and initial biofilm formation on membranes was known, as well as the fittest membranes for biofilm development, we tested the proof-of-concept of the hypothetical large-scale R-MBfR. To simulate it on the laboratory scale, we created a simulator cell based on previous designs (Villacorte, 2017). Fig. S1 shows a schematic diagram of the experimental setup used to reproduce biofilm formation in the R-MBfR simulation cell. The size of the employed discarded RO flat sheet membrane was 9 cm2 and the internal volume of the cell was 3 mL. Hence, the effective area of the simulator cell was 5 × 10−4 m2. MC-degrading bacteria Sphingopyxis sp. strain IM-1 was grown by a continuous culture flowed tangentially at a rate of 1.4 mL·min−1

788

J. Morón-López et al. / Science of the Total Environment 647 (2019) 785–793

through the simulator cell. This cell contained a flat sheet of the SWd RO membrane and followed the same culturing condition as explained above (Section 2.6.2).

Hence we assumed the slope of the MC degradation constant in that experiment. 2.11. Statistical analysis

2.7. Confocal laser scanning microscopy (CLSM) The confocal biofilm images were obtained under a confocal laser scanning microscope (CLSM Leica SP5, Leica Microsystems). Next the 5 mm × 5 mm membrane area was stained with the Live/Dead BacLight Bacterial Viability Kit (Molecular Probes™) according to the manufacturer's instructions. For each membrane sample, 10 different areas were observed by CLSM.

Statistical analyses were conducted using the Statistical Package for Social Sciences, v.17 (SPSS, Inc.). Shapiro-Wilk and Levene tests were performed to determine the normality and homogeneity of variances, respectively. One-way analysis of variance (ANOVA) and Post Hoc Turkey HSD analyses were run to determine the significant differences between bacterial colonization rates and zeta potentials. For the variables of nonhomogeneous variances, hydrophobicity and roughness data, one-way ANOVA and Tamhane post hoc analyses were performed.

2.8. Biomass quantification 2.12. Economical assessment The bacterial coverage percent (%) on the membrane surface was analyzed by the ImageJ software. Biomass (μm3·μm−2), average thickness (μm2), maximum thickness (μm) and roughness coefficient (Ra) were analyzed by ImageJ with COMSTAT plugin (Version 2.1). The 3D images of the biomass were constructed using the Volocity® Demo software (Version 6.3, PerkinElmer). 2.9. Scanning electron microscopy (SEM) A scan electron microscope (Zeiss DSM 950, Germany) was used to examine both previous fouling and the biofilm morphology on the membrane surface. Prior to the SEM analysis, the membrane area was washed with a 0.01 M PBS to remove unbound cells, fixed with 2.5% (v/v) glutaraldehyde at 4 °C for at least 30 min and rinsed in 0.01 M PBS. Subsequently, membranes were dehydrated by increasing ethanol concentrations (30, 50, 70, 96 and 100%). Then the dehydrated membranes were dried by the critical point dryer (FISON, CPD7501) and further coated by cathodic spray with gold (Polaron, SC7640) for the SEM observations (Mogana and Ramasamy, 2016). 2.10. MC degradation analysis After generating biofilms on membranes (Section 2.6.2 for the initial biofilm formation and Section 2.6.3 for the biofilm developed by the simulator cell), the MC-degradation tests were run in minimal salt medium (hereafter MSM) enriched with the MC mixture as follows: a) the initial biofilm formation on membranes was tested in batches for MCdegradation ability in flasks at 27 °C with air bubbling; b) the degradation ability of the biofilm developed by the simulator cell was verified by the continuous tangential flow of the MC mixture at 1.4 mL·min−1 and recirculating by an external flask at the 71.42 min·cycle−1 rate at room temperature (Fig. S2). In both experiments, a pre-incubation time with 1 mg·L−1 of the MC mixture was required to acclimate the MCdegraders. Afterward, a pulse of 1 mg·L−1 of the MC mixture was added. Two technical replicates were collected for MC quantification (MC-LR, -RR and -YR) at 0, 3, 6, 9, 24 h, and were filtered immediately through 0.45 μm sterile filters (Econofltr PTFE 13 mm, Agilent Technologies) and stored at −20 °C until analyzed. The MC concentration in the collected samples was quantified according to Morón-López et al. (2017) in an HPLC-MS-TOF (Agilent 6230 accurate mass TOF Agilent Technologies, Santa Clara, CA, USA). The MC degradation kinetics of the biofilms formed on the discarded RO membranes (Section 2.6.2) were calculated following the first-order model described by Xu et al. (2011) and the MC degradation kinetics proposed by Li et al. (2014). Given the lag phase during preincubation, k constants were determined by taking into account the first point at which biodegradation began (Ho et al., 2012b). In contrast, when the MC degradation experiment was performed in the simulator cell, it matched a linear trend, but not with the above model. This could be due to the continuous recirculation at 1.4 mL·min−1, which could mask the natural degradation kinetics.

