Bubbles emerge from a methane seep offshore Virginia. (Image: NOAA Okeanos Explorer) Stable isotope probing proteomics (proteomic SIP) combines advanced analytical technology, computational wizardry, and experimental design to address one of microbial ecology’s most fundamental questions: which organisms are present and what exactly are they doing? In a recent study, some colleagues and I gave this method one of its biggest tests to date, adding isotopically labeled nitrogen to methane-infused deep-sea sediment from the Hydrate Ridge methane seep. By tracing this heavy nitrogen into the enzymes that enact a cell’s metabolic agenda, we were able to see which biochemical responses were most associated with high-methane conditions. Overall, nearly 3500 unique proteins were identified, everything from proton-channeling ATPases that act as cellular power generators to recycling centers that degrade old enzymes. However, just 11% of these proteins had the 15N label, highlighting the low turnover of these very slow-growing microbes. Other lines of evidence revealed active growth (for example, we saw more of the cell clumps typically involved in methanotrophy over time) but it appears that most of the proteins in the initial sediment samples, made before we ever scooped them off of the seafloor, stuck around for several months. Old proteins are often juicy fodder for other organisms, and this long protein retention time could point to a microbial habitat tuned to subsist on vent-derived chemicals rather than organic material. In other words, microbes around methane seeps may not die fast enough to support scavengers.
The detection of 15N-enriched proteins is strong evidence that such proteins are useful under the seep-simulating conditions of our experiments. In this context, it wasn’t too surprising to find a lot of methane metabolism proteins showing up in the enriched fraction. But a closer look revealed some unexpected details. Eight proteins are directly implicated in methanogenesis, the biological production of methane from carbon dioxide and hydrogen. Are the methane-eating microbes at seafloor seeps somehow performing the reverse reaction with the same types of enzymes? Many researchers have believed this to be the case, but one key enzyme - abbreviated as Mer - had evaded detection until our study. With several versions of methanotroph-derived Mer enriched in 15N, it appears that “reverse methanogenesis” is indeed a prominent form of methane consumption. As important at Mer may be, the dominant protein in methane-makers and methane-eaters is Mcr, which performs the most biochemically complicated step. We found that 10.4% of protein fragments across all experiments could be attributed to Mcr; that number rose to a whopping 35% when only 15N enriched proteins were considered. What’s more, we weren’t seeing a single methanotroph dominate the system; rather, it seems that there is room for several distinct methane-consuming organisms. Whether this is a widely accessible niche or there are subtly distinct niches among the methanotrophs - each dominant in its own specific realm - remains to be seen. A critical caveat of any proteomic study is that you’re only going to find what you’re looking for. Peaks in the mass spectral data were cross-referenced with an enormous database of genomic information - roughly 1.4 million genes sequenced from Hydrate Ridge, other methane seeps, as well as full genomes from cultured organisms. A protein fragment mass leads you to a gene, but if this gene isn’t associated with any particular function, then you’re left with a “hypothetical protein”, molecular dark matter with an unknown role. Of the proteins found in all analyses and enriched in 15N (the subcategory of proteins most reliably linked to methanotrophy), 26 had this dubious designation. It doesn’t really help us figure out what these mysterious proteins are doing, but they seem to be pretty important, and are compelling targets for future analysis. SIP proteomics is a promising way to link identity to function in mixed microbial communities, and as computational and analytic methods become equal to the task, the challenge of making sense of big data becomes paramount. To Chongle Pan, a Staff Scientist at Oak Ridge National Lab's Computer Science & Mathematics and BioSciences Divisions who developed our study’s computational architecture, “it really takes a tremendous amount of work to mine some biological stories out of these results.” He hopes to “eventually create a workflow from the proteomic SIP results to a metabolic model of a microbial community” alongside system-specific scientific experts. Combining analytical precision, computational innovation, and biological interpretation, a future of understanding microbial communities in modular, functional detail is tantalizingly close.