Article pubs.acs.org/est
Relative Insignificance of Virus Inactivation during Aluminum Electrocoagulation of Saline Waters Charan Tej Tanneru,† Jothikumar N.,‡ Vincent R. Hill,‡ and Shankararaman Chellam*,†,§ †
Department of Civil and Environmental Engineering, University of Houston, Houston, Texas 77204-4003, United States Centers for Disease Control and Prevention, National Center for Emerging and Zoonotic Infectious Diseases, Atlanta, Georgia 30329, United States § Department of Chemical and Biomolecular Engineering, University of Houston, Houston, Texas 77204-4004, United States ‡
S Supporting Information *
ABSTRACT: Combined removal and inactivation of the MS2 bacteriophage from model saline (0−100 mM NaCl) waters by electrochemical treatment using a sacrificial aluminum anode was evaluated. Both chemical and electrodissolution contributed to coagulant dosing since measured aluminum concentrations were statistically higher than purely electrochemical predictions using Faraday’s law. Electrocoagulation generated only small amounts of free chlorine in situ but effectively destabilized viruses and incorporated them into Al(OH)3(s) flocs during electrolysis. Low chlorine concentrations combined with virus shielding and aggregation within flocs resulted in very slow disinfection rates necessitating extended flocculation/contact times to achieve significant loginactivation. Therefore, the dominant virus control mechanism during aluminum electrocoagulation of saline waters is “physical” removal by uptake onto flocs rather than “chemical” inactivation by chlorine. Attenuated total reflectance−Fourier transform infrared spectroscopy provided evidence for oxidative transformations of capsid proteins including formation of oxyacids, aldehydes, and ketones. Electrocoagulation significantly altered protein secondary structures decreasing peak areas associated with turns, bends, α-helices, β-structures, and random coils for inactivated viruses compared with the MS2 stock. Quantitative reverse transcription polymerase chain reaction (qRT-PCR) measurements showed rapid initial RNA damage following a similar trend as plaque assay measurements of infectious viruses. However, ssRNA cleavage measured by qRT-PCR underestimated inactivation over longer durations. Although aluminum electrocoagulation of saline waters disorders virus capsids and damages RNA, inactivation occurs at a sufficiently low rate so as to only play a secondary role to floc-encapsulation during residence times typical of electrochemical treatment.
■
small packaged plants to treat seawater and brackish water15−17 because their high conductivity (i) decreases power consumption, (ii) reduces residence time by allowing operation at elevated current densities (i.e., smaller reactors), and (iii) generates oxidants that can disinfect microorganisms. To our knowledge, electrochemical disinfection has been investigated only using dimensionally stable electrodes (i.e., no coagulation).18−21 In contrast, electrocoagulation is performed by corroding sacrificial anodes to form coagulants in situ so that a wide range of dissolved, macromolecular, and suspended contaminants can be nonspecifically removed. Since electrolysis of saline water is expected to concurrently generate chlorine and aluminum,2,22 we hypothesize that electrocoagulation will enhance virus control in such systems by a combination of “physical” removal and “chemical” inactivation.
