Wednesday’s sessions were both joint session with the Innate Immunity conference. The morning’s session was titled “Microbiota regulation of the innate immune system” and the evening’s session was “Lymphocyte-microbiota interactions”.
– The mornings session’s lead off talk was given by Andrew Macpherson and entitled “Immune compartmentalization in mutualism with the intestinal microbiota”. Macpherson began by noting that one of the drawbacks to our current germ free systems is that once an organism is colonized by bacteria, it is more or less impossible to make them germ free again (at least with any amount of certainty) because it’s not guaranteed that even ridiculous doses of broad spectrum antibiotics will kill everything. So, his group derived a strain of E. coli, ha107, that lacks the ability to make specific amino acids that it requires; these amino acids are also not supplied by the host. The bacteria are cultured in a medium that supplies these amino acids, but once they colonize mice, they eventually die out. What they found was that colonization with E. coli induced IgA production, as has been shown with other monoassocation models, but interestingly the IgA production remained at normal conventional levels even after the E. coli had all died. The IgA were also E. coli specific, and this specificity persists through the life of the mouse unless a new challenge is presented to the immune system. The mice’s IgA did not bind to Schaedler flora (which lacks E. coli, or Salmonella). Oddly (to me, at least) this persistent specificity seems to be due to long-lived plasma cells, and not memory B cells, meaning when they rechallenged them with E. coli, the mice did not mount a classic secondary response normally associated with, say, vaccines. Another interesting thing about this whole situation is that live bacteria were required to induce the IgA response, and there was a very high abundance threshold (about 10^9 bacteria or more).
Macpherson then went on to discuss the importance of the mesenteric lymph nodes as a firewall to keep bacteria out of the central organs once they’ve made it past the the intestinal epithelium.
Enterobacter cloacae gavaged into mice end up in the mesenteric lymph nodes, but not in the spleen, however, if they are introduced intravenously, they make it to the spleen quite easily. Bacteria that penetrate the epithelium are sampled by dendritic cells which can induce IgA production. This compartmentalization of bacteria in the mesenteric lymph nodes keeps the innate immune system ignorant of them. In mice without mesenteric lymph nodes, the immune response is much greater. But, while the bacteria are confined to areas such as the mesenteric lymph nodes, their products are not. In fact, Macpherson’s group found large quantities of bacterial molecules in and on the liver and other central organs of mice.
The innate immune system sets the threshold at which there is a systemic response to the microbial. In innate-deficient (MyD88 -/-) mice, the IgG response to commensals is much greater. Additionally, the bacteria can penetrate to sterile organs such as the spleen. Furthermore, mice primed by gavaging with a bacterial species, are able to abrogate an intravenous challenge with the same bacterium, whereas mice that are not first challenged are not.
So, over all, mucosal immunity requires the presence of the microbiota for initiation, but it is the innate immune system that sets the threshold for activation, which as a result, affects the degree to which the adaptive mucosal immune system responds. The mesenteric lymph nodes are firstly an important firewall to keep bacteria from penetrating to the central organs, and secondly an important area where commensals are sampled to protect from possible bacteremia.
– In the two slot for the morning was Maria Resigno, with the talk “Dendritic cells in host-microbe interactions in the intestine”. She had quite a lot of information, and talked rather fast, so I apologize in advance if my summation has some holes in it.
There are two major antigen presenting cells (APCs) in the gut: CX3CR1+ (phagocytic restimulating cells) and CD103+ dendrictic cells (DCs) which are tolerogenic. Mouse intestinal epithelial cells can induce CD103 expression in DCs through either direct contact or signaling. CD103+ DCs produce TSLP which acts directly on T cells by ihibiting IL-17 production (required for Th17 cells, a type of T cell that can induce inflammation as well as other immune responses), and fostering the development of regulatory T cells (Tregs). When TSLP is not expressed, colitis is worsened.
It turns out that patients with Crohn’s disease (a intestinal bowel disease [IBD] involving chronic inflammation) do not condition tolergenic DCs, and the question Resigno’s group had was whether probiotics could participate to re-establish control of homeostasis in the intestine via the tolerogenic DCs.
The took surgical specimens from healthy and IBD tissue and affixed them to a mount. They then glued a cylinder to the apical surface of the tissue, thus preventing bacteria from circumventing the mucosal surface and epithelium. They tend tested the effects of different probiotics on the healthy and inflammed tissues. One thing they found (and is relevant to Colin Hill’s talk, among others) is that not all Lactobacilli are innocuous. While some species, such as L. paracasei excrete cell products that are anti-inflammatory and can protect from inflammation induced by Salmonella, other species when added to already inflammed tissue actually made it worse. Their cellular products, though, harvested from the supernatant of pelleted cells can still reduce inflammation. This, to me, underscores the increasing evidence that the physical location of the bacteria is the main determining factor for how they can affect their host. Bacteria that associated directly with the epithelium cause inflammation, and those that stay out in the mucus layer or lumen generally do not.
