The Role of Cell Geometry and Adhesion Ligand Presentation in Guiding Stem Cell Differentiation
1 Video View·Sep 5, 2023
Okay, we're going to get started then, thanks everybody for coming, today we're really happy
to welcome the Department of Materials Science and Engineering, he's also an affiliate member
of our department, and as you'll see he does things very similar to many of the things
that we do, so Chris began life as a chemist, we won't hold that against him, and actually
worked as a chemist in industry for several years, actually over the time he was getting
his master's degree as well, so he had a lesson, right, he did many things simultaneously,
then went off to Australia to get a PhD, returned to the University of Chicago for a postdoc,
along the way somewhere he picked up all this cool stuff about how materials interact with
cells, which is what he's going to tell us, Chris, thanks for coming. Thank you Michael,
it's my pleasure to be here, thanks for the invitation, and so I've spent most of my time
since I started my faculty position around material scientists, so it's great to be able
to come and talk to the bioengineers, because my group does a lot of fundamental biology,
and as I'll try to, can everybody hear me, should I scream louder, testing the acoustics of the
room, as I'll try to show you throughout the talk, you know a lot of what my lab does is it
looks at biological systems, and is motivated by some hypotheses based on biology, to then develop
materials and use some of these materials-based platforms to answer some very fundamental
biological questions, and as Michael mentioned, I'm a professor in material science, but I'm also
a affiliate here of bioengineering, so I like being involved with bioengineering, I also have
an affiliation in the micro nanotechnology lab, where I've got a stem cell lab, where we do a lot
of our culture, and we use a lot of the great fabrication infrastructure there to do a lot of
patterning, and things of that nature, and I'm also involved with the IGB in their regenerative
biology and tissue engineering theme, and so you know what I think we bring to the table is we are
really interested in some of these, how we can really breathe life back into some two-dimensional
assays, where there's very talented individuals on campus, and all throughout the world, that
build very beautiful complex three-dimensional scaffolds, and so we're still at
the 2D level, but hopefully some of the insights that we gain, we're hoping to translate into
three-dimensional materials for next generation type applications. Okay, so here's just
one of these montage slides of overview of my lab. We do a lot of micro patterning, and the
reason we like micro patterning is there's a lot of interesting things you can do when you capture
a single cell, or two cells, or multiple cells in various architectures to study how they signal
across junctions, how they communicate with their extracellular matrix that they're sitting on top
of, and how we can control that in a rational way, such that we can study the role that things like
geometric cues and actomyosin contractility play in guiding ultimately gene expression,
which will tell a cell what to do, whether it's going to hang out, or in the case of stem cells,
turn into something that you're interested in. We're also interested in nanostructured materials.
Biology has a lot of nanostructure in it, so if you look at a lot of the membranes in biological
systems, and a lot of the hard materials that are already there in biology, often the nanoscale
presentation of things like the mechanical environment, or the presentation of ligands, is
crucial for conveying signals to cells to tell them to either migrate, to stay put, to die, to
differentiate, all across the board. We're also interested in molecular engineering. As Michael
suggested, I am a chemist, and it's true, from my early days, and there's a lot of things we can do
with surface chemistry, utilizing platforms where we can vary the presentation of cell adhesion ligands
and growth factor ligands, to see how these can signal to cells to get them to
perform certain functions, to understand how they work. We're also interested in hydrogel
materials. This is where we start getting closer to 3D, and the reason we're interested
in hydrogel, just like so many people in this department and others are, is biological
materials are comprised largely of polymeric architectures with biopolymers, and so understanding
things like the elasticity of the microenvironment, or the viscosity of these things, is
important. Combining some of these platforms together, I feel, is really exciting, because we
may be able to develop models that more closely recapitulate the way that cells and tissues form
during development, and this will help us understand developmental processes, as well as ways that we
can design new materials for applications like tissue engineering. All right, so here's an
overview of what I'm going to show you today. I'm going to talk about some of this work, because I
think it builds a nice cohesive picture of how we can use these material science-based platforms to
study cells and cell signaling, and so we're going to explore the cues in the microenvironment.
I'm going to tell you a little bit about geometry, how the geometry of single cells can influence
the cytoskeleton, which can then influence gene expression. I'm going to tell you a new story
that's emerged that I think is really exciting, and that is when you study these individual cues,
we're often missing the big picture, which is that in vivo, you know, you're not just going
to have a cell responding to something like stiffness, or a cell responding to something
like nanostructure. There's going to be an integration of these cues, and there's going
to be a synergy and complementarity that arises, and we're really excited about that. Some of our
recent work has shown that biological signaling, so paracrine and autocrine events, can play into
some of these pathways. And then finally, I'll talk a bit about some chemistry and how we can
modulate stem cell differentiation using the presentate, by modulating the presentation of
adhesion ligands on a surface. And then I'll talk about some of our current efforts in trying
to combine these things, with this idea that, you know, if we can incorporate multiple cues
together, we can maximize the outcome that we want, and to more clearly understand how cells
do this naturally in vivo, as well as design new materials. And then I'll tell you, you know,
a bit of a different story, but using the same techniques, is something that's recently come
out where we found that patterning may be able to be used to prolong multipotency in stem cells
and culture. I'll show you a bit more about that. Okay, so today I'm primarily going to talk about
mosaical stem cells. I imagine everybody in this room has heard of these cells. I'll just, so I'm
not going to belabor it, but these are an exciting cell for a number of reasons, primarily because,
you know, they're a very good model system for studying stem cell biology experiments, whether
that's mechanobiology or differentiation type studies, because they're relatively easy to work
with. You can get them from commercial vendors these days. And, but in addition, they're also
still a pretty exciting candidate for regenerative therapies, because of this reason. They can be
isolated from bone marrow, as well as fat aspirates. A lot of people are using these adipose
derived stem cells, and in fact, the literature is suggesting now that these cells may be a pericyte,
and reside on your vessel walls, and could be present in every tissue throughout your body.