The first insight into the R-MBfR technology potential was estimated by cost criteria. According to this estimation, technological aspects of scaling-up, such as the effectiveness of the potential modules and required auxiliary components, were taken into account. Next an economical assessment was made by calculating the cost effectiveness of the Capital Expenditure (CAPEX) and Operational Expenditure (OPEX). Hypothetical R-MBfR effectiveness was estimated by transferring our laboratory simulator cell data (the system's flow, volume and MC degradation rate) to real scale membrane modules. We extrapolated our laboratory scale membrane flat sheet efficiencies to a real-life membrane module of a 37 m2 membrane size and with a 50 L internal volume (detailed explanation in the Supplementary material, Table S1). A real-life membrane module with the attached MC-degrading biofilm was considered an R-MBfR module unit (Fig. 1). As the simulator cell worked as a recirculation system, uncertainty about the MC degradation location (inside or outside the simulator cell) was also considered. To address this issue, each possible specific MC degradation rate of the simulator cell (degrading between 0% and 100% of MC inside the cell) was calculated (Table S2 and Fig. S3). To simplify the civil work calculations and to focus on a system for a small population, we considered a modular container with four water lines. As we proposed a scenario based on scale-out, cost-effectiveness remained constant to larger populations. We hypothesized a scaled-up R-MBfR system which, by each water line, would include (Fig. S4): a) a system of eight R-MBfR units; b) four main pipelines (two for flow-in and two for flow-out); c) two microfiltration units to remove the bacterial cells placed both upstream and downstream of the RMBfR system; d) a recirculation tank. This design was used to calculate the flow design, and to approach the initial capital investment and operational cost associated with our system. These calculations were also compared to other studies. The initial capital investment and operational costs were estimated by Capital Expenditure (CAPEX) (Table S3) and Operational Expenditure (OPEX) (Table S4), respectively. These calculations were based on the hypothetical scaling-up estimation and European Market components prices. In order to compare our estimation to other technologies, we included some important aspects from Mulder et al. (2015), such as the annual discount rate (4%) for the calculations related to Annual Equivalent Cost (AEC) and Unitary costs (Formulas S1, S2 and S3). Detailed calculations and considerations are provided in more detail in the Supplementary Material. We compared our economical estimations to other technologies already implemented for MC treatment (Alvarez et al., 2010; Rodríguez et al., 2007; US EPA Office of Water, 2016). To complete insufficient published data, we also included information about micropollutant treatments (hereafter MP) (Bui et al., 2016; Mulder et al., 2015) by adapting all the costs and doses for 1 mg·L−1 of MC in incoming water, and removal efficiencies of at least 95%. We assumed a direct and proportional relation between MP and MC (Table S5). Uncertainties about our technology effectiveness were also evaluated by a sensitivity

J. Morón-López et al. / Science of the Total Environment 647 (2019) 785–793

789

Table 1 Membrane surface characterization presenting data on the fouling type, Zeta potential and roughness of the membranes tested in this study. Different letters in each columns indicate significant differences between membranes at p b 0.05 after the one-way ANOVA. Type

Fouling Organic (%)

BWc BWd SWc SWd a

– 17.28 – 66.27a

Zeta potential (mV)

Contact angle (°)

Inorganic (%) – 82.72 – 33.73a

Roughness Ra (nm)

A

−33.66 ± 4.62 −31.68 ± 4.89A −31.08 ± 5.04A −30.53 ± 3.40A

B

61.29 ± 1.01 50.74 ± 2.96C 66.78 ± 2.14A 29.48 ± 1.54D

Rq (nm) B

51.30 ± 6.71 32.01 ± 4.77C 69.77 ± 5.66A 88.75 ± 28.89A

62.28 ± 3.66B 42.69 ± 5.26C 97.23 ± 14.81A 98.71 ± 30.03A

These results have been reported in Molina et al. (2018).

