Tip:
Highlight text to annotate it
X
Dan Rudolf Littman: -- thank the organizers for the invitation
and giving me the chance to participate and learn from all of you here. I am an immunologist,
and I'm going to give you a little bit of background as to where we come from. And that
needs to be switched, but I should be connected. I was before just fine. Should I close? Yeah,
okay, thank you.
So I'm going to talk about the relationship between the microbiota and the immune system,
and give you some examples of homeostatic interactions and what we think may occur in
states of disease. And David Relman showed the entire painting earlier, but years ago,
when I was in the Prado, I found a Hieronymus Bosch painting, and found this little fragment
of it where this person is analyzing a model animal system. And David, if you look carefully,
I think you'll find that somewhere in the middle of the painting.
And so we've been studying for a long time the experimental mouse to try to give us insights
into human physiology. And been focusing, in particular, on different types of T-lymphocytes
in the immune system, and more recently, at mucosal surfaces. So this is just a very general
review of some of the more important immune cell types at the mucosal border in the gut.
We know that there are signals that are transmitted through different types of dendritic cells
that can influence the balance of potentially pro-inflammatory and anti-inflammatory cells.
The anti-inflammatory cells are regulatory T cells that express FOXP3. The pro-inflammatory
cells can be TH1 or TH17 cells. I have the TH17 cells listed here, which are particularly
abundant in the lamina propria. Then there are also innate lymphoid cells that make very
similar types of cytokines, as the different types of T cells. And all of these function
together, normally, to protect the barrier, and, in large part, through production of
anti-microbial peptides that can regulate the levels of the commensal microbiota, as
well as, potentially, any potential pathogens.
So I'm going to just very briefly tell you about how we think different modes of interaction
of microbiota with the host immune system can occur. So we believe there are certain
bacteria, such as the segmented filamentous bacteria that I showed you an EM [spelled
phonetically] of at the outset, that can homeostatically induce cells of the immune system, presumably
for beneficial purpose, both to the microbiota as well as, hopefully, to the host. And I'll
show you an example of this through T helper 17 cells, but there are other examples, very
likely in the case of regulatory T cells, and maybe Sarkis Mazmanian will tell us a
little bit about B. fragilis and its role in regulatory functions.
But then we also think there are -- there is a process of homeostatic inhibition of
the activation of the immune system in which microbiota can signal to prevent the activation
of effector responses at inductive sites. And in the absence of this kind of inhibitory
signal that presumably puts a stop signal to certain cells of the immune system, such
as dendritic cells, these would now become released, taking up components of the microbiota
and leading to an inflammatory induction of immune responses. And maybe something like
this could be going on in some of the IBDs.
So we have certain examples, and I'll just cite a couple from our own laboratory, of
how the immune system is compartmentalized from the microbiota. So one of this is this
avoidance of the microbe-elicited immune responses, in which a particular type of myeloid cell
that expresses the chemokine receptor CX3CR1 seems to be particularly important in preventing
-- in avoiding activation of immune responses. So these cells seem to be inhibited by MyD88
regulated microbial signals to -- from migrating to inductive sights such as the mesenteric
lymph nodes. So that is a kind of a functional compartmentalization.
Then there's another kind of compartmentalization that's enforced by regulatory T cells, recruitment
of TRegs to the lamina propria. And then another mechanism is this selective micro-specific
immune response that I mentioned. We know from work of Kenya Honda that there can be
regulatory T cell expansion in response to certain Clostridium. And then Che Shay's [spelled
phonetically] group in St. Louis, Lathra [spelled phonetically] put out a paper a couple years
ago showed that there can be microbe-specific regulatory T cells, although their specificity
was not demonstrated. And I'm going to tell you about segmented filamentous bacteria-mediated
induction of antigen-specific T helper 17 cells.
So just an example of this compartmentalization by negative signals, Gretchen Diehl in our
laboratory found very surprisingly that when mice are treated with antibiotics, and then
given bacteria like a noninvasive salmonella, or, in this case, just E. coli K12, and then
looked at E. coli titers, in this case, as well as an IgA antibody response in the feces,
surprisingly after antibiotic treatment, there was now E. coli in the mesenteric nodes, as
well as the induction of a E. coli-specific IgA response.
