Hypersaline sapropels act as hotspots for microbial dark matter

Present-day terrestrial analogue sites are crucial ground truth proxies for studying life in geochemical conditions close to those assumed to be present on early Earth or inferred to exist on other celestial bodies (e.g. Mars, Europa). Although hypersaline sapropels are border-of-life habitats with moderate occurrence, their microbiological and physicochemical characterization lags behind. Here, we study the diversity of life under low water activity by describing the prokaryotic communities from two disparate hypersaline sapropels (Transylvanian Basin, Romania) in relation to geochemical milieu and pore water chemistry, while inferring their role in carbon cycling by matching taxa to known taxon-specific biogeochemical functions. The polyphasic approach combined deep coverage SSU rRNA gene amplicon sequencing and bioinformatics with RT-qPCR and physicochemical investigations. We found that sapropels developed an analogous elemental milieu and harbored prokaryotes affiliated with fifty-nine phyla, among which the most abundant were Proteobacteria, Bacteroidetes and Chloroflexi. Containing thirty-two candidate divisions and possibly undocumented prokaryotic lineages, the hypersaline sapropels were found to accommodate one of the most diverse and novel ecosystems reported to date and may contribute to completing the phylogenetic branching of the tree of life.


Analytical techniques
The extraction and estimation of dissolved CH4 concentrations at the sediment-water interface were determined at the sampling site, using the head-space method on an instrumental package (West System, Pontedera, Italy) equipped with semiconductor (range 0-2000 ppmv; lower detection limit of 1 ppmv; resolution 1 ppmv), catalytic (range 2000 ppmv -3 % v/v) and thermal conductivity (range 3 % -100 % v/v) detectors.
The organic matter content (% OM) of sapropel samples was determined by using the loss-onignition (LOI) and wet oxidation techniques. LOI determination was executed as described by Ball 1 , with the assumption that the total organic carbon content (% TOC) is 50 % of the OM.
The dichromate oxidation followed by titration was performed as described by Walkley and Black 2 , following a % TOC correction estimation by a factor of 1.16 (equivalent to ca. 86 % organic carbon recovery).
As most of the organic nitrogen in aquatic sediments seems to be deposited under the form of amide bonds 3 , the total alkaline extractable protein content was estimated in the sapropels by the Lowry method 4 modified by Rice 5 . Ammonium nitrogen was quantified by colorimetry using a Lambda 25 spectrophotometer (Perkin Elmer, Beaconsfield, UK). The contents of total nitrogen (TN) as bound nitrogen (including free ammonia, ammonium, nitrite, nitrate and organic nitrogen, but not dissolved nitrogen gas) were assessed by thermal oxidation, conversion of nitrogen oxides to NO, followed by NO oxidation with ozone and subsequent chemiluminescence detection using the Multi N/C 2100 S Analyzer (Analytik Jena, Jena, Germany). Organic nitrogen (ON) in the form of dissolved organic nitrogen was calculated by subtracting ammonium nitrogen, nitrate and nitrite nitrogen from the total nitrogen. Total carbon (TC), total dissolved carbon (TDC) and dissolved organic carbon (DOC -calculated as difference between total dissolved carbon and dissolved inorganic carbon measured upon sample acidification) were quantified by combustion and NDIR detection of CO2 released using the multi N/C 2100 S Analyzer. Chloride (Cl -), carbonate (as CaCO3) and bicarbonate (HCO3 -) anions were measured by titrimetric methods. Phosphate (PO4 3-), sulfate (SO4 2-), nitrate (NO3 -), and nitrite (NO2 -) were measured by ion chromatography on a 761 Compact IC (Metrohm, Herisau, Switzerland). The concentration of sulfides was determined by methylene blue method after fixation of samples with 2 % (v/v) Zn-acetate 6 . Concentrations of major metals (Mg, Ca, Na, K, Fe, Mn) were measured by inductively coupled plasma atomic emission spectrometry (ICP-AES) using OPTIMA 5300 DV spectrometer (Perkin Elmer, Norwalk, USA) after homogenization of sediments and digestion in concentrated aqua regia. Relative expanded measurement uncertainty (UErel.) was calculated according to ISO-GUM using a coverage factor (k) of 2 (k = 2), equivalent to a confidence of approximately 95 %. Uc ranged from 4.5 to 13 % depending on the compound as follows: 8 % for TC, HCO3 -, Fe, Mn, Mg and Ca; 9.5 % for PO4 3-; 9.8 % for Na + ; 10 % for TN, TDC, CO3 2-, SO4 2-, HSand K; 11 % for NO3and ON; 13 % for NO2 -, NH4 + and Cl -.
For X-ray fluorescence analysis (XRF), sapropel samples were dried to constant weight at 60 °C and grinded to pass a < 200 µm sieve. Major and trace elements (Cl, Fe, Ca, K, Ti, S, Mn, Cu, Zn, etc.) were determined by using a dispersive X-ray fluorescence spectrometer (Innov-X Alpha 6500, The Netherlands) equipped with X-ray tube, W anode and Si PiN diode detector. XRF spectra were recorded at Kα line on samples placed in small plastic cups covered by Mylar X-ray window film. The energy-dispersive X-ray (EDX) spectroscopic analyses of dried sapropel samples were performed over an area of 250 x 200 µm 2 , at 20 keV beam energy using a FEI Quanta 3D FEG dual beam SEM equipped with an energy dispersive X-ray detector.
The mineral composition of bulk sediment samples was assessed by X-ray diffraction (XRD), using a Bruker D8 Advance diffractometer with CoKα1 with λ = 1.78897, Fe filter and a onedimensional detector using corundum (NIST SRM1976a) as internal standard. The data were collected on a 5 -64 o 2θ interval, at a 0.02 o 2θ, with the measuring step of 0.2 seconds. The clay fraction was analyzed on oriented samples (pipetted from suspension in distilled water), which were further treated with ethylene glycol. Data were collected on the same interval as mentioned above, with a measuring step of 0.5 seconds. The identification of mineral phases was performed with the Diffrac.Eva 2.1 software from Bruker AXS, using the PDF2 (2012) database.

SEM, cell counts, chlorophyll a and total carotenoids analyses
For SEM analyses, samples of sapropel and plant material recovered from sapropels were fixed in 4 % glutaraldehyde, filtered onto 0.25 µm polycarbonate Millipore filters (Bedford, MA, USA) and subsequently washed with 1 x PBS (pH 7.4). Henceforth, they were subjected to an alcohol dehydration series after which they were fixed with 100 % hexamethyldisilizane (Electron Microscopy Sciences, Hatfield, Pennsylvania) and air-dried. After mounting on SEM specimen holders they were sputter-coated (Agar Automatic Sputter Coater, Agar Scientific, UK) with 7-10 nm of gold and visualized with a Jeol JEM 5510LV scanning electron microscope operated at 15 kV. For total cell counts, sapropel pore water samples were fixed in 1 % glutaraldehyde and filtered through black, gridded cellulose ester membrane filters (0.45 µm, d=47 mm).

Fungal community analyses
The collected sapropel samples were subsampled (i.e. 3 subsamples from each sapropel sample) and small pices of decaying plant material (leaves) were recovered from Ursu and Fara