INTRODUCTION Electrocoagulation, the in situ addition of metal-ion coagulants by anodic dissolution is well-suited for small-scale and mobile water treatment because it can be highly automated, is portable, and reduces handling of corrosive chemicals.1,2 Although a large body of literature has documented its capability to remove physicochemical constituents,2,3 very little information is available on its ability to control viruses, which is critical for drinking water applications. Limited data suggest that aluminum is a better electrocoagulant than iron to remove viruses from surface water since electrolysis releases insoluble Al(III) rather than soluble Fe(II).3−5 Conventional alum coagulation and aluminum electrocoagulation destabilize viruses through a combination of sweep flocculation and charge neutralization with insignificant inactivation during treatment of surface waters laden with natural organic matter (NOM),5−11 whereas polyaluminum chloride is capable of simultaneously coagulating and inactivating viruses.6−8,12−14 Given the increasing pressures on existing fresh water supplies, electrocoagulation is beginning to be evaluated for © 2014 American Chemical Society
Received: Revised: Accepted: Published: 14590
September 5, 2014 November 15, 2014 November 18, 2014 November 18, 2014 dx.doi.org/10.1021/es504381f | Environ. Sci. Technol. 2014, 48, 14590−14598
Environmental Science & Technology
Article
effluent can be consumed. In this work, centrifugation was employed and other options include flotation, sedimentation, media filtration, or low-pressure membrane filtration.10,11,15,25 Glassware was made to be chlorine-demand free by overnight soaking in free chlorine, rinsing 5-times with nanopure water, wrapping in aluminum foil and baking at 400 °C for 3 h. Free chlorine was measured using N,N-diethyl-p-phenylenediamine (DPD) colorimetric method using a Hach DR-4000 spectrophotometer. Duplicate experiments were performed on different dates during the course of this study at pH 6.2 and 100 mM NaCl concentration. Measured inactivation rates were statistically identical at 95% confidence (Supporting Information (SI) Table S1) demonstrating the reproducibility of our experimental protocols. Virus Propagation, Purification, and Enumeration. MS2 (ATCC 15597-B1; host Escherichia coli ATCC 15597) was propagated using the double-top agar layer technique. Organic carbon carryover was minimized and stock titer was increased to order of 1011 PFU/mL by successive polyethylene glycol precipitation and chloroform extraction.5,10 Viruses were mixed with 9% 8 kDa polyethylene glycol and 1 M NaCl and stirred overnight at 4 °C. Following centrifugation (8000g/45 min), equal amounts of resuspended pellet (in 4 mL PBS) and chloroform were mixed and centrifuged (5000g/45 min). Purified stock was collected from the supernatant freshly before each experiment. Extraction of Viral RNA and qRT-PCR Protocol. qRTPCR was performed on the MS2 stock and inactivated viruses to estimate the amount of intact RNA. To avoid RNA degradation, equal volumes (400 μL) of sample and UNEX lysis buffer26 (Microbiologics, St. Cloud, MN) were mixed thoroughly and stored at room temperature until extraction. A 700 μL sample was passed through a silica spin column (Omega Biotek, Norcross, GA) by centrifugation (10 000g for 1 min). The silica column was washed twice, once with 500 μL of 100% ethanol and then with 500 μL 70% ethanol. The column was centrifuged at 10 000g for 1 min after each wash and followed by one final centrifugation again to remove any excess ethanol. The column was then transferred to a clean microfuge tube. Nucleic acid was eluted by adding 70 μL of 10 mM Tris−1 mM EDTA (pH 8.0) buffer and centrifugation (10 000g for 1 min). MS2 RNA was detected by a one-step qRT-PCR assay using the following primers and probe sequences.27 The TaqMan assay amplified a 77bp fragment with a forward primer 632F, GTCGCGGTAATTGGCGC, reverse primer 708R, GGCCACGTGTTTTGATCGA and a fluorescent labeled probe 650P, 5′-AGGCGCTCCGCTACCTTGCCCT-3′, labeled on the 5′ end with FAM (6-carboxy-fluorescein) and 3′ end with the black hole quencher (BHQ). A 20 μL TaqMan reaction mixture for the assay consisted of 5 μL of 4X TaqMan Fast Virus 1-Step Mastermix (Life Technologies), 0.8 μL primers each (250 nM), 1 μL probe (100 nM), 10.4 μL of nuclease free water and a 2 μL of sample RNA. All qRT-PCR reactions were performed on an ABI 7500 system (Applied Biosystems) with slight modifications to the reported amplification protocol.27 Reverse transcription was performed at 50 °C for 5 min to obtain complementary DNA (cDNA), and denaturation at 95 °C for 20 s. PCR amplification performed on cDNA consisted of 45 cycles of denaturation at 95 °C for 5s and annealing at 60 °C for 30 s followed by fluorescence acquisition at the end of annealing. Cycle threshold (CT) values were collected based on the fluorescence increase during amplification and cycle
The objectives of this work are to empirically demonstrate that viruses are simultaneously inactivated and destabilized during aluminum electrocoagulation of saline solutions and to elucidate inactivation mechanisms. Loss of infectivity of the Fspecific ssRNA coliphage MS2 was monitored in NaCl solutions by the plaque assay and compared with RNA damage measured by quantitative reverse transcription polymerase chain reaction (qRT-PCR). Free chlorine is shown to be generated during electrolysis and evidence for oxidative capsid modifications is provided using attenuated total reflectance− Fourier transform infrared spectroscopy (ATR-FTIR). Decelerating disinfection kinetics in the presence of Al(OH)3(s) precipitates is discussed in terms of virus shielding and occlusion during electrocoagulation as virions are aggregated and taken up by flocs.
