A Benefit of the Human Microbiome Project: Putting the "99% of Bacteria Are Unculturable" Canard to Rest

One of the common sayings in microbiology that drives me up a wall is the notion that 99% of all bacteria can't be grown in the lab. This false statement stems from the observation that if you take any sample (soil, water, clinical samples) and look under a microscope we see many more bacterial cells that contain DNA than we can grow. The problem is that, if you look at the paper that claimed this, they attempted to grow bacteria on a single, rich medium.

One weekend, when I was a post-doc, I did a very simple comparison. I took standard rich lab medium ('nutrient agar' which is basically one of those high protein nutrient bars minus the artificial flavorings. That's why I don't eat them. Seriously, if you dilute one bar in a liter of water, you basically have nutrient agar). Anyway, the other medium was a home-brew very-low nutrient agar that had one-thousandth of the ingredients in the standar nutrient agar. I also used extremely high-grade agar (electrophoresis grade) to solidify the medium (standard agars actually contain a fair amount of nutrients). I did this because we know that many bacteria can't grow in environments with a lot of carbon or protein sources--they basically end up choking on their own excreta. I also added cycloheximide to kill off fungi, as well as catalase to protect the bacteria from their own metabolisms.

After 72 hours, my special plates had 20-40 times as many bacteria as the standard plates. So I've never bought into the 99% percent.

One of the interesting developments from the Human Microbiome Project is that we're discovering that, in fact, we can culture many human-associated bacteria. Since we have culture-independent, DNA sequencing-based methods of determining which bacteria live in and on you, we can also see if, when we try to grow things, if those bacteria show up.

Indeed, they do. Several groups, working with gut and oral microbes, are able to culture most of what the DNA surveys are yielding--which is also a good confirmation of the molecular methods. Yes, they have to be clever: Tom Schmitt of Michigan State University grows bacteria from your gut (yes, shit) at levels of oxygen typically found in the intestinal tract and finds all sorts of 'uncultured' bugs. Floyd Dewhirst, who works on oral microbes, and who has done extensive surveys of bacteria from the mouth says that 92% of bacteria for which we have sequence data exist in culture collections (again, you have to be clever). And Jeff Gordon reported that 70% of genera found in the gut, and ~60% of species could be grown without too much work (you did have to wait a week, however).

These are very cool results, and I hope they put the 99% of bacteria are unculturable myth to rest.

Seriously, have some pride, fellow microbiologists, and be clever when you isolate....

More like this

Preach it, brother! I did a similar experiment, as I'm sure many others did - Kim Lewis has done similarly interesting things with environmental samples. In his work, it's iron acquisition that is the problem. Unculturable means "didn't really try to culture it".

By Paul Orwin (not verified) on 03 Sep 2010 #permalink

Yep, I did some very similar experiments during my grad work. You have to be very creative with your media, and be very patient. I was able to get an "unculturable" bacterium growing in media that was highly specialized to its environment. Unfortunately, many microbiologists today are really bacterial geneticists. The tedious art of perfecting culture media is slowly fading away.

Part of the problem, I think, is that culturing these organisms is a lot of work with very little payoff. Perfecting a culture medium and growth conditions takes a lot of time, especially if the organisms are slow growers. And the payoff? Practically nothing. I doubt you can get a publication in a decent journal for culturing an unknown bug these days. So... is there really any incentive for this kind of work?

I thought the 99% rule came from sequencing ribosomal RNA sequences directly from environmental samples, and finding that most of them had never been recovered before. This is still a long way from being 'unculturable under any circumstances.' I suspsect that there may be lots of syntrophic organisms that don't do well in isolation. Of course, if we knew what they needed, culturing them should not be a problem.

I agree that "unculturable" isn't the appropriate term. I suspect that if good enrichments (say 10% of total) are included 100% success is possible with enough effort. Effort being the key point.

That said, the HMP isn't the best place to argue this point. Most of what is associated with the human body comes from a few bacterial phyla where, incidentally, we have the most success in obtaining isolates. So, how success in growing human associated bacteria relates to the full diversity of bacteria in the environment is unclear.

The bigger picture, I think, is that many fields of biology suffer from a lack of creativity and thinking outside of the box. There's a lot of old dogma out there and untested superstition that hasn't been carefully scrutinized. Many biologists don't have broad enough training and/or are too timid to think outside the box.

