Oct 15, 2020
This month on Episode 17 of the Discover CircRes podcast, host Cindy St. Hilaire highlights four featured articles from the September 25 and October 9 issues of Circulation Research. This episode features an in-depth conversation with Drs David Dichek, Sina Gharib and Tomáš Vaisar regarding their study titled Parallel Murine and Human Plaque Proteomics Reveals Pathways of Plaque Rupture.
Article highlights:
Cai, et al.
Single Cell RNA-Seq in Arteriosclerosis
Schuhmann, et al. CD84
in Ischemic Stroke
VanOudenhove, et al. Gene
Regulatory Dynamics of Developing Human Heart
Nie, et al. Periostin
is a Target to Treat PH
Dr Cindy St. Hilaire: Hi, welcome to Discover CircRes, the podcast
of the American Heart Association's journal Circulation Research.
I'm your host, Dr Cindy St. Hilaire, from the Vascular Medicine
Institute at the University of Pittsburgh, and today I'll be
highlighting four articles selected from the late September and
early October issues of CircRes. I'll also be interviewing Drs
David Dichek, Sina Gharib, and Tomáš Vaisar regarding their study
titled Parallel Murine and Human Plaque Proteomics Reveals Pathways
of Plaque Rupture.
The first article I want to share is titled Single Cell RNA
Sequencing of Allograft Cells in Transplant Arteriosclerosis. The
first author is Jingjing Cai and the corresponding author is Qingbo
Xu from Zhejiang university in Hangzhou, China. Arteriosclerosis is
a major contributor to organ transplant failure. The thickening and
stiffening of arteries within the grafts results in diminished
blood flow supplies and diminished organ function. While it is well
understood that atherosclerosis is an inflammatory disease, the
details of the cellular and molecular players on this
transplant-specific pathology are lacking. Now, Cai and colleagues
used single cell RNA sequencing to identify a consensus of cells
and cytokines in sclerotic transplanted aortas in mice. Two weeks
after transplant, the grafted vessels exhibited signs of
arteriosclerosis and by four weeks, this remodeling had worsened
significantly.
Analyzing the RNA transcripts of over 12,000 individual cells
isolated at both two and four weeks, the team discovered that the
number of T-cells was greatly increased throughout the process. An
early abundance of macrophages gave way to a later wave of B cells
and there was also evidence of the development of tertiary lymphoid
tissue. They further found that chemokine CCL121 was up regulated
after transplant, both the mRNA in the tissues, as well as the
protein levels in the animal's blood. The authors then went on to
show that blocking CCL121 or its attracting partner, CXC3,
significantly delayed arteriosclerosis in the grafted vessel.
Hence, this work not only defines the cellular and molecular
drivers of arteriosclerosis in grafted vessels, but highlights
potential molecular targets for future therapeutic
interventions.
The second article I want to share is titled CD84 Links T-cell and
Platelet Activity in Cerebral Thrombo-inflammation in Acute Stroke.
The first author is Michael Schuhmann and the corresponding author
is David Stegner, and the work was completed at University of
Würzburg in Germany. Ischemic stroke is caused by the occlusion of
cerebral blood vessels and it is a leading cause of death and
disability worldwide. Despite treatments to degrade or remove clots
such as mechanical thrombectomy, infarct size itself can continue
to grow even when blood perfusion is re-established.
Thrombo-inflammatory processes are thought to mediate this
worsening injury, with both T-cells and activated platelets playing
a role. Because both T-cells and activated platelets express CD84,
which is a self-binding adhesion molecule involved in lymphocyte
activation, this team tested the hypothesis that CD84 might mediate
stroke inflammatory processes.
They went on to show that mice lacking CD84 have smaller infarct
sizes with reduced T-cell inflammation after stroke than wild-type
animals. Furthermore, mice that specifically lack CD84 in either
T-cells or platelets also experienced smaller infarcts. The team
went on to show that CD84 promotes T cell migration in vitro. And
then in patients with stroke, high expression of CD84 in platelets
was associated with poor outcomes. Together, these results suggest
that activated CD84-secreting platelets encourage inflammatory T
cell migration to the infarct site. And that blocking CD84 activity
could be a novel therapeutic strategy for minimizing inflammatory
injury after stroke.