■
EXPERIMENTAL SECTION Electrocoagulation. Batch electrolysis of NaCl solutions (1 mM, 10 mM, and 100 mM) was performed in a 450 mL cylindrical Plexiglas cell with an annular electrode configuration; 15 cm aluminum anode (Puratronic grade 99.9965% as Al, Alfa Aesar) surrounded by a perforated 316-stainless steel cathode. Before each experiment, the cell was soaked overnight in free chlorine solution, rinsed 5-times with nanopure water, and thoroughly air-dried to remove any chlorine demand. The anode was mechanically scrubbed prior to electrolysis, and periodically cleaned with HCl to avoid passivation. All experiments were performed in buffered (10 mM NaHCO3) solutions at pH 6.2 ± 0.2 with the targeted NaCl concentration. A few selected experiments were also performed at pH 8.2 ± 0.2. Electrolysis did not appreciably increase the pH, which remained within 0.2 units of the initial value. Total aluminum concentrations were analyzed using atomic absorption spectroscopy (Flame AA-AAnalyst 300, PerkinElmer) according to Standard Method 3111 after acidifying samples to pH 2 using 11.5 N HCl. After seeding (spiking) with virus stock and thoroughly mixing the feedwater, a sample was collected to enumerate viruses before electrolysis to measure the initial concentration. Electrolysis was performed for predetermined durations based on Faraday’s law to dose the appropriate amount of aluminum (0−30 mgAl/L) at a constant current density of 20 mA/cm2 corresponding to a potential of 1.2 V. The entire cell contents were then stirred slowly in the dark for a total flocculation duration of 5 h and hourly samples were collected to which sodium thiosulfate was added to halt disinfection by quenching free chlorine. A sterile 10% (w/v) stock of thiosulfate was prepared 1 mL of which 2 mL was added during sampling.23 Viruses associated with the flocs were separated by centrifugation (10 000g/20 min) and dissolved in 6% beef extract after elevating the pH to 9.5 and vortexing manually for 15 min.4,5,7,8,12,24 Hourly samples were collected during the 5 h period, including one immediately after electrolysis (193 s, 386 s, and 579 s) which was centrifuged (10 000g/20 min) to separate solids. Each pellet was dissolved in beef extract at high pH to measure plaque counts, (i.e., electrocoagulated infective viruses over time). Plaques were also measured in the centrifuged supernatant corresponding to “free” infective viruses remaining in the water column that were not incorporated into flocs. Viruses in supernatant and flocs were also assayed by qRT-PCR. It is emphasized that similar to conventional coagulation liquid−solid separation needs to be performed to remove electrocoagulated flocs before the purified 14591
dx.doi.org/10.1021/es504381f | Environ. Sci. Technol. 2014, 48, 14590−14598
Environmental Science & Technology
Article
Figure 1. Aluminum electrochemical generation was augmented by chemical dissolution to release more coagulant than predicted by Faraday’s law (solid brown line in (a)). Expected (electro)chemical and hydrolysis reactions are also shown.2,3,28,29 Note that free chlorine is predicted to be generated via oxidation of dissolved chloride ions. Electron micrographs of the surface of a new aluminum electrode (b), aluminum anode after electrolysis showing pitting (c), and a magnified image of the corroded pit (d) are also shown. The scale bar is 20 μm.