And so, this is exciting for obvious reasons, because these cells, or closely related cells,
that we can identify with our marker schemes, may be in every single tissue, and they may have
slight differences, but they all follow a similar set of rules to get them to regenerate tissues
using standard homeostasis processes. One thing that has emerged from the use of these, particularly
in clinical settings, there are a number of clinical trials underway for using these, primarily for
a trophic response, where these cells are shown to secrete molecules that may promote angiogenesis,
for instance. But we've been investigating these, but really with the state that things are at right
now, we have a difficult time controlling the way they specify, and what they turn into when
they differentiate. And so, really understanding the microenvironment better, so that we can design
better materials for therapy, and things like this is important. All right, so I like to make
things simple, because these systems are incredibly complex, as I think everyone understands who's
studied biology. So, when we talk about signaling in the microenvironment, there's a number of
signals that we can consider. Physical signals is one that is important, and as a material
scientist, we're interested in exploring some of the physical material properties and how they
influence cells. So, these are things like the shape of cells, the shape of tissues, how those
arise, and how you may be able to manipulate cells that way. The mechanical properties, so
stiffness or softness of the matrix. Stretch, you know, we're all moving all the time, and our matrices
are stretching, and strain is being put on various tissues, and this clearly has an influence on
signaling. Chemical cues is my second area that I break things down into, and these include things
like small molecules, but more importantly for the purpose of this talk, is the precise chemical
moieties that you see in the extracellular matrix. So, you know, when you have a cell sitting on top
of a bed of fibronectin protein, there's a lot going on. It's not just that they're interacting
with the protein, they're recognizing very distinct moieties in the ECM. So, things like our friend
RGD peptide that everybody in biomaterials is aware of, but there's other synergistic ligands,
and other ones that haven't really been fully explored that have a function, and so we're
also interested in exploring what the function of these precise ligands are, and I think there's
a lot of room for discovery in this, because there's so many integrin ligands on the surface
of the cell, and so many different ECM proteins, and then the one the area that's probably been
studied the most is biological signaling, as I call it, and so I put into this little bucket
things like paracrine and autocrine signaling, growth factors, hormones, etc., as well as signaling
across junctions, so it could hear in junctions and gap junctions, and how this might influence
cell fate, and really the theme I think that's coming out in the field, and you hear more and
more about this integration of these cues, and trying to understand how they play into
one another, I think is really important, because clearly things like shape and stiffness is going
to influence the presentation of the adhesion ligands that a cell responds to, and evidence,
some of which I'll show you today, suggests that these things can lead to differential signaling of
biological factors. Okay, so we'll talk about physical cues first, and so there's a lot of
things that have been done, and a lot of studies that I think are important for exploring this
space. Nanostructured surfaces, clearly I mentioned that a lot of biology is nanostructured, but a lot
of these techniques we may be able to use to make new tissue engineering scaffolds, for instance,
that have this component. In this example here, these are titanium oxide nanotubes that are grown
in different sizes, and I know it's probably tough to see back there, but essentially when they put
mesenchymal stem cells on these surfaces, they found that cells adherent to these small nanotubes
were able to spread out, get a high degree of sub-skeletal tension, and a lot of robust focal
adhesion formation throughout the cell, compared to cells that are adherent on these large nanotubes,
where they remain rather spindly, and they don't adopt a stable cytoskeletal configuration.
And then after several weeks, when they went in and stained for markers associated with bone cell
differentiation, they found that cells that were spread out at this healthy, well-developed
cytoskeleton expressed higher levels of differentiation markers for bone. Similarly, you
can do similar studies with strain, and so here's showing some mesenchymal stem cells that are being
stretched on a polymeric substrate, and by doing this you can see a differential regulation of
genes associated with different differentiation outcomes. So here we have alkaline phosphatase
activity in cells that were stretched for a period of time, have higher levels of bone cell
differentiation, and that's true for several other markers, while they see a decrease in
genes associated with fat. And so what a lot of these studies are pointing to is that cytoskeletal
tension appears to be something that's important for bone cell differentiation. A real seminal work
came out of Dennis Disher's lab at Penn with Adam Engler as the first author, who's now at
University of California, San Diego, and he's visited us last year, I think. Adam and Dennis
published this paper in 2006 that I think really changed the way a lot of the material
scientists in bioengineering thought about these things. What they did is they did a simple
experiment by varying the stiffness of polyacrylamide gels, so standard gels used for
running protein gels, from very soft gels that approximated the stiffness of brain tissue,
so less than one kilopascal, and then they increased the cross-linking density of those gels
to make them more rigid, to approximate that of muscle tissue and bone tissue, and they made a
really startling observation, which I think makes a lot of sense, in that as you change
the mechanical properties of the matrix of which these stem cells are sitting, you change the way
that they differentiate. So here you see cells on the soft matrix are expressing markers associated
with neurons. They're developing neuronal-like phenotypes, too, by extending out processes.
