The impact of environmental parameters on microcystin production in dialysis bag experiments

It is important to understand what environmental parameters may regulate microcystin (MC) production and congener type. To determine if environmental conditions in two hydraulically connected lakes can influence MC production and congener ratios, we incubated dialysis bags containing phytoplankton from mesotrophic/eutrophic Muskegon Lake into hypereutrophic Bear Lake (Michigan, USA) and vice versa. Strong cyanobacteria growth was observed in all dialysis bags with Bear Lake phytoplankton in July and August. Phytoplankton communities were dominated by Aphanizomenon aphanizomenoides, Microcystis wesenbergii, Limnothrix redekei. MC concentrations were correlated with M. wesenbergii and A. aphanizomenoides biovolume. MC concentrations in bags incubated in the Muskegon Lake with Bear Lake water were significantly higher than the other bags. The higher light intensity and total nitrogen concentration may have caused the increase of MC production. The MC-LR/MC-RR ratios varied with sample origin but not with lake of incubation, indicating that physical environmental factors (water temperature and turbidity) were not the reasons for different toxin production ratios. Differences in total phosphorus concentrations might be one reason for the dissimilarity of the MC-LR/MC-RR ratio between the two lakes. The higher light intensity and NO3-N concentration in Muskegon Lake are two factors contributing to an increase of MC production.

In August, the mean concentrations of total MC (7.04 ± 0.73 μ g·L −1 , range: 5.77-7.97 μ g·L −1 ) in bags in Muskegon Lake initiated with Bear Lake phytoplankton (MKBL1-3) were also significantly higher than the MC in other bags (p < 0.001) (

Environmental factors. Physicochemical parameters showed little temporal and spatial variation in Bear
Lake and Muskegon Lake ( Table 2). In both months, the SRP concentration was below the detection limit during the sampling period. The concentrations of nitrate (NO 3 -N) and ammonia (NH 3 -N) were higher in Muskegon Lake and the corresponding bags (MKMK; MKBL) than in Bear Lake and the corresponding bags (BLBL; BLMK) (p < 0.010 and p < 0.030, respectively). The MC concentrations were not correlated with the nitrate concentration (R 2 = − 0.422, p = 0.509) or ammonia concentration (R 2 = − 0.616, p = 0.150). The concentrations of TP and TN Muskegon Lake Initial; MKF: Muskegon Lake Final; MKMK: Muskegon Lake with Muskegon Lake water; MKBL: Muskegon Lake with Bear Lake water; BLI: Bear Lake Initial; BLF: Bear Lake Final; BLBL: Bear Lake with Bear Lake water; BLMK: Bear Lake with Muskegon Lake water. The taxonomic analyses were conducted with three replicates in July and five replicates in August.

