Materials
Microcystin reference standards for MC-LR, MC-YR, MC-RR, MC-LF, MC-LA, MC-LW, MC-LY, and nodularin (all > 95% purity by HPLC) were obtained from Enzo Life Science (Lausen, Switzerland) and [D-Asp3,E-Dhb7]MC-RR (> 95% purity by HPLC) from CyanoBiotech GmbH (Berlin, Germany). Bioreagents for aeruginosin 98B, cyanopeptolin A, cyanopeptolin D, anabaenopeptin A, anabaenopeptin B, anabaenopeptin NZ857, and oscillamide Y (all > 90% purity by HPLC) were obtained from CyanoBiotech GmbH (Berlin, Germany). Aerucyclamide A was obtained as purified bioreagent in dimethyl sulfoxide by Prof. Karl Gademann (University of Zurich, Switzerland) [37]. Additional materials are listed in the Supporting Information (Additional file 1: Text S1).
Cyanobacterial cultures
Microcystis aeruginosa PCC7806 was originally isolated from Braakman reservoir in The Netherlands (1972) and was obtained from the Pasteur Culture Collection of Cyanobacteria (France). Dolichospermum flos aquae NIVA-CYA 269/6 was originally isolated from Lake Frøylandsvatnet in Norway (1990) and Planktothrix rubescens K-0576 was originally isolated from Lake Borre Sø in Denmark. Both strains were obtained from the Norwegian Culture Collection of Algae (NORCCA). Microcystis aeruginosa UV006 was originally isolated from Hartebeespoort Dam in South Africa and an inoculum was provided by Prof. Jakob Pernthaler (University of Zurich, Switzerland). Primary cultures were kept in 75-mL modified WC medium (Additional file 1: Table S1) at 20 ± 2 °C and irradiated at 12 μmol photons m−2 s−1 on a 12:12-h light/dark cycle [38]. To produce significant amount of biomass, 4.5 L of sterile WC medium was inoculated with 10–15% inoculum every four weeks. The 5-L Schott bottles were cultivated at the same conditions described above and aerated with filtered air (GE Healthcare, Whatman, HEPA-VENT, 0.3 µm). All materials used for culturing were autoclaved before use and all the subculturing was performed under sterile conditions.
Cyanopeptide extraction for photochemical experiments
The cells were harvested by centrifugation (rcf of 4000 g at 10 °C, 10 min, Herolab HiCen XL), lyophilized (− 40 °C, − 3 mbar, 24 h, Lyovac GT2, Leybold) and stored at − 20 °C until further analysis. For the extraction, the weight of the dry material was recorded and MeOH/H2O (70/30% v/v) was added at a ratio of 200 µL mgdry,wt−1. The suspension was homogenized by vortexing, incubated under sonication (VWR, Ultrasonic cleaner USC-THD, level 6, 10 min at 15 °C) and pellets were separated from supernatant by centrifugation (rcf of 4660 g at 10 °C, 10 min, Megafuge 1.0 R). The supernatant was transferred to a new glass vial, the extraction was repeated twice and supernatants were combined. The solvent was evaporated from the pooled extract under a gently stream of nitrogen (40 °C, TurboVap®LV, Biotage) to reduce the methanol content to less than 5%. The extracts were then purified by liquid–liquid extraction (LLE). Therefore, extracts were diluted to a total volume of 25 mL with nanopure water in a separatory funnel containing 25 mL of hexane. The funnels were shaken vigorously for 3 min before allowing separation of the two phases again. The water phase was then collected and the hexane phase was discarded. The water phase was extracted with hexane two more times. The extraction allowed to remove a large portion of chromophoric matter with > 90% reduced absorbance at 665 and 613 nm indicative of dominating pigments (chlorophyll-a and phycocyanin, respectively) that would otherwise interfere with the photochemical tests (absorbance spectra of extracts in Additional file 1: Figure S1). The water fraction was concentrated to 300 μL by vacuum-assisted evaporation (Syncore® Analyst R-12, BÜCHI Labortechnik AG, 40 °C, 120 rpm, 20 mbar). Each volume was adjusted gravimetrically to 1.0 mL in nanopure water and stored in the fridge at 4 °C if used within 24 h or in the freezer at − 20 °C until its further use for photochemical experiments.
Cyanopeptide profile in four strains
To analyze the cyanopeptide profiles from different strains, the biomass was extracted as described above with the difference that MeOH/H2O (70/30% v/v) was added at a ratio of 15 µL mgdry,wt−1. The extracts were then individually purified by solid phase extraction (SPE). For the SPE (12-fold vacuum extraction box, Visiprep, 12 ports, Sigma Aldrich) the extracts were diluted to a total volume of 3 mL with nanopure water. The SPE cartridges (Oasis HLB 3 cc, 60 mg) were consecutively conditioned with methanol and water (9 mL each). The extracts were loaded onto the cartridges, washed with 9 mL nanopure water followed by 9 mL MeOH/H2O (20/80% v/v) prior to elution with 9 mL MeOH/H2O (85/15% v/v) at a flow rate of 1 mL min−1. The eluted fraction was concentrated to 300 μL by vacuum-assisted evaporation (Syncore® Analyst R-12, BÜCHI Labortechnik AG, 40 °C, 120 rpm, 20 mbar), and each volume was adjusted gravimetrically to 1.0 mL in nanopure water. The cyanopeptides were analyzed as described below.
