Metagenomic Evidence for Sulfide Oxidation in Extremely Acidic Cave Biofilms

Abstract

Snottites are extremely acidic (pH 0–2) biofilms that form on the walls and ceilings of hydrogen sulfide-rich caves. Recent work suggests that microbial communities including snottites and related cave wall biofilms accelerate cave formation by oxidizing sulfide to sulfuric acid. Therefore, we used full-cycle rRNA methods and metagenomics to explore the community composition and sulfur metabolism of snottites from the sulfidic Frasassi and Acquasanta cave systems, Italy. Acquasanta snottites were dominated by strains of Acidithiobacillus thiooxidans, with a smaller population of Ferroplasma sp. Frasassi snottites were also dominated by At. thiooxidans but with a more diverse community including relatives of ‘G-plasma’ (Thermoplasmatales), Acidimicrobium, and rare taxa. We identified diverse homologues of sulfide:quinone oxidoreductase (SQR) in the metagenomic datasets. Based on phylogenetic analysis, the numerically dominant At. thiooxidans populations have four different types of SQR, while Ferroplasma has two and Acidimicrobium and G-plasma each have one. No other genetic evidence for sulfur oxidation was detected for either Acidimicrobium or G-plasma, suggesting that they do not generate sulfuric acid. Our results confirm earlier findings that At. thiooxidans is the dominant primary producer and sulfide oxidizer in sulfidic cave snottites.

Keywords:

• archaea
• biofilm
• community structure
• hydrogen sulfide
• karst
• subsurface microbiology

Acknowledgments

We thank A. Montanari at the Osservatorio Geologico di Coldigioco for logistical support during field expeditions, S. Mariani, S. Carnevali, S. Ceroni, M. Mainiero, and S. Recanatini for technical caving expertise, and S. Dattagupta and R. McCauley for field assistance. We also thank D. Grove for advice concerning pyrosequencing methods, and the Genomics Core Facility of the Huck Institutes of the Life Sciences of Penn State (University Park, PA) for performing the sequencing. We also acknowledge T. Canich and D. Futrick for computing support, M. Mansor and B. Flood for insightful discussion, and external reviewers whose input substantially improved the manuscript.