Monthly Archives: November 2013

The turkey microbiome – a fitting Thanksgiving tribute

Since turkeys are on the mind this time of year, I thought it would be fitting to write a little more about our work on the turkey bacterial microbiome. The link describing this project is here, and the direct link to a manuscript in revision at PeerJ is here. The project data is available here through MG-RAST. The entire point of this project from the beginning was to 1) identify the bacteria in the gastrointestinal tract of the commercial turkeys over time, 2) identify differences in the microbiomes of high-performing versus poor-performing birds, and 3) develop diagnostic tools to predict performance outcome in a bird based upon ileum microbiome. The long-term goal of this project is to identify antibiotic-free alternatives to modulate the gut and improve turkey health and performance.

I’m writing this post, in part, because we are nearly finished with the descriptive portion of the project and moving on to the animal experiments aimed at modulating the turkey microbiome. First, what have we learned? Much of what we have learned is not so surprising, but some of what we have learned is quite interesting. We looked at the bacterial populations in the turkey ileum via 16S rRNA profiling using MiSeq on the V3 hypervariable region. First, the not-so-surprising. The ileum microbiome diversifies with age. It also stabilizes with age. And age is more of a driver of the ileum microbiome than are environment or treatment effects.

Now, the interesting. There were a number of specific markers (note I say markers and not drivers) of gut development in the bacterial microbiome, including several notable Lactobacillus species (L. aviarius, L. johnsonii, etc.). More prominent of a marker was a segmented filamentous bacteria (SFB) known as Candidatus division Arthromitus. These SFB bacteria were positively associated with high-performing flocks. What was interesting was that SFBs were of very short duration in the ileum, less than two weeks on average. After SFBs appeared, and disappeared, along came the other notable Lactobacillus species. This same pattern was observed in all birds in multiple flocks studied. However, the timing of this succession differed from bird-to-bird and flock-to-flock. The pattern that stood out was that the shift in microbiome occurred earlier in flocks performing better than their counterparts, suggesting a correlation between bacterial community succession and flock performance.

Now we are faced with a lot of lingering questions. There might be a cause-effect relationship between the ileum microbiome and immune system development, nutrient utilization, and ultimately growth of the bird. If there is a causative effect of modulating the microbiome, then it should be relatively straightforward to test such a hypothesis through animal inoculation experiments with cultured bacteria. But, animal experiments are costly and time consuming. We don’t know the best timing of inoculations, best combinations of bacteria, best dosage of bacteria, etc. Not to mention that we need to culture these bacteria and find representative isolates to use for the challenges. And some bacteria such as SFBs are non-culturable and will require other approaches to collect and inoculate them. We are currently looking for some in vitro screens that can be used to better refine the list of microbes to study in the animal. This too can be challenging, as there are limited cell lines and immunological tools available for turkeys. This is an exciting project with huge potential, but a great deal of challenges lie ahead….

Comments and suggestions welcomed!!

Thanks hypothesis-generator, now back to the bench (or, why I love microbiome projects).

The collective turkey ileum microbiomeYes, I have fully jumped on the microbiome and metagenomics bandwagon.

In fact, I know of very few labs in my field that have not in some way latched on to studies involving the 16S rRNA surveys in the past few years. We watched as the pioneers developed some fantastic protocols for applying this method to next-gen sequencing. Then, as the Human Microbiome Project emerged and revealed a plethora of associations between the microbiome and health/disease. Next, as even bigger projects were unveiled, such as the Earth Microbiome Project, and sought to sequence the world’s microbes. All of which have been very fruitful and extremely cool.

Sure, there have been modest efforts to characterize the bacterial microbiomes of agricultural animals, but funding for animal ag (mostly through USDA or industry support) has always been a small fraction of what is available through NIH. However, just like with genome sequencing, 16S rRNA surveys are now very much in the hands of the people instead of the large centers and research groups. So, here we are now with the ability to perform these surveys for an extremely low cost.

This brings me to the point of this post. A few years back we got into the 16S rRNA surveying using 454 technology, then it quickly translated into the same using Illumina MiSeq. Sure, we went in with hypotheses (yes, they are necessary even for pilot projects and small industry dollars) – but the hypotheses were more of the “duh, yeah” questions. We hypothesized that differences would be observed between commercial turkeys of differing weights and flocks of different average daily weights. And, we hypothesized that a predictable succession of bacteria would occur in the turkey small intestine. But, we really did not know what those specific changes would be and why they might be important.

So, we collected samples and sequenced. And sequenced. And sequenced some more. We learned how to analyze these data. We looked at the obvious things. And then we looked at the not-so-obvious things. It was a fantastic fishing (or maybe hunting is more appropriate here) expedition. And it totally paid off. We found some really interesting differences in the species of Lactobacilli and Clostridium that change over time in the turkey gut. We found that the successional shifts in the bacteria in the ileum are predictable, and they occur earlier in research flocks where the birds perform better. The work is nearly published in PeerJ. Overall, it was a great success.

As we all have learned, correlation does not equal causation. We have a lot of correlative data to suggest how the turkey microbiome can be manipulated to improve health and performance. Now, we head back to the benchtop to validate these findings and go after the handful of interesting hypotheses that came from this work. I feel like this is the trend of science lately. Technology drives new hypotheses, then the classical approaches are necessary to figure things out (Look to Jeff Gordon’s recent paper for inspiration). There are scientists on both fringes of this, though. Some refuse to use the technology and actually lambast anyone who uses it as pseudoscientists without hypothesis-driven research programs. Then, there are those who do nothing but use the technology and probably deserve to be lambasted. Me? I don’t really care what anyone says. 16S rRNA microbiome surveys have generated or refined more questions for us than I could have dreamed up during the same time. I suspect that most of us fall somewhere in this zone, where we are using the approaches to generate interesting questions from correlative observations, then to go out and get funding to address those questions. 16S rRNA surveys (and probably soon to be better metagenomic approaches) have changed the way that I operate for the better, and I can’t wait for the data from our next survey study.