Tip:
Highlight text to annotate it
X
Gary Huffnagle: All right, I'd like to start off by thanking
the -- Lita and the organizing committee for this invitation to speak. The goal of my presentation
today is both included in my title, which is to bring the audience here up to speed
on this rapidly-changing field of the study of the microbial communities, the microbes
themselves that we're finding in lungs both during health and disease, but then also identify
the gaps, challenges, and needs.
And so to start with, I would like to take you a little bit back in history because I
think this puts it in the context. And so this summer actually pushes, I was just thinking
about this, this is the 10-year anniversary of when my post‑doc Maury Nover [spelled
phonetically] in my lab was finishing studies that -- in which we were able to demonstrate
using mice, that if you treated mice with an oral, poorly-absorbed antibiotic, you could
change immune responses in the lungs, a distal site. And the importance of that set of observations
is because, at that point, in this idea of the study of the hygiene hypothesis, which,
by the way, had its origins in the study and -- of the etiology of allergic disease. One
of the caveats is -- let's see if I got a pointer here -- is ‑‑ had something to
do with antibiotic use, so high antibiotic use, low antibiotic use. And this is coming
on about 15 years or so of epidemiologic data that pointed stronger and stronger and stronger
that antibiotic use correlated with the development of allergic disease, but not a single study
that could actually demonstrate that that could be the case. And so that's what that
was.
The -- interestingly, the results as I presented them were well received at the meetings I
was at. We were actually asked to write a review on the topic. Okay, a review on the
topic that never existed. That was actually quite an interesting challenge to us. So we
kind of made it more of an opinion piece back then, of like, "Okay, here's where the field
stands, here's the epidemiology; what are the possible mechanisms that could -- that
this -- that could drive this?"
Since the studies, there have been now a number of studies that have come out that are all
addressing the same issues and have shown the same sort of thing, that whether you use
poorly-absorbed antibiotics or you use germ-free mice, that if you disturb or alter the gut
microbiome, the microbial interactions in the GI tract, you can change immune response
in the lungs, not just allergic, which is on this slide, but on also immune responses
to viruses.
So the question was, then, how's this work? And so just a little interesting side note,
so as we put these studies out for publication, Nature turned -- their response was, "These
are not broadly applicable. Go away." So it was a 39-hour rejection. Science did not send
us out even for editorial review. And later on as we went through our review processes,
this is the question that keeps coming up again and again, "How's it work?" Okay.
And I love this slide actually from Jacque and Larry's studies. And so Jacque showed
the other day, or yesterday, about the bouncing ball along trying to get from one point to
another. And this is basically how this field has moved for the past decade. Which is, as
we try to go from one point here, to trying to get to the question, how do things work,
we've run into many obstacles. And when I say "we," it's not just us, it's many other
laboratories as we bounce around. And what are those obstacles?
Well, the gap is trying to understand it, and the option, the obstacle, is the type
of science we need to do when we're talking about distal site communication. And that's
this idea that this is a complex in vivo system, and how do you get this by study sections
in which they're dominated by reductionist biologist? And so I know I'm preaching to
the choir here because we deal with interdisciplinary research all the time and things like that.
But this has been a huge, huge challenge to this field. And so -- but we're not without
ideas of how this works.
And so the idea is that -- so as you inhale antigen we know a lot about how immune responses
develop in the lungs and their outcome of that pulmonary challenge. But studies showed
many years ago that sampling itself, it's hard to deliver something directly to the
lungs, whether it's intranasally, inhalation, et cetera, things get swept up and swallowed,
and they go to the GI tract where the immune system in the GI tract picks it up, and now,
as a number of studies have been highlighted so far in this meeting about the potential
of immune regulation in the GI tract affecting sites, whether it be the lungs or the brain
or whatever, the idea you can move around. And then, of course, the idea that the microbiota
might actually influence how immune responses develop in the gut. We know that, or you know
a lot of studies here are talking about this for local responses, but then the potential
that this could affect this immune regulation this way.