assessment (Fig. S3) to observe the competitive limits of our technology and to define future objectives. 3. Results and discussion The human impact on the environment through climate change and water eutrophication has led the cyanoHABs phenomenon to become a recurrent worrying issue. New strategies and tools must be developed to mitigate rising cyanotoxin production. Of all the different existing management models, water treatment by biological activity has been widely recognized as a potential enviro-friendly alternative for cyanotoxin removal in the water industry (Li et al., 2017). Several support materials have been described to attach these active bacteria and to conduct treatment. Indeed the reuse of EoL membranes modules could be an interesting substitute for conventional carriers given the inherent characteristic of RO membrane technology to suffer fouling. Moreover, current concern about the management of discarded membrane modules makes their recycling even more appealing as it prolongs their life span and reduces their environmental impact as waste. Among the diversity of the bacteria described as MC-degraders (Lezcano et al., 2017; Li et al., 2017; Mou et al., 2013), we selected high-efficient aerobic MC-degrading bacterium Sphingopyxis sp. strain IM-1. This strain possesses the mlr A-D gene cluster, which encodes the MC degradation pathway (Bourne et al., 2001). The fist enzyme in the pathway cleaves the Adda-Arg peptide bond from cyclic MC molecule and produces a linear molecule that is 160-fold less toxic (Bourne

et al., 1996). The subsequent steps degrade the linearized product to smaller peptides and amino acids (Dziga et al., 2016). This bacterium was grown as a biofilm on BWd and SWd RO membranes. In order to understand the possible factors that may affect biofilm formation on these membranes, we analyzed membrane surface characteristics in detail. This is also important for the system to be replicated: a) under other laboratory conditions; b) with other MC-degrading strains; c) to remove other types of organic water contaminants. 3.1. Cell–cell interaction In biofilms, the biomass attachment to the support material depends on cell-cell and cell-substrate interactions (Donlan, 2002). To enhance the cell-cell interaction of strain IM-1, experiments involving various biofilm promoters were performed, as previously described in the literature (He et al., 2016; Monzon et al., 2016). Addition of 0.01 M of Ca2+ greatly enhanced the aggregation capacity of strain IM-1 (Fig. S5). Our results agree with those of previous studies (He et al., 2016; Mangwani et al., 2014), in which adding divalent cations promoted biofilm formation by stimulating EPS induction. 3.2. Cell–substrate interaction To evaluate cell-membrane interactions, the discarded membranes were compared with the new commercial membranes. The frequently considered requirements for bacterial deposition and biofilm formation

Fig. 2. AFM visualization of membrane roughness of (a) BWc; (b) BWd; (c) SWc and (d) SWd.

790

J. Morón-López et al. / Science of the Total Environment 647 (2019) 785–793

Bacterial coverage (percentage of the total area)

1

3 a

20% 18% 16% 14% 12% 10% 8% 6% 4% 2% 0%

2 BW model

SW model

A B B, C C

New commercial RO

Discarded RO

b

c

d

Fig. 3. Bacterial colonies and biofilm on different membrane surfaces after 96-hour growth: (1) Percentage of bacterial coverage. Different letters indicate significant differences between bacterial coverage on the membrane at p b 0.05 after the one-way ANOVA; (2) SEM visualization of strain IM-1 associated with the membrane surfaces of (a) BWd and (b) SWd. Bar shows 5 μm. (3) CLSM image after staining membranes with a life/dead bacterial kit. (a) BWc, (b) SWc, (c) BWd, (d) SWd. Bar shows 25 μm.

3.3. MC degradation capability The MC degradation assays were run to study if strain IM-1 maintained its ability to remove MC when attached to these membranes. Fig. 4 illustrates how the initially formed biofilm on membranes BWd and SWd completely removed the added MC concentration, while its MC degradation rate improved after the pulse. These results agree with previous reports, where prior exposure to MC stimulated

250%

Residual percentage of MC

were hydrophobicity, charge and roughness (Herzberg et al., 2009; Li et al., 2016; Sehar and Naz, 2016; Vu et al., 2009). However, these surface properties can be modified by organic and inorganic matter deposition during the life span of membranes during water treatment in desalination plants, and this deposition process can even promote it (Chen et al., 2013; Halan et al., 2012; Subramani et al., 2009). The presence of Ca2+ has also been demonstrated to enhance bacterial deposition by nutrient absorption onto the membrane surface. For all these reasons, the roles of previous fouling and Ca2+ in bacterial association and initial biofilm formation on membranes were taken into account as key conditioning factors. The discarded membranes, which previously had treated brackish water (BWd type) and seawater (SWd type), were analyzed to study the influence of fouling type on bacterial association. As expected, the presence/absence of fouling, plus its nature, on BWd and SWd significantly affected surface characteristics (Table 1, Fig. 2). BWd contained mainly inorganic fouling with a low hydrophobicity and roughness, while SWd carried organic fouling, hydrophilic surfaces and higher roughness. The differences in fouling compositions were also evidenced in the SEM images (Fig. S6), with a shaped-scab surface seen for inorganic fouling, and a shaped-mucus surface when fouling was mainly organic. As shown in Fig. 3, these surface properties triggered higher bacterial deposition compared to BWc and SWc (no fouling), which resulted from the association of strain IM-1 (Fig. S7(1)). Besides, the CLSM analysis showed that a biofilm did not equally form on all the conditioned surfaces, as shown by the highest affinity of SWd. Hence our results indicated that organic fouling high roughness and hydrophilic surfaces improved the surface conditioning for bacterial association with surfaces, which has also been reported in previous studies (Chen et al., 2013; Guo et al., 2013; Subramani et al., 2009). It is worth mentioning that the numerous imperfections could also affect the bacterial deposition on discarded membranes, as shown by the hole with bacteria accumulation (Fig. 3(2-a)) versus those uniformly associated (Fig. 3(2-b)).

