Guarding Drinking Water Safety against Harmful Algal Blooms: Could UV/Cl₂ Treatment Be the Answer?

Abstract

Frequent and severe occurrences of harmful algal blooms increasingly threaten human health by the release of microcystins (MCs). Urgent attention is directed toward managing MCs, as evidenced by rising HAB-related do not drink/do not boil advisories due to unsafe MC levels in drinking water. UV/chlorine treatment, in which UV light is applied simultaneously with chlorine, showed early promise for effectively degrading MC-LR to values below the World Health Organization’s guideline limits. Still, much is unknown regarding potential disinfection byproduct formation and associated toxicity, which can occur from the reaction of chlorine and other reactive species with MCs and algal and natural organic matter. To ensure UV/chlorine guarding drinking water for human consumption, the degradation and detoxification of four of the most problematic MC variants, namely, MC-LR, -RR, -YR, and -LA, which differ in amino acid substituents, were evaluated using UV/chlorine and compared to results from chlorination. Overall, UV/chlorine effectively enhanced MC degradation kinetics and generated less halogenated disinfection byproducts in the target analysis of 11 types of DBPs_C_1–3 from 7 classes, total organic chlorine, and nontarget analysis revealing 35 higher molecular weight DBPs_C_46–52, which maintained the MC structures. Reactivity and cytotoxicity changes varied based on the individual amino acid moieties within the cyclic heptapeptide structure common to all MCs. Analogous trends in MC reactivity were observed in degradation kinetics and mixed MC competition reactions, aligning with individual amino acid structure–reactivity. Cytotoxicity results indicated no significant unintended toxic consequences from MC_DBPs. Our results suggest that UV/chlorine treatment offers an efficient strategy for treating MCs in drinking water.

Graphical Abstract

1. INTRODUCTION

Nutrification of water bodies and climate change trigger excess growth of algae and cyanobacteria, leading to harmful algal blooms (HABs)^1–5 and the production of cyanobacterial toxins, known as cyanotoxins.^6,7 Notably, Toledo, Ohio, issued a “do not drink/do not boil (DND/DNB)” order in 2014 due to cyanotoxin detection in drinking water,⁸ and similar orders have occurred in several U.S. states,^4,9–12 underscoring the urgent need to manage cyanotoxins in drinking water sources and treatment plants.

Microcystins (MCs), a group of hepatotoxic cyclic heptapeptides, are important cyanotoxins produced by microcystis blooms and are often difficult to remove in treatment. During extreme HAB events, MCs can exceed the EPA’s health threshold by 100-fold in source waters. For example, MCs reached 16,920 and 1449 μg/L at a major drinking water intake in Clear Lake, CA in 2014 and 2021 and 860 μg/L in Lake Okeechobee, FL in 2021.^13–15 MC structures consist of a heptapeptide ring with variable AAs at positions 2 (AA₂) and 4 (AA₄). Over 300 MC variants have been identified. During cyanotoxin-related water crises, multiple MC congeners have been detected in source and treated waters (Figure S1 and Table S1).^16–18 For example, HABs occurring in Lake Erie are typically dominated by four MCs, including MC-LR (leucine in AA₂ and arginine in AA₄), MC-RR (arginine in AA₂ and AA₄), MC-YR (tyrosine in AA₂ and arginine in AA₄), and MC-LA (leucine in AA₂ and alanine in AA₄).^16,19 MC structures also have the 3S-amino-9S-methoxy-2S_,6,8S-trimethyl-10-phenyldeca-4E,6E-dienoic acid (Adda) group containing a diene that can be oxidized and compete for an available oxidant.

Drinking water treatment plants impacted by HABs regularly monitor for MCs and explore potential treatment enhancements,^20–22 such as enhanced coagulation and/or prolonged chlorine contact time and increased dosages. However, an overlooked source of MCs is intracellular toxins from cyanobacteria, which accumulate in clarifier sludge and filter backwash.¹⁸ Accumulated MC concentrations can surpass those in source water by up to 12 times and can be released due to eventual cell death or lysis, often occurring later than the detection of blooms at the plant intake.¹⁷ Therefore, the burden of cyanotoxin management relies heavily on the final barriers following filtration [e.g., granular activated carbon (GAC) and chlorination].

UV-advanced oxidation processes have been explored and implemented at drinking water plants as a barrier against cyanotoxins.^9,21,23 In 2017, UV/H₂O₂ was employed at a drinking water plant to treat MCs and taste-and-odor issues during a year-round HAB event at a lake in Ohio. Compared to UV/H₂O₂, UV/chlorine is a relatively new and less-studied UV-AOP, offering promising advantages for tackling HABs events.²⁴ The quantum yield of HO^● produced in UV/chlorine is comparable to UV/H₂O₂.^25,26 Moreover, MCs with variant amino acid moieties are more susceptible to reactive chlorine species (RCS) generated by UV/chlorine.

We previously confirmed the susceptibility of MC-LR to UV/chlorine²⁷ and UV/H₂O₂,²⁸ with UV/chlorine demonstrating greater efficiency and a wider application range. However, the vulnerability of other ubiquitous and toxic MC congeners (e.g., MC-RR, MC-YR, and MC-LA) to UV/chlorine remains undemonstrated. The AA residue of MC congeners with different structures and pK_a values influences their reactivity against oxidants (e.g., chlorine,²⁹ ozone,³⁰ and HO^●28). Moreover, UV/chlorine has the potential to generate chlorine radicals that could produce toxic DBPs; thus, it is important to also thoroughly evaluate DBPs and toxicity following these reactions.