Discussion
Several research groups have studied how environmental parameters affect the dominance of cyanobacteria and total MC concentrations in lakes 11,13,27 . Also, some studies described the relationship of bloom community Muskegon Lake with Bear Lake water, BLI; Bear Lake Initial, BLF; Bear Lake Final, BLBL; Bear Lake with Bear Lake water, and BLMK; Bear Lake with Muskegon Lake water.) dynamics and the MC congener concentration and composition 6,13,28 . About MC congers, Tonk et al. 17 suggested that the ratio of MC variants changed in response to differing light intensities; de Figueiredo et al. 29 found out that higher temperatures enhanced MC-RR production, whereas lower temperatures favored MC-LR synthesis. While Monchamp et al. 13 suggested that environmental factors did not appear to affect MC congener composition directly but there were significant associations between specific MC congeners and particular species. In our experiment, there was a significant difference in total MC concentrations between all the treatments. Total MC concentrations in the bags incubated in Muskegon Lake with Bear Lake water (MKBL) were significantly higher than the other treatments. No significant differences between the cyanobacteria biovolume in all the bags with Bear Lake water were observed and MC-LR/MC-RR ratios from the treatments with the corresponding lakes were similar during the study period. According to previous studies [30][31][32] , MC production was correlated with algal species and cell growth. M. aeruginosa has been classified as a major MC producer in previous research 33,34 . In July, the greatest total MC concentrations (20.1 ± 3.88 μ g·L −1 , range: 14.97-24.32 μ g·L −1 ) were found in bags without M. aeruginosa present. In addition, MC concentrations were not correlated with M. aruginosa biomass in both months, indicating that there were other cyanobacteria strains producing MC. MC concentrations were found to be correlated with M. wesenbergii in the current experiment. In term of MC production by M. wesenbergii, previous studies yielded contradictory conclusions. Henriksen 35 found that M. wesenbergii was dominated in hepatotoxic Microcystis blooms of Danish lakes. While Watanabe 36 concluded that M. wesenbergii has generally been considered as nontoxic. By both molecular and chemical methods, recent studies showed that M. wesenbergii lacked MC production genes in Germany and other European lakes 37,38 and in China 39 . Also, in our early MC investigation in seven lakes of Michigan, the MC concentrations were not correlated with the biomass M. wesenbergii (unpublished data). Based on the literature findings in spite of the observed correlation, it was likely that M. wesenbergii was a nontoxic species in our experiments.  Table 2. Chemical data (mean ± SD, n = 3 (July); n = 5 (August)) for dialysis bag experiments and ambient lake water (MKI: Muskegon Lake Initial; MKF: Muskegon Lake Final; MKMK: Muskegon Lake with Muskegon Lake water; MKBL: Muskegon Lake with Bear Lake water; BLI: Bear Lake Initial; BLF: Bear Lake Final; BLBL: Bear Lake with Bear Lake water; and BLMK: Bear Lake with Muskegon Lake water; MKI, MKF, BLI and BLF represented the ambient samples; "<"represented the concentration was below the limit of the detection).  The traditional genus Aphanizomenon comprises a group of filamentous nitrogen-fixing cyanobacteria of which several members are able to develop blooms and to produce toxic metabolites (cyanotoxins), including hepatotoxins (microcystins), neurotoxins (anatoxins and saxitoxins) and cytotoxins (cylindrospermopsin) 40 . The species of Sphaerospermopsis aphanizomenoides isolated from Lake Oued Mellah was reported to contain MCs, namely four compounds displaying a retention time similar to that of MC-LA, LY, LW or LF in HPLC-PDA chromatograms 41 . In this study, MC concentrations correlated with the biomass of A. aphanizomenoides in both months indicating that A. aphanizomenoides is a potential MC producer. A. aphanizomenoides was considered to be salinity-tolerant 42 , requires high water temperature 43 , and the biomass of A. aphanizomenoides was found to be significantly related to the water temperatures 44 . This cyanobacterium has been detected in water bodies in several countries 44 and has been expanding its range into more half regions of European 45,46 . A. aphanizomenoides has not been linked to MC production with the exception of a study also conducted in Bear Lake where the organism was listed as the dominant cyanobacteria species and a suspected MC producer 47 . In consideration of the strong statistical correlation between A. aphanizomenoides biovolume and MC production occurring in the same lake, our study assumes that A. aphanizomenoides may be a MC producer. Genetic studies still need to be performed to determine if toxin producing genes are present in this organism.