And, in brief, what we think happens is that the microbiota give this kind of a negative
signal through MyD88 thorough, presumably, through toll-like receptors, keeping the CD31-positive
mononuclear phagocytes in the -- attached to the epithelium in the lamina propria. And
in the absence of this signal, there's a CCR7-mediated migration of these cells. These cells basically
become untethered, they pick up bacterial content, and they transport, actually, full
bacteria that we can now culture from the mesenteric nodes, they transport these through
the A form lymphatics, where they can induce both the T cell response and an antibody IgA
response. And we'd like to propose a dysregulation of this kind of a process, in a dysbiotic
gut, may result in inflammation through this kind of a mechanism. So this is just one example.
And Sang-*** Kim [spelled phonetically] has been doing work on regulatory T cell recruitment
to the lamina propria, and he found that a orphan G protein cover receptor GPR15, that's
expressed preferentially on regulatory T cells, is involved in the migration of these cells
to the large intestine lamina propria, where we think that these promote homeostasis.
So I'll now tell you a bit about selective microbe-specific immune responses. So a few
years ago, Evo Evanov [spelled phonetically] in our laboratory, found that animals from
Jackson Lab versus Taconic -- the Taconic Farms providers of mice differed in the level
of TH17 cells. And he found that that was because of the absence versus the presence
of segmented filamentous bacteria shown here attaching to the ileum. Now this is -- this
was present only in the Taconic black six mice here and not in the Jackson mice. And
these, just to remind you, these are Gram-positive anaerobes, they're spore forming, they have
yet to be cultured, and they most closely resembled Clostridium but they're still very,
very different from typical Clostridium. And they have a reduced genome size. And we collaborated
with Kenya Honda and Yoshio Misaki's [spelled phonetically] group in Japan because they
were able to provide us with mono-associated feces from mice colonized just with SFB. So
when these were provided to germ-free mice, within a week to 10 days we could now see
the induction of T cells, CD4T cells, that make both interleukin 22 and interleukin 17.
So you can see this induction occurring over here.
So that proved, really, that SFB itself is the inducer of TH17 cells. So how does this
now influence TH17-mediated processes? We know that TH17 cells are involved in a number
of autoimmune diseases, a number of inflammatory processes, and can just also be beneficial.
And this is just showing a couple of examples. So, in colonization of animals with Citrobacter,
and Citrobacter will grow primarily in the colon, you can see that animals that have
SFB are relatively protected from the growth of Citrobacter in the colon. On the other
hand, as shown here, with Diane Mathis and Christopher Benoist in a arthritis -- spontaneous
arthritis model that they have developed, you can see the germ-free mice basically do
not get arthritis, and animals that have now been given SFB, within days begin to develop
the arthritis. And this is a TH17-mediated process. It's absolutely dependant on T helper
17 cells, and Sarkis has shown in a different model, in the EAE model, something very similar,
that SFB greatly exacerbates the disease process in EAE.
So that indicates, then, that TH17 cells that are induced locally in the gut can contribute
somehow, at a systemic level, to autoimmune disease, and potentially can also contribute
at a systemic level to protective processes because SFB is present in the small intestine
and may induce a protective response also for the colon. But that is something that
still needs to be studied.
So how do SFB-elicited TH17 cells then provide specific protection or exert pathogenic functions?
So there are several questions we can ask within this larger question here. And one
is, what is the antigen specificity of the SFB-elicited TH17 cells? And we can think
of several possibilities. One is that these cells are nonspecific, so that any T cell
in an SFB condition microenvironment could become a T helper 17 cell. Alternatively,
these could be cells that are particularly reactive to self-antigen but the SFB-dependant
microenvironment may sensitive this, may actually lower the threshold for their activation as
autoreactive TH17 cells. Or, alternatively, this microenvironment may curb regulatory
T cell-mediated tolerance. And the third possibility is that these T cells are specific for SFB
or other commensal antigens. And I'll tell you about some of the experiments we did to
look at this.
Another question which we have not addressed is, do SFB-induced T helper 17 cells circulate
to lymphoid tissues that are draining the specific organs involved in autoimmunity?
And if so, how do these circulate out of the gut, out of the inductive site to go to, say,
tissues that drain the CNS synovium for rheumatoid arthritis, or the skin in, for example, cirrhosis?