The third article I want to share is titled Epigenomic and
Transcriptomic Dynamics During Human Heart Organogenesis. The first
author is Jennifer VanOudenhove and the corresponding author is
Justin Cotney. And they're from the University of Connecticut.
Congenital heart defects, or CHDs, are common birth abnormalities
and while some genes have been linked to congenital heart defects,
the majority, close to 60%, have unknown etiologies. It's thought
that multiple genetic and environmental factors contribute to
congenital heart defects. One of which could be variations in both
cis and trans regulatory regions of the genome. To find such heart
specific regulatory regions, this team examined heart tissue from
human embryos obtained four to eight weeks after conception.
They performed chromatin immunoprecipitation experiments to scour
the heart genomes for histone modifications associated with
increased or decreased gene transcription. They also performed
transcriptome analysis to see whether the genomic regions
identified by chip corresponded with the activity status of nearby
genes. In total, the team found more than 12,000 previously unknown
enhancers that were enriched for binding sites for heart specific
transcription factors, some of which included GATA, MEF2 and Nkx.
These binding sites tended to be close to genes activated in the
heart. Many of the regions also contain sequence variations that
have been associated with atrial fibrillation. These newly
identified sites are potential congenital heart defect candidate
loci and the authors have now made their data readily available so
that other investigators may study it.
The last article I want to share with you before we switch to our
interview is titled Periostin: a Potential Therapeutic Target for
Pulmonary Hypertension? The first author is Xiaowei Nie from the
Shenzhen Third People's Hospital and the corresponding authors are
Jingyu Chen and Jin-Song Bian from the Wuxi People's Hospital and
the National University of Singapore, respectively. Pulmonary
hypertension, or PH for short, is a life-threatening disease where
an excess in the proliferation of vascular smooth muscle cells and
the deposition of extracellular matrix thickens the walls of the
lung vasculature, which leads to an increase in pulmonary blood
pressure and ultimately contributes to right heart failure.
Vasodilatory medications can be used to treat the symptoms of the
disease. However, these medications do not prevent or reverse the
underlying pathogenic remodeling. This study now suggests that
drugs targeting the secreted extracellular matrix protein,
periostin, might be a potential therapeutic strategy for the
treatment of pulmonary hypertension.
Periostin is an abundant protein in the lung arteries of pulmonary
hypertension patients. And it is thought to be involved in cell
adhesion and wound healing mechanisms, such as the proliferation
and the migration of smooth muscle cells. The team confirmed
increased production of periostin in patient lungs, and also found
the same to be true for mice with an induced model of pulmonary
hypertension. They also showed that genetic deletion of periostin
attenuated pulmonary hypertension in mice, while suppression of
periostin via RNA inhibition could even reverse pathological vessel
thickening and the subsequent right ventricle hypertrophy. The team
went on to identify factors HIF-1a and TrkB as factors that mediate
periostin's effects in cultured arterial cells. And they suggest
that blocking either of these factors or by blocking periostin
itself could be a novel strategy for the treatment of pulmonary
hypertension patients.
Drs Tomáš Vaisar, Sina Gharib, and David Dichek from the University
of Washington in Seattle, Washington are here with me today and
we're going to discuss their recent study titled Parallel Murine
and Human Plaque Proteomics Reveals Pathways of Plaque Rupture.
Thank you all so much for joining me today and congratulations on
this beautiful and interesting study. So this is an atherosclerotic
study, but unlike many in the field, it's really looking at the end
stage event called plaque rupture. So for those listeners who are
unfamiliar with the term, plaque rupture is when an atherosclerotic
plaque degrades and its contents are exposed to the circulation,
which can then induce a clotting event and lead to all sorts of
adverse pathologies, myocardial infarction, transient ischemic
events, stroke. So I'm wondering if we if maybe you can give us a
little bit of background about what's known and what really is
unknown at least before your study regarding plaque rupture.
Dr David Dichek: So the pathogenesis of acute myocardial infarction
and stroke was really unknown for many years. And the idea that it
was due to acute thrombosis was really confirmed by a study
probably 30 years ago, that did angioscopy in the coronary
arteries, proximal to a myocardial infarction, and visualized
actual clot so that the clot was confirmed to be associated with
the acute event. At that point, the question became why would a
coronary artery form a clot? And that led to identification of, or
histologic studies that identified ruptured caps of atherosclerotic
plaques, exposure of the blood to the thrombogenic contents of the
lesion, and a thrombus.