BWd

SWd

k= 2.39

k= 0.15

T1/2=0.29

T1/2=4.55

200%

150%

100%

50%

0% -8Pre-incubation -4 0 time

4

8

12

16

20

24

Time (h)

Fig. 4. Graph of the MC degradation trend by the biofilm formed on the BWd and SWd RO membranes during the first MC exposure (pre-incubation period) and the second one (after pulse). Arrow indicates the MC pulse. MC error bars represent the standard deviation of two technical replicates. Degradation rates were determined from the point where biodegradation began (both cases 3 h after the pulse).

J. Morón-López et al. / Science of the Total Environment 647 (2019) 785–793

1

2

x y

791

3

a

Simulator cell

Average

Biomass (µm3/µm2)

0.198 ± 0.075

2500

Average thickness (µm) 0.340 ± 0.127 Roughness coeficient (Ra) 1.816 ± 0.058 Maximum thickness (µm) 12.774 ± 2.888

MC concentration (µg . L-1)

2250

z b

y = -83.492x + 1950.4 R² = 0.987

2000 1750 1500 1250 1000 750 500 250 0 0

4

8

12

16

20

24

Time (h)

Fig. 5. Biofilm formation and MC removal by the simulator cell on a lab scale. (1) Biofilm visualization by (a) the CLSM image from a cross-sectional xy-view (on top) and z-projection (bottom) (bar shows 25 µm); and (b) SEM image (bar shows 1 µm). (2) Biofilm characterization by COMSTAT and 3D visualization; (3) MC degradation in µg·L-1 by the biofilm formed on the SWd membrane. Dotted line indicates the regression line.

a large surface-to-volume active biofilm to remove MC. Fig. 1 represents the R-MBfR concept for MC removal.

degradation rates (Ho et al., 2007; Holst et al., 2003; Morón-López et al., 2017; Shimizu et al., 2011). No MC removal was observed in the negative controls (Fig. S7(2)), which indicates that MC elimination was due to biological activity. The interesting results obtained with the discarded membranes led us to address a proof-of-concept of a bioreactor to biologically remove MC. Based on the total recycling of the discarded RO membrane module, we proposed a new bioreactor concept named Recycled-Membrane Biofilm Reactor (R-MBfR). R-MBfR takes advantage of the inherited flow of RO membrane technology, i.e., fouling, and uses it to generate

3.4. R-MBFR proof-of-concept for MC removal To validate the feasibility of this proof-of-concept, we used a simulator cell to reproduce a hypothetical large scale on a laboratory scale (Fig. S1). As Fig. 5 depicts, this simulator cell allowed bacterial association, initial biofilm formation and development on a discarded membrane by tangential flow with a long contact time. SWd was selected

0.90 Present study

0.80

Alvarez et al., 2010

Bui et al., 2016

Mulder et al., 2015

0.60 0.50

*

0.40 0.30 0.20

450,000

450,000

1,625

372,000

24,000

126,000 GAC

not-define

PAC

450,000

372,000

126,000

* 24,000

372,000

126,000

24,000

Ozone

450,000

*

0.10 0.00

*

*

450,000

Cost (€ · m-3)

0.70

RO

NF/RO

UV

RMBFR

Technology & population design Fig. 6. Comparison of annual costs per m3 of MC treatment with degradation efficiency above 95%. In gray, the costs adapted by Mulder et al. (2015) (MP removal from different water sources); in blue, the costs adapted from Alvarez et al. (2010) (revision based on lab results for MC removal from American DWTP); in yellow, the costs adapted by Bui et al. (2016) (MP removal); in green, our estimations. Asterisks mean that no uncertainty was quantified in this study. The population design was calculated by assuming 200 L·person−1·day−1 and a 65% operation rate. The population for our technology is related to an 50% efficiency. Value presented in euros by March 2015. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