At the typical pH range used in drinking water treatment (pH 6.5–8.5), HOCl and OCl⁻ (pK_a = 7.5) are the most abundant RCS, surpassing others by at least 3 orders of magnitude (e.g., Cl₂O, H₂OCl⁺, and Cl₂).^31,32 While present at significantly lower concentrations, molecular chlorine [Cl_2(aq)] has been identified by Raman spectroscopy³³ and is also a highly reactive chlorinating agent. Even at low concentration, Cl_2(aq) is able to significantly influence the chlorination kinetics of some organic compounds, particularly in the presence of excess chloride ions,^31–36 following the equilibrium in eq 1.

$$\text{HOCl}+\text{C}{\text{l}}^{-}+{\text{H}}^{+}\leftrightarrow \text{C}{\text{l}}_{2(\text{aq})}+{\text{H}}_2\text{O},\text{p}{K}_a=3.4{,}^{33,36}$$ (1)

Lau et al.^31,32 demonstrated that Cl_2(aq) enhanced the chlorination kinetics of various phenols and alkenes, with chloride ions as a potential catalyst at pH < 7.5. Rose et al.³⁷ also referred to Cl_2(aq) as exotic electrophiles that selectively enhance the chlorination rate of less nucleophilic moieties. As the ionizable residues (i.e., Arg, Tyr) and aromatic moiety (i.e., Adda) of MCs enable their chlorination,²⁹ it is reasonable to hypothesize that chloride ions can enhance MC degradation kinetics by generating Cl_2(aq). Huang and MacKay³⁸ reported that without the addition of chloride at pH 7–9, the role of Cl_2(aq) in MC-LR chlorination was negligible. A closer look at the reactivity between MCs and Cl_2(aq) is needed. Although some studies reported that Cl⁻ affected the degradation of contaminants by UV/chlorine to various extents,^39–44 the role of Cl_2(aq) has not been clearly determined.

Significant increases in freshwater salt concentrations have resulted in various environmental and public health consequences, known as freshwater salinization syndrome (FSS).^45,46 FSS is particularly prevalent in densely urbanized and agricultural areas due to anthropogenic activities, such as deicer applications,⁴⁷ agricultural runoff,⁴⁸ and wastewater discharges.⁴⁹ Salinization exacerbates HABs by mobilizing nutrients in soil and sediments⁵⁰ and aggravating eutrophication.⁵¹ Therefore, it is important to investigate the impact of rising Cl⁻ concentrations, the primary FSS anion, on the chlorine chemistry during HAB events.

The objectives of this study are to (1) systematically investigate the feasibility of UV/chlorine treatment for the removal/degradation of MCs, with a focus on the AA residue-reactivity relationship; (2) explore the roles of water matrices in HAB-impacted water; (3) investigate, for the first time, the influence of Cl⁻ and generated Cl_2(aq)) on MC chlorination, and the role of Cl_2(aq) in UV/chlorine treatment; (4) investigate the formation of potentially hazardous DBPs in UV/chlorine treatment; and (5) assess MC detoxification efficiencies in both UV/chlorine and chlorination treatment. This study evaluates the overall merits and concerns of using UV/chlorine in controlling MCs to ensure DW safety against HABs. Additionally, it sheds light on potential chemistry changes in HABs with increased salinity.

2. MATERIALS AND METHODS

2.1. Chemicals.

Ultrapure water (18.2 MΩ·cm) was used in all stock solutions and reactions; additional autoclave sterilization was performed for cytotoxicity measurements. Free available chlorine (FAC) stock solutions (100 mg/L Cl₂) were prepared by diluting sodium hypochlorite solution (active chlorine 4.5–5.5%, Acros Organics). The working FAC solution was standardized prior to each batch using the N,N-diethyl-p-phenylenediamine (DPD) colorimetric method.⁵² Four MC solid standards (MC-LR MC-RR, MC-YR, and MC-LA), with ≥95% purity, were obtained from Enzo Life Sciences, Inc. (East Farmingdale, NY) and were fully dissolved in ultrapure water for stock solutions (100 mg/L, stored at −20 °C). To minimize interference from other halide ions, 99.999% pure sodium chloride (trace metal basis, Thermo Fisher, Waltham, MA) was used. Standards of l-arginine (Arg), l-tyrosine (Tyr), l-leucine (Leu), and l-alanine (Ala) (≥98%) were obtained from Thermo Fisher Scientific for fresh working solutions prepared within 4 h before testing. Additional information on reagents and field water samples is available in Supporting Information (Section S1).

2.2. Reactions and Kinetic Model Simulations.

UV/chlorine and chlorination reactions were performed in Petri dishes (Pyrex, 10 × 60 mm) with a quartz and black cover, respectively. Reagents prewarmed to room temperature (21 ± 1 °C, e.g., pH buffers: acetate for pH 4–5, phosphate for pH 6–8, and borate for pH 8–10, NaCl, and MCs or AAs) were dosed into Petri dishes with magnetic stirring. Chlorination reactions were initiated by adding a chlorine stock solution and were maintained in the dark with a black cover except during sampling intervals. To initiate UV/chlorine reactions, the well-mixed solution with a quartz cover dish was exposed to UV irradiation within 15 s after chlorine dosing under a LP-UV collimated beam system (0.100 ± 0.005 mW/cm²), as described elsewhere.^27,53 Six 150 μL aliquots were periodically taken from each reaction (<10% of total volume 10 mL) and immediately quenched by vigorously mixing with ascorbic acid to achieve a 5:1 molar ratio between ascorbic acid and initial FAC concentration. More details are available in Section S2. The potential formation of unstable N-chloro-MCs was excluded from consideration by quenching with excess ascorbic acid. Therefore, kinetic constants reported represent toxicity-attributing side-chain changes through UV photolysis⁵⁴ and oxidation (Figures S2 and S3).