Okay, so in order to do this we decided to look at the repertoire of the TH17 cells in
the gut. This is work that Benny Yang in our laboratory did, and we're very fortunate to
get mice made by Muhammad Hoque and VJ Cutru [spelled phonetically] in which the green
fluorescent protein was knocked into the R23 receptor locus. So the wonderful thing about
this tool is that all TH17 cells express GFP now. And so we can look at GFP-positive TH17
cells and at GFP-negative non-TH17 cells, and look at various T cell receptor subunits
on the surface, in this case, Vbeta14.
And I show Vbeta14 in particular because you see here that there is a prevalence of Vbeta14-positive
cells among TH17 cells compared to non-TH17 cells. So looking at many animals here, you
can see that it's roughly a 3 to 1 preference for TH17 in this particular type of T cell,
whereas Vbeta6 cells, you can see, are at a 1 to 1 ratio. Initially we thought this
might be due to some kind of a super antigen, but upon much more work, we figured that this,
indeed, has something to do with a predisposition of this subset of T cell receptors for antigens
on SFB, as I will show you.
So the way that we learned this, and this is work done by Benny with help from Mariam
Torchinski [spelled phonetically], was to take a reporter T cell hybridoma that reports
on activation to the T cell receptor by expressing GFP driven by an FAT. So when these cells
have an introduced T cell receptor, along with co-receptor CD4, if they can be activated
as a control with anti-T cell receptor antibody, or with antigen-presenting cells and colonic
contents, we can look for the presence or absence of GFP induction. And so we introduce
then into the cells, various pairs of alpha-beta T cell receptors that are cloned from individual
T cells that were either GFP plus or minus, in the colon; either TH17 plus or non-TH17
cells.
And something very surprising, to me at least, happened when we looked at this. First of
all, the non-TH17 cells T receptors did not respond at all, or responded very poorly here
to colonic content, but the TH17 cell receptors, you can see, almost all of them respond, and
some very, very strongly, only, in this case, to SFB mono-associated fecal content. And
this then allowed us to look in more detail here at activation of T cells that were either
GFP-positive TH17 or non-TH17. And you can see, essentially, all the Vbeta14-positive
cells in the gut, if they are TH17, now respond to SFB containing microbiota, whereas the
non-TH17 cells did not respond. But this is not restricted to Vbeta14. You see Vbeta8
cells also, a large proportion of these cells respond. Also Vbeta6 cells are shown here.
So this is just showing non-Vbeta14 cells responding as well. But only the TH17 cells
are responding.
So this is really telling us there's something about SFB that elicits a T helper 17 response
and not other types of T cells. So to really nail this -- try to narrow this down a bit
more, we then did shotgun cloning of the SFB genome into E. coli and looked to see whether
there were colonies that could stimulate these T cell reporter hybridomas in the presence
of syngeneic splenocytes. And what you see here, for example, is a negative well where
CDC3 cells that are not activated. And this is a positive well here in which GFP is being
turned on. And that allowed us to show that among the 11 or so T cell receptor hybridomas,
they fell about half and half into two different categories. Again, these are just Vbeta14
hybridomas, but they recognized two different proteins: one a very large protein that we
predict to be extracellular that is expressed at a fairly high level within SFB in the gut
of the mouse, and one that's a smaller protein, also thought to be extracellular, that's expressed
at a much lower level, but yet there are a lot of T cells specific for this.
So that, then, gave us some tools that we could make. One of which was to make a T cell
receptor transgenic mouse with one of these receptors that was specific for this more
prevalent protein. And this is the 3340 protein here. So what we could do is basically inject
these T cell receptor transgenic cells into a mouse that either is given SFB or is not.
And then we can look at the lamina propria T cells or mesenteric nodes at various times
after transfer. And one of the things that we found right away is that when we do this
kind of a transfer, we see expansion of these SFB-specific T cells only in SFB-colonized
animals. So you can see these are donor-derived cells, these are host-derived cells, and we
see this expansion in the lamina propria of the small intestine only in the SFB-plus mice
and not in the SFB-negative mice.
And if we now look at, in this case, we look with just very small numbers of cells that
are injected that are now expanding there, and we can look at also different times, what
we see is that, essentially, all the donor-derived cells become RPR gamma T-positive cells, which
is a mark -- transcription factor that marks the TH17 cells, whereas in the host-derived
cells, typically, about 20 to 30 percent of the cells will be TH17 cells, ROR gamma T-positive
cells. So the reactivity to SFB makes these cells become TH17 cells. This is looking at
multiple animals, and you can see always this kind of a distribution, which the T cell receptor,
transgenic-derived cells all become T helper 17 cells.
