Versatile cyanobacteria control the timing and extent of sulfide production in a Proterozoic analog microbial mat

Cyanobacterial mats were hotspots of biogeochemical cycling during the Precambrian. However, mechanisms that controlled O2 release by these ecosystems are poorly understood. In an analog to Proterozoic coastal ecosystems, the Frasassi sulfidic springs mats, we studied the regulation of oxygenic and sulfide-driven anoxygenic photosynthesis (OP and AP) in versatile cyanobacteria, and interactions with sulfur reducing bacteria (SRB). Using microsensors and stable isotope probing we found that dissolved organic carbon (DOC) released by OP fuels sulfide production, likely by a specialized SRB population. Increased sulfide fluxes were only stimulated after the cyanobacteria switched from AP to OP. O2 production triggered migration of large sulfur-oxidizing bacteria from the surface to underneath the cyanobacterial layer. The resultant sulfide shield tempered AP and allowed OP to occur for a longer duration over a diel cycle. The lack of cyanobacterial DOC supply to SRB during AP therefore maximized O2 export. This mechanism is unique to benthic ecosystems because transitions between metabolisms occur on the same time scale as solute transport to functionally distinct layers, with the rearrangement of the system by migration of microorganisms exaggerating the effect. Overall, cyanobacterial versatility disrupts the synergistic relationship between sulfide production and AP, and thus enhances diel O2 production.


RNA extraction and library preparation
Samples for 16SrRNA sequencing were immediately stabilized by collection of ~2 mL mat into RNAlater (Ambion) in 15 mL sterile conicals to a final volume of 10 mL. After a 15 min incubation time at room temperature, samples were transported and stored at -20°C. Prior extraction, samples were thawed at room temperature and centrifuged at 2,000 x g for 30 min at 4°C. The supernatant was discarded by aspiration. Samples were extracted with FAST RNA Pro Soil direct kit (MP Bio). Lysis buffer was directly added into the falcon tube and all the pellet was transferred to lysing matrix E tube with cut filtered tips. Samples were mixed by vortexing and then were beat for 40 s/6.0, twice with a 3 min pause on ice. Downstream steps were according to the manufacturer protocol. Precipitation of nucleic acids was overnight at -20°C, followed by centrifugation at 14,000 x g at 4°C for 30 min. Pellets were washed twice with 70% ice-cold ethanol. After a brief drying at room temperature to remove all residual ethanol, pellets were dissolved in 1000µl TE buffer (pH 7.5). DNA was digested with RQ1 RNase free DNase (Promega), with the addition of 40 units ribonuclease inhibitor (RNasin, Promega) according to the manufacturer instructions. After digestion, RNA was isolated with a standard RNA PCI/CI purification kit (ROTH). Precipitation of RNA was overnight at -20°C with Isopropanol/Na-Acetate, followed by centrifugation at 14,000 x g at 4°C for 30 min. Pellets were washed twice with 70% ice-cold ethanol. After a brief drying at room temperature to remove all residual ethanol, pellets were dissolved in TE buffer (pH 7.5). Samples went through a second DNase digestion followed by a purification with a RNeasy Minelute kit (Qiagen). RNA was eluted with 30µl TE buffer (pH 7.5), passed through the column twice. RNA quality was checked with bioanalyzer using RNA chip Nano (Agilent) and Nanodrop (Thermo). Quantification of RNA was performed with Ribogreen quantification kit (Invitrogen) according to the manufacturer protocol. Pyrosequencing libraries were constructed as described previously [49] with modification. Briefly, the Ultrafast III qRT-PCR kit (Agilent) was used with Ba27f/519r primers, with initial 20 min at 45°C for reverse transcription. Each sample reaction was performed in triplicate. To ensure absence of residual DNA, for each sample was performed a reaction omitting reverse transcriptase. Additionally, for each sample a negative control was used. All triplicate amplicons for each sample were pooled and purified with Nucleospin kit (Macherey-Nagel) according to the manufacturer. Purified libraries were evaluated with Bioanalyzer on 7500 DNA chip and quantified with Picogreen dsDNA quantification kit (Invitrogen) Libraries were pooled in an equal molar ratio of 10 9 molecules µL -1 .

16S rRNA analysis
The next-generation sequencing analysis pipeline of the SILVA project (available at www.arbsilva.de/ngs) [50] allowed sequence and alignment quality-based filtering of the amplicons, to obtain aligned sequences, and a taxonomic classification to identify sequences belonging to taxa of interest. The quality cut-offs used were 50 for alignment and sequence quality, and classification cut-off of (percent query coverage + percent alignment identity)/2 > 95% was used to assign sequences to taxa. -NP‖ reads were not included in any further analysis due to the low sequencing output (Table S3). 12,058 cyanobacterial and 1428 deltaproteobacterial aligned sequencing reads were selected for further oligotyping analysis. Following the entropy analysis, oligotyping was performed with 26 and 25 components using the version 2.1 (available from https://meren.github.io/projects/oligotyping/), for cyanobacteria and deltaproteobacteria, respectively. To reduce noise, we imposed requirements that each oligotype must (i) appear in at least one sample, and (ii) have a most abundant unique sequence with a minimum abundance of 2. Final number of quality-controlled oligotypes for cyanobacteria and deltaproteobacteria were 75 and 11, respectively. Oligotype representatives were further checked for their taxonomic affiliations by adding them to the SILVA RefNR 132 guide phylogenetic tree using ARBparsimony addition tool [51]. Four oligotypes from Deltaproteobacteria had disagreements with the SILVA ngs classifications, and were eliminated from further downstream analysis (Supplementary Table S3). Further processing of the data was performed in R environment for statistical computing (https://www.R-project.org/), using package phyloseq (version: 1.19.1) [52].
In phyloseq, oligotypes were further filtered by removing those oligotypes not observed more than twice in at least 20 % of the samples.

FA-SIP
The total lipids extracts (TLE) of freeze dried mat samples were obtained following a modified procedure [34]. Namely, combusted sand (30-40 g per sample) was used to adapt this method originally designed for lipid extraction in sediments. The extraction consists of four steps using dichloromethane/methanol with phosphate-and trichloroacetic acid buffers (each twice). 2methyl-octadecanoic acid was used as internal standard.
After removal of elemental sulfur, an aliquot of the TLE was saponified [36]. The procedure includes a base saponification using potassium hydroxide in methanol, base extraction of the neutral lipids and acid extraction of the free fatty acids.         Table S2: Fluxes of O 2 and S tot measured in 1-3 replicate spots in the first and second run of the flow chamber experiment together with the theoretical stoichiometry of aerobic sulfide oxidation (aSOx).Concentrations are in µM. Fluxes including depth integrated gross rates (DIR) are in mmol m -2 h -1 and shown as negative when consumptive. Shaded rows refer to measurements in the mat spot where gross rates were assessed. These rates are plotted in Figure 1.   Table S3: Available as a separate supplementary file. Results of 16SrRNA sequence analysis using the next-generation sequencing analysis pipeline of the SILVA project (available at www.arb-silva.de/ngs), including sequence classifications and oligotyping results (https://github.com/merenlab/oligotyping). MID08 was obtained from a mat sample taken at Main Spring prior to the experiment. MID17, MID15, MID11 and MID16 were obtained from samples taken during the first flow chamber run in the dark, during exposure to 19 µmol photons m -2 s -1 , during first exposure to 315 µmol photons m -2 s -1 and after addition of DCMU (Fig. 3c). MID14 and MID13 correspond to samples taken in the dark and during exposure to 315 µmol photons m -2 s -1 in the second experiment (Fig. 4c).