This study employed both second-order degradation kinetics and pseudo-first-order kinetics to investigate the role of Cl⁻, pH, and UV irradiation in chlorination and UV/chlorine under diverse conditions and models (Section S4). Specifically, degradation of MCs in buffered ultrapure water (21 μM FAC, 1 μM MCs) was investigated with pseudo-first-order reactions (eq 2), where samples were collected prior to decay of chlorine, reaching 10%.

$$\text{ln}\frac{{[\text{MC}]}_t}{{[\text{MC}]}_0}=-k_{\text{obs},\text{MC}}t$$ (2)

For reactions with chlorine consumption exceeding 10% and those with chlorine present at less than 10-fold excess (such as the slow stage chlorination of AAs and field water samples), second-order kinetics³⁸ were employed. Our results demonstrated a precise fit with eq 3, where FAC exposure (C_t values, M·s) was calculated by simulating the nonlinear time-course FAC depletion as explained in Figure S4.

$$\begin{gathered} \ln \frac{{[C]}_t}{{[C]}_{60\text{s}}}=-k_{\text{app},C}\times \underset{60}{\overset{t}{\int }}[\text{FAC}{]}_t\text{d}t,\text{where}C \\ \text{refers to any AA or MC} \end{gathered}$$ (3)

Observed first-order-kinetic rate constants (k_obs, s⁻¹) and apparent second-order-kinetic rate constants (k_app, M⁻¹ s⁻¹) were obtained by least-squares regression fitting of ln([C]_t/[C]₀) vs time and vs FAC exposure, respectively, based on measured concentrations and reaction time of six samples from each triplicate batch reaction. Error bars represent the standard deviation of three rate constants obtained from the reaction with r² > 98%. Examples of model fitting are shown in Figures S5 and S6 with shadow lines representing 95% confidence intervals.

2.3. Quantification of 66 DBPs and Total Organic Halogen Analysis.

Sixty-six DBPs, including regulated and unregulated DBPs, were measured using liquid–liquid extraction with GC–MS (Agilent 7890 GC, 5977A) with electron ionization in selected ion monitoring mode as described previously.^55–57 DBPs included 10 trihalomethanes, 13 haloacetic acids (HAAs), 10 haloacetonitriles, 9 haloketones, 7 halonitromethanes, 13 haloacetamides, and 4 haloaldehydes (Table S6). For the HAA analysis, extracts were derivatized with diazomethane. Total organic chlorine (TOC1), bromine (TOBr), and iodine (TOI) were measured using combustion-ion chromatography with a Mitsubishi total organic halogen analyzer (Cosa Instruments; Mandel Scientific) coupled to a Dionex Integrion ion chromatograph (Dionex, Sunnyvale, CA) as described previously.^56–58

2.4. Target and Nontarget High Resolution Mass Spectrometry Analysis.

An ultrahigh performance liquid chromatograph (UHPLC, UltiMate 3000) coupled to a triple quadrupole mass spectrometer (TSQ_ Quantis, Thermo Scientific) was used for MC qualification and quantification with positive ion electrospray ionization (ESI+). An Orbitrap Fusion Lumos mass spectrometer (Orbitrap MS, Thermo Scientific) using both positive and negative ESI, a UHPLC (Agilent 1290 Infinity) coupled with quadrupole time-of-flight mass spectrometer (QTOF-MS, Agilent 6540), and a UHPLC coupled to an Agilent 6545 QTOF-MS with negative ESI were applied for the disinfection byproducts analysis. Further details are found in Section S5.

Nontarget LC–MS analysis was conducted using an Agilent 1290 Infinity II UHPLC coupled to an Agilent 6545 quadrupole time-of-flight (QTOF) mass spectrometer with negative ion-electrospray ionization (ESI), using UHPLC and MS parameters similar to EPA Method 544.¹⁷ Parameters were set as following: 30,000 resolution, capillary voltage 5500 V, gas temperature 125 °C, drying gas 12 L min⁻¹, nebulizer pressure 40 psi, m/z 50–1700 scan range, and gradient LC program (Table S7).

2.5. Cytotoxicity Measurements.

HepaRG cells, a human hepatoma cell line, were cultured and challenged with two of the more toxic MCs (MC-LA and -LR) and their DBPs during treatment. Cytotoxicity was evaluated by comparing cell viabilities of treatment and control groups (positive controls with untreated MCs, negative controls with toxicity medium only), each in triplicate on the same plate using the cell proliferation assay (XTT). Further details are given in Section S6.

