Harmful Algal Bloom Toxins in Aerosol Generated from Inland Lake Water

Abstract

Harmful algal blooms (HABs) caused by cyanobacteria in freshwater environments produce toxins (e.g., microcystin) that are harmful to human and animal health. HAB frequency and intensity are increasing with greater nutrient runoff and a warming climate. Lake spray aerosol (LSA) released from freshwater lakes has been identified on lakeshores and after transport inland, including from lakes with HABs, but little is known about the potential for HAB toxins to be incorporated into LSA. In this study, freshwater samples were collected from two lakes in Michigan: Mona Lake during a severe HAB with microcystin concentrations (>200 μg/L) well above the Environmental Protection Agency (EPA) recommended “do not drink” level (1.6 μg/L) and Muskegon Lake without a HAB (<1 μg/L microcystin). Microcystin toxins were identified in freshwater, as well as aerosol particles generated in the laboratory from Mona Lake water by liquid chromatography–tandem mass spectrometry (LC–MS/MS) at atmospheric concentrations up to 50 ± 20 ng/m³. Enrichment of hydrophobic microcystin congeners (e.g., microcystin-LR) was observed in aerosol particles relative to bulk freshwater, while enrichment of hydrophilic microcystin (e.g., microcystin-RR) was lower. As HABs increase in a warming climate, understanding and quantifying the emissions of toxins into the atmosphere is crucial for evaluating the health consequences of HABs.

Graphical Abstract

INTRODUCTION

The frequency and intensity of harmful algal blooms (HABs) from cyanobacteria in freshwater are increasing globally due to increased anthropogenic nutrient loading.^1–4 Warmer global temperatures are also expected to increase HAB frequency and intensity due to longer growing seasons and amplified algal growth rates.^5–14 Intracellular and extracellular algal toxins from cyanobacteria pose a threat to human and animal health.^16,17 In 2014, a HAB with high concentrations of microcystin in Lake Erie infiltrated the public drinking water treatment facility for the city of Toledo, Ohio, causing a do not drink posting for >400 000 people due to toxin levels unsafe for ingestion.^18,19 The immediate impact of HAB toxins on human health necessitates a deeper scientific understanding of their chemical structure and mechanisms of transport. Variations in HAB toxin concentrations and the relative amounts of different toxin congeners (i.e., chemical structures) have been observed that are independent of overall HAB biomass concentrations.^20–23 The physical size of a HAB does not correlate with the concentration of toxins produced, which creates difficulty in predicting and mitigating HAB-related health impacts.

Many genera of cyanobacteria, commonly known as blue-green algae (BGA, e.g., Microcystis), and associated cyanotoxins (e.g., microcystin) have been identified within freshwater HABs, but there is limited understanding of their impact on human health.^24,25 Microcystin is a class of cyclic heptapeptides named by the combination of amino acids in their structure,²⁶ with over 200 microcystin congeners identified to date.^27–31 However, despite differences in the acute toxicities of different congeners, there is a minimal overall understanding of HAB toxicity.^32,33 Laboratory studies examining the toxicity of microcystin-containing aerosols administered to mice have shown 10 times higher sensitivity to inhaled microcystin compared to orally ingested microcystin.^34,35 Therefore, toxic effects from exposure to aerosolized microcystin likely occur at lower doses than that for microcystin ingestion,³⁶ on which the Environmental Protection Agency (EPA) recommendations are based. This raises concerns regarding unexpected exposure that could occur for populations living near or downwind of HABs, for occupations that interact with HABs, or for recreational users on HAB-impacted water. Studies have observed microscopic lesions in both the nasal cavity³⁶ and liver³⁵ of mice after microcystin inhalation. It is believed that the microscopic lesions observed in the nasal cavity from microcystin inhalation enhanced absorption into the blood-stream leading to systemic impacts.³⁶ The diversity of cyanobacteria and associated toxins, as well as our limited knowledge of exposure routes and acute toxicities, necessitate further study as this has the potential to negatively impact populations living and working near HABs globally.³⁷

Lake spray aerosol (LSA) is produced by freshwater wave breaking and bubble bursting,³⁸ similar to the production of sea spray aerosol (SSA) from wave breaking in marine environments.^39–42 LSA from freshwater lakes have been observed along lakeshores on the ground,⁴³ after transport inland,⁴⁴ when lofted to cloud heights,^45,46 and incorporated into cloud water.⁴⁶ Modeling of LSA concentrations shows that freshwater aerosol production contributes significantly to particle number concentrations over the Great Lakes region.⁴⁷ LSA is generated in a size range important for inhalation exposure with modes in the ultrafine (30–80 nm) and accumulation (200–300 nm) size ranges.^38,45 LSA has a distinct chemical composition in comparison to SSA^41,48 and is composed primarily of calcium carbonate and organic carbon.^38,43 Recently, May et al.⁴⁹ showed that LSA generated from HABs contained greater organic and biological materials than LSA generated during non-HAB conditions, demonstrating the incorporation of the biological material from HABs into LSA. However, little is known regarding the incorporation of toxins from freshwater HABs into LSA in the Great Lakes region.

