Eczema, Immunity and the Skin Microbiome – Heidi Kong
Articles,  Blog

Eczema, Immunity and the Skin Microbiome – Heidi Kong

Heidi Kong:
Good morning. I’d like to thank Lita for the invitation to come here today and talk to
you about our work and how it relates with our HMP demonstration project on atopic dermatitis. We’ve heard a lot about the gut microbiome,
and I think it was nicely organized that Susan’s talk about the skin and wound healing segues
very well into my focus on the skin microbiome. And one of the overarching questions that
has developed in our time in looking at the skin microbiome is: Do skin microbes influence
host skin immunity? I’m going to go give some background for those who are not familiar
with the skin microbiome work in order to be able to talk about the microbiome in eczematous
skin. I’m highlighting here a series of experiments
by a collaborator, Yasmine Belkaid’s group, to highlight some of the differences that
we see in that — this — not only is the skin microbiome different than other body
sites, but that the skin immunity and the skin microbial relationship is also distinct.
Here, Trudy, who was a graduate student in Yasmine’s lab, she treated SPF mice — these
are all experiments in mice with antibiotics — and what we can see in the gut, that after
antibiotic treatment, there is a significant shift in the microbial communities, but you
don’t see those significant changes in the skin microbial communities. Also, Trudy looked
at the immune cells that were in the gut: IL-17 LMA, as well as interferon gamma-producing
cells in the gut. And, as compared to the skin, once there was treatment with antibiotics,
there was a significant reduction in the interferon gamma in IL-17A-producing cells. So there
are, in distinct epithelia, there are differential responses to antibiotic treatment, at least
in these mice. So, Trudy then went on, and in comparing SPF
mice and germ-free mice, SPF mice here, germ-free mice, looking at IL-17A producing T-cells
in the gut and the skin, we see here that there are reduced IL-17A producing T-cells
in the germ-free gut and skin. However, what she then did was topically monoassociate staph
epidermidis in the germ-free mice, and interestingly, there were no changes in the IL-17A-producing
T-cells in the gut; however, in the skin, there was an increase in the IL-17A-producing
T-cells in the skin after application of staph epidermidis. So we can see here that skin
microbes do influence the immune system, but is there a function? Is there a biological
result when you do that? So, what then Trudy did was use a Leishmania
major infection model. She infected the ears of these SPF mice and germ-free mice, and
in comparing these two, you see that there is a local cutaneous response in the SPF mice,
as compared — much larger than the germ-free mice, but also so you see increased interferon
gamma-producing T-cells in the SPF mice, as compared to the germ-free mice. So then when
she topically associated staph epidermidis in these germ-free mice, you see an increase
in the local cutaneous response, as well as interferon gamma-producing T-cells that are
in the ear. And these resemble what you see in the SPF mice. But also, here we see in
parasite burden, that you have a decrease in the parasite burden when you treat or topically
associate staph epidermidis. So skin commensals can restore immunity to Leishmania major infection
in germ-free mice. So summarizing this part, the host immunity
and microbial interactions in skin are distinct. In mice, skin microbes can tune the skin level
of activation and function of skin-resonant T-cells, promote immunity to pathogens, and
drive responses locally that are distinct and independent from the gut flora. But what about human skin immunity and microbial
interactions? Unfortunately, we don’t have a lot of germ-free humans walking around that
we can do these types of experiments, but what we have done is started to look at what
are the microbes that are on human skin. So, again, this is a background for those who
are not as familiar with human skin microbiome surveys. These are highly selected due to
limited time that I’m not going to go into all the surveys that have been done. You’ve seen this figure before, but the reason
why I’m showing it again is I want to emphasize that in skin, here in pink, there is a wide
variation in the microbiome that you see in skin, as compared to the other body sites.