On the intermediate stiffness gels, you see muscle markers, and finally on the stiff gels,
where the cells are very spread, they have high cytoskeletal tension again,
they start expressing bone markers. So I think it's a really exciting demonstration of the
importance of biomaterials in stem cell fate decisions. Another important story that will
play into some of the work I'm going to show you today is this work that came out of Chris Chen's
lab, also at Penn, where they explained something that had been observed with some of the early
reports of mesenchymal stem cells. Basically, when they first started calling these mesenchymal stem
cells, they showed that if you seed them at low density on tissue culture plastic, and then you
seed them at high density on tissue culture plastic, the cells that were initially seeded at
low density after several weeks showed a much higher degree of bone cell differentiation.
And no one was really clear why that was. I think a lot of the initial arguments was that clearly
there's a paracrine signaling mechanism, where biological signals are diffusing across the dish.
But what Chris Chen and now others have done is they use micropatterning to culture cells in
either small islands, where they remained as big balls just sitting on the surface with very low
cytoskeletal tension, or they put them on these large micro islands that they patterned on glass
such that they can spread out and get a high degree of cytoskeletal tension. And then they
mixed two medias together, one that promotes fat and one that promotes bone. And after about a week
in culture, they stained them for lipid droplets, which are shown in red, and they show that cells
in these small islands almost exclusively turn into fat, while cells in these large islands that
are well spread almost exclusively turn into bone. And so clearly the spreading of these cells is
important for differentiation, and it makes sense in terms of other observations that have been made
with other material systems that show that, you know, when a cell is very well spread on a
nanostructured surface or on a stiff matrix, and that they prefer to differentiate to bone than if
they're balled up or rather spindly. They went and did some very nice elegant work to show that
this involved some signaling and pathways that utilize a feedback loop between
this RhoA, small GTPase, as well as RhoKinase, and that in the absence of high levels of RhoA,
cells prefer to go turn into fat. Okay, and so I think another exciting thing about patterning
that we can do outside of the spreading thing is we can explore other geometric cues that can
influence these cells. And this is a picture that I really like that came out of a review from Manuel
Thierry in 2010, and he showed, he basically explained how by micro patterning cells you can
recapitulate some of the aspects of the native microenvironment, whether that's geometry,
architecture, you know, so connection with other cells, composition, mechanics, dynamics, etc.
And so this is a immunofluorescence image of some mesenchymal stem cells on plastic,
just about everybody uses plastic, including us, for culturing cells, but in no way reflects what
a cell sees in the body, right? I mean, you have this rigid gigapascal synthetic material and this
sea of nutrients sitting above the cells. And so with micro patterning, we can at least start
breaking apart some of these things. And so here's a picture of cells that have been micro patterned
in our lab in a range of different geometries. And the reason we do this is now instead of just
changing area, we manipulate some of the subcellular curvatures and aspect ratios and
things like that. And we can get very different organization of the filaments within the cells.
So this is a flower shape, and you see here we got the nuclei in the center, and we get this nice
ring-like filamentous actin network, and it's just this beautiful web formation of microtubules
throughout the cell. And by modulating that architecture, changing it up, using different
shapes, you can influence the way that these things align and use this as a method to study
the importance of geometry and how it influences the cytoskeletal architecture. And so the platform
we use to do this is microcontact printing on thin layers of gold. We like gold because it's
such a well-developed chemistry for self-assembled monolayers. And so in essence, we make a PDMS stamp,
so just a regular silicone rubber stamp. We can ink that with a chemistry of choice. In this case,
we use this octadecane thiol molecule, which can chemisorb to the gold surface and form a very
stable gold-sulfur bond. And then by patterning that, we can then backfill the remaining regions
outside with a non-adherent oligoethylene glycol moiety, which will prevent protein adsorption
and cell adhesion. We can then physisorb to the surface of these hydrophobic regions some protein
of interest. We usually use fibronectin, but we've also looked at laminins, collagens.
And then because we have these ethylene glycol regions, we can capture cells specifically on the
islands that we stamped. And so we can do this across a range of geometries. This is some work
that I helped out with during my postdoc, where we looked at motility features and contractility
features using just a wide array of different geometries to see how you could influence,
and these are with cancer cells, the migratory features of these adherent cancer cells by
changing some of the geometric cues. So the question we're asking is, can we utilize some
of these to look at the influence that some of these more subtle cues have on differentiation?
And so the first thing that we looked at was aspect ratio. And so using a similar protocol
to what Chris Chen had devised in their 2004 work, we just took two bottles of media and mixed them
one-to-one. One would promote fat, one would promote bone. And when you do that on plastic,
you get this crazy heterogeneous mixture of cells. You have little pockets where you've got a lot of
fat, a lot of bone in some regions, and it's quite variable. When you pattern the cells,
you can just go across the surface and count cells that either express bone markers or fat
markers or both. And there tend to be a small amount that presents both, usually less than
five percent. And so when we did this with rectangles of the same area, but with increasing
aspect ratio from one-to-one, three-to-two, and four-to-one, and we counted the cells, we see an
interesting trend that emerged, where that as we increase aspect ratio, we increase the proportion
of cells in the culture that are turning into bone cells. And so we see in the square, we see about
equal numbers of fat and bone cells, but as we go to this longer rectangle, we see more bone.