The few studies of aerosolized HAB toxins in freshwater environments include Backer et al., which observed microcystin in ambient LSA emitted from small, inland lakes in Michigan^50,51 and California.⁵² Wood et al.⁵³ detected microcystin-containing aerosols at a freshwater site 20 m from the shore and 30 m high, demonstrating that freshwater aerosol can be transported aloft and inland. Cheng et al.⁵⁴ identified microcystin-containing droplets produced from freshwater bubble bursting in the laboratory. These studies confirmed that freshwater toxins can become aerosolized, however they all used enzyme-linked immunosorbent assays (ELISA) that are unable to distinguish between the congeners of aerosolized microcystin.^55,56 Additional questions remain about the relationship between freshwater and aerosolized toxin concentrations. Further studies are needed to assess aerosolized toxin congeners with respect to toxin concentrations found in bulk water. Toxin incorporation into aerosol is not expected to be uniform at the particle-to-particle level, which makes understanding its presence as a function of size and at the single-particle level (i.e., toxins as a function of the aerosol mixing state)^57–59 important for distinguishing the health implications in the Great Lakes region or other regions with HABs (e.g., Florida).^{44,46,60–63} Characterizing the aerosolization of HAB toxins in freshwater environments is crucial for understanding the impact of toxin inhalation on public health.

Given the limited information on freshwater HAB toxin aerosolization, it is useful to consider what is known about aerosolization of toxins in oceanic environments from red tides and other marine blooms. Studies examining toxin aerosolization in oceanic blooms by wave breaking^64–69 and recreational activities^70,71 show adverse respiratory effects after as little as 1 h exposure.^68,71 Exposure to toxin-containing aerosols particularly impacts people with pre-existing breathing diseases, such as asthma.^69,72,73 Gambaro et al.⁷⁴ detected microcystin in SSA particles generated from seawater artificially spiked with microcystin, showing algal toxin transfer from seawater to aerosol particles. Similarly, the organic material increases in SSA during blooms,^41,48,75 impacting water uptake,⁷⁶ incorporating metals,⁷⁷ and even emitting whole, intact bacterial cells.⁷⁸ Blanchard et al.⁷⁹ discovered elevated bacterial cells in aerosol particles compared to bulk water concentrations due to aerosolization of the surface-active organics present in the sea surface microlayer.⁸⁰ Brevetoxins produced from red tides contain hydrophobic functional groups⁸¹ that impact the toxin type and potency,⁸² but further information is needed regarding the impact of hydrophobic structures on the aerosolization efficiency of HAB toxins. Pierce et al.⁶⁷ observed marine toxin aerosolization up to one mile inland, suggesting inland transport of aerosolized toxins in addition to previously reported inland transport of SSA.^44,83 Toxins in atmospheric particles were transported many kilometers inland without degradation,⁵³ consistent with the stability of microcystin under a range of chemical and physical conditions.^84,85 The well-established aerosolization, transport, and health consequences of toxins from marine blooms highlight the need to study these properties in freshwater environments.

In this study, freshwater was collected in Michigan from Mona Lake during a HAB with high microcystin concentrations and Muskegon Lake in which the microcystin concentrations were below EPA-recommended levels. Freshwater samples were analyzed for the presence of BGA and microcystin toxins, after which LSA was generated using established methods in the laboratory³⁸ to gain an understanding of fundamental emission processes of nascent LSA in a controlled setting.⁴⁰ Microcystin present in the freshwater samples was also detected in aerosol samples. However, microcystin congeners were not transferred from freshwater to the aerosol phase uniformly, leading to greater enrichment of hydrophobic microcystin congeners in aerosol particle samples. Freshwater samples were also analyzed for insoluble organic particles using nanoparticle tracking analysis (NTA).⁸⁶ A relationship between increasing freshwater organic particle concentrations and higher aerosol number concentrations was observed. Overall, this study demonstrates the emission of microcystin within particles and suggests that the relative amounts of toxins present in the aerosol phase are distinctly different than those present in the water column. These key findings improve the currently limited understanding of freshwater HAB toxins in size-resolved aerosols, which will ultimately lead to a better understanding of this potential route of exposure for HAB toxins.