792

J. Morón-López et al. / Science of the Total Environment 647 (2019) 785–793

4. Conclusion

Table 2 CAPEX & OPEX by the annual equivalent cost. Index

Concept

Cost (€·year−1)

1 1.1 1.2 1.3 1.4 1.5 2 2.1 2.2 2.3 2.4 2.5 2.6

CAPEX Naval container Baseline components Modules (36 units) Engineering Assembly and construction OPEX Component maintenance Personnel cost Microfiltration replacement Modules bioactivation Catchment energy cost Recirculation energy cost Total annual cost

5357.85 515.07 2,068.65 159.79 370.10 2,244.24 5,373.85 690.00 2,666.67 235.80 1,280.00 231.41 269.97 10,731.69

for this proof-of-concept experiment given its bigger recycling difficulty for other applications (García-Pacheco, 2017) and, therefore, for the high pollution it produces compared to other RO waste membrane modules. After surface conditioning (previous fouling), bacterial association and initial biofilm formation, the biofilm development results showed stable adhesion to the surface with EPS production (Fig. 5(1, a–b)). This biofilm was heterogenic, as indicated by the roughness coefficient (Fig. 5(2)), with a maximum thickness of 12.774 μm. What these results indicated is that a more advanced biofilm stage was achieved by the simulation cell, which supports the potential of using the total recycling of RO membrane modules as a reactor for biofilm development. The R-MBfR simulator cell was able to remove 2 mg·L−1 MC within 24 h (Fig. 5(3)). Hence these results confirmed the feasibility of proof-of-concept on a laboratory scale. Nevertheless, no mature biofilm was achieved after 96 h (Fig. 5(2)). Therefore, more experiments are needed to enhance our current outputs. The possibility of future improvement could be R-MBfR simulator cell optimization by allowing, for example, a thicker active MCdegrading biofilm to develop by metabolic stimulation through oxygen supply. Oxygen supply through an inactive inner layer of polyester (Fig. 1, gap A–B) could generate a high-performance system, as applied in membrane aerated biofilms reactors (MABR) technology (Martin and Nerenberg, 2012). Indeed our preliminary results indicated that the discarded membranes used herein supported oxygen diffusion (data not shown).

3.5. Economic feasibility Finally, we made a cost assessment to evaluate the economic competitiveness of our hypothetical R-MBfR technology. Our preliminary data showed a 0.140 € m−3 of unitary cost estimation for removing N95% of extracellular MC (Fig. 6). As far as we know, very few reports have been published on costs associated with current MC removal technologies (Hamilton et al., 2013). Based on previous studies (Alvarez et al., 2010; Bui et al., 2016; Mulder et al., 2015; US EPA Office of Water, 2016), the comparative cost analysis indicated that if our proposed system was extrapolated to a large scale (Table S5), it could be economically competitive compared to conventional treatments, such as ozone, powdered activated carbon (PAC), granular activated carbon (GAC) and UV radiation, and was even less costly than conventional membrane filtration (RO or NF/RO) (Fig. 6). This was especially true for small populations where the R-MBfR system could be highly competitive due to the expected low capital investment (Table 2). It is noteworthy that, although our cost estimation was hypothetical, it clearly represented the order of magnitude in costs of already implemented MC removal technologies. We also made an upward estimation, thus the cost reduction predicted by the economy scale principle during scaling-up was not taken into account.

Based on our promising outcomes, future higher scale experiments will be necessary to reduce uncertainty and to evaluate the hypothesis herein posed. Nevertheless, as far as our results show, the R-MBfR concept combines three essential aspects; a) the biological removal of potent toxins that pose worldwide environmental, health and economic problems; b) the conversion of waste that causes global concern into a valuable product; c) the empowerment of green bioreactor technology in drinking water treatment. Therefore, if further developed and implemented, our concept will open up a new horizon to use recycling methods in water treatment.

Acknowledgements The financial support provided by the Spanish Ministry of Science and Innovation through project INREMEM (CTM2015-65348-C2-1-R) is gratefully acknowledged. The biofilm visualization has been performed by ICTS “NANBIOSIS”, more specifically by the Confocal Microscopy Service: Ciber in Bioengineering, Biomaterials & Nanomedicine (CIBER-BNN) at the Alcala University (CAI Medicine Biology). Conflict of interest The authors declare that they have no conflict of interest. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.scitotenv.2018.07.435.