Figure 2. Summary of infective viruses extracted from the flocs and those present in the water column after centrifugation at different aluminum dosages (electrolysis times) at pH 6.2 and 100 mM NaCl; (a) no electrolysis (0 mgAl/L) control (b) negative control with 386 s electrolysis with simultaneous thiosulfate quenching (c) 193 s electrolysis for target 10 mgAl/L dose (d) 386 s electrolysis for target 20 mgAl/L dose (e) 579 s electrolysis for target 30 mgAl/L dose. Within each subfigure, the set of six bars on the left-hand side corresponds to viruses sorbed onto the flocs (extracted from pellets) and the other set of six bars on the right-hand side corresponds to free viruses (in the centrifuged supernatant). The horizontal dashed line near the top of each graph is the measured initial virus concentrations in the feedwater before starting electrolysis (order of 108.5). Note that the black colored bar on the far left of each data set represents the sample collected at the end of electrolysis and the other 5 bar to its right represent the hourly samples during extended contact (flocculation). Error bars represent one standard deviation of a minimum of three plaque assays.
14592
dx.doi.org/10.1021/es504381f | Environ. Sci. Technol. 2014, 48, 14590−14598
Environmental Science & Technology
Article
as NOM.5,25 Free chlorine is also predicted to be generated (discussed later). Total Virus Reductions (Removal and Inactivation). After electrolysis, the current was turned off and the entire suspension was slowly mixed (flocculated) for a total of 5 h to provide extended contact between electrolysis products and viruses. Infective virus concentrations measured in the flocs and supernatant along with two controls are summarized in Figure 2. Results from similar experiments at pH 8.2 are depicted in SI Figure S1. Figure 2a shows the first control experiment performed by adding viruses to the cell and simply stirring the feed suspension; in other words electrocoagulation was not performed. As observed, viruses were quantitatively recovered from the aqueous phase for the entire 5 h flocculation duration following electrolysis demonstrating no inactivation or loss of viruses on the cell components. Figure 2b is another control experiment where the chlorine quenching agent sodium thiosulfate was initially added along with viruses to the feedwater prior to 386 s electrolysis targeting 20 mgAl/L. Spiked viruses were quantitatively recovered at each time point (constant magnitude of the left set of bars in Figure 2b all equaling the feed concentration) and supernatant concentrations were ∼2.5−5.0 orders of magnitude lower than those recovered from the flocs. Therefore, viruses were effectively uptaken by Al(OH)3(s) and they were not inactivated or lost in the apparatus. Hence, our extraction and plating protocols were accurate and reproducible and any measured loss of plaque forming units can be attributed to inactivation. The monotonic decrease in supernatant concentrations in Figure 2b signifies progressive coagulation/flocculation of viruses and uptake by Al(OH)3(s) over the entire flocculation duration of 5 h. Since we have recently reported sweep flocculation and charge neutralization as dominant virus electrocoagulation mechanisms,5 they are not discussed in this manuscript; we rather focus on inactivation. Figure 2c−e correspond to electrolysis durations of 193 s, 386 s, and 579 s, respectively, from which two important trends can be discerned. First, similar to the negative control described in Figure 2b, for all three coagulant doses at each time point, infective virus concentrations extracted from flocs were orders of magnitude higher than the corresponding supernatant. For example, in Figure 2d after 2 h flocculation 106.5PFU/mL viruses were present in the flocs whereas only 103.8PFU/mL was measured in the supernatant, demonstrating highly effective electrocoagulation. Second, for any electrolysis duration or aluminum dosage, the number of infective viruses in the flocs decreased monotonically when additional contact time was provided during flocculation. For example, in Figure 2e 108.2PFU/mL viruses were seeded, which decreased to 107.8PFU/mL immediately after the electrolysis duration of 579s. Infective virus concentrations extracted from electrocoagulated solids over the flocculation duration further decreased to 106.7PFU/mL, 106.3PFU/mL, 105.3PFU/mL, 104.4PFU/mL, 104.1PFU/mL, respectively for 1, 2, 3, 4, and 5 hours of contact, respectively. This suggests that viruses were inactivated during electrocoagulation. Similarly, infective virus concentrations in the solid phase also decreased over the flocculation duration when extended contact was provided for other electrolysis durations as shown in Figure 2c,d indicating inactivation. Decreasing supernatant concentrations in Figure 2c−e over time denote progressive virus control by combined coagulation