MATERIALS AND METHODS

Freshwater Sample Collection and Aerosol Generation.

Freshwater samples were collected from the surface of Mona Lake (43.1856, −86.2360) on July 11, 2018 and September 4, 2018 and from Muskegon Lake (43.2325, −86.2677) on June 11, 2015 and October 25, 2015 (Figure 1A). Freshwater samples were stored in 8 L carboy LDPE containers (United States Plastic Corp) and frozen (−20 °C) prior to analysis. After thawing, freshwater samples were then used to generate aerosol particles using the method described by May et al.³⁸ Briefly, 4 L aliquots of freshwater were cycled through four plunging jets at 2 L/min, creating bubbles that burst at the air–water interface to generate aerosol particles. Particles were sampled from the headspace of the tank after passing through two silica gel diffusion driers to achieve a relative humidity (RH) of ~15%, a standard established for SSA.⁴² The tank was kept at room temperature (23 ± 1 °C) for all experiments. Particle-free air (Pall, HEPA Capsule Filter) was cycled through the LSA generator to test for leaks before LSA generation. Background particle concentrations were minimal (<20 particles/cm³) compared to the particle concentrations generated from freshwater samples (~500–1500 particles/cm³). The aerosol number size distributions for each LSA sample were measured by a scanning mobility particle sizer (SMPS) consisting of a differential mobility analyzer (DMA, TSI Inc., model 3082) and a condensation particle counter (CPC, TSI Inc., model 3755) for particles ranging from 14.1 to 736.5 nm diameter. Particles with diameters from 0.52 to 19.8 μm were measured by an aerodynamic particle sizer (APS, TSI Inc., model 3321).

Figure 1.

(A) Map of Mona and Muskegon Lakes located in Western Michigan, with the inset showing the location of inland lakes relative to the surrounding Great Lakes (Map images from NASA Worldview). (B) Total microcystin and phycocyanin (blue-green algae, BGA) concentrations for each freshwater sample. The inset shows microcystin for Muskegon Lake samples. The dashed red line shows EPA drinking water advisory of 1.6 μg/L total microcystin.

Bulk Measurement Techniques and Toxin Characterization.

A spectrophotometer (AquaFluor 8000, Turner Designs) measured phycocyanin fluorescence of freshwater samples, serving as an indicator of BGA.⁸⁷ ELISA kits (Thermo Fisher Scientific) measured total microcystin concentrations in freshwater samples. For comparison, 12 microcystin congeners were also measured using the liquid chromatography triple quadrupole mass spectrometry (LC–MS/MS) method developed by Birbeck et al.⁸⁸ Laboratory-generated LSA was impacted onto glass fiber filters (Whatman, grade GF/c, 47 mm) using a cyclone (URG Crop., model 2000–30ED) and single-stage impactor (URG Corp., model 2000–30 FV) that operated at 3 L/min and impacted particles <2.5 μm aerodynamic diameter (d_a). Impacted particles were extracted following the method described by Wood et al.⁵³ Filters and 5 mL of 100% methanol were placed into 50 mL beakers covered by Parafilm. Samples were sonicated for 30 min, after which the supernatant of each sample was placed into a glass vial. The extraction procedure was repeated for a total of three times. The supernatants from each filter were combined, dried under nitrogen, and solubilized in LC/MS grade water before LC–MS/MS analysis using a Thermo TSQ Quantiva triple quadrupole MS (Thermo Scientific) equipped with an online concentrating column (Thermo Scientific Hypersil GOLD aQ 2.1 × 20 mm, 12 μm).⁸⁸ The extract from three filters, each representing 60 min LSA experiments, was analyzed by LC–MS/MS, as well as aliquots from each corresponding freshwater sample. The mobile phases consisting of 0.1% formic acid in water and 0.1% formic acid in acetonitrile separated the microcystin congeners present using gradient analysis with a C18 column (Thermo Accucore aQ, 50 ×2.1 mm, 2.6 μm) and then detected using selective reaction monitoring (SRM) in positive electrospray ionization (ESI) mode. Retention times and concentrations of each measured microcystin congener were determined by calibration using commercially available standards.

Single-Particle Analysis.