And so we went to explore why this was, and one of the things that really stood out was,
as you increase aspect ratio, while keeping area constant, so there's no difference in spreading
here, we see an increase in cytoskeletal tension, where now we get very large stress fibers along
the long edge of the cell with more focal adhesions on the short end. And so similar to the spreading,
we have an increase in actomyosin contractility in these geometries, which seems to be a cue to
promote bone cell fate. Similarly, we did the same thing with a five-fold symmetric geometry,
where we have a flower-shaped cell, and what we have here is these convex regions of curvature on
the perimeter, and then a simple pentagon, and then this star-shaped geometry. And so looking
under a microscope, at first they all look the same, and so this was really startling when we
first made this observation. And so when you go, when you expose them to the differentiation
conditions, we find that going from this flower to a star shape, you see almost a complete
lineage shift, where now on the star we see most cells are differentiating to bone cells. And
consistent with the spreading and with the aspect ratio, when we do immunostaining for actin and
vinculin for focal adhesions, we see that cells in the star geometry have a much higher degree of
these bundle-like stress fibers between the points of the star. And the reason is that there's these
non-adhesive spaces between these points where the cell has to overlap, and this is a cue to
generate a lot of stress in the cells, and so they develop these strong stress fibers that are then
anchored by very large focal adhesions, sometimes on the order of tens of microns, right on the
tips of the cell. One thing that's really terrific about doing these pattern cultures is when you
have unpatterned cells, they're all over the place. Every cell has a different shape. It's
really tough to make any morphological conclusions about them. But when you do patterning, you're
constraining every cell in the same geometry, so you can do cool things like make heat maps.
And so what these are is we took 80 and 86 images of single cells that were stained for myosin 2a
and then overlaid them in ImageJ, a program you get from the NIH, and then you can generate an
average immunofluorescence heat map, which can give you somewhat of a semi-quantitative indication
of a feature. And in this case, it's contractility, where we're looking at myosin 2, and you see that
cells in this flower geometry have a diffuse myosin staining all across the cell, while in the star
geometry, you see a very high localization of myosin between the points of the star,
which is consistent with this idea that it's in a more of a contractile phenotype.
So we thought this was pretty cool, so we went to try to figure out what was causing some of this,
and whether we could attribute this directly to cytoskeletal tension, so we used some
pharmacological inhibition studies. And so here's our control condition, where you get fat on the
flowers and bone on the stars. When you treat these cultures with nicotinazole, which inhibits
microtubules and can lead to a high degree of cytoskeletal tension in the cells, but using a
concentration where the cells remain viable and they stick to the shapes, we see that we get
abrogation of this trend, where now all of the cells are turning into bone. And so we've lost
the shape dependence by basically using a sledgehammer to knock the contractility up in
these cells. And in contrast, you can do it with other molecules, so we inhibited myosin, a light
chain with blevistatin, and there we saw the same trend, but this time we see all of the cells turn
into fat. And by blocking rilkinase, which is an important player in the contractility of cells,
we also saw abrogation of this trend in all fat. So this is consistent with our story,
that these subtle geometric cues can influence actomyosin contractility, and by knocking it out,
we can get rid of the shape-directed effect. Okay, one of the things that came out that was
somewhat surprising from this is, and this happens when you run microarray experiments,
anyone who's done them will know this, you just get a big old barrel full of data,
and you have to spend a lot of time digging through it. And so, you know, I spent several
months staring at microarray data, and there are a lot of pathways that seem to be differentially
modulated by shape. But one thing that really stood out, that I thought was interesting,
is cells in the star geometry had a high expression of these wingless-type soluble
proteins that are secreted, and they're involved in all sorts of stages of development, as well as
cancer and bone cell formation. And we also saw increased expression of a lot of the mitogen
activated protein kinase pathways that are involved in mechanotransduction, like the
extracellular-related kinases, and C-June N-terminal kinases. So, we were curious about
what role they played in this shape-directed effect, and at the time, you know, it wasn't
so clear how these mechanochemical pathways influenced the lineage programs. So, we wanted
to try to figure out if we can get a direct link to gene expression. And so, we inhibited Wnts
extracellularly using just small protein molecules that inhibit Wnt signaling. So,
DKK1, which inhibits the canonical Wnts at the membrane. Now, when we add DKK1, we see a slight
increase in adipogenesis and decrease in osteogenesis. Not very significant, though.
When we add secreted frizzle-related protein, which will block all Wnts in solution, as well
as at the membrane, we see a bigger increase in adipogenesis. And then, when we mix those two,
we see pretty much, we lose all of our shape-directed differentiation, suggesting that
with this contractility, the Wnts are really playing a serious role. And so, you know, the
picture that I paint with this is that when you have a contractile cell and a cell that is not as
contractile, the contractile cell up-regulates a lot of the players that are involved in Wnt
signaling. So, the small GTPases, the Rho kinase, and some of these pathways, because there's a
convergence between many of these signal transduction pathways. So, Wnt signaling to
the membrane, where it reacts with these frizzle receptors, will lead to an increase and an
activation of Rho kinase and some of the MAP kinases. And so, we feel these pathways are
working together to enhance this shape-directed differentiation. Okay? So, you know, this,
we thought this was really interesting. And so, how can we utilize micropatterning to
explore this further? And so, we went back to this argument that the spreading of the cell
influences osteogenesis. And I think a lot of biologists, and many I'd spoken with felt this
way, still believe that paracrine signaling is very important and is likely to play a significant
role in these different density cultures. And so, to remind you, at low-density seeding,
you get a high degree of osteogenesis, shown here with the blue. Well, and when you initially
seed at high-density, you get a low degree of osteogenesis. And Chris Chen and others suggested
this was because of this mechanical aspect of spreading. So, we wanted to look at paracrine
signaling. And so, the way we devised to do this, very simply, is by patterning cells at a
different distance. So, you can reduce the diffusional path length between the single
cells. And so, going from about 100 microns separating out every cell, to 200, to 400,
to 800. So, they're nearly a millimeter apart in your dish. And so, we did the analysis that way,
and we dumped in the chemicals as before. And something really startling came out of this,
where at very high density, we see most of the cells are turning into fat. But as we increase
the spacing between them, we see an enrichment in cells that are differentiating to bone.