And we developed another tool in collaboration with Mark Jenkins, which was to make a tetramer
because we mapped the peptides that are being recognized within the, for example, the 3340
protein, that are being recognized with MHC IALB [spelled phonetically] by the T cell
receptors. And these tetramers now allow us to stain unmanipulated animals for the presence
of T cells specific for this antigen. And you see now, when we look at the lamina propria,
that ROR gamma T-positive cells and negative cells down here, and tetramer-positive cells,
you see, essentially, all the tetramer-positive cells are in the ROR gamma T-positive TH17
compartment, whereas we see here in the negative, ROR gamma T-negative compartment, typically
we do see some, of course, some TH17 cells that have other specificities, obviously.
So when we do this again in multiple animals, we see again the 20 to 20 percent distribution
of ROR gamma T-positive cells in the tetramer-negative cells, and more than 90 percent in the tetramer-positive
cells.
So, surprisingly, then, when we asked, "What is the antigen specificity of SFB-elicited
TH17 cells?" We find that they are -- these are SFB specific by and large, and maybe in
other cases there could be specificities for other commensals. That does not rule out that
these other mechanisms may not be involved, and, in fact, if we look in colon, and I'll
show you the data here, we find that there are very few TH17 cells in colon in animals
that have this tetramer positive kind of specificity. So there may be different mechanisms, perhaps
some kind of a -- there could be environmental-related induction of TH17 cells as well. So that really
remains to be figured out.
So let me move on here, quickly, because I want to make sure I have time for some relevant
questions here. So what we think, then, is that there may be specific niches where particular
types of antigen-presenting cells will bring microbial antigens to the inductive site,
in this case, say, mesenteric lymph node, where there will be induction, for example,
of TRegs if we start with Clostridium here. And if SFB are stimulating maybe a different
type of niche, where a different antigen-presenting cell would be present, that would now induce
the -- make the types of cytokines that would induce TH17 cells.
By using the TCR transgenic mice, we find that this induction occurs first in the mesenteric
lymph node around day one to three. And what then would happen is, presumably, these cells
would redistribute by turning on homing receptors. And after day four, we see them scattered
through the lamina propria, and with TRegs, induction of TCR15 at least, would lead to
these cells going to the large intestine in large number, where we find them in the largest
numbers.
And then what happens in autoimmunity? Well, one possibility is that these cells are redirected
apparently, to -- either to joints, for example, for arthritis, or to lymphoid tissues that
are draining these sites, and, for some reason, these would, either through cross-reactivity
or through near-neighbor kind of functions, would induce disease at these sites.
So, of course, then, we wanted look at TH17 cells in human disease. And I won't belabor
this point, but we know that many autoimmune diseases in humans and asthma, steroid-resistant
asthma, are associated with TH17 cells. So Jose Scher, who's a rheumatology fellow at
NYU, along with Steve Abramson, who is now chair of Medicine, but he was head of rheumatology,
joined us in looking at commensal 16S in rheumatoid arthritis patients. And we collaborated with
Carlos Abata [spelled phonetically] and Eric Paymer [spelled phonetically] initially to
do the 16S analysis. And we're very fortunate that we had access to new outset RA patients
at Bellevue Hospital but we also said chronic RA patients and psoriatic arthritis, along
with healthy control.
And what came out immediately, though, was really striking, is that only in this new
onset RA patients, NORA patients, there was a very high abundance of a Prevotella in the
feces of these patients, whereas in the healthys, and the chronic RAs and psoriatic RAs, we
saw only 15 to 20 percent of these with high levels of Prevotella.
And if we look, then, more closely, by metagenomic sequencing, we saw something that was very
similar to what was observed in the HMP analysis. And Michael Fischbach provided me this analysis
here, but Curtis, who is following me in this talk, he may have one slide that shows this
as well, there really is a bimodal distribution of Prevotella copri in the healthy population.
Typically around 15 percent of people have it, but 85 percent of it do not. And in our
case, we see about 70 percent of the NORA patients who have this. And it is the most
closely-related to P. copri based on the metagenomic analysis.
What we also find is that there are some sequences that we find associated with the patient sample
Prevotella, and others primarily with the healthy sample Prevotella, bringing up the
possibility that, actually, there may be some virulence associated with this particular
species in the patients.
