Description of the Five Sisters Site
The FS site (RCN Database LWCG023A, LWCG023B, LWCG023C) consists of a group of alkaline-chloride springs located in the south-east corner of the Lower Geyser Basin in the WCD (44.5325oN, 110.7971oW) (Figure 1A). White Creek flows down the drainage and is accompanied by several thermoalkaline springs including Spindle Geyser (LWCG149) and two of the best studied sites in YNP, Octopus Spring (LWCG138) and Mushroom Spring [12]. The FS site is located a few meters south of White Creek against a steeply inclined hill and consists of three springs and five distinct pools, labeled 1 to 5 from East to West, with variable interconnectivity (Figure 1B). FS1 is fed by a small geyser and is connected above-ground to FS2 and FS3. There is no visible above-ground connection between FS3 and FS4, although there may be below-ground connectivity between the two springs. FS4 and FS5 are connected above-ground and FS5 outflow continues away from the spring group and feeds into White Creek. FS1 is the largest spring at several meters wide and deep, followed by FS5, FS3, FS2 and FS4.
Sample collection
Samples were collected within one week of the first day of March in 2017, 2018 and 2019. Stainless steel cups on extendable poles (~0.25L total volume) were used to collect samples from each spring. The cups were rinsed with spring water first and then used to collect a small amount of sediment slurry. Samples were collected in triplicate and were taken at three distantly located locations within each spring with paired subsamples for metabarcoding and mass spec analysis collected at each sampling location. Sediment slurry samples were composed of ~8mL of sediment with ~7mL of thermal water and were placed in sterile 15mL conical centrifuge tubes after collection (Corning, Corning, NY). Slurry samples were immediately frozen in a dry ice and ethanol bath in the field and stored on dry ice until transported to Montana State University (MSU) and placed in a -80°C freezer until genomic or mass spectrometry analysis.
Geochemical analysis
Aqueous geochemistry was monitored with each sampling event. Briefly, the temperature and pH of each site was measured in situ using a combined pH-temperature probe (Hach HQ30d, Hach Co., Loveland, CO). Dissolved oxygen was measured in the field using the High Range Dissolved Oxygen method and a portable colorimeter (Hach DR900, Hach Co., Loveland, CO). Total dissolved metals were analyzed using 0.22 µm filter-sterilized water acidified with 5% trace metal-grade nitric acid. Concentrations of total metals were quantified using an Agilent 7500ce ICP-MS by comparing to certified standards (Agilent Technologies, Environmental Calibration Standard 5183-4688) at MSU’s Environmental and Biofilm Mass Spectrometry Facility. Samples for anion analysis were filtered through 0.22 µm filters, and the filtrate was analyzed using a Dionex ICS-1100 chromatography System (Dionex Corp., Sunnyvale, CA) equipped with a 25 μL injection loop and an AS22-4x250 mm anion exchange column, using an eluent concentration of 4.5 mmol∙L-1 sodium carbonate and 1.4 mmol∙L-1 sodium bicarbonate flowing at a rate of 1.2 mL∙min-1. Samples for total carbon (TC), total nitrogen (TN), non-purgeable organic carbon (NPOC), and dissolved inorganic carbon (IC) were filtered through 0.22 µm filters, and the filtrate deposited in ashed glass vials filled with no headspace and capped with a septum. A Shimadzu TOC-CSH instrument with an attached TN module (Shimadzu Scientific Instruments, Columbia, MD) was used to measure TC/TN/NPOC/IC at MSU’s Environmental Analytical Lab (EAL). Filtered samples acidified with sulfuric acid (final pH<2) were also sent to the EAL for ammonium analysis using a Lachat QuickChem 8500 flow injection analyzer (Hach Co., Loveland, CO).
