Contesting the evidence for gene expression in lower oceanic crust

Li et al [Nature 579, 250-255 (2020)] report microbial gene expression in deep ocean crust, from microbial communities that have concentrations as low as 100 cell per cubic cm. This would be incredible, since that would be the lowest biomass sample ever analyzed for gene expression for a deep biosphere sample by several orders of magnitude. However, I have reanalyzed the data and show that the authors data are derived from contamination via seawater, humans, human skin associated bacteria, drilling fluid, and molecular reagent kits. The method that the authors used to produce their gene expression data is highly sensitive to DNA contamination, which is the case here. The authors claim that Pseudomonas, a known widespread contaminant in molecular reagent kits, is one of the most active groups in their lower crust samples and base their metabolic analysis on this contaminating organism. Here, I show that the gene expression data are derived from contaminating Pseudomonas (as well as other groups), and I show that the groups claimed as being active in the rock samples are also present in the negative controls at similar or higher relative abundance. For example, methane production in long term incubations was reported but gene expression from methanogens was 10 times higher in the drill fluid negative controls compared to the rock samples, demonstrating that the methane measured is likely derived from methanogen contaminants introduced from the drill fluid. There are no signs of active life here, but lipid biomarkers were found preserved in the rocks. Thus, the authors have sampled a fossilized biosphere, not a living one.