References Alvarez, M.B., Rose, J.B., Bellamy, B., Water Research Foundation, Cocoa (Fla.), St. Johns River Water Management District (Fla.), Melbourne (Fla.), 2010. Treating Algal Toxins Using Oxidation, Adsorption, and Membrane Technologies. Water RF, Water Research Foundation. Bourne, D.G., Jones, G.J., Blakeley, R.L., Jones, A., Negri, A.P., Riddles, P., 1996. Enzymatic Pathway for the Bacterial Degradation of the Cyanobacterial Cyclic Peptide Toxin Microcystin LR. 62 pp. 4086–4094. Bourne, D.G., Riddles, P., Jones, G.J., Smith, W., Blakeley, R.L., 2001. Characterisation of a gene cluster involved in bacterial degradation of the cyanobacterial toxin microcystin LR. Environ. Toxicol. 16, 523–534. https://doi.org/10.1002/tox.10013. Bui, X.T., Vo, T.P.T., Ngo, H.H., Guo, W.S., Nguyen, T.T., 2016. Multicriteria assessment of advanced treatment technologies for micropollutants removal at large-scale applications. Sci. Total Environ. 563–564, 1050–1067. https://doi.org/10.1016/j. scitotenv.2016.04.191. Chen, Y.P., Zhang, P., Guo, J.S., Fang, F., Gao, X., Li, C., 2013. Functional groups characteristics of EPS in biofilm growing on different carriers. Chemosphere 92, 633–638. https://doi.org/10.1016/j.chemosphere.2013.01.059. Donlan, R.M., 2002. Biofilms: microbial life on surfaces. Emerg. Infect. Dis. 8, 881–890 http://www.cdc.gov/ncidod/EID/vol8no9/02-0063.htm. https://doi.org/10.3201/ eid0809.020063. Dziga, D., Zielinska, G., Wladyka, B., Bochenska, O., Maksylewicz, A., Strzalka, W., Meriluoto, J., 2016. Characterization of enzymatic activity of MlrB and MlrC proteins involved in bacterial degradation of cyanotoxins microcystins. Toxins (Basel) 8, 76. https://doi.org/10.3390/toxins8030076. Eleuterio, L., Batista, J.R., 2010. Biodegradation studies and sequencing of microcystin-LR degrading bacteria isolated from a drinking water biofilter and a fresh water lake. Toxicon 55, 1434–1442. https://doi.org/10.1016/j.toxicon.2010.02.020. García-Pacheco, R., 2017. Nanofiltration and Ultrafiltration Membranes from End-of-Life Reverse Osmosis Membranes a Study of Recycling (Doctoral thesis). Geise, G.M., Lee, H.-S., Miller, D.J., Freeman, B.D., Mcgrath, J.E., Paul, D.R., 2010. Effect of molecular weight and temperature on physical aging of thin glassy poly(2,6-dimethyl-1,4-phenylene oxide) films. J. Polym. Sci. B Polym. Phys. 45, 1390–1398. https://doi.org/10.1002/polb. Guo, K., Freguia, S., Dennis, P.G., Chen, X., Donose, B.C., Keller, J., Gooding, J.J., Rabaey, K., 2013. Effects of surface charge and hydrophobicity on anodic biofilm formation, community composition, and current generation in bioelectrochemical systems. Environ. Sci. Technol. 47, 7563–7570. https://doi.org/10.1021/es400901u. Halan, B., Buehler, K., Schmid, A., 2012. Biofilms as living catalysts in continuous chemical syntheses. Trends Biotechnol. 30, 453–465. https://doi.org/10.1016/j. tibtech.2012.05.003.