However, it was not known what the initiating event was in rupture
of the plaque cap. And there were a lot of hypotheses and a lot of
nice work, but it does still remain an unknown. A significant
amount of focus has been devoted to the possibility that
proteolysis is the initiating event. And that was sort of the
takeoff point from our study because we had developed a mouse model
where proteolysis clearly was associated with rupture of plaque
caps. And we decided we wanted to get more into the biochemistry of
what was going on and go beyond the histology. So that was really
what led up to our study.
Dr Cindy St. Hilaire: It's really interesting. So, mice are really
good and obviously really useful, very well-known model systems to
study atherosclerosis and particularly the initial drivers and
maybe the mechanisms of the disease pathogenesis, but like many
models systems, they're not perfect. So I'm wondering if you could
discuss the limitations of murine model systems and specifically
for this study, how you were able to overcome some of those
limitations.
Dr David Dichek: So the limitations of mouse models of plaque
rupture are that essentially none of them duplicate the histology
of human plaque rupture, particularly the thrombus that occurs on
top of the plaque rupture. So there are various mouse models where
caps are disrupted, but there's not acute thrombosis. It has been
argued in the vascular community as to whether these models are
authentic models of plaque rupture, because they don't have the
superimposed thrombosis. And the counter argument is well, mice
aren't people, they have different hemostatic and coagulation
factors that may be differentially regulated. The hemodynamics of
small mouse arteries is different from human mouse arteries. And
the fact that you don't get a thrombus doesn't necessarily mean
that you're not modeling the process that would cause it.
So we really accepted that argument as being valid and felt that
the occurrence of frank plaque rupture, and that was in our
Circulation paper in 2010, in these lesions in the mice, really
validated it as an authentic model of cap disruption. And so I
agree it's arguable, whether this is an authentic model. But we
actually took that issue head on by saying, well, is the
biochemistry of the ruptured plaque similar to the biochemistry of
a ruptured human plaque? And that if there were similarities that
we would gain more confidence in our model being an authentic model
of plaque rupture and that it matched not only the histology, but
also the biochemistry.
Dr Cindy St. Hilaire: One of the main tools you used, you used
shotgun proteomics, which I think is just a great name for it. And
also a algorithmic learning tool or analysis tool called
proteomaps. I was wondering if you could give us a little bit of
background about the proteomic aanalysis involved, what does that
entail? Especially, when you're comparing a teeny tiny mouse plaque
to a larger human plaque and then how that analysis was done?
Dr Tomáš Vaisar: So the shotgun proteomics term was coined by John
Yates, but way back in the early 1990s. And it's essentially the
way how you enumerate the proteins and in more recent forms, you
even quantify the abundance of the proteins in a very complex
mixture. So shotgun proteomics essentially takes a protein sample.
And in this case it was an extract of the tissue and uses protease,
trypsin typically, to cut the proteins down to peptides, which are
relatively small and relatively well
behaved compared to intact proteins. And then using tandem mass
spectrometry combined with liquid chromatography separation,
basically aims to sequence every single of those peptides or
majority of the peptides. And then based from the identification of
the sequence of individual peptides piece back together like a
jigsaw puzzle, what was the protein present in the original
mixture?
Dr Cindy St. Hilaire: That's so interesting. I know it's been
around for a while. I'm always impressed by it.
Dr Tomáš Vaisar: And then the other approach we use it's called
Proteomap developed by Ben Cravatt, collaborator on this paper at
La Jolla. And that approach uses basically the shotgun approach but
as a first step uses gel electrophoresis, SDS PAGE electrophoresis
to fractionate the very complex mixture to size segments.
Dr Tomáš Vaisar: So you run your complex mixture on a gel, you
slice it by size and then run shotgun proteomics experiment on each
of those slices after gel digestion. And then Ben developed a set
of tools where you identify the proteins and their abundance in
each of the bands based on the size and the way it's applied to
formation or mapping of proteolytic events is based on the idea of
that intact protein will show up at the molecular weight of the
intact protein but if it was cleaved by a protease, it will also
show up at the molecular weight, which is smaller corresponding to
the fragments formed by proteolysis. Then you use set of
bioinformatics tools to piece this all together and generate the
Proteomaps.