3. RESULTS AND DISCUSSION

3.1. Reaction Overview of Free and Combined Amino Acids with HOCl and Cl_2(aq).

In the UV/chlorine process with elevated [Cl⁻], HOCl, RCS, Cl_2(aq), HO^●, and UV photolysis may contribute to MC degradation. In the less varied parts of the MC structure, the benzene ring and conjugated diene on the Adda residue, and the exocyclic C–C double bond of the Mdha site are potential oxidizable sites by oxidants, such as HOCl and HO^●.^28,59 Meanwhile, the peptide bonds (backbone sites) composed of amides are much less reactive with HOCl (k_app, 10⁻³ −10⁻¹ M⁻¹ s⁻¹),⁶⁰ especially in the case of peptides incorporating charged side chains. Thus, Adda, Mdha, and oxidizable residues of AA₂ and AA₄ are considered the primary target sites during UV/chlorine treatment. Based on the group contribution method theory, the reactivities of AA₂ and AA₄ residues are expected to contribute to the different reactivities of MCs.^28,61 All potential reactive sites compete to react with available reactive oxidants.⁶² Thus, the degradation kinetics of AAs and MCs in UV/chlorine processes was systematically compared.

Chlorine speciation simulations shown in Figure 1A (based on eq 1 and details in Section S7) reveal that [Cl_2(aq)] increases linearly with [Cl⁻], while other active chlorine species (HOCl, OCl⁻, Cl₂O, and H₂OCl⁺) remain relatively stable. Figure 1B shows the impact of [Cl⁻] (0.3–10 mM)⁶³ and pH (7.4 and 8.4).

Figure 1.

Role of chloride ions in chlorine speciation ([FAC] = 21 μM). (A) Chlorine speciation from pH 6.4–9.4. (B) [Cl₂(aq)] vs [Cl⁻] at pH 7.4 and 8.4. T = 20 °C. Simulation of chlorine speciation was based on equilibrium constants corrected by Davies equation at different chloride concentrations, detailed in the Section S7.

With increasing [Cl⁻], the equilibrium of eq 1 shifts to the right, with 1 mol of HOCl and Cl⁻ consumed for each mole of Cl_2(aq) generated. Raising [Cl⁻] from 0.3 to 5 mM at pH 7.4 leads to a 14-fold increase in Cl_2(aq) (0.3–5.2 pM, Figure 1B), with neglectable changes in [HOCl] and [OCl^-], and decreases in [Cl₂O] and [H₂OCl⁺] by 12% and 5%, respectively.

3.2. Reactivity of Selected Amino Acids with HOCl and Cl_2(aq).

The apparent rate constants (k_app, M⁻¹ s⁻¹) were obtained through the kinetic experiments as described in the previous section. Control experiments with Cl⁻ as only variable changing from 0.1 to 15 mM were conducted at pH 7.4 and 8.4 to exaggerate the impact of Cl⁻ and resulting Cl_2(aq) on the chlorination.^31–36 The species-specific kinetics related to the four specific AA residues (i.e., Tyr, Arg, Ala, and Leu) and their associated MCs are shown in Table 1.^64–67

Table 1:

Reactivities of MCs and Specific Residues with HO Cl, Cl_2(aq), and HO^•, Species-Specific Second-Order Rate Constants, M⁻¹ s⁻¹ at pH 7.2–7.5, 20 °C

The calculated apparent second-order rate constants represent the covalent modification rate for both −NH₂ and oxidizable side chains in the tested AAs. At pH 7.4, rate constants followed the order: Tyr (40.1 M⁻¹ s⁻¹) > Ala (35.2 M⁻¹ s⁻¹) > Arg (24.4 M⁻¹ s⁻¹) > Leu (18.3 M⁻¹ s⁻¹). Ala and Leu, having inert aliphatic chains [−CH₃ and −CHCH-(CH₃)₂], primarily undergo covalent modification on R–NH₂. These rate constants are consistent with previous studies,^68–70 where Ala is more reactive than Leu, possibly due to steric hindrance, particularly the hyperconjugative effect. In other words, larger residues on the α-carbon lead to more stable N–Cl–AAs.⁷¹ For Tyr and Arg, the residue oxidation rates are predominantly determined by the functional groups within their residues (phenolic group in Tyr, guanidino group in Arg), not the amino group. This has been confirmed by the fact that oxidation rate constants we determined for Tyr and Arg are comparable to those of the AA residues (<10% error) determined by N-acetylated amino acids (R–NHCOCH₃), as the residues are the primary reactive sites.^65–67 Furthermore, How et al.⁷² reported products with chlorine substitution on the Tyr aromatic ring providing further evidence that oxidation primarily targets the phenolic group in Tyr.

The roles of pH and Cl⁻ were assessed by chlorination (initial FAC concentration, [FAC]₀ = 28 μM) of AAs (Tyr, Arg, Ala, and Leu, 4 μM each, respectively) at 25 °C, maintaining a constant ionic strength of 30 ± 1 mM and pH, as shown in Figure 2.

Figure 2.

Role of pH and chloride ions on AA chlorination kinetics. Apparent second-order-rate constants, k_app (M⁻¹ s⁻¹), obtained based on eq 3 at 20 °C, [FAC]₀ = 28 μM (containing 0.1 mM Cl⁻). AAs = 4 μM each. The shadow shows 95% confidence band, dotted line shows predicted k_app using acquired constants in Table 1, as Section S4.