J. Morón-López et al. / Science of the Total Environment 647 (2019) 785–793 Hamilton, D.P., Wood, S.A., Dietrich, D.R., Puddick, J., 2013. Costs of harmful blooms of freshwater cyanobacteria. Cyanobacteria An Econ. Perspect , pp. 245–256. https:// doi.org/10.1002/9781118402238.ch15. He, X., Wang, J., Abdoli, L., Li, H., 2016. Mg2+/Ca2+ promotes the adhesion of marine bacteria and algae and enhances following biofilm formation in artificial seawater. Colloids Surf. B: Biointerfaces 146, 289–295. https://doi.org/10.1016/j.colsurfb.2016.06.029. Herzberg, M., Kang, S., Elimelech, M., 2009. Role of Extracellular Polymeric Substances (EPS) in Biofouling of Reverse Osmosis Membranes. Environ. Sci. Technol. 43, 4393–4398. https://doi.org/10.1021/es900087j. Ho, L., Hoefel, D., Saint, C.P., Newcombe, G., 2007. Isolation and identification of a novel microcystin-degrading bacterium from a biological sand filter. Water Res. 41, 4685–4695. https://doi.org/10.1016/j.watres.2007.06.057. Ho, L., Sawade, E., Newcombe, G., 2012a. Biological treatment options for cyanobacteria metabolite removal e a review. Water Res. 46, 1536–1548. https://doi.org/10.1016/ j.watres.2011.11.018. Ho, L., Tang, T., Hoefel, D., Vigneswaran, B., 2012b. Determination of rate constants and half-lives for the simultaneous biodegradation of several cyanobacterial metabolites in Australian source waters. Water Res. 46, 5735–5746. https://doi.org/10.1016/j. watres.2012.08.003. Holst, T., Jørgensen, N.O.G., Jørgensen, C., Johansen, A., 2003. Degradation of microcystin in sediments at oxic and anoxic, denitrifying conditions. Water Res. 37, 4748–4760. https://doi.org/10.1016/S0043-1354(03)00413-5. Kinh, C.T., Suenaga, T., Hori, T., Riya, S., Hosomi, M., Smets, B.F., Terada, A., 2017. Counterdiffusion biofilms have lower N2O emissions than co-diffusion biofilms during simultaneous nitrification and denitrification: Insights from depth-profile analysis. Water Res. 124, 363–371. https://doi.org/10.1016/j.watres.2017.07.058. Landaburu-Aguirre, J., García-Pacheco, R., Molina, S., Rodríguez-Sáez, L., Rabadan, J., García-Calvo, E., 2016. Fouling prevention, preparing for re-use and membrane recycling. Towards circular economy in RO desalination. Desalination 393, 16–30. https://doi.org/10.1016/j.desal.2016.04.002. Lezcano, M.Á., Morón-López, J., Agha, R., López-Heras, I., Nozal, L., Quesada, A., ElShehawy, R., 2016. Presence or absence of mlr genes and nutrient concentrations co-determine the microcystin biodegradation efficiency of a natural bacterial community. Toxins (Basel). 8. https://doi.org/10.3390/toxins8110318. Lezcano, M., Velázquez, D., Quesada, A., El-Shehawy, R., 2017. Diversity and temporal shifts of the bacterial community associated with a toxic cyanobacterial bloom: an interplay between microcystin producers and degraders. Water Res. 125, 52–61. https://doi.org/10.1016/j.watres.2017.08.025. Li, J., Peng, L., Li, J., Qiao, Y., 2014. Divergent responses of functional gene expression to various nutrient conditions during microcystin-LR biodegradation by Novosphingobium sp. THN1 strain. Bioresour. Technol. 156, 335–341. https://doi. org/10.1016/j.biortech.2013.12.118. Li, C., Yang, Y., Ding, S., Hou, L.A., 2016. Dynamics of biofouling development on the conditioned membrane and its relationship with membrane performance. J. Membr. Sci. 514, 264–273. https://doi.org/10.1016/j.memsci.2016.04.066. Li, J., Li, R., Li, J., 2017. Current research scenario for microcystins biodegradation – a review on fundamental knowledge, application prospects and challenges. Sci. Total Environ. 595, 615–632. https://doi.org/10.1016/j.scitotenv.2017.03.285. Mangwani, N., Shukla, S.K., Rao, T.S., Das, S., 2014. Calcium-mediated modulation of Pseudomonas mendocina NR802 biofilm influences the phenanthrene degradation. Colloids Surf. B: Biointerfaces 114, 301–309. https://doi.org/10.1016/j.colsurfb.2013.10.003. Martin, K.J., Nerenberg, R., 2012. The membrane biofilm reactor (MBfR) for water and wastewater treatment: principles, applications, and recent developments. Bioresour. Technol. 122, 83–94. https://doi.org/10.1016/j.biortech.2012.02.110. Merel, S., Clément, M., Thomas, O., 2010. State of the art on cyanotoxins in water and their behaviour towards chlorine. Toxicon 55, 677–691. https://doi.org/10.1016/j. toxicon.2009.10.028. Merritt, J.H., Kadouri, D.E., O'Toole, G.a., 2011. Growing and analyzing static biofilms. Curr. Protoc. Microbiol. 1–18. https://doi.org/10.1002/9780471729259.mc01b01s22.