DNA extraction and metabarcoding analysis
DNA was extracted from each replicate using the FastDNATM Spin Kit for Soil kit (MP Biomedicals) according to the manufacturer’s instructions, with an additional 40s bead beating step. The V4 region of the 16S rRNA gene was targeted using the latest versions of the 515F-80R primers [13], 515F-A (GTGYCAGCMGCCGCGGTAA; [14]) and 806R-B (GGACTACNVGGGTWTCTAAT; [15]) using Phusion Hot Start II Hi Fidelity polymerase in 25 ul reactions. Thermocycling conditions were an initial denaturation at 98oC for 30 s, 22 cycles of 98oC denaturation for 15 s, annealing at 58oC for 30 s, extension at 72oC for 20 s, with a final extension at 72oC for 5 minutes. To facilitate multiplexing, dual-index barcodes were added in a second PCR using the Nextera kit (Illumina Inc.) with 10 cycles as above, but with an annealing temperature of 55oC. PCR reactions were quantified using the Quant-It HS dsDNA kit (Invitrogen) and measured using a Biotek H2 plate reader. Reactions were pooled at equal concentrations and sequenced on a MiSeq using V3 600 cycle kits.
Paired-end reads were merged, primers were removed, and sequences were quality filtered with an expected error rate of 0.5 using USEARCH [16,17]. Operational taxonomic units (OTUs) were identified using UNOISE3, which identifies biological sequences clustered into zero-radius OTUs (ZOTUs), similar to amplicon sequence variants. ZOTU tables were generated by mapping reads back to representative ZOTU sequences. Taxonomy of ZOTUs was identified using the IDTAXA online classifier via DECIPHER (http://DECIPHER.codes) [18]. Archaeal sequences were added to a reference tree generated using full and near full-length 16S sequences downloaded from GenBank and the Genome Taxonomy database [19]. Sequences were aligned using mafft and the reference phylogeny was constructed using maximum likelihood with RAxML[20,21]. Environmental sequences were aligned to the reference and added to the tree using pplacer, and the combined trees were annotated using the Interactive Tree of Life [22,23].
Sediment small molecule extraction
Sediment samples were extracted using several methods resulting in small molecule fractions that were characterized by LCMS analysis. Extraction began by thawing frozen samples in a 40ºC water bath and dislodging microbes from the sediment layer. Microbes were dislodged through two one-minute rounds of agitation on a vortex machine. Well-mixed samples were then centrifuged for five-minutes at 400 RPM to create a sediment pellet. The clear supernatant was then collected and placed in a clean vial and the procedure was repeated with the addition of a 5mL Milli-Q water wash. Wash supernatant was added to the original sediment-free supernatant. Combined supernatant was then centrifuged at 10,000 RPM for 15 minutes to create a cell pellet. The supernatant was collected in a clean vial for extracellular solid-phase extraction (SPE) analysis and enough phosphate buffer solution (1X PBS) was added to the cell pellet to completely cover the pellet.
The extracellular layer was first acidified to a pH of 2 using formic acid (Fisher Chemical, Hampton, ND) and Agilent Bond Elut PPL solid phase extraction cartridges (Agilent Technologies, Santa Clara, CA) were prepared. Preparation involved two cartridge volume additions of methanol (Fisher Chemical, Hampton, NH) followed by two cartridge volume additions of Milli-Q water and a final cartridge volume addition of methanol. SPE cartridges were then connected to a SPE manifold (VacMaster 10, Biotage, Uppsala, Sweden) and a vacuum pump to selectively concentrate and isolate extracellular small molecules. Acidified samples were passed through the cartridges and then placed under nitrogen to dryness. Clean vials were placed below the cartridges and captured extracellular small molecules were eluted using 1mL of methanol. Extracellular small molecules were further concentrated under negative pressure using a Concentrator Plus (Eppendorf, Hamburg, Germany) until dry and then stored in a -80 ºC freezer until ready for LCMS analysis.