Simulated sunlight exposure
Irradiation experiments were carried out in a benchtop xenon instrument that simulates sunlight (Heraeus, Suntest CPS + , 700 W m−2, light emission spectrum in Additional file 1: Figure S2). Cyanopeptide degradation over time was studied in lake matrix, using lake water collected from Greifensee (06/08/2019, 47.3663°N, 8.665°E), and in buffered nanopure water at pH 7 and 8 (5-mM phosphate buffer) and pH 9 and 10 (10-mM carbonate buffer) with constant ionic strength (13 mM adjusted with sodium chloride). Aqueous cyanobacterial extracts (200 µL) were added to either lake matrix or buffer solutions (total volume of 4 ml). Furfuryl alcohol (FFA, 40 µM) was added for quantification of singlet oxygen. The solutions were exposed to simulated sunlight for three hours in open quartz vials (Pyrex, 7.5 cm, inner diameter 1 cm), positioned at 50° angle from the horizontal plane, assuring that the solutions were completely submerged in a temperature-controlled water bath (20 °C ± 1 °C). The experiments were conducted in experimental duplicate for the buffered solutions and in triplicates for the lake matrix. To monitor the light flux, the chemical actinometer system PNA-PYR was used and solutions in nanopure water (10 µM para-nitroanisole, 0.5 mM pyridine) were irradiated along with the experiment in simulated sunlight. During the irradiation experiment, three technical replicates of 150 µL were collected for time point 0 and two technical replicates were collected at different time points (0.5, 1, 2, 3 h). To account for transformation independent of light, aliquots of each solution in glass vials were covered from light with tin foil and positioned next to the other exposure vials serving as dark controls. All the samples were immediately analyzed for FFA and PNA upon sampling. These samples were then frozen at − 20 °C for further cyanopeptide analysis as detailed below.
FFA and PNA analysis
To assess the steady-state concentration of singlet oxygen [1O2]ss and the photon fluence rate, the degradation of FFA and PNA was monitored, respectively. Both FFA and PNA analyses were performed by high-performance liquid chromatography (HPLC) coupled to a UV–VIS/DAD detector (Dionex UltiMate3000 HPLC, Thermo Fischer Scientific). Chromatographic separation was carried out on an Atlantis T3 C18 column (3 µm, 3 × 150 mm, Waters) with pre-column (VanGuard® Cartridge, Waters) and inline filter (BGB®). The mobile phases consisted of (A) sodium acetate buffer (pH 5.9; 15.6 mM; 10%ACN) and (B) acetonitrile. Isocratic elution was carried out at a flow rate of 350 µL min−1 with an isocratic ratio of 90:10 (A:B) for FFA and 40:60 (A:B) for PNA. The injection volume was 20 µL and detection occurred at 219 nm for FFA and at 316 nm for PNA. Measured peak areas of both FFA and PNA at each timepoint (At) were normalized to their initial concentration (A0) and the natural logarithm of this ratio was plotted against time. The observed degradation rate constants (kobs in s−1) were calculated as the slope of a linear regression. Steady-state concentrations of singlet oxygen [1O2]ss (M) were determined as [39]:
$$\left[ {1O2} \right] = \frac{{k_{obs} ,{\text{FFA}}}}{{k_{rxn} ,{\text{FFA}}}} ,$$
(1)
where kobs,FFA (s−1) is the observed degradation rate constant of FFA of each sample vial and krxn,FFA (M−1 s−1) is the temperature-dependent second-order reaction rate constant of FFA with singlet oxygen that can be calculated according to:
$$\ln {k_{\text{rxn,FFA}}} = {\rm{ }}\frac{({1.59 \pm 0.06 )\times {{10}^8}}}{{273.16 + {\rm{T}}[^\circ {\rm{C}}]}} + (23.82 \pm 0.21)$$
(2)
The photon fluence rate was calculated based on the kobs of PNA in the actinometer solutions according to established procedures (details in Additional file 1: Text S2) [40].