And that is because the skin’s surface is highly heterogeneous, and that is one of the
reasons why, in this study, we selected 20 skin sites. These are distinct microenvironments
on the skin’s surface, but I also selected them because they were sites of predilection
for skin disease, specific skin diseases for which microbes have been postulated to potentially
have a role. And David Relman showed this figure earlier, so I won’t go into it in detail,
but relative abundance of predominant bacteria appear to be dependent on the microenvironment. So we’ve talked about this throughout yesterday
and some this morning, how there are carefully-defined cohorts that we’ve looked at; most of these
are adults. We’ve talked about what happens, there have been the studies by Maria Gloria,
looking at the neonatal period, but what happens after that? Ruth Ley talked about the — at
least for the gut microbiome, how that can transition in the first two and a half years,
but what about the skin? We don’t know that information. In the other — next transition:
puberty. This is a turbulent period of time for our bodies. What happens between — as
we transition into adulthood? There’s something called a Tanner stage. For those of you who
aren’t clinicians, the Tanner stage is used by pediatricians to assess the level of sexual
maturation in an individual child. This is based on development of breast tissue, as
well as genitalia, pubic hair, but it ranges from one to five: one being pre-pubertal,
five being fully-developed adult. So, what we have here is we had cohorts of
patients that were — ranging from two to adulthood, so what Julia Oh did in our — in
Julie Segre’s lab was take these data and compare them with each other. And based on
Tanner stages one, two, three, four, and five, we see the striking difference between Tanner
stages one, two, three, and tanner stages four and five. This is from the nares, but
we see a similar separation in other body sites — in other skin sites. What drives that? Here we see a relative abundance
chart, which you’ve seen a lot in these talks leading up to now. And here is Tanner stages
one, two, three, and then here is four and five. I do acknowledge that the ends are pretty
low in these populations, but what we see in Tanner stages one, two, three, that we
have increased proteobacteria, and more lipophilic bacteria in Tanner stages four and five. That
makes sense for anybody who’s gone — everybody who’s through puberty; you recognize that
your body is changing dramatically, and one of the things that changes is the further
maturation of our sebaceous glands, or oil glands in our skin. So it’s very possible
this is one of the reasons why we have more lipophilic bacteria, because they can thrive
in that type of environment. So we’ve talked a lot about 16S taxonomic
surveys. We’ve heard some about the virome, and several of the investigators looking at
that. But what about the fungal microbiome? So, more recently, we’ve published a project
looking at sequencing the fungal organisms on the human skin surface, but in order to
do this, we had to go back from the beginning and figure out how do we optimize sample collection.
And this refers back to the conversation we had in the open floor session yesterday: What
are the protocols to use? Optimizing DNA extraction, we had to deviate from the standard protocols
for extracting DNA in order to optimize extractional fungal data using bead-beating, opening the
fungal cell walls. What primers would we use? There are individuals who use 18S, some who
use 28S. There is the ITS region here: ITS1, ITS2. That stands for internal transcribed
spacer region. And so based on our analyses, for us, at least for skin, that ITS1 provides
more taxonomic resolution. This alludes back to Jacques Ravel’s comment yesterday that
just because it works for the gut or for the vagina, it may not work for the skin. And
so this is for, at least in our experience, ITS1 worked better for taxonomic resolution. So what we did was we had 10 healthy volunteers.
I selected 14 sites. These are not identical to the ones that we did in the bacterial survey,
because these are, again, sites that reflect sites of predilection for skin diseases for
which fungal contribution may be playing in a role. So in 10 healthy volunteers we see
here. Then we — if you look at the relative abundance charts, each of these horizontal
lines represent one body site. Each of these vertical bars represents one of the healthy
volunteers. And one of the most striking this that you can observe here is that malassezia
predominates in 11 out of the 14 sites. That’s the purple. That’s why there’s so much purple.
But on the heel, toenail, and toe web space, so on the feet, there is much greater diversity.
We have fungi everywhere, but it’s just a different population depending on where you
are. One of the things that you may notice is that
healthy volunteer seven is a bit different. This individual, again talking about protocol
standardization, our eligibility was no one could take — have taken an oral antifungal
or an oral antibiotic within six months of being sampled for this protocol. And this
individual had completed a course of oral antifungal six, seven months prior to being
sampled. And so it’s not clear whether the differences we see are related to residual
effects of taking an oral antifungal, or whether or not the person’s predisposition — they
were taking for a toenail infection — whether a predisposition for a toenail infection represents
some difference in their fungal microbial communities. Interestingly, I don’t have the
data right here, but their 16S survey for this particular individual, HV7, was similar
to the others. So the bacterial microbiome was — resembled any other healthy volunteer. But it was frustrating to see so much purple.