Dr Cindy St. Hilaire: You pick it apart, throw it in the machine
and then put it back together. That's so cool. It's so amazing.
So what were the main findings that you were able to pull out of
your comparisons? And I think you had three main groups, if I
understood it right? There's the transgenic mouse that has the
plaques that don't rupture, and then there's the atherosclerotic
mouse that had the transgenic bone marrow, and then you had the
human. Can you tell us a little bit about the different groups you
compared and then what ultimately you found?
Dr David Dichek: Sure. Yes. You are absolutely right. We had what
we refer to as the straight transgenic mice that are either
transgenic for macrophage overexpressed uPA or not. We also had
older mice who had advanced atherosclerosis and receive bone marrow
transplants from mice that either had uPA overexpression in bone
marrow or not and then we had the human plaque. So those were the
three groups.
So what we found was that looking at the proteome of those three
groups, we were able to find some common biological processes, and
this was really Sina's work, but taking the proteomics data and
analyzing it with sophisticated bioinformatics tools. We looked not
only at the overlap in specific proteins among the models, but the
overlap in biological processes, because it may be in different
species that there are different proteins, different actors
carrying out the same roles. And that's been described in other
systems as well.
So we were able to identify not only common biological processes,
but surprisingly, we were able to identify decreases in specific
category of proteins, basement membrane proteins that were common
to two of the models, the straight transgenic and the human model
and loss of these proteins certainly has a plausible role in
precipitating plaque rupture. So I think one aspect of the analysis
that's worthy of note is that we initially thought we would observe
more profound changes in the bone marrow transplant mice because
they had more advanced atherosclerosis. And in fact, we found fewer
changes than in the straight transgenic mice, but thinking about it
after letting the data talk to us, rather than trying to impose our
own on the data-
Dr Cindy St. Hilaire: Always a good idea.
Dr David Dichek: ... was that the straight transgenic mice were
telling us we've been overexpressing urokinase for 20 weeks since
we were conceived, and the bone marrow transplant mice had it for
only eight weeks. And indeed they had far less loss of basement
membrane proteins and far fewer changes in their plaques than the
mice that had expressed it for a longer time. And so when one
placed the three groups in a chronology of exposure to protease
activity with the bone marrow transplant mice, being the shortest
exposure than the straight transgenic mice, and then the humans
who've had decades of exposure, it really tells you a nice
chronological story about the biological processes leading to
plaque rupture. And I think that's a generally applicable lesson
and can be applied to other problems in cardiovascular biology. And
that is when you have a biological process for which you can't get
human tissue until after it's occurred because you can't go in and
biopsy.
Dr Cindy St. Hilaire: I have that problem with valve calcifications
that you can't take your valve out early.
Dr David Dichek: If you can get a mouse model that duplicates the
pathology, then you have access to the steps leading up to the
event. And that's what we tried to construct in this study. And
really it was really Sina’s analyses that allowed us to make those
connections.
Dr Sina Gharib: Yeah. Of course, David was kind of the mastermind
behind the design of the experiments on the developing of genetic
models and Tomáš is a renowned expert in proteomics analysis. And I
kind of joined more on the bioinformatics component of this study,
tried to put some of the large data that was being generated
together. And as David and Tomáš mentioned, of course
atherosclerosis is a very complex disease with many, many
components. And then of course the mouse model doesn't quite
capture all the different pathophysiological events that happen. So
one of the aims of this study was to try to integrate and merge the
findings from these model without coming a priori with a bias or a
pathway or a candidate gene, we decided to do a relatively unbiased
shotgun proteomics approach, we actually do for everything.
So the challenge then was how to put it all together. And as Tomáš
mentioned, there are statistical tools to try to identify a
relative abundance of proteins. But, a few things that pure
biologists often don't have to encounter is, you're not one or two
different proteins, you're looking at thousands of proteins. So
there are issues, statistical issues, such as multiple comparisons.
If you looked for changes, you're going to find changes just by
random chance. So a lot of statistical adjustments had to be made
to ensure that those were adjusted for. This are also
pathways and processes that were coming out of these results.
There's many different pathways that were interrogated. And again,
statistically, you want to adjust for the fact that many of those
could have been there by random chance.