Cyanopeptide analysis
Cyanopeptide analysis was performed by HPLC (Dionex UltiMate3000 RS pump, Thermo Fischer Scientific) coupled to a high-resolution tandem mass spectrometer (HRMS/MS, LumosFusion Orbitrap, ThermoFisher Scientific). Chromatographic separation was carried out on a XBridgeTM C18 column (3.5 μm, 2.1 × 50 mm, Waters) with pre-column (VanGuard® Cartridge, Waters) and inline filter (BGB®). The mobile phases consisted of (A) nanopure water and (B) methanol both acidified with formic acid (0.1%). Binary gradient elution was carried out at a flow rate of 200 μL min−1 and increasing eluent B from 10 to 95% between 0 and 25 min. The injection volume was 20 μL. Detection of analytes was achieved by HRMS/MS with electrospray ionization (ESI), 320 °C capillary temperature, 4 kV electrospray voltage and 3500 V capillary voltage in positive ionization mode. Full scan accurate mass spectra were acquired from 450 to 1350 m/z with a nominal resolving power of 240,000 referenced at m/z 250, automated gain control (AGC) of 5⋅104, maximal injection time of 100 ms with 1 ppm mass accuracy. Data-dependent high-resolution product ion spectra were obtained by stepped normalized collision energy for HCD (10, 20, 30, 40 and 50%) and CID (30 and 35%), at a resolving power of 15,000 at 400 m/z, AGC of 1⋅104 and maximal injection time of 22 ms. For triggering data-dependent MS/MS acquisition we included cyanopeptides from the publicly available list CyanoMetDB [18]. The suspect screening included 1219 cyanopeptides in total with 160 microcystins, 177 cyanopeptolins, 73 anabaenopeptins, 65 cyclamides, 78 microginins, 79 aeruginosins and 587 other compounds, accounting structural isomers and the mass window of 450–1350 m/z.
Cyanopeptide identification
Data evaluation and peak area extraction were performed with Skyline 20.1 (MacCoss Lab Software). Charge states (z = 1 and z = 2) and adducts (H+, Na+) were considered for all compounds. The identification of most cyanopeptides needed to be carried out without available reference standard materials. Thus, a comprehensive data analysis workflow established for suspect screening of micropollutant was modified and applied [41]. One major difference to micropollutant suspect screening is the fact that no spectral libraries exist for most cyanopeptide suspects. Therefore, we used in-silico fragmentation predictions to facilitate compound identification (Mass Frontier 7.0, mMass 5.5.0) and the confidence level scheme widely used for mass spectrometry by Schymanski et al. [41]. Herein, only those cyanopeptides were reported that could be identified as one of the following criteria: a cyanopeptide was identified as a tentative candidate (Level 3) based on exact mass (< 5 ppm mass error), accurate isotopic pattern (Skyline idotp value > 0.9), and evidence from fragmentation data; a cyanopeptide was identified as probable structure (Level 2) based on complete fragmentation information confirming the connectivity of the building blocks of the peptide; and a cyanopeptide was identified as confirmed structure (Level 1) when these parameters were in agreement with available reference standards or bioreagents. The fragmentation spectra of reference standards (or bioreagents) were compared to confirmed structures in our experiments (i.e., cyanobacteria extracts) with head to tail plots using the R packages RMassBank [42] and MSMSsim [43] (Additional file 1: Figures S3–S15). Data analysis was performed in RStudio with R version 3.6.1 [44]. Therefore, the HRMS measurement data files were converted to open.mzXML data format using the msconvert tool from ProteoWizard [45].
The peak areas of selected ion chromatograms were extracted for all cyanopeptides identified with Level 1–3 in Skyline (Version 20.1). For all cyanopeptides the M + H ion was dominating with the exception of microcystins that contain two arginine moieties, here the M + 2H was selected for area extraction. The identified cyanopeptides were quantified by external calibration curves of available reference standards and bioreagents in the range of 0.5–500 µg L−1. Concentrations were only reported when the peak area was above the limits of quantification (LOQs), defined as 10 times the ratio of standard deviation of the response over slope of the logarithmic calibration curve (details in Additional file 1: Table S2). For those cyanopeptides for which no reference standard or bioreagent was available, we calculated class-specific equivalents, which were calculated from external calibration curves with the structurally most similar bioreagent or standard assigned for each compound (details in Additional file 1: Table S3) according to previous work [46].
Photodegradation assessment
All first-order degradation rate constants, kobs (s−1), were assessed as the slope of a linear regression of natural log-transformed normalized peak area (ln(At/A0)) versus irradiation time. A kobs was only reported for regressions with a correlation coefficient r2 > 0.6 and when the final concentration after irradiation was statistically significantly different from both, the dark control and initial concentration (t-test, p-value < 0.05). If a compound degraded (i.e., significant difference between final concentration and dark control and initial concentration) but the loss did not follow pseudo-first-order kinetics (i.e., r2 < 0.6), we report “no first-order kinetics” or “n.f.k.”. If we did not observe any significant loss of concentration relative to the dark control and the initial concentration during irradiation, we report “no degradation detected” or “n.d.”.
Statistical analysis
One-way analysis of variance (ANOVA) followed by Tukey pairwise comparison was employed to detect statistical significance influence of amino acid moieties and cyanopeptide class on the observed degradation rate constants by comparison of the 95% confidence intervals in RStudio (version 3.6.1).