There’s a lot of malassezia; that’s not that helpful. But one of the questions we had was,
what happens if you speciate? Getting down to the speciation level, how do we do that?
Well, we couldn’t do that based on the databases that were available, so we had to cultivate
our own malassezia and do genomic sequencing to actually populate the databases for us
to be able to speciate. So when you do that, you again see there is striking site specificity
for skin. Sarkis was talking about how there’s specificity in the gut, but also we see that
for the skin, with regards to the fungal microbiome as well. So we can see in this area here,
the external ear canal, behind the ear, the forehead, that malassezia restricta predominates.
But in other sites, we have the upper back, the back of the scalp, the inside hip, those
are — you have malassezia globosa. So even though they’re all malassezia, there is specific
species that we’re seeing at these different body sites. So I did mention that we did 16S, so from
these same clinical samples, we did ITS1 sequencing as well as 16S sequencing from these samples,
and what you see here is interesting anatomically, in that those central body sites from the
head and neck, or core body sites listed here, they have a relatively limited number with
regards to richness, relatively limited numbers of different types of bacteria and fungi.
Whereas on the arms — here’s the palm, the forearm, and the inside elbow — there’s a
higher richness, a greater richness, with regards to bacteria, but still relatively
limited for fungi. That’s different for the feet. Here, the heel, toe web, and toe nail
— relatively limited richness for bacteria, a much greater richness for fungi. So there
are these regional differences that cannot be explained just by sebaceous, moist, and
dry, and that it’s going to require a lot more understanding about human physiology
and skin physiology to be able to explain these differences. So just summarizing this beginning portion,
the skin bacterial microbiome is highly dependent on the sampled skin site; the neonatal skin
bacterial microbiome varies based on mode of delivery. We’ve talked about that. That
there are dramatically major shifts that we see in the Tanner stages one, two, three,
versus Tanner stages four and five. And that fungal communities over the skin’s surface
differentially vary from the bacterial microbiome. Now, I’m going to shift into what my charge
was, into talking about eczematous skin, specifically atopic dermatitis, and then briefly I’ll talk
about some data from other groups and our group about primary immunodeficiency syndromes
and why that’s interesting. So for background, those who are not familiar
with atopic dermatitis, it’s a chronic, itchy inflammatory skin condition.  It is not considered
an infectious disease, it’s an inflammatory skin disease, yet these individuals respond
relatively well to anti-microbials, so there is something going on that suggests a role
of microbes.  It affects 15 percent of U.S. children at a high cost financially as well
as socially.  The quality of life in these children and in these families is severely
adversely affected.  And I mentioned before, there is this association with microbes we
observe, and when we take care of these patients, that disease flares are associated with increased
colonization and infections of staph aureus, but also a subset of these patients are at
high risk for severe spread of herpes simplex virus infection, and if they come into contact
with smallpox vacinis [spelled phonetically].   There is something that’s been termed the
atopic march, in that there are 40 to 70 percent of those with severe atopic dermatitis over
time go on to develop asthma and then hay fever.  So these three diseases are termed
the “atopic triad.”  The incidence of these atopic diseases have doubled in the last three
decades in industrialized countries, suggesting there may be a possible external factor, and
this alludes to some of the comments yesterday by Marty, and again today, that it is unlikely
that our human genome can change in that period of time, and that an external, possibly microbial,
contribution may be playing a role.  Interestingly, in mirroring studies, skin exposure to antigens
can result in subsequent mucosal sensitization to those antigens, suggesting that if we could
somehow modify what happens in atopic dermatitis, could we then go on and abrogate the development
or the disease severity of asthma, which has significant morbidity and potential mortality,
as well as hay fever.  So understanding the triggers of atopic dermatitis may allow us
to modify the development of AD and atopic disorders, and potentially develop therapeutic
targets.   So atopic dermatitis is a complex disease.
 It looks very simple here, but this is not the whole story.  But just emphasizing, we’ve
talked about barrier, the gut barrier earlier today, but emphasizing — I won’t talk about
it here, but it is important in atopic dermatitis.  The skin barrier, we know if you have a
mutation in the skin epidermal protein filaggrin, it’s highly associated with the development
of atopic dermatitis, particularly the kind that goes on to develop asthma and hay fever.