An aerosol time-of-flight mass spectrometer (ATOFMS) measured the size and chemical composition of individual LSA particles ranging from 0.1 to 1.5 μm.⁸⁹ Briefly, particles entered the instrument and were focused into a narrow particle beam by passing through an aerodynamic focusing lens. Particle sizes were determined by measuring the time it takes an individual particle to pass through two continuous wave lasers (wavelengths of 405 and 488 nm, respectively) separated by 6 cm. Particle d_a was determined after calibration using polystyrene latex spheres of known diameters ranging from 0.1 to 1.5 μm. Individual particles were desorbed and ionized by a 266 nm Nd:YAG laser upon entering the mass spectrometer source region, generating positive and negative ions for individual particles. Ions were detected using a dual polarity time-of-flight mass spectrometer. A total of 5277 particles were chemically analyzed by ATOFMS from the Mona Lake—July sample, 2332 particles from the Mona Lake—September sample, 1616 particles from the Muskegon Lake—June sample, and 1007 particles from the Muskegon Lake—October sample. Single-particle mass spectra were analyzed using the MATLAB toolkit FATES (a flexible analysis toolkit for the exploration of single-particle mass spectrometry data).⁹⁰ Mass spectral peak assignments correspond to ions that were identified from previous studies.^{40,43,44,49,59,77,91–97}

A three-stage microanalysis particle sampler (MPS-3, California Measurements, Inc.) impacted LSA particles onto Formvar-coated copper microscopy grids (Ted Pella Inc.) for scanning electron microscopy coupled to energy-dispersive X-ray spectroscopy (SEM–EDX) analysis. The MPS-3 operated at 2 L/min and impacted particles with diameters 2.5–5.0, 0.4–2.5, and <0.4 μm onto stages 1, 2, and 3, respectively. SEM analysis was performed using an FEI Helios 650 Nanolab Dualbeam electron microscope that operated at an accelerating voltage of 20.0 kV and a current of 0.40 nA. The Helios microscope utilized a high angle annular dark-field (HAADF) detector to provide contrast between areas of differing chemical composition.⁹⁸ EDX spectra were acquired for 20 s using an EDAX detector and GENESIS EDX software version 5.10 (EDAX Inc.). Raman microspectroscopy was performed on a Horiba LabRAM HR Evolution Raman spectrometer (Horiba Scientific) to analyze droplets of freshwater placed onto quartz substrates (Ted Pella Inc.). The Raman spectrometer contained a 50 mW 532 nm Nd:YAG laser, CCD detector, 600 gr/mm grating, and 100× objective on a confocal microscope (Olympus Life Science). Raman spectra were collected using 3 accumulations at 15 s acquisition times for the range 500–4000 cm^–1.

Analysis of Insoluble Residues.

Concentrations and number size distributions of the insoluble residues present in each freshwater sample were analyzed by NTA using a NanoSight LM10 (NanoSight Ltd.) traditionally applied to nanoparticle analysis,^58,99,100 following the method for environmental insoluble residues described in Axson et al.⁸⁶ NTA determines the size-resolved number concentration (i.e., size distribution) of individual particles in a liquid medium by illuminating them with a laser and monitoring the Brownian motion of particles to determine their two-dimensional (2D) displacement. The Stokes–Einstein equation is then used to determine the individual particle hydrodynamic diameter.^101,102 Aliquots of each freshwater sample were filtered prior to analysis using a 2 μm glass syringe filter before 500 μL was loaded into the LM10 cell housing a 405 nm laser. The freshwater sample flowed through the instrument using a syringe pump that operated at 60 rpm. Light scattering was measured using an sCMOS camera (Hamamatsu, Orca) coupled with a 20× objective microscope. Ten 60 s videos of each freshwater sample were analyzed using the NTA 3.2 (Build 60) software. The average of the ten videos generated an average size distribution in the range 10 nm to 1 μm for each freshwater sample. Droplets of Mona Lake—September freshwater were placed on Raman and SEM substrates following previously established methods,¹⁰³ and the insoluble residues were analyzed by SEM–EDX and Raman microspectroscopy after the droplet dried.^41,75 The insoluble residues were composed of particulate organic carbon (POC, Figure S1)^104,105 and will be referred to as POC.