threshold (CT) values >40 were classified as negative. Two replicates of each sample extract (2 μL) were assayed by qRTPCR and runs included an MS2 RNA positive control and a notemplate negative control. The negative control did not show any amplification. The CT values, the fractional cycle number reported by real-time PCR instruments indicates the point at which the fluorescence associated with a positive DNA amplification reaction increased beyond the threshold fluorescence associated with negative reactions. The sensitivity of TaqMan assays was evaluated using 10-fold serial dilutions of RNA extracted from MS2. A sample of known concentration was used to construct the standard curve. By running standards of varying concentrations, the standard curve was plotted, and the results interpolated for unknown samples. A 2 μL RNA per 20 μL reaction was added to express in PFU/mL. For reproducibility, duplicate samples were tested at each dilution. The detection limit was established based on the highest dilution with a positive signal. ATR-FTIR Spectroscopy. ATR-FTIR spectra were collected by placing viruses on the transmissive Ge window of a Nicolet iS10 spectrometer equipped with an Ever-Glo MIR source, DTGS detector, KBr beam splitter and Omnic 8.5 software. Reported spectra are averages of six measurements from 512 coadded scans collected at 4 cm−1 resolution and a zero filling factor of 1 using Happ-Genzel apodization and Mertz phase correction. The resolution in the amide I and II regions (1700−1500 cm−1) was enhanced by taking the second derivative after nine-point Savitzky-Golay smoothing to deconvolute overlapping peaks. Maximum absorption intensity, band frequency, and bandwidth obtained from second derivative spectra were used to curve-fit the original spectra by assuming a Gaussian−Lorentzian shape for the amide I band and for each peak. This procedure gave insights into secondary structures of capsid proteins.
■
RESULTS AND DISCUSSION Aluminum Dissolution. Electrolysis was performed for the requisite times calculated from Faraday’s law4,25 (SI equation S1) to generate the targeted coagulant dosages (193 s, 386 s, and 579 s for 10 mgAl/L, 20 mgAl/L, and 30 mgAl/L, respectively). However, as seen in Figure 1 slightly more aluminum was reproducibly dissolved in replicate experiments than expected from purely electrochemical considerations. Measured concentrations followed a linear trend with a statistically significant (95% confidence) higher slope (0.0558 ± 5.81 × 10−4mg/L-s) than Faraday’s law (0.0518 mg/L-s). Hence, chemical dissolution of the electrode augmented aluminum dosing in the presence of NaCl,2,3,22 which is in contrast to low salinity surface water where coagulant dosing quantitatively follows electrochemical considerations alone.5,25 Electron micrographs revealed electrolysis-induced pitting corrosion on the anode surface22 (Figure 1c,d), which is consistent with chemical dissolution, compared to the relatively smooth surface of the new aluminum electrode (Figure 1b). Hence, chloride ions appear to have penetrated any existing passivation layer on the anode leading to pitting thereby increasing aluminum release. The associated (electro)chemical reactions2,3,28,29 are summarized in Figure 1a. Only a slight increase in the bulk pH was measured during electrolysis (∼0.2units) attributed to generation of OH− ions. Gas bubbles were visibly released as expected from cathodic H2(g) evolution, which however did not cause significant floc flotation presumably due to the absence of hydrophobic moieties such 14593
dx.doi.org/10.1021/es504381f | Environ. Sci. Technol. 2014, 48, 14590−14598
Environmental Science & Technology
Article
Figure 3. (a) Virus inactivation over 5 h of contact or flocculation following electrolysis for 193 s, 386 s, and 579 s corresponding to target coagulant dosages of 10, 20, and 30 mgAl/L respectively. Two negative controls are also depicted showing no inactivation with no electrolysis or when electrolysis was performed for 386 s in the presence of sodium thiosulfate. (b) Virus coagulation and uptake by Al(OH)3 flocs is compared with inactivation immediately at the end of electrolysis with no additional contact time (flocculation). (c) Virus coagulation and uptake by Al(OH)3 flocs is compared with inactivation after electrolysis and 4 h of additional contact time (flocculation). Error bars represent one standard deviation of a minimum of three plaque assays.