Thank goodness many now use "not yet cultured." I work in the ocean and found that they key to "unculturable" SAR11 was pyruvate, glycine, and methionine. The key to OM43 was methanol. Their genomes told the story. The truth is out there!

Saddest note: I have been asked more than once by undergrads why I spend so much time trying to culture bacteria that everyone already knows are unculturable.


I had reviewers raise major objections to a paper (this was 10-15 years ago) because our plate counts (on a dilute medium) were half of our direct microscopic counts, and they thought this was an impossible high percentage for culturability. Luckily, we had already established credentials and could argue for our data's reliability. Having said that, I think we are in fact talking about two separate issues, both interesting in their own right. Is the difference between direct counts and plate counts due to a limited taxonomic range of culturable organisms? Or is the discrepancy because some of the microscopically visible cells, though taxonomically affiliated with the organisms growing on the plate, are in a physiological state that limits their ability to form colonies on a plate? The former suggests we are missing many taxonomic clades of organisms, while the latter suggests we need to understand microbial physiology better. Also, you seem to be arguing that the discrepancy between microscopic counts and direct counts doesn't exist - while your dilute medium results support that claim, the microbiome results don't really address that claim, if the discrepancy is real and due to physiology.

Finally, given that I have just spent oodles of time isolating, and am now characterizing what I think are some novel Archaea, I sure hope IJSEB is still interested in publishing on new isolates.

By Mark Wilson (not verified) on 03 Sep 2010 #permalink

There's part of me that thinks everything is TOO genetic oriented. I have a lot of respect for a neighboring lab which is trying to cultivate sponge symbionts. I'm working with anaerobes, which have their own share of headache, but at least I know what growth conditions they require.

By grad student (not verified) on 03 Sep 2010 #permalink

Lest graduates students think everything is already cultured, I think it is important to note that most clone library surveys are turning up lots and lots of unaffiliated phylotypes. For example, at our site, Boiling Springs Lake in Lassen Volcanic National Park, more than 75% of the environmental 16S rRNA gene sequences shared less than 85% identity with their closest cultured BLAST match. Once people know that new organisms are there to find, they look more closely - there are several recent examples.

By Mark Wilson (not verified) on 03 Sep 2010 #permalink

The bacteria I am working with, ammonia oxidizing bacteria as commensal organisms are unculturable using every single heterotrophic media. Dilute, concentrated, they won't grow because they can only derive ATP by oxidizing NH3 to nitrite. They are also slow growing with a doubling time of ~10 hours under optimum conditions.

Since every bit of nitrate in the environment has cycled through them they are not rare or unabundant.

grad student #8, have then look for nitrifiers. I have found them on multiple invertebrates and vertebrates. Mussels, clams, lobsters, worms, reptiles and mammals. I find it inconceivable that they are not present as commensals in sponges.

These bacteria are well known from soil and water chemistry, and they were one of the first important bacteria utilized industrially to make potassium nitrate for gunpowder. But because they are obligate autotrophs they don't grow on any of the media used for disease isolation which is why (I think) they have not been recognized as commensals.

As I remember those genetic surveys still suffer a lot from being pretty sparse. When you only sequence a few thousand organisms out of a few trillion you are bound to miss a few.

Another issue with this sort of thing is the academic rat race. A number of factors drive people toward the genetics/genomics approach rather than culture (of course, they are complementary, but leave that aside)
1) publish or perish - as someone said above, does anyone want to read a paper about designing media to culture new bacteria? According to reviewers, no ;). More accurately, it will probably be published in a journal that won't get you tenure at big research universities.
2) students don't think it's cool - this is a failure by science educators (and I include myself in this) to get students excited about the idea of culturing new organisms. I did a neat enrichment as an undergrad, then promptly forgot about it until I was a post-doc. I use it in my micro class I teach, and in my lab.
3) this is a subtle one, but probably an important one - post-docs in particular need two things, a high quality pub and something they can take with them. If culture stuff doesn't get them the high quality pub, and they generally aren't that interested in it (or would rather do genetic/molecular stuff) then the incentives are all wrong for it.

Of course, I see HMP and other projects (Tree of Life type stuff) and think that counterintuitively they may aid the cause of promoting culture based work. Now there are thousands of things that have been identified by sequencing but there's no info on what they do. So we culture them! Exciting times...