Tested AAs showed the following reaction rates at pH 8.4: Tyr (50.5 M⁻¹ s⁻¹) > Arg (21.4 M⁻¹ s⁻¹) > Ala (12.4 M⁻¹ s⁻¹) > Leu (10.6 M⁻¹ s⁻¹). At pH 8.4, chlorination rates of Ala, Leu, and Arg were lower than those at pH 7.4, which aligns with lower fractions of the conjugate acid (HOCl) in total FAC (f_{HOCl,pH 7.4} = 58%, f_{HOCl, pH 8.4} = 12%), as HOCl is a stronger oxidant due to its higher electrophilicity than OCl⁻.^73–75 Conversely, Tyr exhibited a greater reactivity at pH 8.4. This indicates that pH dependence of AA chlorination is determined by ionizability of both AA residues and HOCl at different pH, as shown in Figures S7 and S8. Among tested AAs, Arg and Tyr have ionizable side-chain moieties: (1) Arg (pK_a = 12.5) remains positively charged at pH 7.4 and 8.4; (2) Tyr (pK_a = 9.9) has nearly constant fractions of Tyr-OH (99% and 97%, respectively); and (3) as an initial byproduct from chlorine substitution on the Tyr phenol ring, 3-Cl-Tyr pK_a = 8.5) has a smaller fraction of ArOH (55%) at pH 8.4 than at pH 7.4 (92%). According to Lau et al.³² and Gallard and von Gunten,⁷⁶ phenolate anions (ArO⁻, Cl–ArO⁻) readily react with HOC1 compared to phenol (ArOH, Cl–ArOH), exhibiting 3–5 orders of magnitude larger rate constants because the −O⁻ substituent makes the phenol ring more activated for electrophilic substitution compared to −OH.⁷⁷ This explains the increased reactivity of the Tyr residue and chlorine substituted intermediates (3-Cl-Tyr) at pH 8.4 compared to that at pH 7.4, despite the lower fraction of the more reactive HOC1.

Subsequent experiments evaluated the impact of Cl⁻ at concentrations typical for DW and extreme cases in FSS on the chlorination of tested AAs (Figure 2). At pH 7.4, all four AA kinetic constants increased linearly with increasing Cl⁻ concentration and were correlated (Pearson’s r > 0.99, p-value < 0.01). In contrast, no appreciable increase was observed at pH 8.4. With the addition of 15 mM Cl⁻ at pH 7.4, the rate constants significantly increased 2–5 fold with estimated 13.7 pM Cl_2(aq), while at pH 8.4, minimal impacts were observed with 0.3 pM Cl_2(aq) (Figure 1B). By analyzing the apparent second-order rate constant changes with the relevant Cl_2(aq) concentrations at tested Cl⁻ dosages, the best-fit estimates of the second order rate constants between AAs (Tyr, Arg, Ala, and Leu) and Cl_2(aq) were obtained for the first time (Table 1) and were validated by predicting the concentration of AAs as a function of time at pH 8.4 (Figure 2) and 7, 12 mM [Cl⁻] at pH 7.4 (Figure S9). These values were approximately 5 orders of magnitude greater than those of HOC1. Similarly, Lau et al.³² reported that phenolic moieties (ArOH, ArO⁻, Cl–ArO⁻) readily react with Cl_2(aq), displaying 5–6 orders of magnitude larger rate constants compared to HOCl. These results emphasize the underestimated significance of Cl_2(aq) in the chlorination of amino moieties, even at neutral pH (7.4). In the presence of 7 mM Cl⁻ (the 7 mM chloride is equivalent to 250 mg/L as suggested by the nonmandatory standard from national secondary drinking water regulations, USEPA⁷⁸), the generated Cl_2(aq) enhanced the AA chlorination rate constants 1.4–2.6 fold, supporting the notion that increased Cl_2(aq) formation is responsible for the elevated rate of chlorination. This aligns with the findings of Rose et al.³⁷ regarding chlorine degradation of electron-rich contaminants.

3.3. Reactivity of MCs in UV/Chlorine + Cl⁻.

In the absence of Cl⁻, a faster degradation was observed for the reaction of MC-RR with UV/chlorine (Figure S6). Under the experimental conditions (1.0 μM MC, 21 μM FAC, and 0.1 mW cm⁻² UV_{254 nm} fluence), the degradation kinetics of UV/chlorine treatment (0.21 min⁻¹) exceeded the sum of UV (0.01 min⁻¹) and chlorination (0.09 min⁻¹) individually. With Cl⁻ increasing from 5 to 15 mM, the rate constants increased linearly for both chlorination and UV/chlorine treatment (Figure 3). This suggests two key points, (1) Similar to the susceptibility of the four tested AAs to Cl_2(aq), MC-RR exhibited substantial reactivity with Cl_2(aq); (2) the similarity in rate constants implies that Cl⁻ enhanced MC-RR degradation to the same extent in both UV/chlorine treatment and chlorination. The broader confidence interval in the UV/chlorine-related kinetics suggests that potential Cl⁻ derived radical species contributed to the degradation, while determining whether Cl_2(aq) is actively involved requires further investigations.

Figure 3.

Role of chloride on MC-RR degradation kinetics in chlorination and UV/chlorine. Conditions: 6–12 mM phosphate buffer (pH 7.4) with a fixed ionic strength of 30 ± 1 mM, [MC-RR]₀ = 1.0 μM, [FAC]₀ = 21 μM containing 0.1 mM Cl⁻, UV_{254 nm} fluence rate = 0.1 mW/cm⁻². UV dose: 12–96 mJ/cm².

The presence of 7 mM Cl⁻ enhanced the chlorination efficiency in tested conditions, which lower the chlorine demand for 1.0-log (90%) degradation of MC-RR significantly (Table S4). (1) For chlorination, the required chlorination C_t is lowered from 39 to 22.5 mg min/L; and (2) for UV/chlorine treatment, the required C_t and UV dose can be reduced from 18 mg min/L with 72 mJ/cm² to 13.5 mg min/L with 54 mJ/cm². By integrating UV and chlorination within the practical range of UV doses (40–150 mJ/cm²) and C_t values (2–30 mg min/L) for drinking water disinfection purposes, MC degradation can be significantly enhanced, especially with abundant Cl⁻ in the water.