Finally, we don't have any proof of causality here, but the way we've been looking is by
introducing Prevotella, in this case, a reference strain P. copri, into mice that have been
first treated with antibiotics. We confirm colonization, and then we treat these mice
with disulfate sodium, which is basically an irritant. And we can see that there is
a more severe weight loss and more severe colonic inflammation as well, that Randy Longman
in the lab found, if the animals have received Prevotella. We see something very similar
also with collagen-induced arthritis.
So, then, to close here, what we think, then, is that there may be different mechanisms
of achieving homeostasis, achieving a balance between TRegs and effector T cells by having
a response to different types of bacteria in different niches within the gut, but in
addition, there is a way of restraining bacterial content from reaching the inductive sites.
So a few questions, then, that we can ask right away. Are there microbiota-specific
TH 17 cells expanded in RA, and do they contribute to pathogenesis? We don't have answer to this.
But also another what I think is an interesting questions is, "Could commensals specialize
for homeostatic activation of adaptive immune cells be used for protective or tolerogenic
vaccination? As an example, to control early *** induced depletion of TH17 cells that occurs
in the intestinal lamina propria."
So these are just some of the questions that we're interested in, but we think that there
are many outstanding questions here in the field. So in terms of what might be the microbiota
influence on the immune system? And I'll just read quickly through these because I'm going
a minute or two over here. Is there a subset of microbes that influence differentiation
of these three components of the immune system? So I showed you an example for SFB. SFB may
exist in human, and I'd like to have a discussion about that later, but is SFB an outlier, or
are there many such bacteria like this? And this would involve induction of T cell subsets,
B cells in immunoglobulins, of course, in A cells, there are both lymphoid and myeloid.
And do specific microbial metabolites, such as secondary bile acids, influence immune
cells? And we think that that is going to be the case.
What sets apart these microorganisms that influence immune responses, such as the bacteria
that induce the TH17 cells? Are there regional differences in inductive events, say TH17
cells in the small intestine versus the large intestine? And what's the role of the microbiota
in the TB cell functional repertoire? I showed you one example here, but are there circulating
the B and T cells specific for antigens encoded by the microbiome. Are there associated effector
functions? What's a proportion of such cells in our circulation? Could such cells account
for the repertoire of T cells and potentially B cells with effector or memory phenotypes?
Okay, so these are experienced cells, presumably that are anticipatory for potential pathogens.
And Mark Davis recently showed that there are anti-*** memory T cells in people who
are naïve to ***. Could these be induced by microbiota?
And could there be evolutionary pressures for these kind of microbiota interactions
with the host? And could individual commensals be exploited, then, to induce specific anti-microbe
protective immune response, such as commensal *** vaccines I mentioned. And some other questions
is, what is the relationship of microbiota composition to host genetics at steady state?
And we heard a little bit about this from Ruth Ley, but, in particular, is there a link
of MHC haplotypes or immune system gene polymorphisms to the composition of the microbiota? And
do dysbiosis and the abundance of particular microbes contribute to diverse autoimmune
diseases and other inflammatory conditions? I gave you an example of Prevotella and the
association with new onset RA, but, as David Relman told us this morning, there can -- you
know, we don't know if there's cause or effect, if it's initiating or propagating; in our
case, we think it's very likely it would be initiating if it is causal, because we don't
see it in the chronic RA patients. Is it necessary and/or sufficient? And can micro-specific
T cells and antibodies be detected at steady state and in dysbiosis-associated disease?
And are these present systemically?
I think these are all questions that we are now in a position to have the tools to begin
to answer.
I've mentioned many of the people who have been involved in this work already. I want
to particularly point out Nicolas Sagata and Curtis Huttenhower have helped us enormously
on the Prevotella work most recently. And these folks here have helped us a great deal
with the SFB. So thank you very much, and sorry for going over.
[applause]
Female Speaker: Yeah, thank you so much, Dr. Littman. And
since you already asked himself the questions and put on the slides, I think we're going
to move on to the next speaker. And at the end, when we have the open floor discussion,
maybe you can ask other questions.
So our next speaker is Dr. Curtis Huttenhower from Harvard School of Public Health, and
he's going to present Functional Analysis of Human Microbiome, Metagenomes, Metatranscriptomes,
and Multi-omics.