But now there's more evidence coming out that microbial metabolites circulate. And so it
opens the possibility that metabolites might also be affecting this response here. And
there's actually a wonderful set of studies by Ben Marzlin in Switzerland that was on
-- that I've seen at meetings so far where he's looking at short-chain fatty acids and
their ability to affect pulmonary immune responses when delivered systemically.
But one of the sets of reviews that came along during this whole process was, "Well, what
-- you know, are your antibiotics affecting the microbes in the lungs?" Okay, well, besides
the fact that these are poorly-absorbed antibiotics, and later on we've shown that they actually
would not have any effect even on an infection, what a stupidly absurd questions, okay, 10
years ago, because look at a text book. The normal lung is free from bacteria, okay? And
so, in fact, is that one of the drivers that when you look at the Human Microbiome Project,
there is nothing there in the lungs. There is no sampling of the lungs in the original
surveys.
And so the question about paradigm shifts is what I'm trying to focus on here. And so
a few years ago, then, a landmark paper came out from Markus Hilty and Bill Cookson over
in England that actually showed that in asthmatic airways -- well, first off, they showed that
-- this is using clone libraries -- that in allergic disease, a) there was ‑‑ you
could isolate microbial signatures from the lungs of healthy individuals. They changed
during asthma. There is also some COPD data in there. And so this was pretty fascinating.
And so since that point in time, there have been a number of diseases in which lung microbiomes
have been reported to be altered. I put cystic fibrosis on here because, actually, that disease
in and of itself has a really long history of having microbial colonization of the airways.
And it kind of -- it's cheating. It doesn't really count because we know that with all
that mucous there, it really changes the architecture, and you've been able to culture a lot of things
out. But for a lot of other diseases, culture-based processes have been ‑‑ fell short. But
now we know in asthma, COPD, and in bronchiolitis obliterans syndrome that follows lung transplantation,
that we can find altered microbial communities.
So let me tell you a little bit, then, about airway anatomy, because this is very important
when we talk about where are the microbes or where could they be. And so you need to
realize that for anything that's coming down here, this tube that starts in the mouth and
the nasal pharynx, so it's a fairly significant microbial load up in the mouth and nose. And
then it takes a 90-degree turn. Okay. And then you head down the throat, you get a split,
whether you go down the esophagus or into the lungs. And then the lungs begin to branch,
and branch, and branch, and branch, and at the very end of those branches are the alveoli.
And so there is one little area here, so where the larynx is, I mean, could potentially serve
as kind of a dam between the upper and lower airways. We know that there is cilia, ciliad
[spelled phonetically] epithelium that are pushing the mucous up, and there's lots of
turns and branches. And, in fact, it kind of -- this is what the lung looks like. So
you get this tree, and you get things that are subject to gravity, some areas that are
not. And actually, if you look at a diseased lung, we later on published a study looking
at explant [spelled phonetically] lungs. These are lungs that are pulled out of individuals
who had advanced COPD. We pulled the tissues sterilely, look at different regions, and
you can actually see regional heterogeneity in disease of actually the microbes that are
growing there.
So what are the potential sources of microbes in the lungs? They include nasopharyngeal
aspiration, inhalation -- the air that is full of bacteria -- reflux and aspiration,
and then probably only if you're ‑‑ if there's a problem would you actually get microbes
coming into the lungs via the bloodstream.
So can lungs be considered sterile? And if so, what do we mean by sterile? In other words,
does it mean that they lack microbial exposure? Well, no way. I mean, they're at the end ‑‑ they're
the cul-de-sac at the end of a very busy street.
Of the microbes we find, is there actually microbial metabolism going on? Are they alive,
are they metabolizing? That's a question that we're ‑‑ that is an active area of pursuit
in this field.
What about replication? We're finding microbes there. Are they actually replicating? So if
they're alive, or are they just persisting. Again, an unanswered question.