By Paul Orwin (not verified) on 04 Sep 2010 #permalink

A lot of good points about an important topic - regardless of what journals say. A few other points with respect to the Human Microbiome Project and Uncultured (not unculturable) organisms:

There are surely biases due to the 16S primers being used. These were selected to capture as much diversity as one PCR reaction can, but they are known to miss some things with more divergent sequences. They are also not targeting the Archaea. These are known issues that hopefully will be addressed now that the initial survey of bacteria is completing.

There is also the question of the DNA prep being used - are there organisms that aren't being lysed? Most probably so. This would also add to the bias.

Maybe factors like these don't explain the Great Plate Count Anomaly, but perhaps there's not a single explanation but a lot of quantitative effects that add up.

Despite the lack of enthusiasm for funding this area, the HMP has an RFA funding studies of dealing with uncultured organisms, both how to grow them and how to sequence them without requiring growth media, and hopefully this will momentum to many of the points being raised in this thread and enliven the field.

This is a very important topic for several reasons, many of which have already been presented in this forum. I will add a few more.
Interestingly, we keep discovering new bugs via culture-based methods. Just open every new issue of IJSEM. Moreover, how often we sequence the 16S of isolate we grow from an environmental matrix. If we do, we might discover a species that we have never been seen before (from a 16S seq standpoint). At a first glance discovery of novel diversity happens at a higher frequency when we try to sequence 16S directly from environmental samples. We need to keep in mind that nextgen sequencing methods are making so much easier to go after environmental samples and that culturing efforts cannot keep with this particular high rate. We need to conside that nextgen is introducing quite a few less than perfect sequences into public databases, to the point that we might be overestimating in situ diversity. Sequencing errors are less frequent when you go after a pure culture since in most cases we use Sanger chemistry.
Another issue relevant to this discussion is that it is possible that a good percentage of the unculturable fraction is actually dead. Additionally, there might be some that are viable but not culturable and others that require a combination of factors that are difficult to reproduce in the lab. We simply cannot estimate the contribution of any of these populations very easily from one environmental sample (or matrix) to another. Seasonal and geographical differences may add to this.So the great âPlate Count Anomalyâ needs to be revisited to see if we are talking 99% or 1%.
Lastly, we need to consider the limitation of using the one gene approach to estimate diversity. Using 16S we can have an idea of the type of microbes, and in some cases the overall function potential, within a sample. However, you might recover several nearly identical sequences from the same sample and it might come from different populations. How important is this within a sample is unclear, but definitely more relevant when you are comparing different samples.
Surely there are other points that we are missing but this forum has covered quite a few of the most important ones from both theoretical and applied standpoint.

By Jorge Santo Domingo (not verified) on 05 Sep 2010 #permalink

George makes an excellent point about Archaea. As powerful as genomic methods are, they are not magic and being aware of their limitations is especially important in understanding what they are missing so the next generation of tests can do better.

Tom Schmidt taught one of my phylo courses; his name is spelled thus (not Schmitt).

This is an interesting blog. When I did my PhD, I showed that even domesticated bacteria can switch from an active to a nonculturable state from which we were not successful in growing them back (PMID: 16751541 and 17293521 in Pubmed).
Kindly forgive my skepticism, but I wonder how many of the newly identified previously considered unculturable but now cultured bacteria are genera that have never been isolated historically (apart from what Paul mentioned). More numbers of colonies do not necessarily rule out outgrowth of known genera (more of the same old) as opposed to truly unknowns showing up in diluted media. If anyone would willingly share your publications towards new genera isolated and IDed using diluted media, I would happily be convinced.
While I respect the concerns of my fellow scientists about genetic roots, I would also point that genetics has brought us much forward in terms of being able to understand what mechanisms drive bacterial physiology, exemplified by the Nobel awarded to Jacob & Monod. As gut microbiologists can certify, the roles of sugar metabolism and diauxie that originated from their work pervade our ideas today in understanding pathogenicity, probioticity, and commensalism for aiding gut health. Even more importantly, they linked together the role of small molecule substrates in aiding bacterial growth, which is essentially what the bacterial isolation idea is based on.

Mike, try culturing Nitrosomonas, and then get back to me with how easy it is to grow some bacterial species. Try culturing Anammox in pure culture too.