Resigno’s group also looked at products from the bacteria that could possibly help ameliorate inflammation (called them postbiotics as opposed to pharmabiotics). They took colonized tissue and made metagenomic clones in E. coli which they screened for their effects on tissue. The found 18 clones that either up or down regulated NFkB as well at a particular molecule from a Clostridium species that affected the tolerogenecity of DCs.
Overall, they Resigno restated the importance of knowing which probiotic species are actually probiotic, and which were actually inflammatory/pathogenic. One way to get around this is with purified bacterial products, and they invented a pretty neat system to do this on live intestinal tissue without having to do it in vivo.
– Third in the lineup for Wednesday morning was Kiyoshi Takeda with a talk titled “Regulatory mechanisms of immune responses to intestinal bacteria”. There are several subsets of innate immune cells that are associated with the gut, and Takeda’s group wanted to determine how probiotic species can influence them. They orally treated mice with Bifidobacterium breve or Lactobacillus casei every day for three months. There was no detectable change in the composition of the microbiota. However, B. breve increased IL-10 production (an anti-inflammatory IL) as well as the development of Tr1 cells (suppressor T cells) and Tregs (which suppress inflammation as well). These changes in the host were mediated by CD103+ DCs.
Then, to be honest, I rather lost track of where he went. I mostly gathered that Stat3 mutant mice have increased DC activity that causes inflammation, and some cells called Mreg cells aggregate around T cells, preventing from getting near hyperactive DCs and keeping them from developing into pro-inflammatory T cells.
– The cleanup talk for the morning was given by Sarkis Mazmanian and titled “Learning to tolerate our microbial self”. Mazmanian approached his talk from the the point of view of the IBD intestine, and focused on the bacterium, Bacteroides fragilis. This is a prominent commensal that is Gram (-) and an obligate anaerobe. It makes a polysaccharide capsule containing two molecules, polysaccharide A (PSA) and polysaccharide B (PSB). PSA, but not PSB can protect a mouse intestine from chemically induced inflammation. In the case of IBD, colitis is driven by pro-inflammatory T helper cells and elevated levels of IL-17. PSA mediates the expansion of suppressive Tregs during chemical colitis, and this protection requires signaling by TLR2.
B. fragilis delivers PSA to the immune system by packaging it into outer membrane vesicles (OMVs). PSA in OMVs signal tolerogenic dendritic cells via TLR2, but PSA can directly activate regulatory T cells as well.
When B. fragilis and E. coli are co-colonized into mice, E. coli is maintained in the lumen and outer mucus layer, but B. fragilis is able to invade to the epithelium without producing inflammation. PSA defective B. fragilis mutants cannot invade to the epithelium, but they are rescued by the addition of PSA. The numbers of B. fragilis are the same regardless of the presence of PSA. In mice lacking TLR2, B. fragilis is also not able to penetrate to the epithelial surface. Thus, Mazmanian claims PSA in B. fragilis evolved to promote mucosal tolerance during colonization.
– The Wednesday evening session began with David Artis with a talk titled “Microbial regulation of immune cell homeostasis”. His talk focused mainly on type 2 immunity and allergic inflammation. These related responses are integral for the body’s defense against helminth parasites as well as the repair and maintenance of epithelial barriers. However, as it rather obvious in the case of allergies, they can pose problems as well. There is evidence linking commensal dysbiosis and susceptibility to allergic inflammation. Antibiotics treatment temporally and spatially alters the microbiota structure, and these changes are associated with severe allergen induced type 2 inflammation to papain (an enzyme present in papaya) and house dust mite.
Artis focused his talk on the role of granulocyte lineages (specifically basophils) in maintaining homeostasis of type 2 immunity and allergic inflammation. Antibiotic related alterations result in elevated levels of secreted IgE (in mammals) and basophilia, but no significant change in mast cell or eosinophil numbers. House dust mite (HDM) treatment results in elevated Th2 (T helper 2) cells as well. Depletion of basophils ameliorates allergic inflammation, though a detectable (but lowered) response of Th2 cells is still present. Anti IgE treatment normalizes the basophil response in antibiotics treated mice.
Artis then moved on to talking about human patients with hyper IgE syndrome. These people have mutations in Dock8 (involved in intracellular signalling) and have up to 10 times the normal IgE levels. Artis and his group decided to look at what was possibly going on with the basophils in these patients, using mice as a proxy. Germ free mice have elevated frequencies of basophil progenitors in the bone marrow, and their basophils have an increased IL-3 response which leads to increased proliferation. Elevated levels of IgE also cause an increase in basophil progenitors. Which could explain the issues faced by hyper IgE syndrome patients.
Next, they asked whether commensal derived signals influence mucosal or systemic anti-viral immunity. Mice infected with influenza A are usually not killed by the infection. However, if they are pre-treated with antibiotics, the mortality rate goes up significantly. The mice’s response to murine norovirus is also defective following depletion of commensal bacteria, as is their response to lymphocytic choriomeningitis virus. He then linked this association of presence of microbiota with resistance to viruses with the protection that Wolbachia confers to their insect hosts. However there are big differences in the relationship between Wolbachia and the members of the microbiota, the most important of which is that Wolbachia in an intracellular parasite/symbiont and is often transmitted vertically, unlike, as far as we can tell, any of the usual gut bacterial community members.