And what about colonizers? Like in health, certainly in disease we can find them, but
at what point in transition from health to disease do we actually now get colonization?
So it really raises a question. If there's ‑‑ if the lungs, the healthy lungs, are
exposed all the time to microbes or microbial signatures, and there are live microbes going
down there, what do we call this constant, this persistent low-grade flow of microbial
immigrants into the lungs? What do we call this microbial flux? The microbiome of the
lungs?
So now how do you sample the airways? This is another issue in this field. So, first
off, you can get sputum, which is basically a mucous plug, in a sense. It's mucous from
the sort of upper airways. It can come from a few branches down but you hack it up. And
it can be spontaneous or induced. So spontaneous is usually if you're sick. Otherwise, you
squirt some hypertonic saline in the back of your throat and you hack it up. But then
there's bronchial lavage. So sticking a tube down there to rinse out the airways, but in
that, as you stick a bronchoscope down there, you could put a protected brush. There's a
little cap at the -- little waxy sort of plug that sort of pops out as you get down there,
and you can brush the epithelium. You could do a biopsy, either through a bronchoscope,
or you could come, potentially, through the chest cavity. And then there's the concept
of sterile tissue sampling, which largely is only going to probably occur in either
cadaverous lungs or ones removed for lung transplant.
So the gap is determining the degree of bacterial transience versus persistence versus colonization
in the lower airways. And the challenges are the types of sampling. To study the lung,
everything is invasive. Okay? So the rules are just different right from the get-go of
how we can handle samples. We can't do longitudinal sampling, I mean, when you stick bronchoscope
down there, with the exception of lung transplants where you do surveillance bronchoscopies,
you can't really go more than two or three samplings in an individual over a period of
a couple of years even. And then, of course, as I mentioned, the bronchoscopes go through
the nose, they go through the mouth; there's some potential for contamination samples from
nasal or oral microbiota. And so a couple ‑‑ to deal with that last one, a couple
of bioinformatic options have been put forward which are very good. So one from Ric Bushman's
lab on using a single-sided outlier test, and another one that Tom Schmidt in our group
is actively using. It's been published in the Lung *** Microbiome Project, which is
called the neutral community model. Which is ‑‑ I mean, the idea is that you're
at the end of a flow, okay, so how can you tell the difference between a scope going
down or just what naturally flows down? And then is there selective pressure in that site?
But you want to know what's down there? Well, this gives you kind of an idea. And so if
we do a rinse of the bronchoscope before it goes down, we get, using 16S PCR ‑‑ QPCR
in a 5 ml bronc sample, it's about 1,000 copies. In a healthy, non‑smoker it's about tenfold
higher than what we see over base line, so maybe about 104 copies per 5 ml of BAL. If
you have disease, or interstitial pulmonary fibrosis, or lung transplant, you can see
the numbers go up.
So to give you an idea what do these communities look like compared to, like, the mouth. And
so this is from a study in our group in which we looked to the left BAL, right BAL, and
then other, and then the oral wash. And so they're color colored. So now here's the blue,
which is the left BAL, and so is the centroid of all the BAL samples from our subjects.
Here's the black, this is the right BAL. You can see it doesn't matter if you're left or
right, that the centroids are smack on top of each other in terms of what the populations
look like, but here's the mouth, and the centroids are different. You can run statistics on it,
and this, as a population, as a collection, is significantly different than what you see.
So the mouth is significantly different than the lavage.
