Sample collection and storage
Samples of cHABs were collected from the shallow waters of Snow Lake (New Mexico, USA) on 10/8/2020. Specifically, samples were collected in the West from Loco Mountain (33°25'37.1"N 108°29'54.1"W) and observed as a visible bloom close to the shore. Snow Lake is a small artificial reservoir located in the Catron County, New Mexico. It is situated in the relatively high altitude: 2229 m. It formed by the dam of a Middle Fork Gila river. Snow Lake is a generally oligotrophic lake with relatively low N, and P concentrations (data is not shown). Previously, occurrence of a cyanobacteria were detected in the Snow Lake (summer of 2011, 2019 years) by New Mexico Environment Department Surface Water Quality Bureau (NMED-SWQB). Dominated species were Aphanizomenon spp. and Dolichospermum spp. (https://www.waterqualitydata.us/) that is similar to our summer observations (data is not shown).Water and biomass were collected with plastic bottles from the surface of a shallow water. More detailed information about sample site is available on the web via CRIS (Melekhin et al. 2013) and in L. (Melekhin et al. 2019). Samples were kept on ice, in the darkness, until light microscopy observation, DNA, and toxin extraction. Samples for negatively stained TEM and pigment analysis were frozen at -80 C°. Samples for thin section TEM, SEM, and AFM were fixed with 2.5% solution of glutaraldehyde (several weeks storage in 4 C°). Water and biomass samples for confocal analysis were kept in a glass bottle for 1 month. During this month, multiple confocal observations were recorded.
Molecular methods and phylogenetic analysis
DNA was extracted soon after sample collection using DNeasy UltraClean Microbial Kit (Cat ID: 12224-50, QIAGEN, Venlo, Netherlands). Samples were sent to MR DNA to perform bTEFAP® PacBio Sequel analysis of the long (about 1400 bp) 16S rRNA gene region (www.mrdnalab.com, Shallowater, TX, USA). Bacterial specific primers: 27F/1492R with attached barcodes were used for amplification. Analysis utilized 10 thousand sequences per assay. The 16S rRNA gene amplification was done using a 35-cycle PCR (5 cycle used on PCR products). It used the HotStarTaq Plus Master Mix Kit (QIAGEN, USA) utilizing following steps: 94°C for 3 min., followed by 35 cycles of 94°C for 30 sec, 53°C for 40 sec. and 72°C for 90 sec., after which a final elongation step at 72°C for 5 min. The PCR products were observed in the 2% agarose gel, following amplification. The PCR pool was purified using Ampure PB beads (Pacific Biosciences). The SMRTbell libraries (Pacific Biosciences) were prepared following the manufacturer's guide and sequencing done at MR DNA (www.mrdnalab.com, Shallowater, TX, USA) on the PacBio Sequel following the manufacturer's guidelines. Sequences were processed using the MR DNA analysis pipeline (MR DNA, Shallowater, TX, USA). All sequences were denoised, and chimeras removed (Edgar 2016). Two Microcystis sequences were deposited into NCBI GenBank with following accession numbers: OK160994, OK160995. Megaphylogeny containing 9700 (including 2 sequences of Microcystis) sequences was performed in ARB software package (Ludwig 2004). Sequences of major representatives of nine cyanobacterial orders for the megaphylogeny were collected from Silva (Quast et al. 2012) and GenBank (Clark et al. 2016). During analysis, sequences were automatically aligned according to secondary structure of 16S rRNA molecule -SINA (Pruesse et al. 2012). Before ARB import, maximum likelihood phylogeny was calculated in the FastTreeMP software (Price et al. 2010) on XSEDE (Towns et al. 2014) based on alignment exported from SINA, with 1000 replicates.
Microscopy and Spectroscopy
Fresh bloom material was observed on the same day as a collection using a Zeiss Axioscope light microscope with Differential Interference Contrast (DIC) optics (Zeiss, Oberkochen, Germany). Morphometric features were measured using Zeiss AxioVision LE (Zeiss, Oberkochen, Germany). Morphological identification was accomplished following Komárek and Anagnostidis (1998).
Leica TCS SP5 II confocal microscope (Leica Microsystems, Germany) was utilized for the living cells observations. Argon laser (488 nm) at 16% intensity was used for the excitation (for the determination of the special pigment distribution), fluorescence emission spectra for carotenoids ranged from 510 to 605 nm, fluorescence emission spectra for chlorophyll a ranged from 680 to 690 nm. Sequential scan of these two spectra was implemented using HyD 4 channel in order to reduce photobleaching. Emission spectra were determined based on xyλ scan and (Vermaas et al. 2008). Pictures showing pigment distribution were taken with the following lens: 63x/1.4 Oil. For the aerotopes observation different settings were used, they were: 1) reflection signal collected at 484-493 nm using PMT2 channel, 2) PMT3 was used to capture all photosynthetic pigments at 582–717 nm 3) HC PL FLUOTAR 20x 0.5 DRY lens was utilized. The same laser was used for this analysis with different intensity that was 45%. Lastly, for the pigment profiling, xyλ scan was utilized starting from 490 nm and ending with 700 nm. Pigments were named according to (Vermaas et al. 2008).
