By Noah Fierer
Let’s say a microbial ecologist wants to identify which bacteria are active in a given environmental sample at a given point in time. For example, you may want to determine which bacteria are active in a desert soil after a large rainfall event. This is often done by comparing the representation of microbial taxa in the RNA pool versus their representation in the DNA pool. The assumption is that the DNA captures both active and inactive (or dormant) taxa while analysis of the RNA pool allows one to identify the more active members of a given community. In other words, taxa that are over-represented in the RNA pool (i.e. those with high rRNA:rRNA gene ratios) should be more active. This paradigm is firmly entrenched and rarely questioned. However, this paradigm is also likely false – or at least a gross oversimplification.
A number of scientists have already pointed out the problems associated with blindly assuming that rRNA:rRNA gene ratios indicate the activity of bacterial or archaeal taxa in environmental samples – most notably this beautiful paper by Steve Blazewicz et al. Nevertheless, these points are worth further discussion as this paradigm is still treated as dogma in many recent proposals and papers.
So – why is it an oversimplification to blindly assume that rRNA:rRNA gene ratios can be used to differentiate between ‘active’ and ‘inactive’ microbial taxa when analyzing environmental samples?
The assumption that rRNA content can be used to infer microbial activity is, like many assumptions in microbiology, rooted in studies of E. coli and other model species where cells growing rapidly under controlled conditions often have higher rRNA concentrations than cells growing under resource-limited conditions. However, if microbial ecologists have learned anything over the years, it is that few environmental microbes behave like E. coli growing in the lab. The relationship between rRNA content and growth rate is often inconsistent and highly variable. Starved cells can maintain high rRNA concentrations and the correlation between rRNA content and growth rate can vary dramatically depending on the taxon in question. Even dormant cells will have rRNA and the rRNA concentrations in these ‘inactive’ cells can be surprisingly high. References to support these points are provided in the paper by Blazewicz et al – a ‘must read’ for anyone interesting in using rRNA to infer microbial activity levels. As they state:
“…. conflicting patterns between rRNA content and growth rate indicate that rRNA is not a reliable metric for growth or activity and in some cases may be grossly misleading”
In addition to all of these reasons not to trust rRNA:rRNA gene ratios as a robust indicator of activity levels, ‘relic’ DNA (sensu Carini et al.) can persist in some environments and this extracellular DNA or DNA from dead cells may further obscure attempts to identify active or less-active populations (e.g. this study). Plus, undersampling of rRNA or rRNA gene pools can further complicate efforts to identify active populations in those environments with highly diverse microbial communities (see this study).
On a related note, it is worth highlighting that an ‘active population’ can be defined many ways (like other vague terms frequently used by microbial ecologists, including ‘core microbiome’ and ‘phylotype’). Active compared to what? Actively replicating? Actively producing extracellular enzymes? Actively generating ATP? Any intact cell in the environment will have some low level of metabolic activity to support basic cell maintenance, so what delineates an ‘active’ population from an ‘inactive’ population needs to be defined carefully. Typically, ‘activity’ is defined by the representation of a given taxon in the RNA pool, but this is a tautology. Moreover, ‘activity’ is usually assumed to be synonymous with growth – but a microbe can be ‘active’ and contributing to key processes without growing. I would argue that, in many environments, activity without replication is the norm – conditions in soil, water, sediments, and other environments are often sufficiently favorable to maintain some level of metabolic activity, but insufficient to sustain consistent cell division.
The aforementioned problems with using rRNA:rRNA gene ratios to infer taxon-specific levels of activity in environmental samples do not imply that metatranscriptomic analyses are useless. There is a lot that can be learned by analyzing the pool of RNA in a given sample – including the use of mRNA analyses to track microbial responses to changes in biotic or abiotic conditions (but see this paper). Moreover, there are other approaches (including stable isotope probing) that could provide alternative strategies for discriminating between active and dormant microbial taxa.
I recognize the utility of differentiating between active and less-active (or dormant cells) for linking microbial communities to microbial processes of interest. However, until the assumptions have been validated – it is not appropriate to use rRNA:rRNA gene ratios as a metric of microbial activity, especially when ‘activity’ is not explicitly defined. The paradigm is simply not robust.