However, we talk about individual-to-individual variation. And so what we can also do is go
back to these samples and measure a distance metrics. So Bray-Curtis, thetayc, morisitahorn,
you name it, but in this case, we did a Bray-Curtis distance between samples, so the oral wash
of a single individual versus her left BAL, or the oral wash versus the right BAL, and
what you see is a spread. So you see some individuals in which a rinse of their mouth
and what we get out of the BAL is wickedly different, and you get some that are very
similar. What does that mean? We don't know. But if we run some of these tests, like, for
example, the single-sided outlier test with it -- here's one with the low Bray Curtis
dissimilarity, so therefore they're very similar. We go up here to the outlier test, and you
can see most of what we find, this is oral versus lung, is being found in both. And so
here's another example. Okay, everything's falling along the line. But what about a sample
up here? You can see that what we find in the oral and what we find in the lung are
very different. There's another example of that.
Okay, so we don't know what it means when a healthy individual comes in and their bronchoalveolar
lavage looks like an oral rinse, or if it doesn't look like an oral rinse. And again,
we can't follow over time what that actually means.
So what I think the field so far, the investigators are involved, because there's, obviously,
as you can imagine, there is some debate about sampling, about how to handle contamination
and things like that. But I think what we can all agree on at this point is that when
healthy, the microbial load in the lungs is low, and the BAL samples contain a predominance
of bacterial taxa that's also found in the mouth. We know the bacterial diversity in
the lungs is very low. Actually, we can use that to our advantage because we can do pyrosequencing,
and we don't get that many types of OTUs, and we can actually take consensus sequences
from those OTUs from the 16S rRNA gene amplicons, blast them backwards, and really come up with
only one genus that it could be. And sometimes we can get as low as a species because there's
no other options around it. And so it's actually very useful that way. So we -- again, and
we know in some individuals there's some differences, suggesting that selective pressures can exist
in the lungs for elimination, persistence, colonization, and growth.
Now if we move to disease. So here's our healthy individuals in red. Here's individuals who
had a lung transplant and these are bronchoalveolar lavages, and this light green, hopefully you
can all see it, is interstitial pulmonary fibrosis. And you can see that if we call
this side the normal cluster, you can see there's some lung transplants that are up
in the normals. There are some IPFs that are up in the normals. But if we focus on interstitial
pulmonary fibrosis for just a second -- so we'll call this the IPF, the healthy group,
and this is the other group that doesn't look healthy -- and we actually ask who are the
bugs that are there? Who are the bacteria that are there? What you can see is in healthy
individuals, we get this signature that not only we get, but they get it in Europe, they
get it in labs all around the country, which is the dominant organisms that are coming
out of a bronchoalveolar lavage are prevotella, veillonella, and streptococcus. And then,
quite often, things like fusobacteria and neisseria. And so when we look at our, quote,
unquote, "normal," or IPF1 group, we get the same sort of organization. But if we look
at this other group, suddenly we get pseudomonas, we get Escherichia. So we're suddenly getting
some gammaproteobacteria, some Gram negatives that are coming in there.
Now interestingly, if we go looking at the lung transplant recipients, or healthy controls,
we actually found, surprisingly, an OTU that when we blasted, it came back as pseudomonas
fluorescens. And interestingly, it was in our ‑‑ only in the lung transplant. We've
never found it in our healthy controls. And what it ‑‑ we find it only when, in our
lung transplants, when the number -- when, basically, the relative ratio of prevotella
reads is low. So in other words, when it stops looking kind of like that mouthy-lungy sort
of pattern, suddenly we get pseudomonas fluorescens there in a large number of individuals. And
so pseudomonas fluorescens should perk the ears of a few people in the audience because,
like Baufer [spelled phonetically], and Jonathan Braun, and anyone else who works with Crohn's
disease because you make antibodies to it at a very high rate if you have Crohn's disease,
yet it's not a pathogenic organism, or at least it's not believed to be. So why is it
there? Why is it in these lungs? And so to get to that question let me summarize this
point and say that when diseased, the microbial load in the lungs increases and BAL samples
now often contain numerous bacterial taxa that are not found in the mouth, indicating
that there are selective pressures in diseased lungs.