Parameters
MC production also was influenced by environmental parameters 32 . Some studies suggested that the environmental parameters, i.e., phosphorus, nitrogen, temperature, light etc., affect the MC production and the growth of M. aeruginosa in continuous cultures, laboratory batch, or in the field 11,12,48 . Environmental parameters may affect MC concentrationsin two principal ways: regulating MC production by the toxigenic strains or regulating the population of MC-producing strains 49 . Sivonen 10 indicated that MC production by Oscillatoria agardhii correlated with high nitrate concentration (0.42-0.84 mg·N/L) and low light intensity (12-95 μ mol·m −2 ·s −1 ). While Jiang et al. 34 suggested that light and iron had significant interactive effect on MC production. For Microcystis PCC 7806, Wiedner et al. 49 indicated that the maximum MC concentrations were reached at light intensities of 40 μ mol·m −2 ·s −1 but a decline in MC production and cellular MC content were observed by further increasing the irradiance during lab experiments. In addition, for M. aeruginosa W334, Hesse and Kohl 16 found that celluar MC-LR concentrations decreased at a growth rate at 80 μ mol·m −2 ·s −1 , but for M. aeruginosa W368, MC-LR and MC-YR, cellular contents increased at 100 μ mol·m −2 ·s −1 . Yang et al. 50 found out that MC production decreased significantly when the strain was exposed to UV-B radiation. For P. agardhii, Sivonen 10 noted that higher MC concentrations were produced at lower irradiances (12 and 24 μ mol·m −2 ·s −1 ) rather than at higher numbers (50 and 95 μ mol·m −2 ·s −1 ). Monchamp et al. 13 indicated that water temperature, TN, ammonium and DON can influence the cyanobacterial population structure, which resulted in the differences of the dominant MC congeners and the toxicity. It seemed that the diverse effects of light on the MC production depend on the cyanobacterial species and on the MC analogue. Currently, although opinions vary, MC production appears to be linked to N availability 27,51,52 and functions to alleviate oxidative stress during high light conditions 53,54 .
In this study, MC-LR/MC-RR ratios varied with sample origin but not with lake of incubation, indicating that water temperature, light and turbidity were not the reasons for the difference of the MC-LR/MC-RR ratio. Van de Waal et al. 27 studied how nitrogen pulse affect the MC variants of P. agardhii and found out MC-RR increased strongly, while MC-LR increased weakly after the nitrogen pulse. They speculated Microcystis and other MC-producing algae would respond similarly. In this study, we observed that the biovolume of A. aphanizomenoides followed the increase of MC production. A. aphanizomenoides is able to fix molecular nitrogen (diazotrophy) and in this study, we found low levels of NO 3 -N and NH 4 -N along with high levels of TN (Table 1). These numbers are typical for an environment in which N 2 fixation takes place. Hence, it is possible that with fixed N 2 made available for MC producing strains, both the overall MC content and the MC-LR/MC-RR ratio should be expected to change. With A. aphanizomenoides present, the limiting nutrient is supplied by N 2 fixation may have resulted in the relative increase of MC-RR and MC-LR (Fig. 2). In Muskegon Lake water, the nutrient balance may not be suitable for N-fixation due to higher NO 3 -N concentrations since nitrate can suppress nitrogenase in some cyanobacterium 55 . Hence, the higher NO 3 -N concentrations were a possible factor for the increase of MC concentrations in the dialysis bags. Also, light was considered an important factor affecting MC production as light intensity can regulate the transcription of the MC-synthesizing gene 56 . In the present study, Muskegon Lake had lower temperature, higher light intensity, and lower turbidity than Bear Lake. Since the growth of A. aphanizomenoides requires higher water temperatures, the lower thermal profile observed in Muskegon Lake might not be conducive for the increase toxin production. In this study, the light intensity of Bear Lake (average: 397.2 μ mol·m −2 ·s −1 ) was significantly lower than Muskegon Lake (800.1 μ mol·m −2 ·s −1 ). Low-light conditions were generated by two main factors: water depth and turbidity 57 . Since we incubated all the dialysis bags in the same depth (1 m) of the two lakes, the higher turbidity of Bear Lake appears to be responsible for the lower light intensity. The high light intensity of Muskegon Lake appears be another reason for the increase of MC concentrations in the dialysis bags with Bear Lake water incubated in Muskegon Lake.
Oh et al. 58 suggested that MC-LR/MC-RR ratio can increase with severe P-limited conditions. Sas et al. 59 indicated that phytoplankton growth was P-limited if FRP was < 10 μ g·L −1 of the growing season. In this study, SRP of the two lakes and all the dialysis bags were less than 5 μ g·L −1 , TP in Muskegon Lake and the bags with Muskegon Lake water were all less than 50 μ g·L −1 , while TP in Bear Lake and bags with Bear Lake water were ~100 μ g·L −1 . The difference in bioavailable TP concentrations may be one reason for the dissimilarity of the MC-LR to MC-RR ratio of Muskegon Lake and Bear Lake. Furthermore, other factors which were not specifically investigated during the present study (e.g. turbulence, zooplankton predation) could also have an influence on the abundance of different microcystin congeners and we will do the further research in this field.