 We know the immune system is deranged in these individuals, and that they have extremely
high IgE levels, and antimicrobial peptides in the skin are reduced. But — and we’ve
also talked about the microbes, and that is just one component of this complex disease.
  So our study, we recruited pediatric patients
with moderate to severe disease, and healthy age match controls.  I sampled them in characteristically-affected
sites.  As I mentioned before, dermatological diseases have sites of predilection where
we find them, and that helps us in the diagnosis of these diseases.  But it typically occurs
in the antecubital fossa and the popliteal fossa, that being the inside elbow and behind
the knee.  We selected the volar forearm, or the inner forearm, as a control site, because
it is less often affected in moderate disease, and it’s an adjacent site.  We also sampled
the nares, because that is a site of carriage for staph aureus.  And I sampled them during
the baseline, flare, and post-flare time points, and these are just SCORAD scores, which is
just a method for assessing severity in these patients. And you can see that, over time,
their SCORAD increases during the flare and decreases after they’ve been treated.   This figure just demonstrates that when you
have more severe disease, that we observe a decrease in the bacterial skin communities.
 But that drop in diversity is not everywhere, it is very site specific; again, it is at
the sites where we see disease appear — the inside elbow and behind the knee — but not
at all time points.  This, in particular, are disease flares.  These are the natural
— true natural history of the disease, where they have not been putting anything on their
skin.  These — the blue flares are individuals who put on topical steroids potentially two
days before seeing us, or sometime within the seven days before they were sampled.  So
these flares are truly the natural history of the disease.   But what are the bacteria that seem to be
driving this decrease in diversity?  So looking down to the genus — down to the genus level,
here are the healthy controls.  These are the baseline time points for the atopic dermatitis,
the flare time points, the no treatment ones versus the intermittently treated ones, and
the post-flare time point.  And what you can see is there’s a dramatic increase, a
significant increase in the pink, the staphylococci in the skin of these patients during a flare,
but we do see increased staph in proportion, or relative increase in some of these individuals.
 So that was quite a concern.  Again, we had to go down to speciation level, because
there’s staph epidermidis, staph aureus. Staph epidermidis, again, is a known skin commensal;
staph aureus is a fairly common pathogen.   And so it was important to us, at least at
the genome level — they’re very similar, so we had to know which one was which, because
it makes a difference, at least clinically.  And so what we did, if you focus primarily
on the pink, the staphylococci, and we speciated those, it was reassuring to see that most
of the staph we see in the healthy controls, those are staph epidermidis staph hominis,
known skin commensals.  But what we do observe is, yes, there is an increase in the staph
aureus we see, even at base line, but it is a dramatically increase during the flare,
the natural history of disease we see, increased staph aureus, but also staph epidermidis.
 That was not something we did expect to see, but there is — that is one of the benefits
of looking at the whole microbial community with regards to disease progression.  And
then it decreases once patients had been treated.  And one of the questions that people often
ask:  “Well, you gave them antibiotics,” but often the majority of these patients were
limited — their treatment was limited to topical steroids, or some of them do take
dilute bleach baths, which is like having a small swimming pool in your own tub.   So that’s what we observed, but what is — what
remains a question is our correlations: What happens to get from this point to this point?
 Or what can we do so that these — you never — you remain a control; you don’t go on to
develop atopic dermatitis.  Those are questions that remain. We did look at the fungal communities on the
skin.  Julia Oh looked at this.  These are baseline flare and post-flare for a few of
our patients who had atopic dermatitis, and although you see fluctuations in the bacteria,
we don’t see that type of fluctuation in the fungal communities on the skin. So I’m going to shift gears a little bit and
talk about primary immunodeficiencies.  There are some cohorts of primary immunodeficiency
patients who have eczematous skin disease, and the benefits here are they genotypically
have the same mutation.  They are monogenetic disorders, and some of these patients have
atopic dermatitis-like skin eruptions, and these eruptions can be antibiotic responsive
again.  And so that was one of the reasons why we pursued these cohorts, and asking,
do common and rare diseases with similar clinical phenotypes, do they share skin microbiome
features?  And how does the innate and the adaptive immunity shape the skin microbiome?