Both the presence of Cl⁻ and UV irradiation significantly enhance the chlorination of all MCs (MC-LR, -RR, -LA, and -YR) with various AA moieties (Leu, Arg, Tyr, and Ala). The MC chlorine reactivities follow the order of -YR (Tyr, Arg) > -RR (Arg, Arg) > -LR (Leu, Arg) > -LA (Leu, Ala), which corresponds to a reactivity order of AA residues (Tyr > Arg > Leu ~ Ala). The higher reactivity of MC-YR at pH 8.4 compared to 7.4 is attributed to the increased ionization of the Tyr phenol group, supported by the presence of chlorotyrosine and identical pH dependency of Tyr.⁷⁹ The enhancement of chlorination and UV/chlorine processes by increasing levels of Cl⁻ suggests the potential to utilize Cl⁻ to improve the treatment efficiency of MCs. Given the complex nature of water matrices, optimized conditions for each application scenario require further study.

3.4. MC Degradation in Field Samples by UV/Chlorine + Cl⁻.

To scrutinize the role of the water matrix of HAB-impacted waters on MC degradation in UV/chlorine treatment, four mixed MCs with 100 μg/L each (similar to MC levels in HAB events) were spiked and degraded in HABs-impacted and nonimpacted source waters. The experimental matrices included drinking water collected after GAC filtration from Grand Lakes St. Marys (GLSM) and Greater Cincinnati Water Works (GCWW), as well as two sets of extracted dissolved organic matter from HAB-affected and unaffected source waters, referred to as algal organic matter (AOM) and natural organic matter (NOM), respectively (Table S2). To maintain a FAC residual of at least 0.2 mg/L in the distribution system, [FAC] at the clear wells are usually maintained at 1.0–2.0 mg/L. To mimic the real treatment scenario in HAB events, FAC was normalized to 1.45 (±0.05) mg/L for all tested water matrices before spiking MCs. For comparison, the removal rates were compared both at same chlorine loss (Figure 4A,B) and at the same reaction time (Figure S10). As shown in Figure 4A,B, reaction times were controlled to normalize FAC losses (FAC depletion experiments as Section S4).

Figure 4.

Degradation of mixed MCs in HAB impacted water. Log removal value = −log₁₀([MC]_t/[MC]₀). Mean ± standard error, n = 3. [MC-YR]₀ = [MC-RR]₀ = [MC-LR]₀ = [MC-LA]₀ = 0.1 μM. 0 or 5 mM [Cl⁻], UV_{254 mn} fluence rate = 0.1 mJ/cm².

The FAC loss rates were greater when chlorination was combined with UV irradiation and Cl⁻. The removal rates of four mixed MCs (-LA, -LR, -RR, and -YR) differed from each other and followed the same tendency in the four tested processes and five water matrices: -YR > -RR > -LR > -LA. Through the competition reaction between the four MCs toward reactive oxidants, MCs with greater kinetic rate constants indicated greater removal, in accordance with the previous discussion regarding specific AA moieties, underscoring the significant role of AA moieties in chlorination and UV/chlorine processes.

With the same FAC consumption, no significant difference (p > 0.05) was observed inlog-removal of MCs in AOM, Grand Lakes St. Marys, NOM, and GCWW matrices among the four treatment processes (Figures 4A,B and S10A,B). Following UV irradiation, the reaction times needed to effect removal were shorter than those from chlorination alone. For example, achieving a 1.0-log removal of total MCs in the Grand Lakes St. Marys sample (with 0.6 mg/L FAC loss) took 12 min with chlorination, but only 8 min for UV/chlorine at a UV dose of 48 mJ/cm². This highlights that UV/chlorine can achieve greater MC removal efficiency without the need for additional chlorine dosage, even in matrices with more competition for the oxidant. The presence of 5 mM Cl⁻ has no significant effect in the presence of NOM, because the abundant phenols (>36 folds of [MCs])⁸⁰ and alkenes compete with active sites in the MCs (Figure S11).

With the same reaction time, UV/chlorine (72 mJ/cm²) degraded MCs to a greater extent than by chlorination. In the sample with relatively rich organic carbon content (2 mg/L C for NOM and AOM reactions), the enhancement of UV/chlorine was less remarkable than those with lower DOC, such as GCWW (0.9 mg/L) and Grand Lakes St. Marys (1.0 mg/L) samples. AOM has a larger content of peptides/AAs, resulting in more depletion of chlorine and RCS through the formation of organic chloramines. While NOM with richer aromatic composition exerts a greater hydroxyl radical demand.^81–83

The addition of 5 mM Cl⁻ enhanced total mixed MC degradation removal in chlorine and UV/chlorine treatment for GLSM (Figures S10 and S11). In contrast, no obvious enhancement was observed for pure water, AOM, NOM, or GCWW. The contradictory role of Cl⁻ here can be attributed to the following: (1) the combined chlorine in AOM and GLSM, including organic chloramines and chloramine, also contribute to chlorine and UV/chlorine because the extra chlorine dosage from unstable organic chloramines and the reactivity contributed by UV/chloramine process; (2) phenolic groups from Adda moieties in the MCs and NOM compete for reactive species; (3) HO^● dominates the reaction over Cl_2(aq) in low HO^● demand water. The role of Cl_2(aq) in chlorination with the existing NOM and AOM requires closer investigation.