Methods
Experimental design. Experiments were conducted with water collected from Bear Lake and Muskegon Lake. Bear Lake has a surface area of 1.66 km 2 , an average depth of 2.14 m, and a maximum depth of 3.66 m 23 . Bear Lake discharges to Muskegon Lake through a narrow navigation channel at a rate of 0.9 m 3 /s and has a mean hydraulic residence time of 30 days 60 . Muskegon Lake is a mesotrophic/eutrophic, drowned river mouth system with a surface area of 16.6 km 2 and an average depth of 7.1 m, with a maximum depth of 23 m 61 . Muskegon Lake discharges to Lake Michigan at a rate of 55.5 m 3 /s and has a mean hydraulic residence time of 25 days 61 . The Muskegon River accounts for 95% of the tributary inputs to Muskegon Lake 62 . Both lakes are well mixed 24 .
Dialysis bags were filled with lake water and phytoplankton from five meters away from Bear Lake Dock and 5 meters away from Muskegon Lake Barge (Fig. 4) at 1 meter depth in July 19 th 2010. All measurements occurred between 9:00 and 11:00 AM. The bags were constructed of Spectra/Por 5 dialysis tubing (12-14 K MWCO, 140 mm flat width; Spectrum Laboratories, CA) and contained approximately 500 ml of lake water and were completely sealed. Triplicate dialysis bags of water from each lake were attached to a support cage and incubated for 7 days in Bear Lake and Muskegon Lake at 1 m depth (the maximum depth of Bear Lake shore is 1.5 m). Dialysis bag samples were identified as MKMK (Muskegon Lake with Muskegon Lake water), MKBL (Muskegon Lake with Bear Lake water), BLBL (Bear Lake with Bear Lake water), and BLMK (Bear Lake with Muskegon Lake water). On August 16 th , the samples were taken and incubated in the same location. To confirm the data of July was not random, we use 5 replicates of water from each lake at this time. For chemical and biological analysis, water samples were collected near the support cages at the beginning (MKI and BLI, respectively) and end of the experiments (MKF and BLF, respectively). In addition, daily in situ measurements of Photosynthetically Active Radiation (PAR) were measured with a LiCor Li-193SA (spherical quantum sensor) and temperature, turbidity, and total dissolved solids (TDS) were measured with a YSI 6600. All in situ measurements were conducted adjacent to the dialysis bags at 1 m depth.
After the 7-day incubation period, the bags were mixed well prior to sampling and a 25 ml aliquot from each dialysis bag was withdrawn for phytoplankton analysis. The remaining water was stored immediately in a portable refrigerator (around 4 °C) and composited into a single sample for nutrient analysis.
Chemical analysis. Three 100 ml aliquots from each dialysis bag were immediately placed on ice and returned to the lab for filtration on a 0.7 μ m Whatman GF/F glass microfiber filter (Fisher Scientific cat # 09-874-64) and stored at − 20 °C for cyanotoxin analysis. According to Fastner et al. 63 and Dyble et al. 64 , toxin samples were lyophilized first and then sonicated in 75% aqueous methanol. MC analogues (MC-LR, MC-RR, MC-YR, MC-LA; Sigma-Aldrich) and cylindrospermopsin (CYN) (Sigma-Aldrich) analysis was performed by High-Performance Liquid Chromatography coupled Mass Spectrometry (HPLC/MS) using a Thermo Surveyor MSQ Single Quadrupole Mass Selective Detector and Thermo Spectrasystem gradient chromatographic system according to a method described by Barco et al. 65 . Total MC concentrations were reported as the sum of all congeners (HPLC/MS-Total).
Algae were identified and enumerated utilizing a Nikon Eclipse TE200 inverted microscope 68 . At least 200-300 algal units (cells or filaments) were counted in all the samples. The cell volume of each species was calculated by applying the appropriate geometric formulae 69 . The detailing for cell density calculations please see Table 3.

Statistical calculation.
Statistical analyses were conducted with SPSS version 12.0.1 (SPSS, Inc. Chicago IL, USA). The non-parametric Wilcoxon sign test was used to evaluate MC concentrations differences between the bags and ambient samples in July and August as data were not normally distributed. Differences in cyanobacterial biovolume and MC concenration between the bags and ambient samples were examined with the non-parametric Wilcoxon sign test (a = 0.05). Statistical similarity was evaluated with the Mann Whitney U test (a = 0.05) and multiple correlations were performed with Spearman's Rank-Order Correlation (a = 0.05). To test if the two months (July and August) had significantly different cyanobacterial assemblages, samples were analyzed with the nonparametric-analysis of similarity (ANOSIM, Clarke 70 ). This method tests for significant differences (a = 0.05) between two or more groups using the rank order of the samples similarity matrix based on the Bray-Curtis similarity coefficient. To examine the differences between MC-LR/MC-RR ratio, the Mann Whitney U test was used (differences being significant at p < 0.05). To examine the differences between environmental factors, the Mann Whitney U test was used (differences being significant at p < 0.05).

Conclusion
Our data suggest that differences in total phosphorus concentrations were a reason for the dissimilarity of the MC-LR/MC-RR ratio between Muskegon Lake and Bear Lake. The higher light intensity due to lower turbidity and NO 3 -N concentrations in Muskegon Lake were two factors contributing to an increase of total MC production.