RESULTS AND DISCUSSION

BGA and microcystin concentrations present in freshwater samples were analyzed prior to aerosol generation. Mona Lake—September had the highest phycocyanin/BGA (1166 μg/L) and microcystin concentrations (280 and 230 μg/L for ELISA and LC–MS/MS methods, respectively) of all samples (Figure 1B). For this reason, we focused on this sample for aerosol generation in subsequent laboratory experiments. Mona Lake—July had the second-highest concentrations of phycocyanin/BGA (192 μg/L) and microcystin (21 and 8 μg/L for ELISA and LC–MS/MS methods, respectively). Both Muskegon Lake samples had lower phycocyanin/BGA concentrations (93 and 86 μg/L for the June and October samples, respectively). Muskegon Lake also had lower microcystin concentrations of 0.3 and 0.8 μg/L (using ELISA) and 0.03 and 0.08 μg/L (using LC–MS/MS) for the June and October samples, respectively. The microcystin concentrations present in Muskegon Lake were below the EPA swimming advisory (8 μg/L total microcystin)¹⁰⁶ and drinking advisory (1.6 μg/L total microcystin) recommendations,¹⁰⁷ while Mona Lake samples were above for both guidelines. For all samples, ELISA kits predicted slightly higher amounts of algal toxins more than likely due to the simultaneous detection of toxins and their degradation products and differing responses of each congener to ELISA.^56,88,108 ELISA is not able to differentiate between microcystin congeners like the LC–MS/MS method used.⁸⁸

Aerosolized microcystin congeners from Mona Lake LSA were quantified with LC–MS/MS, as described in Birbeck et al.⁸⁸ Figure 2A shows the molecular structure of microcystin-LR (MC-LR), the congener with the highest concentration in the aerosol generated from Mona Lake (Figure 2B). Chromatograms of the retention times of each microcystin congener measured in Mona Lake LSA are shown in Figure 2B, with the chromatogram of microcystin congeners detected in Mona Lake freshwater shown in Figure S2. Retention times and structures for each congener are listed in Table S1⁸⁸ and illustrated in Figure S3. The commercially available microcystin congeners were previously optimized for selected quantitative and qualitative ion transitions using known fragmentation patterns from the literature.^109–112 The microcystin congeners were separated by gradient analysis HPLC and detected using SRM analysis on the triple quadrupole MS system. Figure 2C shows the mass spectral fragments detected for MC-LR, with fragmentation patterns illustrated in Figure 2A. Quantitative and qualitative fragment ions were observed at m/z +135.07 [Ph–CH₂–CH(OCH₃)]⁺, +155.08 [H + Mdha–Ala]⁺, +163.08 [C₁₁H₁₄O + H]⁺, and +212.97 [H + Glu–Mdha]⁺ and are known fragments of that microcystin molecule.^109–112

Figure 2.

(A) Structure and fragmentation of microcystin-LR, the microcystin congener with the highest concentration in Mona Lake LSA. (B) Chromatogram showing microcystin congeners present in Mona Lake LSA. The inset shows congeners present at lower concentrations. (C) Mass spectrum of microcystin-LR identified from Mona Lake LSA. All congener structures, retention times, and their abundance in the Mona Lake freshwater sample are provided in the Supporting Information.

The LC–MS/MS method allows for the quantification of each microcystin congener based on calibration with microcystin standards. The congeners detected and quantified in Mona Lake—September freshwater are shown in Figure 3A and listed in Table 1. Microcystin concentrations in LSA were determined by extracting LSA impacted onto filters and analyzing the extract with LC–MS/MS. Average aerosolized microcystin concentrations from three experiments are shown in Figure 3B and listed in Table 1. Of the eight toxins detected in freshwater, seven were also detected in aerosol particles. However, the relative concentrations of specific toxins in the water and aerosol phase differed substantially. In our experiments, MC-LR and MC-LA were more concentrated than other congeners in aerosol particles (Figure 3B). Enrichment calculations (details in the Supporting Information) show enrichment factors of 830 and 2000 for MC-LR and MC-LA, respectively, in the aerosol phase relative to the bulk freshwater (Table 1). MC-LR and MC-LA contain the amino acid leucine (Figures 2A and S3), a hydrophobic amino acid.^113–116 Conversely, the amino acid arginine is significantly more hydrophilic,^114–116 and the congener with this structure, MC-RR, was significantly less enriched in the aerosol relative to freshwater (enrichment factor of 10). Hydrophobic molecules have been shown to be enhanced in SSA particles relative to bulk seawater,^{40,80,117–122} which is believed to be due to hydrophobic and low solubility species partitioning to the air–water interface of bubbles as they rise through the water column. Octanol–water partitioning coefficients log-(K_ow) for MC-LR (1.67) and MC-RR (–0.71)¹⁵ support preferential transfer of MC-LR to the aerosol phase, while MC-RR remains primarily in the bulk freshwater. De Maagd et al.¹²³ suggested that the hydrophobicity of MC-LR changes as a function of pH, with higher pH values leading to lower log(K_ow) values. Freshwater sample pH ranged from 6.9 to 7.1 (Figure S4) though it is unclear what the role of pH is with respect to toxin emissions into the particle phase and future work is needed in this area. Though extracellular microcystin excreted from algal cells generally accounts for 20% of total toxins,²⁵ this percentage can increase drastically during bloom senescence, algaecide treatment, and lysis.¹⁸ The aerosolized microcystin detected was likely extracellular because the bubble-bursting process lyses cells, as noted by the observation of aerosolized bacterial cell components from marine algal blooms.⁷⁸ Additionally, the freshwater samples used within this manuscript were collected near bloom senescence and intact cells are unlikely to be aerosolized as the cell size of Microcystis (>4 μm)¹²⁴ is larger than the mass and number modes of the generated aerosol. Therefore, extracellular microcystin likely dominated the measurements.