The vial containing the cell pellet was centrifuged at 400 RPM and the PBS supernatant was removed. Two cell pellet volumes of extraction buffer were added consisting of 8M urea (Fisher Chemical, Hampton, NH), 0.1M Tris-HCL (MilliporeSigma, Munich, Germany), 50mM ethylenediaminetetraacetic acid (EDTA) (MilliporeSigma, Munich, Germany) and 1X protease inhibitor mix (MilliporeSigma, Munich, Germany). Cell pellets were next lysed in a two-part procedure. First, cell pellets in extraction buffer were placed in liquid nitrogen for 10 seconds then removed and allowed to thaw. This procedure was repeated three times. After the freeze/thaw procedure, cell pellets underwent two rounds of sonication using a Biologics Inc. Ultrasonic Homogenizer 3000 (Bioloics, Manassas, VA) set at a 40% duty cycle for three minutes.
After cell lysis, the intracellular samples were centrifuged at 15,000 RPM for 30 minutes to pellet cell debris. The resulting supernatant was removed and placed in a clean vial while the debris were washed with the addition of 50µL of extraction buffer. After agitating the debris with a vortex machine, samples were centrifuged at 15,000 for 30 minutes. The second supernatant was again removed and added to the vial containing the first supernatant. Four sample volumes of ice-cold acetone (Fisher Chemical, Hampton, NH) was then added to precipitate protein. Samples were placed in a -80 ºC for two hours to aid precipitate formation. After two hours, samples were centrifuged at 5,000 RPM for five minutes and the intracellular supernatant was collected and placed in a clean vial. As with the extracellular small molecule layer, the intracellular small molecule layer was concentrated under negative pressure to dryness and stored in a -80ºC freezer until LCMS analysis. When ready for LCMS analysis, both the intracellular and extracellular small molecule samples were reconstituted with 50µL of methanol:water (50:50) and placed in clean MS vials.
LCMS analysis
Samples were analyzed on an Agilent 6538 Q-TOF MS paired with an Agilent 1290 ultra-high performance liquid chromatography (UHPLC) (Agilent Technologies, Santa Clara, CA) using a 132Å, 2.2μm, 2.1mm X 150mm Cogent Diamond Hydride HPLC column (Microsolv, Greater Wilmington, NC) located in the Proteomics, Metabolomics and Mass Spectrometry facility at MSU. Ionization was accomplished via electrospray ionization in positive mode. Mobile phases A and B consisted of water with 0.1% formic acid and acetonitrile with 0.1% formic acid, respectively. A 15-minute UHPLC run time was used, starting with 100% mobile phase B and moving to 30% B in a linear gradient over 14 minutes. At 14 minutes, mobile phase B was increased back to 100% for the final minute of the UHPLC run. Flow was kept at 600µL/minute and the column compartment temperature was constant at 30°C.
LCMS raw data files were converted to .mzML files using MSConvert with Vendor peak picking and data mining was completed using mzMine [24,25]. A minimum intensity of 1,000 counts was used throughout the mining process along with a ppm error of 20 and a time discrepancy of 0.1 minutes for determining unique peaks. Molecules formulas were determined using mzMine’s formula generation identification feature with an error of 15ppm. After data mining, blank samples were used to remove residual features from the datasets. Features were only kept in the experimental data if they had an area five times greater than the blank sample. Using this method, almost 5,000 combined features were found in the extracellular small molecule samples and over 4,500 features were found in the combined intracellular small molecule samples. Datasets were then grouped and analyzed using MetaboAnalyst [26]. Samples were removed if not in over 50% of the samples and the data were filtered using their interquartile range value and normalized using the autoscale function.
Correlation analysis
The geochemical, 16S rRNA, intracellular and extracellular small molecule datasets were compared using correlograms. This was accomplished by utilizing Bioconductor and mixOmics packages in R [27,28]. Two datasets were compared at a time and the top discriminating features were determined using a partial least-squares discriminating analysis (PLSDA). The top features from each dataset were then compared between springs and years to determine their correlative relationship. An ANOVA was also conducted for each relationship to determine the significance of the correlation.
Table 1. Geochemical characteristics from 2017-2019
SWE denotes snow water equivalent percentage of the median for the previous year. DO is
dissolved oxygen. TC is total carbon. TN is total nitrogen. NPOC is non-purgeable organic
carbon.