Fierer Lab

Exploring the structure and function of microbial communities

The Great Plate ‘Debate’ – why we care and what we know about gaps in culture collections

By: Noah Fierer

April 11, 2019

It is a seemingly simple question: What percentage of bacterial and archaeal taxa in a given sample have been cultivated? Or, to phrase this another way: How well do our pre-existing culture collections cover the breadth of prokaryotic1 diversity found in environmental or host-associated samples?

This topic may seem arcane, but it is important. Why?  Because much of what we know about microorganisms comes from studying them in culture. Yes – we can now get bacterial and archaeal genomes from ‘uncultivated’ taxa, but there is a lot of information that is difficult, if not impossible, to glean from genomic data alone (e.g. substrate uptake kinetics, substrate preferences, cell morphologies, temperature optima). Plus, if we want to assign a function (annotate) a gene of unknown or unconfirmed function, it really helps to be able to study an organism in culture. Most of the genes recovered from genomes or metagenomes cannot currently be annotated (and even if they are annotated – the annotations may not necessarily be accurate) – our ability to link DNA sequences to gene products and ‘functions’ is largely built from the study of a few well-described, well-characterized bacterial and archaeal isolates. So – why does the ‘% cultivated’ question matter? Because, if the taxa in your sample are not closely related to those that have been cultivated – you may not necessarily be able to infer the relevant attributes of those taxa. If your vocabulary is very limited – you may be able to follow Goodnight Moon but you will have more trouble understanding The Moon is a Harsh Mistress.

Two recent studies have tried to address this topic: “Phylogenetically novel uncultured microbial cells dominate Earth microbiomes” and “High proportions of bacteria are culturable across major biomes“.  Although the two papers appear contradictory (as evident from the diametrically opposed titles) and there are important differences in the approaches used, many of their main findings are far more similar than the titles would suggest. I don’t want to get into the weeds about why the studies may differ and which study is better (though I note that one is far more comprehensive). It is also important to recognize that this seemingly simple question (What % of percentage of microbes have been cultured?) is actually difficult to answer definitively as a number of factors must be considered, including: Are the cultivation-independent surveys effectively capturing all the taxa in a sample?, How similar does the 16S rRNA gene sequence from a sample have to be to a cultured isolate to flag it as ‘cultured’?, How do we even compile a list of which taxa have been cultured in the first place?, and so forth…..

Instead of discussing the relative merits of one study versus the other – I want to focus on what we know – points that are not debatable.

1) There are major lineages for which few or no cultured representatives currently exist. Culture collections are strongly biased towards a few phyla and it has been estimated that nearly 40% of bacterial phyla have no cultured representatives. Even within well-described genera like the genus Mycobacterium – the majority of mycobacterial strains found in soil have not yet been cultivated. One does not have to search hard across the tree of life to find the obvious gaps where cultivated taxa are few and far between.

2) No strain is necessarily representative of all strains within a given ‘species’. Even if cultured representatives of a given ‘genus’ or ‘species’ exist – there can be an enormous amount of genotypic and phenotypic variation within taxa so those cultured isolates by no means reflect the taxa found in your sample. Just because your 16S rRNA gene sequence is a close match to a cultivated isolate does not mean you can predict with certainty what substrates it consumes, its environmental tolerances, its growth characteristics, or what metabolites it may produce. Don’t underestimate the importance of ‘pan-genomes’.

3) Just because an organism has been cultured doesn’t mean it is easy to culture. Maybe someone just got lucky in culturing a strain previously. Maybe the effective strategy for culturing that strain is lost to history. Maybe the perfect undefined media for growing the strain can’t be recreated (eye of newt, toe of frog, wool of bat….). Maybe you just don’t want to waste years trying to cultivate the microbial equivalent of a snow leopard. Regardless of the reason, never assume that just because someone else has been able to isolate a given organism means that you will be able to do the same from your samples.

4) ‘Cultured’ does not mean fully-characterized. Even if someone, somewhere has cultivated and isolated a given organism – that strain may not still be available for study nor does it mean that strain has been well-characterized. Freezers around the world are filled with undescribed bacterial strains longing for their moment of glory. Even for those strains described in IJSEM (one of the foremost venues for strain descriptions and a good alternative to counting sheep if you are suffering from insomnia) – the phenotypic information collected for those strains is often incomplete, inconsistently-reported, or possibly irrelevant to your research questions (I’m looking at you fatty acid profiles).

5) Some environments clearly have more ‘culturable’ taxa than others. Soils and sediments are often the ‘worst case’ scenarios – the most abundant taxa found in these environments often have no closely related cultivated representatives (a point made clearly in Figs 2 & 3 of this paper). Even for the 511 bacterial and archaeal taxa found to be most abundant and ubiquitous in soil, <30% are remotely similar to representative type strains. If you study cheese bacteria, skin bacteria or, perhaps, human gut bacteria – you may be in luck as more of your taxa may have closely-related cultured representatives (but note point #2).

6) ‘Uncultivated’ is not synonymous with ‘unculturable’. There are many reasons why taxa can be difficult to cultivate (see this nice review of the topic). Often, simply using a broader range of media types and incubation conditions can yield a wealth of previously ‘uncultivated’ lineages. Plus – many of the currently uncultivated taxa may simply just have long generation times making them difficult to cultivate using traditional approaches where the ‘weeds’ will dominate. Using dilution-to-extinction approaches, long incubation times, low-nutrient media, and being able to detect growth of small populations have all been used to cultivate ‘novel’ taxa. Is this process easy? – no – SAR11 wasn’t cultivated in a day – but the taxa that are easy to grow in the lab have been cultivated for decades and a bit of creativity combined with a healthy dose patience can go a long way.

Although culturing of bacteria and archaea is often considered a ‘lost art’ – it is incredibly valuable. If you want test how a given bacteria may affect a host or what it is actually doing in its environment – good luck doing that with sequence information alone – there is no substitute for being able to grow an organism in the lab.

1: Yes – I’m using the term ‘prokaryote’ here – my guess is that the 4 people that will read this blog post know that archaea and bacteria are distinct domains of life.