Figure 3.

Quantification of microcystin congeners present in Mona Lake (A) freshwater and (B) LSA. Concentrations reported in μg/L refer to the mass of toxin per mass of water. Concentrations reported in ng/m³ refer to the mass of toxin per meter cubed of air measured. (C) Schematic showing the aerosolization of hydrophobic microcystin congeners.

Table 1.

Concentrations of Microcystin Congeners

microcystin congeners	microcystin in water (μg/L)	microcystin in aerosol (ng/m3)a	enrichment (aerosol/water)	octanol–water partitioning coefficients log(Kow)
D-Asp3-RR	1.2	not detected	N/A	N/A
MC-RR	114.8	0.7 ± 0.4	10	−0.71 ± 0.05
MC-YR	22.2	1.8 ± 0.5	140	N/A
MC-LR	76.4	40 ± 20	830	1.67 ± 0.09
D-Asp3-LR	1.6	0.8 ± 0.3	800	N/A
MC-HilR	2.0	1.4 ± 0.2	1200	N/A
MC-WR	3.8	0.8 ± 0.1	400	N/A
MC-LA	5.6	7 ± 3	2000	1.30 ± 0.06

^aError bars refer to the standard deviation of triplicate aerosol experiments. MC-HtyR, MC-LY, and MC-LW were below the limit of detection (5 × 10⁻³ μg/L). MC-LF and nodularin were not detected in the freshwater samples. Enrichment factors were calculated by dividing the toxin concentration in aerosol by the toxin concentration in freshwater (details are provided in the Supporting Information). Octanol–water partitioning coefficients log(K_ow) are provided by McCord et al.¹⁵

Toxin concentrations present in the aerosol samples were converted to ng/m³ for comparison to previously reported measurements, as mass per volume of air is the standard unit for particulate matter and specific particle-phase toxins. We detected 50 ± 20 ng/m³ of total aerosolized microcystin (Table 1). Backer et al.⁵¹ detected aerosolized microcystin concentrations up to 23 ng/m³ during ambient sampling on the shore of a small lake in Michigan. However, the freshwater microcystin concentrations in that study were significantly lower than those observed herein (5⁵¹ and 230 μg/L, respectively), which likely accounts for the difference in aerosolized toxin concentrations observed.

The size and concentration of LSA and insoluble residues in freshwater were analyzed to investigate trends between the properties of the organic material from the water and aerosol phases. Mona and Muskegon Lake LSA had a mode at 46 nm in the number size distribution (Figure 4A,B), similar to the Aitken mode observed for LSA generated from Lake Michigan freshwater (30–80 nm).^38,43,45 A larger LSA mode at 270 nm was also observed, corresponding to the accumulation mode previously identified in the laboratory^38,49 and field studies (200–300 nm).^43–46 POC present in Mona Lake—September freshwater as insoluble residues had a number size distribution mode at 155 nm and an average concentration of 1.15 (±0.04) × 10⁹ POC/mL (Figure 4A,C, respectively). Similar concentrations of 1.22 (±0.03) × 10⁹ POC/mL were observed for Mona Lake—July freshwater, with a slightly larger mode at 240 nm. POC concentrations for the Muskegon samples were much lower (3.62 ± 0.08 × 10⁸ and 3.32 ± 0.07 ×10⁸ POC/mL for June and October samples, respectively). These POC distributions correspond to mass concentrations of 68, 38, 6, and 7 μg/cm³ for Mona Lake—July, Mona Lake—September, Muskegon Lake—June, and Muskegon Lake—October samples, respectively. The average LSA number concentrations for Mona Lake—September were 1400 ± 200 particles/cm³ (Figure 4C). LSA number concentrations were lower for Mona Lake—July and both Muskegon Lake samples (260 ± 20 particles/cm³, 270 ± 20 particles/cm³, and 110 ± 6 particles/cm³, respectively). The increase in POC and microcystin for Mona Lake—September corresponded to an increase in overall aerosol production, particularly with enhanced aerosol production in the ultrafine size range (<100 nm), resembling the size of observed POC. Incorporation of insoluble residues into SSA has been observed¹²⁵ with aerosol size shifting to that of the insoluble residues,¹²⁶ similar to the results presented herein. A positive relationship between the organic content in water and increased algal growth has been observed,^127–129 demonstrating the importance of simultaneously measuring POC and microcystin in freshwater environments. This analysis is the first use of NTA to investigate the insoluble residues of freshwater in relationship to aerosol generation and shows that insoluble residues present in freshwater play a role in higher atmospheric aerosol concentrations, as well as impacting the size of particles that form.

