Oh hai everybody. Time to bring you the news from the Keyston conference again. Monday’s morning and evening sessions were respectively titled “Defining the normal and aberrant microbiotas” and “Pathogen-microbe interactions”.
– The first speaker in the morning was Ruth Ley, with a talk titled “Aberrent [sic] microbiotas: the importance of context”. The first major topic was the the role of TLR5 in shaping the mouse microbiota. TLR5 is a receptor for bacterial flagellin, and knockout mice are at risk for increase fat gain and inflammation. Additionally, germ free mice receiving a donor microbiota from T5KO mice gain more weight than with transplants from WT mice.
The microbiota, generally, does not appear to have a significantly different microbial community composition, but rather the diversity of the community is much more volatile (that is to say, beta diversity between sampling time points from the same mouse is much higher than for WT mice). Furthermore, there is no significant change in the gene families present in the metagenomes of the communities. However, there is a large difference in the metatranscriptomes. T5KO metatranscriptomes are highly enriched for motility genes (remember TLR5 targets flagellin) , while WT metatranscriptomes rare enriched for carbohydrate metabolism.
The TRL5 phenotype isn’t mediated by TLR5 alone, either. TRL5 mice have reduced loads of anti-flagellin IgA antibodies as well. They also found that Rag1 -/- mice (which can’t produce any antibodies) and myd88 mice (a protein required for downstream signalling by TRL5) also have increased flagellin loads. Interestingly, and I’m honestly not sure what to make of this, it seems anti-flagellin IgA can cause E. coli to down regulate it’s flagellin production. These findings are a bit odd to me because others have demonstrated significant changes in the composition of the microbiota in mice that are deficient in IgA (through various mutations) implying that the IgA may in fact be removing bacteria that express flagellin. One of the groups that tend to expand in IgA -/- mice are the Proteobacteria, many of which are flagellated. Also, how the bacteria sense the binding of the IgA to their flagella and down regulate flagellin production is quite a mystery.
Ruth’s next big topic was on the gut microbiota associated with pregnancy. During the first trimester (1T), the variation of the microbiota between female mice is very low, but by the third trimester (3T) it is incredibly high. They recorded an expansion of opportunistic pathogens in 3T, as well as less modularity (structure) in the COG network and increased inflammation. Transplantation of the 1T and 3T microbiotas to GF mice maintained the phenotype, implying the changes in the phenotype of the pregnant mouse are possibly due to the microbiota (but of course, it could easily be the other way around). One other thing to point out is an issue raised by an immunologist after the talk to who said that pregnant women are generally found to be in a low inflammation, somewhat immunocompromised state, which is counter to what the Ley group found.
– The next talk was from Rob Knight, titled “Defining an aberrant microbiota: powering microbiome studies” in which he went very quickly through a number of examples to highlight the power of the software tools he has developed for studying the microbiota and microbiome. His first few samples involved microbial biogeography on varying scales. He looked at the bacteria present on keyboards, and found that the communities on the keyboards clustered very nicely with who had used the keyboard. He hinted at the possibility of finding a Wallace Line between the G and H keys, but didn’t
present any evidence of it. He then went on to talk about the fact that there are very few OTUs shared between different body habitats, bacterial communities cluster very well by body habitat (for humans) but not by sex, person, or day. The variation within a habitat, a person, or short time periods is much less than between different habitats, people, or long periods of time (not particularly surprising, but good to know). Interestingly, while there are big taxonomic differences between people, the beta diversity of functional genes is very low, which suggests that it’s more important what the community is doing than who’s there, and there is a lot of functional redundancy the regional/global community. The most shocking claim from all of this was that they could recapitulate these patterns using as few as 10 samples. But of course, methods and metrics matter. Unifrac (a metric he helped devise, a phylogenetic approach) was much better as recaptitulating the patters with low sample numbers than taxonomic approaches.
He then presented results from the biogeography of the face of a post doc in his lab. It turns out the lips are the most different from everything else, and the results can be viewed by different patches on the face, community diversity, and abundance of individual taxa. Also of this was done using a particular piece of software that I honestly just forgot to write down, but all his stuff is open access, so I’m sure it’s easy to find.