 We haven’t been able to ask all of these questions, but these are some of the reasons
we were led down this path.   I’m just going to talk about two of these
diseases in the rest of the time that I have, talking about hyper IgE syndrome, which have
STAT3 mutations, as well as another paper that was recently published by Dirk Gevers
and colleagues, looking at STAT1 mutations with chronic mucocutaneous candidal infections.
 So a STAT pathway is important; it’s a biochemical pathway, and it’s involved in so many things.
 But these patients have — are at risk for infections; on hyper IgE syndrome, they have
staphylococcal skin, and lung infections, candidal infection, and they can develop secondary
aspergillosis lung infections.   So this is the paper I just mentioned, by
Dirk Gevers and colleagues, where they looked at STAT3 and STAT1 mutations.  It’s hard
to see, but these — the left-most groups, these are the STAT1 mutation patients; the
middle group are the hyper IgE syndrome patients, or STAT3 mutations, and these are their controls.
 But, in general, what you see, just summarizing everything this little corner, at the genus
level, you see increased corynebacterium species in the STAT1 mutations. You have decreased
corynebacterium species in the STAT3 mutations. You have increased gram-negatives that you
observe in these patient populations, decreased prevotella and decreased fusobacteriales [spelled
phonetically].  So you — we observe that there are taxonomic differences that we observe
on the skin of these patients with primary immunodeficiency syndrome.   This group went on to do some studies of challenging
PBMCs.  These are PBMCs from healthy volunteers.  What they initially did was they prestimulated
them with either corynebacterium, acinetobacter, or staphylococcus, and then they challenged
them to candida albicans or staph aureus.  And what this essentially shows is that
if they were initially exposed to acinetobacter baumannii, there was a decrease TNF-alpha,
interferon gamma, and IL-22 that was produced upon exposure with staph aureus and C. albicans
in these healthy volunteer PBMCs.  So exposure to certain skin microbes may alter PBMC skin
cytokine response to pathogens, such as C. albicans and staph aureus. This is just a snapshot of some of our work
looking at patients with hyper IgE syndrome, STAT3 mutations, but due to the limited amount
of time, I will direct you to Poster 31, where Julia Oh will walk you through her analyses
that she’s looked at STAT3 mutation patients, but other primary immunodeficiency syndromes
as well. So, briefly, going through summarizing this
part, AD flares are associated with shifts in the skin bacteria.  We talked about the
different species. The specific primary immunity deficiency patients harbor a distinct skin
microbiota, and that altering the skin microbiome may alter the PBMC’s response to specific
microbes, but we need more studies.   Quickly, moving into the gaps, needs, and
challenges, since I’m running out of time.  These are knowledge gaps.  And then I’ll
go into — as Owen charged us with finding things that frustrate us on a day-to-day level.
 But knowledge gaps here are evolution of the skin microbiome over life stages, what
are the physiological factors that contribute to the skin microbiome differences, getting
down to understanding skin microbiome and immunity interactions.  Trying to expand
on what has been done now, but to explore that to a greater degree and fully understand
how our immune system interacts with our skin microbes.  And again, the importance of human
and animal models.   We’ve talked about this, but moving from correlation
to causation, but also, as our recent survey has highlighted, the magnitude of interaction
between fungi and bacteria, and the role in health and disease.  But one of the bigger
challenges that I don’t think I’ve seen anything been able to overcome this one challenge is
skin metagenomics, and that is due to the low biomass on skin.   So I do have a caveat. These are my views;
they do not represent the views of the government. But these are some of the frustrations that
we encounter on a day-to-day basis.  Standardization of protocols.  We’ve talked about this, but
particularly for skin, I did a non-scientific survey of some of the people who do skin work
here, and we have different protocols.  I will admit that we don’t follow the HMP protocol,
we have our own protocol that we’ve been doing, and we’ve had to modify that from time to
time.  So we are just as guilty of not having — following a standardized protocol.  The
importance of phenotyping patients, making sure are the really healthy, or are the normal?