So the challenge, now, to my question that I just asked you is that studies only involving
human subjects will never demonstrate causality, no matter how large the cohort. And so in
vivo animal studies, model organisms, and in vitro experiments are needed delineate
the mechanisms. They have to work hand in hand. They've got to go ‑‑ so translational
research is iterative. You go back and forth, back and forth, back and forth. And so we
need support for animal models in the study of the human microbiome. Otherwise how do
we get back to this whole thing about the hygiene hypothesis, and why these elements
are on the different tip or the balance?
So to address the question, why is pseudomonas fluorescens there in the lungs? We've been
working for a long time on a model in which you can generate allergic or airway inflammation
by multiple exposures to Aspergillus fumigatus spores. And so this shows you after four challenges,
eight challenges, you get lots of leukocytes in lungs, lots of inflammation, TH1, TH2,
TH17. The lungs are kind of really messed up. Well, what we did is we went back and
looked at those animals, say, I wonder what happened to the lung microbiome of those mice?
And if you go over here, this is, again, pyrosequencing. Look, we saw this bloom of gamma proteobacteria
that happened after four and eight challenges, this chronic inflammation that's going on
in the lungs. And if we look, pull the OTUs out, the one that jumps out the pseudomonas
fluorescens. Well, granted, we got lucky, okay, because we didn't -- that wasn't the
point at this point. We've actually gone and taken the inflamed lungs and put pseudomonas
fluorescens in it, and it does indeed like inflamed environments. And so it raises the
question now, gammaproteobacteria ‑‑ so pseudomonas fluorescens is non pathogenic,
but the gammaproteobacteria, Escherichia is a gammaproteobacteria, Gram negatives, okay,
it's a common theme that we've actually seen also in the gut, that inflamed sites favor
the growth of Gram-negative organisms.
So the final case that I want to show you here to show you how our paradigms are changing
and our thought process is changing, is we've known -- we've assumed for a long time that
Klebsiella pneumoniae is an etiologic agent of pneumonia. However, as studies from our
labs and my collaborators, we can take two different strains of Klebsiella pneumoniae
that were isolated from patients with gram negative pneumonia that responded to antibiotics,
et cetera, put them into mice. One will kill the mice, and at the same dose, the other
one, they'll walk away. And so the other way around. One will kill the mice. One will walk
away. Yet they came from human beings that had disease that responded. So what is going
on there?
So an interesting thing popped out of our studies of the lung transplant recipients
if we actually looked -- so this is a single individual, two BALs over six months in time.
At this point they had a pneumonia by CT, and they had all the clinical symptoms. They
were put on Ciprofloxacin. They recovered. They did much better. The culture was negative
the whole way along the way, but when we looked at it ‑‑ this is from the pyrosequencing,
you can see the -- sort of the normal cluster in here, and when they first came in, they
were way different. And, by the way, these individuals all have pseudomonas aeruginosa
up here. And -- but after Ciprofloxacin, they had a microbiome of their lungs that looked
kind of normal.
So what did that look like? Escherichia. Just dominated by Escherichia, but it didn't grow
out.
Okay. So we talk about unculturable bacteria, or difficult in the GI tract, and there are
things that you look at, and you say, "I don't recognize that name," you know? And so what
about the ones whose names we do recognize, that suddenly, I'm going to argue in the airways,
there is something different about microbial growth in the lungs that they are not able
to be cultured under standard microbiological techniques that we use now, which means that
the gap is that we need to understand the implication of culturable and non‑culturable
states of bacteria in the lungs.
And also, I've been talking of bugs; where are they? Okay. So there's some work that's
coming out now, Bill Cookson and things like that. And so whose alive, whose dead, where
are they? But I just gave you an example of somebody who clinically responded to the fact
that they've got a bug in their lung but we couldn't grow it out. And we've actually,
now, we've got one set of pilot experiments where we've done, in a sense, the same thing.
We've taking a culturable bug, put it in an inflamed lung, and watch it become unculturable.
I don't what it means, though. So the bottom line is we need new cultivation strategies
when we start to study the airways.