793

Mogana, D.M., Ramasamy, P., 2016. Sample Preparations for Scanning Electron Microscopy – Life Sciences, Modern Electron Microscopy in Physical and Life Sciences. https://doi.org/10.5772/711. Molina, S., García Pacheco, R., Rodríaguez Sáez, L., García-Calvo, E., Campos Pozuelo, E., Zarco Martínez, D., González de la Campa, J., De Abajo González, J., 2015. Transformation of end-of-life RO membrane into recycled NF and UF membranes, surface characterization. The International Desalination Association World Congress on Desalination and Water Reuse, p. 20. Molina, S., Landaburu-Aguirre, J., Rodríguez-Sáez, L., García-Pacheco, R., de la Campa, J.G., García-Calvo, E., 2018. Effect of sodium hypochlorite exposure on polysulfone recycled UF membranes and their surface characterization. Polym. Degrad. Stab. 150, 46–56. https://doi.org/10.1016/j.polymdegradstab.2018.02.012. Monzon, O., Yang, Y., Li, Q., Alvarez, P.J.J., 2016. Quorum sensing autoinducers enhance biofilm formation and power production in a hypersaline microbial fuel cell. Biochem. Eng. J. 109, 222–227. https://doi.org/10.1016/j.bej.2016.01.023. Morón-López, J., Nieto-Reyes, L., El-Shehawy, R., 2017. Assessment of the influence of key abiotic factors on the alternative microcystin degradation pathway(s) (mlr −): a detailed comparison with the mlr route (mlr +). Sci. Total Environ. 599–600, 1945–1953. https://doi.org/10.1016/j.scitotenv.2017.04.042. Mou, X., Lu, X., Jacob, J., Sun, S., Heath, R., 2013. Metagenomic identification of bacterioplankton taxa and pathways involved in microcystin degradation in Lake Erie. PLoS One 8, e61890. https://doi.org/10.1371/journal.pone.0061890. Mulder, et al., 2015. Costs of Removal of Micropollutants From Effluents of Municipal Wastewater Treatment Plants. Nerenberg, R., 2016. The membrane-biofilm reactor (MBfR) as a counter-diffusional biofilm process. Curr. Opin. Biotechnol. 38, 131–136. https://doi.org/10.1016/j. copbio.2016.01.015. Rodríguez, E., Onstad, G.D., Kull, T.P.J., Metcalf, J.S., Acero, J.L., von Gunten, U., 2007. Oxidative elimination of cyanotoxins: comparison of ozone, chlorine, chlorine dioxide and permanganate. Water Res. 41, 3381–3393. https://doi.org/10.1016/j.watres.2007.03.033. Santiago-Morales, J., Amariei, G., Letón, P., Rosal, R., 2016. Antimicrobial activity of poly (vinyl alcohol)-poly(acrylic acid) electrospun nanofibers. Colloids Surf. B: Biointerfaces 146, 144–151. https://doi.org/10.1016/j.colsurfb.2016.04.052. Sehar, S., Naz, I., 2016. Role of the biofilms in wastewater treatment. Microb. Biofilms Importance Appl. https://doi.org/10.5772/711. Shimizu, K., Maseda, H., Okano, K., Itayama, T., Kawauchi, Y., Chen, R., Utsumi, M., Zhang, Z., Sugiura, N., 2011. How microcystin-degrading bacteria express microcystin degradation activity. Lakes Reserv. Res. Manag. 16, 169–178. https://doi.org/10.1111/ j.1440-1770.2011.00480.x. Subramani, A., Huang, X., Hoek, E.M.V., 2009. Direct observation of bacterial deposition onto clean and organic-fouled polyamide membranes. J. Colloid Interface Sci. 336, 13–20. https://doi.org/10.1016/j.jcis.2009.03.063. US EPA Office of Water, 2016. Water Treatment Optimization for Cyanotoxins. p. 60. Villacorte, L.O., 2017. Biofouling in capillary and spiral wound membranes facilitated by marine algal bloom. Desalination 424, 74–84. https://doi.org/10.1016/J.DESAL.2017.09.035. Vu, B., Chen, M., Crawford, R.J., Ivanova, E.P., 2009. Bacterial extracellular polysaccharides involved in biofilm formation. Molecules 14, 2535–2554. https://doi.org/10.3390/ molecules14072535. Welker, M., Steinberg, C., 2000. Rates of humic substance photosensitized degradation of microcystin-LR in natural waters. Environ. Sci. Technol. 34, 3415–3419. https://doi. org/10.1021/es991274t. Wörmer, L., Huerta-Fontela, M., Cirés, S., Carrasco, D., Quesada, A., 2010. Natural Photodegradation 01 the Cyanobacterial Toxins Microcystin and Cylindrospennopsin. 44 pp. 8–10. Xu, B., Mao, D., Luo, Y., Xu, L., 2011. Sulfamethoxazole biodegradation and biotransformation in the water-sediment system of a natural river. Bioresour. Technol. 102, 7069–7076. https://doi.org/10.1016/j.biortech.2011.04.086.