Overall, UV/chlorine irradiation enhanced the degradation of MCs in all tested waters. Even with the same chlorine loss, integration with UV irradiation shortened the reaction time needed to achieve the same removal. The presence of Cl⁻ improves the efficiency of MC removal by chlorine and UV/chlorine treatment in HAB-impacted waters.

3.5. Disinfection Byproducts.

While the kinetic experiments demonstrate promising efficiency for MC degradation using UV/chlorine, the generation of halogenated DBPs is a major concern,⁸⁴ as chlorine radicals formed in this process could produce halogenated DBPs. To capture a more complete picture of halogenated DBPs, we measured 66 DBPs, along with total organic halogen, which includes larger molecular weight DBPs, beyond the measurement of lower molecular weight, 1- and 2-carbon DBPs.⁸⁵

Results show that the treatment of MC-LR with chlorine or UV/chlorine produces chlorinated DBPs. In total, 11 DBPs were found above detection limits with concentrations ranging from 0.1 to 4.6 μg/L (Table 2). Overall, samples treated with UV/chlorine produced slightly lower DBP concentrations to chlorine. For example, total quantified DBPs were 14.3 μg/L for UV/chlorine reactions of MC-LR but were 15.5 μg/L for corresponding chlorine reactions. It was somewhat surprising that these levels were so similar. This was consistent with the TOC1 results discussed below (Table 2).

Table 2:

Disinfection Byproducts (DBPs) and Total Organic Halogen Formed from the Treatment of MC-LR^a

DBP	chlorine	chlorine + Cl−	UV/chlorine	UV/chlorine + Cl−	no treatment
trichloromethane	1.7	1.4	ND	0.8	ND
trichloroacetaldehyde	ND	0.9	0.9	0.9	ND
dichloroacetonitrile	<0.75	<0.75	1.3	<0.75	ND
chloroacetonitrile	3.2	3.6	3.3	3.5	ND
1,1-dichloropropanone	1.0	<0.25	1.0	0.9	ND
chloropropanone	2.2	4.6	2.2	2.1	ND
trichloronitromethane	0.9	0.9	0.9	0.9	ND
dichloroacetamide	2.6	0.9	1.2	0.9	ND
trichloroacetamide	3.2	3.7	3.3	3.2	ND
chloroacetic acid	0.1	0.1	0.1	0.1	ND
dichloroacetic acid	0.5	0.4	0.1	0.3	ND
trichloroacetic acid	<0.1	<0.1	<0.1	<0.1	ND
total DBPs	15.5	16.4	14.3	13.6	ND
TOC1	47.2	51.2	33.4	39.7	ND
TOBr	ND	ND	ND	ND	ND
TOI	ND	ND	ND	ND	ND

^aND: not detected. DBPs that were not present in any of the sample sets are not listed. Results are the mean of 3 replicates. [FAC] = 1.5 mg/L, [Cl⁻] = 5 mM, UV₂₅₄ dose 70 mJ/cm². TOC1 was detected in degradation experiments of MC-LR (1 mg/L) in four tested processes in pure water, with the following trend: Cl₂ + Cl⁻ (51.2 μg/L) > Cl₂ (47.2 μg/L) > UV/chlorine + Cl⁻ (39.7 μg/L) > UV/chlorine (33.4 μg/L). Thus, when integrating chlorine with UV, lower total concentrations of chlorinated DBPs (TOC1) were produced. No TOBr or TOI was detected (as expected), since bromide and iodide were not added to reactions.

Nontarget LC-QTOF analysis of MC-LR treated with chlorine and Cl⁻ revealed a new DBP that, to the best of our knowledge, has not been reported previously. Its high-resolution accurate mass corresponds to the molecular formula C₄₄H₇₇N₆O₆Cl (observed m/z 821.5668; theoretical m/z 821.5660). The relative isotopic abundance of m/z 821 to m/z 823 is 3:1, indicative of a compound containing 1 chlorine atom, and fragment ions m/z 527 and 233 could be assigned to [C₂₈H₄₁N₅O₅ + H]⁺ and [C₆H₉N₄O₆ + H]⁺, respectively.

Additionally, nontarget analysis revealed evidence of chlorine and hydroxyl group addition and substitution to MC structures. Accurate mass data allowed for 35 unique molecular formula assignments supported by isotopic patterns (including one-chlorine, two-chlorine, and three-chlorine patterns, as well as one-hydroxyl and two-hydroxyl patterns), as shown in Figure 5. Hydroxylated DBPs were more prevalent in the UV/chlorine reactions than chlorine, indicating the susceptibility of MCs to HO^●. Meanwhile, the lower abundance of chlorinated MC-DBPs with UV/chlorine treatment suggests dominance of reactive oxygen species, or efficient decomposition of chlorinated MCs through direct UV photolysis of the C–Cl bond.^27,86 In addition, a variety of 1- and 2-chlorine-substituted MC-YR DBPs were abundantly formed in chlorination, which supports the role of 3-Cl-Tyr as an active intermediate in MC-YR chlorination. Among these transformed MCs, some retain the toxic Adda residue, which is essential in protein phosphatase inhibition⁸⁷ and some form Cl-Tyr-containing DBPs, which feature toxic potencies.⁷⁹

Figure 5.

Heatmap of MC-DBPs with no cleavage of Adda chain. Representing the relative abundance of the MC-DBPs ranging from light red to dark red.