– The second talk of the afternoon was given by Dan Littman, entitled “Microbiota regulation of intestinal lymphocytes”. In this talk he explored the balance that the immune system must maintain between regulatory and inflammatory T cells. Two two T cell subtypes he focused on were Th17 (characterized by expression of IL-17 and RORγt) and Treg (characterized by expression of FoxP3). Both of these T cells are derived from a progenitor T cells which, when exposed to TGFβ, differentiates into an intermediate stage. In this stage, both RORγt and FoxP3 can be expressed; it is the balance of this expression that determines whether the cell become a Th17 or Treg T cell.
Commensal bacteria can influence this differentiation. Mice from different labs have different levels of Th17 T cells, but when co-housed, these levels become similar. It turns out, segmented filamentous bacteria (SFB, a loosely defined set of bacterial species that lives in close association with rodents’ gut epithelia) are present in the mice with high levels of Th17 cells, but absent in mice with low levels of Th17 cells. The monocolonization of germ free mice with even a single SFB induces the proliferation of Th17 cells.
Littman’s group wanted to know how specific the Th17 response was to the SFB. So, they looked for an enrichment of any of the particular V segments that comprised a part of the variable region of the T cell receptor. They found an enrichment of Vβ14 in the Th17 cells.
It also appeared that Th17 cells have a role in autoimmune diseases. Germ free mice colonized with SFB spontaneously developed rheumatoid arthritis in their paws and ankles.
Humans are not colonized by SFB, but there are indications that another species, Prevotella copri could play a similar role in inducing Th17 proliferation. This induction may be dependent on whether or not P. copri penetrates to the mucus (as SFB always do in their hosts). Of course, as we’ve seen from other talks, if anything penetrates through the mucus layers to the epithelium, they tend to cause inflammation (with the exception of a few species such as B. fragilis.
– The final talk on Wednesday evening was given by Kenya Honda. His talk was entitled “Intestinal commensal bacteria-mediated Treg induction”. This talk was well paired with Littman’s talk, as he talked a little bit more about SFB induction of Th17 cells, but then move to the other side of the balance to look at bacteria that induce Tregs.
Mice and rats are both colonized by morphologically similar SFB. However the SFB that colonize each have 5-10% of their genomes that are unique to them, suggesting they may be different species. So, the Honda group decided to see if they were able to function similarly in their non-typical host via monocolonization of germ free rats and mice with SFB. To things easier to distinguish, SFB harvested from mouse guts will be referred to as mSFB and SFB from rat guts will be rSFB. It turns out, rSFB can expand in the intestinal lumen of mice similarly to mSFB, but rSFB cannot adhere to the mouse epithelium. rSFB do not induce IgA production, inter-epithelial leukocytes recruitment, or Th17 proliferation like mSFB do. However, in their usual hosts, rats, rSFB do all the same things that mSFB do in their murine hosts. Honda ended this section of the talk by speculating that these interactions may be mediated by the actin filaments the bacteria use to adhere to the host cells.
In earlier work, the Honda lab had discovered a cocktail of 46 strains of Clostridium species that potently induce Tregs in mice. So, they went looking for a similar cocktail that they could isolate from human guts. Previous work had shown that spore-forming human gut bacteria can induce Tregs in mice. So, they started be chloroform treating the human fecal samples they collected to select for spore-formers. When they inoculated mice with the chloroform-treated fecal samples, Th17 cells were not induced in mice, while non-chloroformed fecal samples did. They then co-housed mice inoculated with chloroformed fecal samples with germ free mice, and found that Tregs were induced in the ex-germ-free mice after transmission of bacteria from the inoculated mice.
Then to get a better idea the potency of these species, they did 20,000 x dilutions of the fecal samples and found they were still able to induce Tregs in mice. So, they performed 16S pyrosequencing on the fecal samples and found that the species left in the chloroformed fecal samples were in the Clostridia class. What had been eliminated by the chloroform were a lot of Bacteroidetes. They identified 97 OTUs from the 20,000 x dilutions, which were lots of Clostridium and Eubacterium species. They cultured these OTUs, and picked 300 colonies, which ended up belonging to 22 species. They then made a cocktail out of these 22 species that induced Tregs in mice very well, and had no effect on Th17 cells.
The lab is currently testing the effect of this cocktail on its ability to regulate colitis and allergic inflammation using mouse models. One thing to note, that is again a recurring trend, is these Clostridia preferentially inhabit the mucus layer and do not regularly penetrate to the epithelium.
So the take home message from Wednesday was that, as we saw on Tuesday, the suppression of inflammation by the immune system can be mediated by bacteria, but they must not penetrate to the epithelium (with at least one exception, of course). However, the immune system must not be completely agnostic to these bacteria, but rather, sampling of bacterial products by APCs such as dendritic cells can ensure the induction of tolerance by the immune system.