Figure 4.

Average aerosol and POC number size distributions for (A) Mona and (B) Muskegon Lake samples. Microcystin concentrations for each sample are listed in parentheses. (C) Quantification of aerosol and POC number concentrations for each freshwater sample.

LSA analyzed by single-particle mass spectrometry (ATOFMS) was classified into three particle types following the criteria used in May et al.:⁴⁹ LSA composed primarily of salt (LSA salt), LSA with organic content (LSA organic), and LSA with the enhanced biological material (LSA biological). Average dual polarity mass spectra of all three particle types are shown in Figure S5 with the major ions labeled. Salt particles contained primarily inorganic ions, with the major ions present being m/z +40 [Ca⁺], +23 [Na⁺], and +24 [Mg⁺] (Figure S5).^43,44,49 LSA organic particles were classified by the presence of m/z +66 [CaCN⁺] and +82 [CaCNO⁺]⁴⁹ (Figure S5), peaks consistent with the presence of organic nitrogen.⁷⁷ Additional organic ions observed were m/z +74 [N(CH₃)₄⁺], −45 [CHOO⁻], −59 [CH₃COO⁻/HCNO₂⁻], and −71 [C₃H₃O₂⁻].⁹¹ LSA biological particles were classified by the presence of m/z −79 [PO₃^–]^49,77,92,93 and +89, the amino acid aspartic acid [Asp–CO₂⁺].¹³⁰

Difference mass spectra, calculated by subtracting the average mass spectra of particles in Muskegon Lake from the average spectra of particles in Mona Lake, highlight the compositional differences between particles generated from each lake (Figures 5A,B, and S6). Both biological markers were observed in higher concentrations in Mona Lake biological LSA (Figure 5B), while organic markers were observed in both Mona and Muskegon Lake LSA organic particle types (Figure 5A). The identification of salt, organic, and biological particles was verified by SEM–EDX analysis (Figure S7). Salt particles were identified with SEM–EDX by containing the elements Na, Mg, Ca, and Cl,⁴⁹ along with an amorphous morphology similar to previous analysis of ambient LSA collected on the shore⁴³ and via aircraft⁴⁶ over Lake Michigan. Organic particles were identified by SEM–EDX by containing only the elements C and O and a circular morphology.⁴⁹ Biological particles were classified by the presence of elemental P and an amorphous morphology similar to previously observed ambient biological particles.^49,131

Figure 5.

Difference spectra calculated by subtracting Muskegon Lake mass spectra from Mona Lake spectra of (A) LSA organic particles and (B) LSA biological particles.