The next plug was for the Earth Microbiome Project, the goal of which is to make connections between very different systems and scales such as bacterial community patterns on leaf surfaces, temperature gradients in Yellowstone hot springs, and the gut. Also, they want to do determine the origins of bacterial community members via source mapping. For example, C-section babies get their first inoculum from their mother’s skin, and the air in Detroit has a high contribution from dog feces. Thus they hope to answer questions such as which environments are the most diverse, and where should we be looking for new diversity?
To finish he plugged the online interface for QIIME, which utilizes cloud computing, and E-vident which you can use to play around with subsets of databases from completed (published?) analyses (the main point is to be able to determine the power a particular design might have to answer a particular question using similar, existing data).
– The third talk, titled “The human virome” was given by Forest Rohwer. The overal gist of the talk was that there are a LOT of viruses (phage and eukaryotic) in the human body, a much greater percentage of them is unknown (70%) than for human microbiota (10%), and they’re probably contributing a lot of diversity to metabolism in the gut.
It turns out virus turnover in an individual is not very high (95% retained over a year) and many of the viruses are temperate.
They also looked at viruses in cystic fibrosis (CF) patients. In CF lungs, the bacterial community is very uniform throughout the lung. But viruses are very spatial heterogeneous with eukaryotic viruses dominating in the upper portion of the lungs and bacteriophage dominating in the lower areas.
Most intriguingly they found a low but significant level of viruses present in the blood (presumed to be sterile) in an entire group of healthy volunteeris, suggesting we migh all have a low asymptomatic level of viremia at all times.
– The morning session regular talks were wrapped up by Erwin Zoetendal with a talk titled “Functional metagenomics in the gut” in which he described an ecosystem biology approach to studying the oft neglected small intestine (’cause it’s hard to get at). He had a group of healthy subjects whose SI were sampled using a catheter (an invasive, disruptive proceedure that precluded resampling) and a group of ileostomy patients (who’ve had their colons surgically removed) from whom he could repeatedly sample from their effluent, which is non invasive.
He found the SI is largely dominated by bacilli and clostridia, especially Streptococcus, E. coli, and Veillonella. The Streptococcus appears to be converting carbohydrates from the host diet and mucus into lactic acid, and E. coli is converting them to acetate. The Veillonella are metabolizing the lactic acid from the Streptococcus. Overall the community is shaped by the fast uptake and conversion of simple carbohydrates.
– The evening session was kicked off by Wendy Garrett, with a talk titled “Disease-associated and beneficial gut microbiota in colitis and colorectal cancer”. The model organism for her studies is the TRUC mouse which is deficient in both innate and adaptive immunity (T-bet -/- x Rag2 -/-). These mice, when raised conventionally develop spontaneous and persistent colitis. They used QIIME and LefSe to analyse the composition of the microbiota in TRUC versus healthy mice. They found that TRUC mice are enriched for Enterobacteriaceae, specifically the species Klebsiella pneumoniae and Proteus mirabilis. Removal of these species with antibiotics ameliorates the colitis. Co-housing and cross fostering WT and TRUC mice can cause healthy mice to develop colitis, and K. pneumoniae and P. mirabilis are present in these mice. However, these two species alone cannot induce colitis; other microbiota species are required. Additionally, not all strains of these species are equally colitogenic, and the age at which GF mice receive microbiota donations from TRUC mice seems to have an effect on the development of colitis (but I’m not sure why). So overall, it appears these species, or even Enterobacteriaceae in general may be keystone species for a colitogenic community, i.e. their effect on the community is disproportionately greater than their relative abundance.
The Garrett lab also found that some species of Bifidobacterium from yogurt helped to ameliorate the colitis in TRUC mice. The mechanism of this is uncertain, but they appear to lower the abundance of Enterobacteriaceae and increase the abundance of other species, possibly through the lowering of the pH via lactate production which recruits species that metabolize the lactate to butyrate. Antibiotics also reduced inflammation in TRUC mice, and different antibiotics resulted in differing communities, which seem to be driven mostly by breeding pair. That is to say, the effect the antibiotics had on the communities were dependent on the initial community supplied by the parents to the offspring.