And so, this, you know, my jaw absolutely dropped on this, because it was unexpected, where we see
at high density, a lot of fat. But as you decrease the density, you get almost a complete switch
here, where now, most of your cells that are very far spaced apart are expressing bone cell markers.
So, clearly, something's going on here. We're still struggling with what it is, but I'll tell
you about some of the, some of the candidates. And so, the first obvious thing to do, change the
media. If there's some secreted factor that's being shot out from cells to their neighbors,
if you take that media out, you'll influence whether or not the cells are seeing that. So,
we changed the media every 12 hours for a week. It was an interesting week. And what we see is,
hopefully, expect to see, where here's our control condition, where we see a decrease in fat
as the spacing increases, and an increase in bone. When you change the media every 12 hours,
you now see more bone on every single shape. The problem with this experiment, of course,
is that you're adding new osteogenic components in there, and there's a lot of other variables
that could play a role. So, it doesn't demonstrate completely solving this issue.
So, we played around some more. The next thing, of course, is take media from these high-density
cultures. So, we cultured cells at high density and took that conditioned media and fed it back.
And we see a slight change in differentiation. So, here's adding 1X media. So, 1X is the 100
microns apart cells to 2X, 4X, and 8X space cells. And we see a slight increase in fat,
a decrease in bone, although it's not statistically significant, but maybe suggestive.
And then, we further took that media and ran some gels to see if we could isolate species,
and actually spent quite a bit of time trying to fractionate the media to determine what could be
causing this density effect, which isn't much fun. It takes a lot of time, actually. It's very messy,
particularly when you've got things like bovine serum albumin in your cell culture media,
and you're trying to deplete that out and just mass this whole region of your gel.
But we did see some interesting candidate species over here in the 30 to 40 kilodalton range. And
they're interesting because of our microarray analysis again, where we saw that cells that
were patterned had a very high expression of these secreted frizzle-related proteins,
which are Wnt inhibitors. And so, we saw greater than tenfold expression compared to a non-patterned
surface with cells in these conditions. And so, we did some Western analysis of our gels,
and they're present, and they increase over time. So, these MSCs are secreting this Wnt
inhibitory molecule over time. And so then, what we did is we purchased some antibodies to block
the secreted frizzle-related proteins in these cultures. And what we saw is that here's our
control again, more fat on the 1x that decreases as you increase the spacing. If you block secreted
frizzle-related protein 2 that we saw high expression of, no real change, maybe some slight
change, but nothing significant. Five, nothing significant. But then here, with the secreted
frizzle-related protein 3, when you block those in culture, we see partial abrogation of this
density dependence. So, right now, this is one of our top candidates for a molecule that may
mediate this paracrine signaling effect. Because here, when we block the secreted frizzle-related
protein 3, we see that pretty much all of our cultures, whether they're 1x, 2x, 4x, or 8x,
separated on the surface are turning into bone. So, we thought that's pretty exciting. We also
supplemented that result by adding recombinant SFRP3. And this wasn't too surprising of a result
because we had seen this with our previous work, where when you add this protein, it promotes fat
cell differentiation. And so, in our control condition, again, we see this trend when you add
recombinant SFRP3, you get rid of the trend, although you still see something going on here
when the cells are really far apart. And so, there's likely other factors that play a role
here, but this is absolutely a candidate. And I think it demonstrates how, if you think outside
of the box with some of these conventional micropatterning technologies, you can do some
really cool fundamental science. All right. So, I'm going to tell you now about a chemistry story
about how we can explore how ligands on a surface can influence stem cell fate.
First, I'm going to have a drink of water. And so, the hypothesis here was similar to
the nanostructured surfaces, stiff matrices, geometry. We could utilize ligands that have
an enhanced affinity for integrin receptors to change the cytoskeletal tension of the cell,
and therefore, influence differentiation. And so, to do that, we used our self-assembled
monolayer chemistry, but this time using this melamide functionalized monomer that we mixed
into the SAM. And then, we immobilized a thiolated peptide, one which has the RGD peptide shown here
in red, which is the linear peptide that's utilized in most biomaterials proof of principle studies.
We also used the cyclic analog that had been explored for years as a therapeutic candidate.
And it's interesting because the difference between these is really just the confirmation.
And what this does is it's really a thermodynamics argument. And so, when the cell, when the peptide
is in this bed confirmation, it'll slide right into the integrin pocket, into the binding pocket.
While if it's this linear peptide, the protein has to do the work and fold it over. And so,
because of this, you get higher focal adhesion turnover as the cells are grabbing onto these
cyclic peptides. You tend to get more focal adhesions throughout the cell and greater spreading.
And so, we thought if you get more focal adhesion to get better spreading, maybe we'll get more
bone cell differentiation. And so, we used an array of gold islands for this where each one has
been modified with different peptides so we can control within the same experiment. And this is
the phenotype of these cells, a linear peptide, where you see they're well spread. They've got
lots of focal adhesions inside. But when they're on the cyclic peptide or on fibronectin,
which has a loop domain where the RGD is, so it's already in that bent shape, right?