So there's a fair amount of statistical methods that need to be
applied for this data. We also did somewhat more sophisticated
pathway analysis where we develop networks based on the
differential expressed proteins between the ruptured and unruptured
plaques to identify connection among these proteins and identify
hubs which are highly connected nodes that could potentially drive
the biology of a network. So other types of kind of deeper
statistical analysis was done, which are maybe more
hypothesis-generating because we actually did not follow up on some
of these candidates, but I think they really do provide a map or
framework to then pursue more mechanistic experiments to see what
happens if we knocked out this highly connected node at the plaque
rupture site to see if we can either stabilize or manipulate the
biology as plaque rupture.
Dr Cindy St. Hilaire: Yeah. I mean, that's really the strength of
these unbiased approaches is you can come up with so many more
novel targets and pathways that might be contributing. So they're
just really great. So one thing I found really interesting, you
mentioned that you saw a clear distinction in the proteome and I
think it was specifically talking about the human samples because
they were large enough to see ruptured area versus non ruptured
area, but you really saw a distinct difference in the proteomes of
the ruptured area of the plaque versus the non ruptured area of the
plaque. And obviously the models you were using are overexpressing
a protease. So of course there's a role for proteolysis in this
process, which you've now firmly established, but I'm wondering if
there are other processes that might also erode the basement
membrane. And did you pull up anything that might suggest of other
things that are happening or even are there other hypothesis out
there that we could test with an approach like yours?
Dr David Dichek: I think the pathways that came up have probably
all been implicated previously. We have processes like inflammation
and complement activation, immune response, thrombosis. That's a
post-hoc event. I think what was most unexpected was the decrease
in the abundance to basement membrane proteins rather than
collagens. So collagen has become the sort of signature protein of
stable and unstable plaques and used as a surrogate, people do
Picrosirius red staining. It's easy to detect with a histochemical
stain, you don't even need an antibody. And surprisingly, we found
very few differences in collagens and actually no differences in
the type 1 or type 3 collagen, which are thought to be the primary
stabilizers of the plaque cap. They weren't significantly different
in between ruptured and stable areas of the same plaque. So that
was certainly a big surprise.
Dr Cindy St. Hilaire: Yeah. Because that would indicate that it's
not necessary... We always say thinning of the cap, which obviously
we know that there's remodeling, it can get thinner. But you kind
of found that the contents were the same, but it's the basement
really that was eroded.
Dr David Dichek: Yeah. The basement membrane proteins were lost. It
used to be said in physiology that if you discovered something new,
you should just go to the German literature and go back 30 years
and it had already been described. And so, looking back in the
literature, there are actually the work of Jean-Baptiste Michelle
in France and a scientist in Finland, Petri Kovanen, have actually
focused on the potential role of basement membrane in unstable
atherosclerosis many years ago, but it was kind of buried in the
collagen hypothesis. And I think it needs resurrection.
Dr Cindy St. Hilaire: Well, I think this paper has done that so
well done there. That's great.
Dr Tomáš Vaisar: Would be worthwhile to know that of course, the
way you prepare the samples may affect what exactly you're seeing.
But we've done very careful characterization of the sample
preparation of the extraction procedure to focus, to enrich the
exosomal matrix proteins because of this collagen hypothesis. And
even with that, we basically saw no difference.
Dr David Dichek: Yeah. I think that's an excellent point. If we
hadn't found collagen in our extracts, we would not be able to
conclude a lot about it. And how you do the extraction, how you
process the samples here can really influence what you find. We
call it unbiased, but there are technical biases that enter,
especially in sample preparation, but our extraction process really
was able to extract collagens as well as elastin, which is really
infamous for being a-
Dr Cindy St. Hilaire: Difficult.
Dr David Dichek: ... I really think we were getting a good sampling
of the matrix here.
Dr Cindy St. Hilaire: I don’t know iif there is an answer for this
question, but it's something I'm always thinking about. We always
talk about athero being so prevalent because there's no kind of
evolutionarily the way to tamp it down, it happens later in life.
But can you think of any advantage that the vasculature would have
in eroding the basement membrane or altering proteases in a
response? I was just trying to think, is this just harnessing a
wound healing process that's gone awry or could this ever be
protective at all in any way?