Lastly Garrett mentioned some work on colorectal cancer. Via 16S sequencing they found an enrichment in Fusobacterium associated with tumors in the colon. Using FISH, they were able to confirm that these bacteria were physically located near and on the tumors. But, their role in the generation of the tumors, if any, has yet to be determined.
– The last full length talk of the day, entitled “Microbiota and IgA: keeping the metabolism up”, was given by Natalia Shulzhenko (currently at Oregon State). Her study system involved common variable immunodeficiency (CVID)’, which can be associated with gastrointestinal sydromes. Within the GI tract, she focused on the jejunum of the small intestine, which is involved in digestion and aborption of cholesterol, fatty acids, monosaccharides, among other nutrients. To mimic CVID in mice, she derived mice that were deficient in B cells (BcKO). Not only do these mice not have an adaptive immune response,mug they are also underweight when compared to WT mice. Natalia then looked at gene expression in these BcKO mice and found a compensatory up regulation of innate immune functions and a down regulation of metabolic genes. A very significant proportion of the effects on gene expression were able to be recapitulated in IgA -/- mice, implying that it is the antibodies produced by the B cells that are mediating the phenotype. Similar to Ruth Ley’s and other’s findings, there was no difference in luminal bacterial numbers in BcKO and WT mice, but there was an increase in luminal LPS load. Additionally the microbiota composition was not extremely altered in BcKO mice, except there were changes in abundance of three minor members: there were fewer Clostridiacea-family bacteria and more of the Paracoccus genus (Rhodobacteraceae family) and of a specific operational taxonomic unit (OTU) comprising the Lactococcus genus subgroup F49WGKH02HS7CN (Streptococcaceae family). The most abundant of these groups amounts to only 0.4% of the microbiota. This implies either that these minor members can have an important role, or that microbiotas with the same composition can behave very differently in different contexts (my money is on the latter). Interestingly, administration of antibiotics did not rescue the BcKO mouse phenotype, but germ free BcKO mice do not exhibit the conventional BcKO phenotype. Furthermore, transplantation of BcKO microbiota to WT mice does not confer the BcKO phenotype, indicating host genotype is more important than microbiota composition.
To determine what’s causing the BcKO gene expression, Shulzhenko looked at Gata4 KO mice. Gata4 is required for normal jejunum function, and Gata4 KO mice expression profiles were 60% to that of BcKO mice. Lipid aborption and body fat accumulation are both reduced in Gata4 and Bc KO mice. To figure out the link between the down regulation of metabolic genes and the up regulation of immune genes, she reconstructed the gene regulation network of the mice. The network was bimodal, with most of the metabolic and immune genes clustering into their own networks. However, they were connected by a few genes, particularly Gbp6 which is positively correlated with a number of immune genes, and negatively correlated with a number of metabolic genes. Additionally the phenotype of the entire jejunum tissue seems to be driven entirely by the epithelial cells. In the absence of adaptive immunity, the epithelial cells are required to compensate by taking over some of the immune function, at the cost of their metabolic functions. This may explain why patients with immunodeficiency diseases such as CVID and AIDS are associated with weight loss.
– Nobuhiko Kamada presented a short talk entitled “Virulence factor expression regulates the interaction between enteric pathogen and the microbiota and pathogen eradication”. The basic phenomenon is that when Citrobacter rodentium infects germ free mice, the mice cannot clear the infection, but they are not killed by it eithe. However when it infects conventionall raised mice, the infection occurs briefly but then is eliminated. Additionally, in both GF and CONV mice, inflammation in the gut subsides over time (even despite the high bacterial loads in the GF mice). It turns out that in the normal course of infection, C. rodentium down regulates its virulence factor. This virulence factor is required for the bacterium to establish even briefly in CONV mice, and when it is down regulated, the microbiota appears to clear the infection. Microbiota transplants from CONV mice to C. rodentium mono associated mice eventually clears the infection. It appears that this competition is mediated by specific members of the microbiota, such as E. coli, which share very similar metabolic requirements. So, when mice are coinfected with C. rodentium and E. coli, the C. rodentium is cleared, but not if coinfected with Enterococcus species which can utilize a broader range of saccharides. This talk presented, I think, some nice evidence that the microbiota really can have a role as an extension or addition to the host immune system by providing colony resistance to some pathogens, and in this case, the resistance is mediated by competition for similar resources.