3.6. Cytotoxicity.

As shown in Figure 6A, no significant difference in cell viability was observed between untreated MC (0.32 and 0.10 mg/L) and UV/chlorine-treated MCs at the two tested concentrations. This suggests that the DBPs from UV/chlorine-treated MC-LR and MC-LA at 0.5- and 1.0-log removal values had no appreciable impact on the cytotoxicity. Notably, cell viabilities in MC-LA treated groups were significantly lower compared to those in MC-LR treated groups at all three concentrations tested. This indicates that the same mass concentration of MC-LA exerts greater cytotoxicity than MC-LR, which is consistent with a previous animal study of oral administration to mouse by Chernoff et al.⁸⁸

Figure 6.

Cytotoxicity evolvement during different treatments (HepaRG cells). (A) Comparing untreated and UV/chlorine treated MCs at same concentration. Log removal value (LRV) = −log₁₀([MC]_t/[MC]₀). (B) Comparison of detoxification efficiency for chlorination (showed as Cl₂), UV/chlorine (showed as UV/C1₂), chlorination + Cl⁻, UV/chlorine + Cl⁻, additional [Cl⁻] = 5 mM. NTC: no treatment control. General experimental conditions: autoclaved Milli-QH₂O, [FAC]₀ = 21 μM with 0.1 mM Cl⁻, ascorbic acid (1.3 fold of [FAC]₀, 28 μM), UV_{254 nm} fluence rate = 0.1 mJ/cm².

To further compare the detoxification efficiencies in the tested processes, 1 mg/L MC-LR and MC-LA were degraded for 12 min in four processes, respectively. Figure 6B indicates the detoxification efficiency as follows: UV/chlorine + Cl⁻ ~ UV/chlorine > chlorine + Cl⁻ > chlorine. For 1 mg/L MC-LR, cell viability increased from 57% to 85% when chlorination was integrated with UV. No significant difference in cell viability was observed between the control group and the samples after 12 min of treatment with UV/chlorine (with or without Cl⁻), indicating that UV/chlorine detoxified almost all of the 1 mg/L MC-LR in the tested condition. For 1 mg/L MC-LA, cell viability increased from 40% to 65% when chlorination was integrated with UV. The detoxification extent of MC-LR and MC-LA aligns with their degradation pattern: MC-LA is more toxic and more resistant to tested processes than MC-LR Further, results suggest that the addition of 5 mM Cl⁻ to chlorine treatment significantly enhances the cell viability in both MC-LR and -LA treated groups.

Interestingly, the cell viability of samples after UV/chlorine treatment appears to be slightly greater (but not significant, Figure S12) than after chlorination + Cl⁻. On the contrary, the residual MC concentrations after UV/chlorine treatment were not lower than after chlorination + Cl⁻, which emphasizes the potential role of chlorinated DBPs in cytotoxicity. Overall, when chlorine was integrated with UV, lower cytotoxicity was observed. However, the presence of Cl⁻ resulted in higher levels of chlorinated DBPs and higher cytotoxicity. These results indicate the greater detoxification potential of UV/chlorine compared to chlorination, with UV irradiation facilitating the decomposition of chlorinated MCs and other organic chloramines.⁸⁹

4. IMPLICATIONS

Unprecedented HABs ring the bell for protecting drinking water safety against cyanotoxins. Conventional physical treatment processes have limited removal efficiencies for soluble MCs and have the potential to act as under-appreciated MC sources by releasing accumulated MCs during maintenance. Chlorination, the most common oxidation practice, has inadequate removal efficiency of MCs at a practical exposure (C_t) range. On the other hand, this study demonstrates that UV/chlorine would significantly enhance degradation of MCs at a working pH range, with lower levels of toxic byproducts formed at practical ranges of C_t (2–30 mg min/L) and UV dose (40–150 mJ/cm²). Our results suggest the integration of UV irradiation and chlorination offers an efficient strategy for the treatment of MCs in drinking water with low chemical demand and energy consumption. However, realistic challenges remain, such as the short hydraulic residence time in typical UV facilities, which may limit the overall treatment efficiency.

In addition, freshwater salinization indirectly aggravates HABs, especially in drought areas and small lakes. This study discovered the interrelationship between an elevated chloride concentration and chlorine chemistry in the case of freshwater salinization concurrent with extreme HAB events. Cl_2(aq), the underestimated chlorinating agent formed by the reverse disproportionation reaction, plays a significant role in the chlorination of amino acids and MCs (heptapeptides), with second-order rate constants reported here for the first time as 4.4 (±1.0) × 10⁶ M⁻¹ s⁻¹ and 8.5 (±2.8) × 10⁷ M⁻¹ s⁻¹, respectively.

In terms of the four most ubiquitous MCs tested in this study, their reactivities follow the trend: MC-YR (with phenolic amino acid Tyr) > MC-RR (with charged amino acid Arg) > MC-LR > MC-LA (with aliphatic amino acids Leu and Ala). The same order of reactivity was observed in competition reactions in water, DOM solutions, and field water samples, which emphasizes the significant role of the amino acid moiety–reactivity relationship. Noteworthily, MC-LA appears to be more toxic and less susceptible to oxidation than MC-LR, -RR, and -YR, and at the same time, its dominance has been reported in small lakes, especially on the West Coast and Midwest of the US. This reveals the importance to prioritize MC-LA in terms of MC control, as well as the potential to utilize MC-LA as an Adda probe to understand MC degradation and detoxification. This study provides a comprehensive assessment of utilizing UV/chlorine as a final barrier against HAB events, especially in dealing with MCs while limiting unintended toxic consequences.