Dr David Dichek: Well, I think you hit the nail on the head, at
least according to my bias. It's a healing response gone awry and
that you can really draw out the pathways, basement membrane
digestion release of chemotactic peptides as part of the
inflammatory response, attraction of more inflammatory cells and
then a potential healing response that unfortunately results in
digestion of the matrix, which has a morbid or fatal consequence
rather than physiologic remodeling. And you're right, that's not
selected against. It's selected for, in settings in earlier life
infections, for example, perhaps neoplasia, but it's not selected
against in late life because people are post reproductive.
Dr Cindy St. Hilaire: So what's next for these studies? What
questions are you going to attack next with either these models or
with some of your proteomic findings?
Dr David Dichek: Well, we were just talking about that recently,
Tomáš and I, and I think we'd like to look at... For one study,
we're interested in doing, plaques that are high risk based on MRI
imaging, which is really very well developed here at the University
of Washington. And many of those patients have endarterectomies and
they don't have ruptured plaques. So they are in a high risk group.
So they undergo a endarterectomy for that. Not because they've had
a plaque rupture and those plaques might be particularly
instructive because they're pre event and won't have the healing
response to thrombotic response. And it would be really interesting
to see if our studies were confirmed. So that's one direction we're
going in.
Dr Cindy St. Hilaire: That would be amazing. Luckily, you have
access to a whole bunch of human tissue for those kinds of really
high impact studies.
Dr Sina Gharib: I just wanted to point out that one of the
advantages of doing proteomics and being part of the scientific
community is that we made all this data available in the manuscript
for other researchers to access and confirm. So, really probably
the best way to procced with this is to have other investigators
replicate our findings and expand on it. So I just want to bring
that up because all of that data that was generated has been
included within the supplements of this manuscripts and it's
accessible to any scientist who wants to pursue further.
Dr David Dichek: Yeah, I would add one other direction we'd like to
go is we still like to know what the substrates are. We think their
disappearance based on their abundance is due to proteolysis. But
boy, would it be exciting if we could detect fragments. We were
unable to do that in the study, probably because they were lost
either in vivo or in processing. Technical advancements in that
area, and maybe Tomáš can speak to that, might enable us to
actually find more direct evidence of proteolysis.
Dr Tomáš Vaisar: Yeah, I mean, to start with, it's really hard to
determine physiological substrates of proteases. There's a huge
amount of literature identifying proteolytic substrates in vitro,
but the physiological substrates are really extremely hard to
determine, and especially physiologic in vivo confirming that
because in vitro, in a tube, you can mix whatever you want and you
modify the ratio of proteins to protease substrate, and you can
cleave almost everything with anything. It's a little exaggeration,
but it's close. While the physiology substrates in the really
complex milia of tissues is extremely hard. And so there has been
several approaches developed and one of them is the Proteomaps. The
other one is an approach called TAILS developed by Chris Overall at
UBC that uses the idea of formation of the neo termini and then
tagging the neo termini. So that in the actual sample, you can
specifically detect these neo termini formed. But even with that
approach, it's really hard to determine what are actual
physiological substrates. And on top of that, what are the cleavage
sites of the proteases?
Dr Cindy St. Hilaire: And I guess the third being, if those
substrates are cleaved, are they circulating and can we detect them
in a blood sample? That would be, I guess, the gold standard. Well,
thank you all so much for joining me today. Congratulations on this
really very cool study. Being into human and translational work, I
always love mouse studies that bring in lots of human samples. So
congratulations on that. And I look forward to your future
publications on this.
Dr Tomáš Vaisar: Thanks a lot.
Dr David Dichek: Thanks.
Dr Cindy St. Hilaire: That's it for the highlights from the late
September and early October issues of Circulation Research. Thank
you so much for listening. Please check out the CircRes Facebook
page and follow us on Twitter and Instagram with the handle
@CircRes and #DiscoverCircRes. Thank you to our guests, Drs David
Dichek, Sina Gharib, and Tomáš Vaisar. This podcast is produced by
Rebecca McTavish and Ishara Ratnayake, edited by Melissa Stoner and
supported by the editorial team of Circulation Research. Some of
the copy texts for the highlighted articles is provided by Ruth
Williams. I'm your host, Dr Cindy St. Hilaire. And this is Discover
CircRes, your on-the-go source for the most up-to-date and exciting
discoveries in basic cardiovascular research.