We see lots of focal adhesions and many more than the linear case and in particular on the
interior of the cell. And they tend to be larger in area, have more cytoskeletal tension, etc.
And so, in our first experiments, thankfully, we decided not to add a bunch of media. We just let
them grow and then we fixed and stained them. And, you know, we got some really interesting
results. So, here we see our control where with these heterogeneous populations of MSCs, anyone
who works with MSCs has seen this. You always get a mixture of cells where some of them express
alkaline phosphatase. And this is likely due to, you know, the heterogeneous nature of the culture,
but also probably because some of these cells are becoming trans and amplifying cells.
They're moving more towards osteoblasts. And so, we always get a mixture of these. But when you
culture them on fibronectin for 10 days, we see that more of them are expressing alkaline
phosphatase. When you compare the two peptide surfaces, both at high density, 1% of the surface
has the peptide or low density, 0.1%. On the cyclic peptides, we see a higher degree of alkaline
phosphatase staining compared to the linear peptide. And that can be shown here where we
see for fibronectin and the high density cyclic, the highest percentage of cells expressing alkaline
phosphatase, which then drops off as you go to lower density and to the linear peptide.
So, this seems to suggest that perhaps the spacing and the affinity of the ligand for
the receptor can influence the fate decisions. So, we did some other staining. This is for RUNX2,
which is a master osteoblast regulator. And we see a peak in the RUNX2 expression. And this
is normalized nuclear to cytoplasmic fluorescence. And we see that for cells on the cyclic peptide
and the fibronectin, we see higher nuclear expression of RUNX2, indicating that you have
more of a proportion of lineage active RUNX2. So, we also isolate RNA from this to verify our
immunofluorescence results. And so, we got RNA and we profiled these and found that we got the same
sort of trend. We're here for alkaline phosphatase. We see robust expression on cells on fibronectin
on cyclic peptide, but then a significant decrease in expression on the linear peptide.
And a similar sort of trend we saw for RUNX2. And one of the things that was interesting with
these experiments, and I started thinking about when I was actually considering some of Adam
Engler and Dennis Disher's work, is the morphological changes that occur on these different materials.
And if you put cells on gels, you'll see, as the cells can exert tractional forces
differentially on these different materials, you can understand and you see a similar thing
here when you change the density of ligand. We're here on the cyclic peptides. Most of the cells
have more of a cuboidal shape, very symmetric. But as you go to lower density cyclic peptide
and the high density of the linear, you start to see elongation where cells are getting more
elongated. They're taking on different shapes. And then finally, at this very low density
peptide surface, many cells start to look a little neuronal. And I'm not going to claim that
these are neurons, but similar to what Adam Engler had shown, these processes being extended was
one of the hallmark characteristics. And so I should point out too, when we quantitated these
shape changes, the low density is really interesting, where we see that after 10 days in culture, we see
a significant decrease in cell area as well as nuclear area. And when I see a change in nuclear
area, that tells me something's going on with gene expression and maybe the chromatin is condensing
and some other program is going on. So we stain these for all sorts of markers. The result,
I'll just give you the punchline, was that on the cyclic peptides, these cells would always
tend to prefer to express markers associated with bone cell differentiation, while on the
high density linear peptide, we started to see expression of markers associated with myogenesis.
So remember on the intermediate stiffness gels that Engler and Disher worked with, they showed
expression of myogenesis. And here at this high density of peptide that has a lower affinity
for the alpha V beta 3 integrin, we started to see expression of muscle markers. And then these
cells that were extending out all these processes on the low density, a good fraction of them were
expressing beta 3 tubulin, which is a marker for neurogenesis. So this is that quantitated,
where here you see it's not an incredible high amount, but it is significant. Here's the RUNX2
data showing that you get preferential bone marker expression in the myoD on the high density
linear peptide and beta 3 tubulin on the low density linear peptide. We did RT-PCR of this
as well, and we see the same sort of trends when we look at the gene expression level.
And so we see higher expression in muscle markers and neuronal markers on these linear peptides.
And so consider this example of stiffness, where people have been talking about for
quite a while now the importance of the material's elasticity and its stiffness in
directing differentiation. So from this work, it suggests that in addition to stiffness, we may be
able to modulate the density and the affinity of ligands on the biomaterial to get similar trends
in lineage specification. And I think this is interesting for a number of reasons. For a
practical reason, for biomaterials design. If you want to promote bone cell fate, have a high
density of cyclic peptides. But also I think it's interesting if you reflect on the nature of the
microenvironment that we're working with. If you're talking about brain tissue, you're going to have
a soft material. The exposure of ligands for adhesion might be very different than if you
have a very rigid calcified matrix, where you're going to have a lot more area for cells to grab
onto, exert forces, and spread out. And so I think in many ways this result reflects that.
In addition, one thing that has become a bit of a controversy just recently that came out in a
paper just several months ago, and I was at a meeting this summer where Adam Engler and some
of the people, the authors of this paper were there, and it got kind of heated. And that's
exciting. It's exciting when science gets turned over and gets argumentative. And I think this has
been in my mind and others for a long time, that when you cross-link polyacrylamide, which is the
gel they used in this study, and you vary cross-linking density, you're not just varying
stiffness, you're varying the porosity. And so this is an SEM image of soft gels and stiff gels
across a range of mechanical properties. And so you get very different pore sizes in these
materials. And if you're cross-linking your protein that the cells are going to grab onto
at the surface, when you look at what the cell's going to see, it's going to see a very different
distribution of protein, depending on what surface it's sitting on, because of this porosity effect.
