Major bacterial phyla and classes identified using 16s rRNA gene mining
We used blast to map high quality short reads to our well known honey bee taxonomy and found that, like our previous studies, these transcriptomes are dominated by three major phyla (the Proteos, the Firmicutes, and the Actinobacteria). The major classes in our dataset were gamma-proteo, Bacilli and Actinobacteria (see Figure below).
If you're interested in "deeper" taxonomic classification, let me point out that short reads are not the best tool for this, as depending on the region they encompass, they can provide taxonomic resolution or not. Given that, here you can see the effect of using the MICROBEnrich kit on one of our samples (bee #2) - yay - more depth!
Predicted metabolic pathways utilized by the honey bee microbiome in vivo
Now for the meat of the study: what exactly are those honey bee microbes doing?? The major signal we identified was carbohydrate utilization and within that, the major contributors were our three dominant classes of bacteria.
As there was already a metagenome published for the honey bee gut, we took advantage of that dataset and used it in combination with the metatranscriptome to put together a genomic + transcriptomic view of metabolism performed by the major bacterial classes. Interestingly, the majority of the pathways were corroborated by each dataset, even though they were collected from bees in different parts of the country, at different times, and likely different ages.
Oh, and in case you were wondering how our dataset mapped to the metagenomic one, here's a nice figure. Essentially, you can find blast hits between our contigs and the published metagenome, but due to variation in microbiome composition and the divergence of these bacterial taxa, we see a very large range of percent identities.
Support for the -omic predictions using community level profiling
One nice aspect to this work is that we went on to validate the -omic data using other kinds of data, in this case, community level profiling. Freddy went on to use BioLog Ecoplates to find evidence that the honey bee gut community could utilize the substrates predicted based on the metagenomic and metatranscriptomic data. In this assay, each well of a 96-well plate contains a single substrate. You essentially inoculate each well with your environmental sample (a mix of microbial members) and look for a color change as the tetrazolium salt in the well is reduced. Freddy incubated his plates at 37C under anaerobic conditions (we saw no color change for aerobic conditions). He did this for 10 individual bee guts and saw a surprising amount of variability bee to bee.
It's easy to overinterpret this kind of variability -- keep in mind that the biolog plates rely on viability of the microbe under those conditions, on serial dilutions from the environmental sample (which introduces variability) and just because you don't see a color change, doesn't mean the community isn't capable of that processing. However, overall, we found that the honey bee microbiome was capable of utilizing almost all the sugars provided, some of the amino acids, and other carbohydrate compounds. This may not be that surprising to some, as the honey bee diet is composed of sugars (such as those found in nectaries) and pollen. In future, we hope to identify which bacterial clades are responsible for each of these metabolic capabilities and which products of metabolism each produces. Understanding how these bacteria interact through co-metabolism of honey bee food will be critical to our understanding of how the microbiome, and dysbiosis of the microbiome, contribute to honey bee health.
<you can check out the official press release here>