So the overall challenge of the field? Yeah, I can't underline this enough, okay? We need
more consistent and supportive peer review of microbiome lung proposals. Granted, this
is all aimed at my world, but my point is whether it's looking at the microbiome that's
in the lung, or the role that distal communication. So I guarantee that anyone who is trying to
do microbiome, nervous system, or gut microbiome nervous system, or gut microbiome, some other
site other than locally, your proposals are going to be open for a field day because there
are so many things that can be poked holes at, and it's hard for the field to move forward.
We also have to accept that sampling of the lower respiratory track in humans will be
imperfect. I cannot even think of how we're going to solve that problem, okay, because
of ethical reasons. And so we are going to need to utilize animal models to close this
gap. And so this is the research group that I showed some other work both in terms of
in my laboratory, but then also my wonderful group of collaborators at U of M. And so I
thank you very much for your time.
[applause]
Female Speaker: Well, thanks, Gary, for that great talk, and
for so clearly articulating some of the gaps and challenges in this area. So we do have
time for a couple of questions. We have -- over there?
Male Speaker: So, just in the spirit of a little bit of
dialogue about this. I -- we -- none of us want to interfere with the exquisite NIH review
process, but --
[laughter]
And I meant that sincerely. But the -- I wonder if one of the things that we could talk about
is the possibility of sort of a trans-institute review panel that was microbiome-friendly,
if -- I would just introduce that idea. I've spoken to an awful lot of people that have
related issues in terms of just, are there panels out there that are sensitive to ‑‑
Gary Huffnagle: So I'll throw my opinion out there, which
is that my first thought is that I like that idea, but then I realize it's a double-edged
sword, okay? And so, like, I sit on study section, so while we're not supposed to talk
about the "F" word when we review a grant, okay. You know the bottom line is, is that
you know that when you sit study section that, if there is a pile of 100 grants that come
in and you know that the institutes are funding at around 10 percent-ish, you know that 90
of those grants are going to go home and 10 are going to stay here. So now, what if we
take everybody in this room, and take all your grants and put them in one study section?
That means that 90 of you are going home.
So there is a danger by lumping everyone together as opposed to moving across, because if we
at least keep things spread, then you're looking within an area, within a discipline, what's
in a sense -- what's the relative important of microbiome research versus cell biology
versus an epidemiology, that sort of thing. But the problem is -- my big beef with the
peer review system is since when is peers three? Okay. You have a study section of 20-something
people, and that very first pass to get through, you have three people read? And then that's
it, that can determine a fate? So basically, it's like the UN Security Council. One individual
can trash something. That part needs to be adjusted a little bit, when you're dealing
with interdisciplinary or risky research, or discovery, okay? Because when it's not
hypothesis testing and it's hypothesis generating, boy, it gets really risky. But I do think
an active conversation needs to be made -- needs to be put up.
Female Speaker: That's probably a good topic for our later
discussion, too. We'll take one more question while our next speaker comes up.
Male Speaker: So regarding your non‑culturable Escherichias,
since they came from a pneumonia which there may have been a biofilm -- so in the biofilm
field, people talk about persisters. Do you think it could be something like that?
Gary Huffnagle: Yes, I do. Actually, I think -- I definitely
think that that's ‑‑ so there's two things at play here. One is the idea of persisters,
and then the other idea is that, you know, the microbial load even in pneumonia is not
necessarily high. And so what we may end up having is sort of, in terms of ecological
sense -- terms, like islands and, you know, or whatever, oases and desserts, okay, so
within the lungs. But I do think that those are mechanisms, those viable but unculturable
mechanisms, are going to be at play here.
Female Speaker: Thank you.
[applause]
Female Speaker: Our next speaker is Dr. Vince Young at the
University of Michigan. He's going to enlighten us with a talk entitled "The Microbiome in
Infectious and Noninfectious Gut Inflammation."