Water column dynamics control nitrite-dependent anaerobic methane oxidation by Candidatus “Methylomirabilis” in stratified lake basins

We investigated microbial methane oxidation in the water column of two connected but hydrodynamically contrasting basins of Lake Lugano, Switzerland. Both basins accumulate large amounts of methane in the water column below their chemoclines, but methane oxidation efficiently prevents methane from reaching surface waters. Here we show that in the meromictic North Basin water column, a substantial fraction of methane was eliminated through anaerobic methane oxidation (AOM) coupled to nitrite reduction by Candidatus Methylomirabilis. Incubations with 14CH4 and concentrated biomass from this basin showed enhanced AOM rates with nitrate (+62%) and nitrite (+43%). In the more dynamic South Basin, however, aerobic methanotrophs prevailed, Ca. Methylomirabilis was absent in the anoxic water column, and no evidence was found for nitrite-dependent AOM. Here, the duration of seasonal stratification and anoxia seems to be too short, relative to the slow growth rate of Ca. Methylomirabilis, to allow for the establishment of anaerobic methanotrophs, in spite of favorable hydrochemical conditions. Using 16 S rRNA gene sequence data covering nearly ten years of community dynamics, we show that Ca. Methylomirabilis was a permanent element of the pelagic methane filter in the North Basin, which proliferated during periods of stable water column conditions and became the dominant methanotroph in the system. Conversely, more dynamic water column conditions led to a decline of Ca. Methylomirabilis and induced blooms of the faster-growing aerobic methanotrophs Methylobacter and Crenothrix. Our data highlight that physical (mixing) processes and ecosystem stability are key drivers controlling the community composition of aerobic and anaerobic methanotrophs.

PE 300 method (V3 reagent kit). Library preparation and sequencing was done at the Genomics Facility Basel (D-BSSE ETHZ and Basel University). Raw sequence data are made available at NCBI under the BioProjectID PRJNA672280 with the accession numbers SRR12936362 through SRR12936382; MH111698 through MH113143.

Sequence analyses
Quality control of raw reads, initial sequence treatment, and taxonomic assignment is described in the main manuscript. Sequencing results are summarized in Table S2. 16S rRNA sequence data were analyzed in R (v3.5.1) (R Core Team, 2014, http://www.r-project.org/) using mostly the libraries: phyloseq [2], vegan [3], and ggplot2 [4]. Rarefaction curves show that 25 cycles yielded better estimates of the species richness than 18 cycles ( Figure S1). ANOVA in R was used to test whether different alpha diversity measures (i.e., Observed richness, Chao1, Shannon and InvSimpson) were affected by either the sequencing depth or the amplification cycle number. Results show ( Table S3) that Shannon is significantly affected by both the amplification cycle number and sampling depth, while Chao1 depends on the sampling depth [5]. Accordingly, samples were rarefied for alpha diversity estimates and for comparing the numbers amplified sequence variants (ASVs) in the two libraries (Table   S2). Principal coordinate analysis (PCoA) shows that both 18 and 25 PCR amplification cycles yielded very similar microbial community structures ( Figure S3). Ultimately we choose 25 amplification cycles for the microbial community analysis, because of the better sensitivity towards minor taxa. Sample Ga_100m in the 25 cycle library had particularly low read and ASV numbers (Table S2) and was excluded from any further analysis. After removing sample Ga_100m and eliminating both mitochondrial and chloroplast 16S rRNA gene reads, as well as empty ASVs, a total of 8717 ASVs remained in the unrarefied 25-cycle library dataset. No further filtering of low abundant reads was done.
In this dataset we identified 41 ASVs of potential methanotrophs (Supplementary Excel File S1) in the two basins of Lake Lugano, North Basin and South Basin. The guild of methanotrophs was dominated by seven highly abundant ASVs with relative abundances of > 1 % in at least one sample, including sequences of uncultured Methylobacter sp. (ASV19, ASV42, ASV5, ASV18), uncultured Crenothrix sp.
Of the 41 ASVs of potential methanotrophs, 20 ASVs were common in both lake basins, whereas 9 ASVs were only detected in the North Basin and another 12 in the South Basin only. The two latter groups consisted of low-abundant taxa, thus the apparent habitat specificity was rather a consequence of incomplete community sampling by sequencing than reflecting true ecological preference for one or the other basin. The fact that the 16S rRNA gene sequences of the MOB common to both basins are identical suggests that the two basins are microbiologically connected.
Clear habitat preference, however, was observed for the highly abundant ASV9 (Ca. Methylomirabilis), which was consistently detected in the water column of the meromictic North Basin of Lake Lugano only. In the South Basin, Ca.
Methylomirabilis was not detectable in any water-column samples. Nonetheless, ASV9 was present in a surface sediment (0-2 cm depth) sample retrieved at 93 m depth, suggesting that sedimentary Ca. Methylomirabilis could colonize the anoxic bottom water for the South Basin, provided that the environmental conditions were favorable. Table S1. Design of the four different sets of forward and reverse primers used for library preparation. The primers contained 0-3 additional ambiguous bases (indicated in bold red) between the adapter sequence and the amplicon PCR primer 515F-Y/926R

Supplementary tables
[1] to increase the nucleotide diversity and improve template generation during Illumina sequencing.

Primer
Sequence