 What diseases, particularly atopic dermatitis, has several different phenotypes, how do we
define what is the phenotype we’re studying, and how do we — especially you do that when
you’re at different, multiple sites. Which sampling sites do you do? How frequently?
 What is the skin prep?  The time since antibiotics?  What sampling method?  Is
swabbing still the best way?  What are the critical metadata fields that are needed?
 And it might change depending on the disease state you’re looking at. DNA extraction.  This is really critical,
looking at — especially when you have low biomass.  It’s easily contaminated when you’re
talking about skin, but then when we’re looking at fungal organisms, we had to modify DNA
extraction.  But the primers, PCR conditions, just looking at various abstracts of the past
and manuscripts, they’re all over the place. Quantitation.  This is a key question that
I always get asked: How do you know?  Can you quantify what’s there?  And we can’t
at this point, and that is a major frustration, I think, not just for skin, but in the entire
microbiome field.  More microbial characterization, where — including genomes, where we had to,
you know, sequence our own malassezia in order to figure out speciation.  Metagenomics analytical
tools, particularly if you have low biomass.  And one gripe about data submission: When
we’re submitting a manuscript and having to do dbGaP and SRA, that is a major frustration.
 Again, my own opinions.   So then, if I can move on to the acknowledgements,
if they will — it’s not — there we go.  I’d like to acknowledge my close collaborator
Julie Segre, and I have underlined here, as well as photos of the post docs’ work who
I’ve highlighted here, but there are many collaborators within my group, but then also
collaborators across different institutions, but especially our patients and volunteers.
 Thank you very much for your time. [applause]   Female Speaker:
Thank you so much for that great talk.  So we do have time for a question if there is
one out there.  Oh, come on.  Go ahead. Male Speaker:
That was a lovely talk.  In your discussion about the areas we need to work on methods,
I think today, as I was talking with a colleague, we’ve seen very few slides, actually, on the
methods that are used in any of the studies, particularly in humans, whether it was a wet
cotton swab, whether it was a brush. You know, I think that as part of the methodological
discovery process, we all need to disclose or discuss exactly what the methods are as
we’re going forward, even in we’re presenting data such as we have here.  I know from the
papilloma virus field, we went through decades of methodological considerations that were
critical.  So I applaud your, you know, pointing these out, and I encourage everybody to start
including more specifics on exactly how they’ve collected the specimens, processed, stored,
et cetera, so we can understand that as we’re interpreting the data. Heidi Kong:
Thank you for your comments.   Female Speaker:
Thanks for a nice talk.  When you showed the results, you show the fungi changes — Heidi Kong:
[affirmative] Female Speaker:
— and you went very fast, basically you didn’t do much comments, but I think I saw malassezia
globosa increasing in the skin of the — in the flare.  Can you comment on that? Heidi Kong:
Are you talking about the atopic dermatitis one?  This one?  Or were you talking about
— Female Speaker:
That one. Heidi Kong:
Yeah.  So this one I will comment, this particular individual was Tanner stage four, and so what
we see is it’s possibility related to, although this was a pediatric patient, that they had
already transitioned through puberty and had more oils on their skin, and so that’s why
we see that malassezia globosa is dominating potentially in this individual.  Or it may
be something unique to this individual, and we just — small, and it’s hard for us to
tease that out.  Thank you for that question. Female Speaker:
Thank you. Female Speaker:
Okay, so for our next talk — [applause] Oh, sorry. [laughs] You can clap.  Our next
talk, we’ll be moving on to the lung microbiome. Our next speaker is Gary Huffnagle.  He’s
from the University of Michigan, and the title of his talk is The Lung Microbiome: Challenging
Old Paradigms about Microbes and the Host Respiratory Tract.”


  • Paul McGlothin

    This is a very insightful and beneficial talk –good to listen to for skin microbiome basics. Thank you Heidi.

  • MrApplewine

    What was the difference in fungi in people who get AD and the controls who don't? Could trigger like IgE (say dust mites) be triggering an already present persistent local fungus to react triggering a bacterial biome disruption that then sets off the immune system producing AD? I'd like to see a list of theories. Thanks for sharing this information.

Leave a Reply

Your email address will not be published. Required fields are marked *