Freshwater toxin concentrations and aerosol particle types were investigated for each freshwater sample. Percentages of microcystin congeners for each freshwater sample were calculated using the concentrations obtained by LC–MS/MS to compare relative toxin abundance for each lake. MC-RR was the toxin of the highest concentration in all samples, accounting for 48% (Mona Lake—July), 50% (Mona Lake—September and Muskegon Lake—October), and 67% (Muskegon Lake—June) of microcystin congeners observed (Figure 6A). MC-LR was the second-highest for each sample at 34% (Mona Lake—July and Mona Lake—September), 33% (Muskegon Lake—June), and 25% (Muskegon Lake—October). The remaining toxins detected were as follows: D-Asp³–RR (1% for both Mona Lake samples, 0% for both Muskegon Lake samples), MC-YR (8 and 9% of Mona Lake—July and—September samples, 0 and 12% of Muskegon Lake—June and –October), D-Asp³-LR (1% for both Mona Lake samples, 0% for both Muskegon Lake samples), MC-HilR (1% for both Mona Lake samples, 0% for both Muskegon Lake samples), MC-WR (2% for both Mona Lake samples, 0% for both Muskegon Lake samples), and MC-LA (5% for both Mona Lake—July, 2% for Mona Lake –September, 0% for Muskegon Lake—July, and 13% for Muskegon Lake—October). Particle types observed by ATOFMS were combined for each freshwater sample. The percentages of particles identified as LSA biological from Mona Lake were 10 and 14% for the July and September samples, respectively (Figure 6B). Biological particles generated from Muskegon Lake freshwater comprised 10 and 6% of all particles analyzed from the June and October samples, respectively. Organic particle types were also at higher percentages for Mona Lake LSA (80 and 75% for July and September, respectively) than those of Muskegon Lake (67 and 60% for June and October, respectively). Salt particles comprised the smallest percentage of LSA generated from Mona Lake freshwater at 10 and 11% for July and September samples, respectively. Conversely, salt particles were 23 and 34% of all particles observed from Muskegon June and October samples, respectively. Mona Lake—September had the highest microcystin (230 μg/L) and phycocyanin/BGA (1166 μg/L) concentrations, corresponding to an increase in the amount of biological and organic particles detected, similar to previous observations.⁴⁹ The significantly lower contribution of salt particles to Mona Lake LSA demonstrates the incorporation of biological and organic materials from HABs into aerosols.

Figure 6.

(A) Fraction of microcystin congeners present in all freshwater samples. (B) Number fractions of ATOFMS particle types generated from Mona and Muskegon Lake freshwater. Average phycocyanin/blue-green algae (BGA) and microcystin freshwater concentrations are shown for each sample.

Atmospheric and Health Implications.

This study identified HAB toxins, specifically microcystin, in laboratory-generated freshwater particles <2.5 μm, a key size range impacting human health.¹³² Microcystin congeners were quantified in freshwater from a HAB in Mona Lake in Michigan using two microcystin measurement methods (ELISA and LC–MS/MS). Aerosol particles were then generated in the laboratory and analyzed for the presence of toxins relative to the toxins identified in the freshwater samples. Of the eight microcystin congeners detected in the freshwater, seven were also detected in LSA particles, demonstrating that toxins are emitted through the aerosolization process in freshwater environments.

For the first time, we show that hydrophobic microcystin congeners were preferentially aerosolized versus hydrophilic microcystin congeners, which suggests that the relative amounts of toxins present in the aerosol phase are distinctly different than those present in the water column. This is analogous to previous observations of the preferentially aerosolized hydrophobic material in SSA from marine environments.^{50,63,84–88} The alignment between aerosol phase enhancement and octanol–water partitioning coefficients log(K_ow’s) for different microcystin congeners¹⁵ supports the finding that hydrophobic congeners partition to the air–water interface of bubbles passing through the HAB and are then transferred to the aerosol phase after bubble bursting. The fact that aerosol phase concentrations of microcystin cannot be predicted solely based on water concentrations, combined with the differing toxicities of microcystin congeners, demonstrates the need for further aerosolization studies aimed at predicting concentrations of toxins emitted to the atmosphere.

To improve understanding of the linkages between the organic material in freshwater and emitted in the aerosol phase, the size and amount of the insoluble organic material in freshwater was tracked and related to the aerosol size. Increased POC number concentrations corresponded to increased aerosol number concentrations and a shift in the aerosol size distribution to resemble that of the POC size distribution, leading to aerosol production at smaller sizes that are more readily inhaled.^127–129 Our results demonstrate an enhancement in ultrafine aerosol particles with an increase in HAB concentration.

Overall this work highlights the potential exposure risks for populations near or downwind of HABs. Understanding these exposures is challenging since we show that the hydrophobicity of microcystin congeners improves their aerosolization efficiency, leading to different relative amounts of congeners in the aerosol phase compared to the freshwater bloom itself. Ambient measurements are needed to quantify HAB toxin aerosolization while accounting for different atmospheric (wind speed, turbulence, temperature, RH) and biological (freshwater nutrient levels, type and amount of cyanobacteria, cyanobacteria buoyancy, and distribution throughout the water column) conditions. Additional questions remain regarding the diel patterns of freshwater toxin production and associated aerosol emission. Future work is needed to determine the relative toxicities of each congener and relate these to the atmospheric concentrations and bloom dynamics to fully assess the health impacts of toxin-containing LSA.