And so I think this is interesting in light of the result that we found, that by varying the
spacing of these RGD peptides, you can vary the differentiation outcome. And so this may clearly
play a role. And so this is the result that they found. This was a big collaboration that was done
with Fiona Watt at Cambridge, as well as Wilhelm Huck and Joachim Speitz, a number of these big
names. And they did this, they published this paper where using polyacrylamide, and this time
they use fat and bone. So fat is the soft tissue, bone is the stiff tissue. And they found that as
you increase the stiffness, you get more cells, which is what others have found. But in terms
of differentiation, you get a decrease in fat and an increase in bone. So this is expected,
that on soft materials, you'll get more fat cell differentiation. On stiff materials,
you'll get more bone cell differentiation. And so this is consistent with this argument that
there's lineage matching between the differentiation outcome and the mechanics of the material.
But then they went and used PDMS, which doesn't have a porosity component. And they showed that
when they modified PDMS of different mechanical properties, that they saw no difference in the
number of cells, no difference in adipogenesis, and no difference in osteogenesis. So here they
have different mechanical properties of their substrates, but the same ligand density, the same
amount of immobilized protein, and they see no change. So this is pretty controversial. There's
a number of problems with the experiment, I think, in terms of the PDMS material and what they're
actually detecting. But I think it's exciting, right, because it opens up new areas for discussion.
In terms of this question, I'm sitting more here, where I think that clearly these things
play different roles. So this was a paper done a couple years ago with Fu and Chris Chen and
Pennegan, where they use these elastomeric micropost arrays that they have. And what they did, a really
cool trick, where they just changed the height of these posts. And so the mechanical properties of
the material is the same, but they're just changing the height such that the stiffness that the cell
sees is different across these. And so what they showed is here cells on very short posts spread
out, and they tend to differentiate in the bone. Cells on these long posts that are very flexible
tend to ball up, and they turn into fat. But here, arguably, the adhesion ligand presentation is
going to be the same across these surfaces. And so I think this is still open debate, but I think
it's definitely an interesting area. Okay, how are we doing on time? Okay, still got a few minutes.
All right, so now I'm just going to show you some of the new things we're doing with combining
these cues. And so as I mentioned at the start of my talk, I think the next generation
of 2D assays is going to be involved, now that we've got lots of control over microstructure
interfaces, surface chemistry, is looking at how these things play together.
And shape and stiffness is one that I've been interested in recently. And so this is the
idea. We can now vary things like shape and stiffness. We can also look at variations in
ECM chemistry and stiffness and shape. And so my lab's actively looking at ways that we can
develop platforms to vary these things rationally and controllably. And shape and stiffness, I got
excited about this. Here's the work by Engler and Disher again, where we see very different
shapes across these different stiffness gels. Here's a neuron, myoblasts, and osteoblasts.
But when I did these experiments, I saw a mixture of fates. And so I just showed you some work where
they showed on soft gels you get fat cell differentiation. When we've cultured MSCs
on these very soft gels, we see that some cells express beta 3 tubulin, a low percentage of them,
while many cells start to accumulate lipid droplets, which appears that they're going
towards an adipocyte fat cell fate. And similarly, on stiffer materials, you tend to get a mixture
of cells that are expressing markers of muscle and bone. And so the question is, can we devise a way
to exclusively promote these different fates on these hydrogel materials? And so one way, perhaps,
is to use shape, since there's so very many different shapes between these cell types.
And so we adapted a technique, and this is work done by a student in my group, Amr Abdeen.
And this is a technique developed many years ago, actually, where polyacrylamide gels are
modified with this hydrazine treatment, and then you oxidize a glycoprotein to have an aldehyde
that you can then immobilize to the gel. And we developed a strategy to robustly do this across
a range of stiffnesses. And it was really tough to get it on the soft materials, but we are able
to pattern cells. These are single mesenchymal stem cells patterned on gels with half a kilopascal of
stiffness. And so we decided to see what patterning would do. Here's an example of
an interesting early development that we're currently investigating, where we found that
when in unpatterned culture, in pattern culture, we see very different mixtures of these lineages.
So here, when cells are patterned, it's tough to see here, but that's because the marker isn't
really there. But these are patterned cells, and they show very low levels of beta-3-tubulin.
While in unpatterned culture, a subset of these cells will elongate, develop these processes,
and start expressing this marker for neurons. And this is that quantitated with this box and
whisker plot, where we see throughout the spread, there's some of these cells that are
expressing higher levels of beta-3-tubulin. Maybe more interestingly, when we pattern
cells, and we stain them for adipogenesis markers, we see that most of the cells across the surface
are expressing markers associated with fat. And so I think this is really interesting.
In a couple experiments, we saw greater than 90% of the patterned cells had lipid droplets.
And my hand-wavy argument right now is that perhaps what we're doing is limiting some of
these morphological changes that occur on these different materials of different stiffness,
where if a cell needs to elongate and form dendritic processes in order to undergo this
neurogenesis pathway, and it's restricted by being patterned in these small shapes,
perhaps it can't do that, and it chooses fat cell fate. And so perhaps this could be a platform to
modulate between these different lineages. And so we're currently exploring this in other systems
to look at other differentiation outcomes, like muscle and bone,
and perhaps finding ways that we can guide neuronal process extension.
On stiffer gels, this is just a snapshot of some recent results from a student in my group,
Jun, Jun Min Lee. And what Jun did is he used this shape that he calls,
affectionately, the Batman shape. He actually sent me a picture for our calendar that has
Batman standing on a tower, and the light is showing this cell up there.