infectivity dominated virus control at short-times (i.e., immediately after electrolysis) as depicted in Figure 3b. Under our experimental conditions, a maximum of only 0.4log inactivation was measured for the highest electrolysis duration studied (579 s), whereas physical removal by coagulation/flocculation was 3log viruses over the same duration. However, inactivation grew in importance with increasing flocculation duration as extended contact time was provided after electrolysis. As summarized in Figure 3c, removal by coagulation/flocculation was comparable to inactivation after 4 h flocculation or contact for all experiments (e.g., 3.6 ± 0.4log removal and 3.4 ± 0.05log inactivation for 386s electrolysis). Evidence for Chlorine-Induced Virus Inactivation during Electrocoagulation. Figure 4 summarizes results from several experiments designed to demonstrate the role of free chlorine in inactivating viruses. First, as shown earlier (Figure 2b and Figure 3a control), infective virus concen-
and inactivation. These values were normalized by initial (seed) concentrations to calculate cumulative virus control by a combination of removal and inactivation (SI Figure S2). As seen, virus control by combined removal/inactivation increased with (i) flocculation time for a given aluminum dosage and (ii) electrolysis duration (aluminum dosage) for a fixed contact time, even reaching >8logs with 30 mgAl/L at ≥4h contact time. Virus Inactivation. It is emphasized that the plaque assay is capable of enumerating only infective viruses. The total number concentration of infective virions at each time point during 5 h of flocculation was obtained by adding the solid and liquid phase concentrations measured using the plaque assay shown in Figure 2. The initial concentration of infective viruses was also measured by the plaque assay. The ratio of these two values was used to calculate the fraction of infective viruses remaining; any associated loss attributed to inactivation (Figure 3a). From the negative controls depicted in Figure 2a,b, the total number of infective viruses remained relatively constant when electrolysis was not performed (black line in Figure 3a) and when a chlorine quencher (Na2S2O3) was added to the feedwater spiked with viruses prior to electrolysis (magenta line in Figure 3a). Therefore, under these conditions, viruses were not inactivated. However, infective virus concentrations decreased monotonically over the flocculation duration for each of the three electrocoagulation experiments targeting different aluminum dosages demonstrating inactivation. For example, approximately 0.5log inactivation was measured after 1 h, which increased to 2.7log after 5 h additional contact during flocculation for 193 s electrolysis time (10 mgAl/L target). Higher levels of inactivation were measured by extending electrolysis durations. Virus removal by coagulation alone was measured as infective viruses uptaken by flocs that could be eluted at high pH onto beef extract and enumerated by the plaque assay. This is compared to inactivation alone (loss of infectious virions) for various electrolysis durations immediately after the current was turned off with no additional flocculation in Figure 3b. Similar data are shown in Figure 3c but after 4 h of flocculation. Importantly, sorption onto Al(OH)3(s) without loss of
Figure 4. Evidence of chlorine-induced inactivation. Results from numerous control experiments are shown along with electrocoagulation of 100 mM NaCl for 386 s with a target of 20 mgAl/L. Error bars represent one standard deviation of a minimum of three plaque assays. 14594
dx.doi.org/10.1021/es504381f | Environ. Sci. Technol. 2014, 48, 14590−14598
Environmental Science & Technology
Article
Table 1. Free Chlorine Concentrations and Corresponding First Inactivation Rate Constants under Different Electrolysis Conditions electrolysis time (s)
NaCl conc. (mM)
pH
0 386 193 193 193 386 579 193 386 579
100 0 1 10 100 100 100 100 100 100
6.2 6.2 6.2 6.2 6.2 6.2 6.2 8.2 8.2 8.2
Al conc. (mg/L)