Glutaraldehyde fixed cells were placed into agarose matrix inside the Eppendorf 2 ml tubes (Mozaffari et al. 2019). Subsequently agarose blocks with the cell pellets were washed with OsO4 (Electron Microscopy Sciences, Hatfield, PA) at 2% dilution with 0.1 M imidazole-HCl at pH 7.2 inside the tubes. After that, agarose blocks were removed from the tubes and cut into thin slices with visible osmium stained regions. Thin slices were placed in to a sterile Petri dish with diH2O and were put on the shaker for 20 mins. Rinsed agarose slices were placed into 10 ml glass vial with plastic caps for the EtOH dehydration series. Series were as follows: 50% EtOH – 20 mins, 80% EtOH – 20 mins, 95% EtOH – 20 mins, absolute EtOH – 20 mins (x2). While dehydrating, glass vials with the samples were placed onto a rotary mixer. Prior to the resin embedding, samples were rinsed with propylene oxide for 20 mins twice on the rotary mixer. Resin embedding utilized Spurr's epoxy resin (Low viscosity embedding kit, Electron Microscopy Sciences, Hatfield, PA). The rest of the procedure was followed as Robert Marc's Lab protocols (The University of Utah, https://marclab.org/tools/protocols/). Resin blocks were cut on the Ultramicrotome in order to obtain thin sections for the TEM analysis. For the on-grid negative staining (pili and cyanophages) observation, fresh grid were placed on the filtered colony and rinsed with 1% aqueous phosphotungstic acid, pH 6.5 (Vaara et al. 1984). TEM H-7650 (Hitachi High-Technologies, Tokyo, Japan) was used to observe both negative staining samples and thin section at 80 kV (high-resolution mode). Digital camera XR 60 (AMT Corp., Woburn, MA) was utilized to capture the images.
For the SEM, dehydrated cells (as it was describe in TEM section) that were not agarose embedded were observed on Tabletop Scanning Electron Microscope, Model TM-1000 (Hitachi High-Technologies, Tokyo, Japan) without sputter coating. In this analysis, accelerating voltage was 15 kV. The rest of the SEM pictures were taken on sputter-coated material utilizing Hitachi S-3400N II (Hitachi High-Technologies, Tokyo, Japan) with the following acceleration voltage – 10 kV. Secondary electron detector was used. Sputter coating was done using Pt/Au source. Energy-dispersive X-ray spectroscopy – EDS (Noran System Six 300, Thermo Electron Corp., Madison, WI) was used to determine elemental composition of the surface of the colony (F-layer). Acceleration voltage also was 10 kV. In order to compare cyanobacterial EPS with chondroitin sulfate, a small portion of a turkey's ligament was dehydrated with absolute EtOH.
Prior to AFM scanning, glutaraldehyde fixed cells were dried out on the cover slip in the room temperature. To acquire AFM images of the pili and cell surfaces, Bruker Dimension Fast Scan (Germany) was utilized with RTESPA probe (part #: MPP-11120-10) on the tapping mode. Amplitude error and height sensors were implemented. Bruker NanoScope Analysis software v1.51 was used for the editing of the obtained images.
Lake water (600 ml) was filtered through Extract-CleanTM SPE C18-HC cartridge (part no. 255350/5122524, Deerfield, IL, USA) for the released toxins, 200 ml of lake water was vacuum filtered through the paper filter for the toxins remained in the biomass. Acetonitrile (H2O:ACN/20%:80%, 2 ml) was used for the toxin extraction. Multiple Reaction Monitoring (MRM) was implemented for high sensitive detection of the toxins as described in (Degryse et al. 2017). Standard ACQUITYTM UPLC (WatersTM corporation, Milford, MA, USA) system together with MICROMASS Quattro Ultima (WatersTM corporation, Manchester, UK) was used in this analysis. Standards for following microcystins: LR, LY, RR were purchased from Sigma-Aldrich (St. Louis, MO, USA). Quantification of the toxins were based on standard curves. Initial quantification analysis was carried out in QuanLynx 4.1 (WatersTM corporation, Milford, MA, USA) where areas on the peak were estimated. The rest of analysis including the generation of standard curves was done in Excel 2021 (Microsoft Corporation). Standard for Microcystin-YR was not available during this study and it was putatively identified based on reported MRM transitions and relative elution order.
Lipophilic pigments analysis with HPLC-MS/MS
Prior to pigments extraction, 300ml of the lake water was filtered on the paper filter using a vacuum pump. Filters with captured biomass were cut with scissors, extraction was held according to Juin et al. (2015) utilizing 2 ml of HPLC grade EtOH (MDL #: MFCD00003568, Thermo Fisher Scientific Inc., Waltham, Massachusetts, U.S). Pigment analysis was performed on Waters Alliance 2695 HPLC (WatersTM corporation, Milford, MA, USA) coupled with MICROMASS Q-Tof UltimaTM ESI (WatersTM corporation, Manchester, UK). Positive mode MSE method described in Juin et al. 2015 was adapted to HPLC/Q-Tof Ultima. Liquid Chromatography (LC) separation was performed on BetaTM C18 HPLC Column (part no. 70105-152130). Modified gradient steps are provided in Table S1. Detailed LC-PDA-MS2 parameters are presented in Table S2. Experimental data was acquired with MassLynx software version 4.0 (WatersTM corporation, Milford, MA, USA). Following masses: 382.2 and 742.4 were excluded from the analysis as carryover artifacts. Daughter ion masses used as a criterion used to switch to MS2 are given in Table S3. MassLynx software version 4.1 was used for pigment identification using daughter ions (Juin et al. 2015) and reference UV absorption spectra (Clementson and Wojtasiewicz 2019; Lopes et al. 2020). Additionally, reserpine (cat. # 83580, Sigma-Aldrich, St. Louis, MO, US) was used for the Lock Mass.
All graphical elements were post-produced in Adobe Photoshop 2021 and Illustrator2021 (Adobe Systems Inc., San Jose, California, USA). For Figures 5B, D super resolution function in Adobe Photoshop 2021 was applied. For Fig. S4 red/green stereo channels were overlaid in Adobe Photoshop 2021. All Fast Fourier Transform (FFT) were made in FIJI 1.53f (Schindelin et al. 2012).