I should have put that in the slides. But the idea of this geometry isn't because we like Batman.
It's that it has some of these cues that we've seen previously on gold to promote bone cell
differentiation. So it's got aspect ratio, and it's got regions of subcellular curvature.
And what Jun found is when he leaves cells just to sit on 10, 20, 30, or 40 kilopascal gels,
he sees expression, and then an increase at 30, just like Adam Engler and Dennis Disher did,
where this seems to be an optimal microenvironment mechanics to promote bone cell fate.
But when cells are a pattern in this Batman geometry, you see a higher expression of Runx2,
this regulator of osteogenesis in this geometry. And so I think this is a pretty cool result,
and we're going to try to find ways that we can do this for myogenesis and these others
in order to maximize the lineage outcome. And the idea in the long run would be able to control
across a material surface differentiation spatially in the way that you want, so that
you could have gradients of cells. Because remember, there's no chemicals added here.
This is just the mechanics and the geometry of the cells. So you might be able to pattern fat
and bone, and integrating it with our ability to use microfluidics and some of these technologies
to form gradients of stiffness, you could envisage all sorts of really cool applications.
All right, I'm pretty much done here, aren't I? I'll just go really briefly through this.
This is just a new story that's come out that I think is interesting, and it revolves around
this idea that kind of came a bit out of a serendipitous discovery where we patterned cells,
and they were left, and these chemistries on gold are not so stable. And so cells can migrate and
leave them after about a week. And these were left for too long, and so they migrated out. And
they were fixed and stayed anyway, thankfully, because we saw something really interesting,
where when cells, as we culture them normally, are allowed to express alkaline phosphatase when
they divide. When they were patterned, after they escaped, they're no longer expressing alkaline
phosphatase for whatever reason. And so we're thinking this could be related to quiescence.
And so we did some staining to see if these cells are synthesizing DNA, and when they're patterned,
they tend to not. If they're in very small patterns, these cells seem to be turning into
a quiescent state. And then one thing that I think is really exciting that's coming out is we see that
in these patterned cells, we're seeing higher expression of stem cell markers. And so we did
these experiments where we patterned the cells, and then after they escaped, we trypsinized them,
and then reseated them at the same density, and looked at stem cell markers. And here for
endogluin and Stro1, we see a higher expression of stem cell markers in the cells that had been
patterned. And importantly, we see high expression of alkaline phosphatase in the
unpatterned cells. So this might be a way to prolong the stem cell phenotype and culture.
And so you can imagine being able to do this to maintain that, and I think it's pretty neat work.
And this is work by Doug Zhang in my group. And they also can still differentiate. I'll show that
before I close here. And so here, this is alkaline phosphatase staining. You see less differentiation,
and that's presumably because many of the cells in the starting population are still
mesenchymal stem cells and not progenitors. We see this with fat as well.
Okay, I'm just going to quickly summarize. Cell geometry and these things influence each other.
This is really at the heart of our group's work, and seeing how we can use these platforms together
and combine these cues to maximize outcomes. And in this new work, I think it's pretty exciting
where we might be able to use patterning to prolong this MSC phenotype. With that,
I'd like to thank my group. This is an older picture, but we've expanded a bit.
In particular, I want to thank Amr, who's developed a lot of the hydrogel chemistry,
Jun, who's done some of the patterning and hydrogels, and Doug with this recent story.
I've got to thank Milan Merksish and his group, of course, for some of the preliminary work that I
did during my postdoc, and these collaborators listed here. With that, I'll be happy to answer
any questions. Thank you. Any questions? Yes? I was really interested in what you mentioned
about the porosity of the polyacomide and then maybe having a role in this sort of stiffness
condition. Would you say that some of the studies that were countered, or the one study that was
kind of countered that argument had some flaws? I'm kind of curious if you could elaborate a bit
more maybe? On the flaws of that study? Or just, yeah. I think, well, one of the main
is that they only looked at the fat and bone cell differentiation, and so I think before we
can relate that directly to a lot of the work that's been done with muscle differentiation
and things like that. I also think that the techniques they used for measuring the stiffness
of the materials, I believe they used AFM in some of these things, which are valid,
but it would have been nice to see some complementary approaches. Also, they didn't
characterize as well as maybe they could have, you know, how the cells are interacting with the
surface. So, I mean, from a surface chemistry perspective, I'm interested in how the cells
are actually grabbing onto these ligands, so I would have liked to see a more detailed analysis
of focal adhesions and tractional forces and things like that. So those types of things. I'm
sure there's other people on this campus that have many other issues too. Yes? So,
maybe Shane, any other? Oh, excellent question. I mean, that's something that I'd love to do.
We haven't, we haven't done any of that. Most of our analysis has been static and measuring
large populations of cells. Clearly, and I think the field's moving in that direction too. We're
doing some live cell analysis, and then, you know, I started a collaboration during my postdoc with
Provost Mody at Rutgers. He does a lot of this high content imaging of cells and looking at
things like cytoskeletal descriptors is something that I think would be really exciting where
after a day or two, you might be able to see rearrangements of the cytoskeleton and be able
to make more firm correlations between these morphological changes in outcome. I mean,
it's absolutely exciting, but we haven't done it. A quick one. I mean, you give a really nice
overview of all the basic science related. What are the important parameters? Perhaps
how some of them correlate in ways. What's sort of the state of the art then in terms of materials
that would, where you can translate a lot of these things into maybe therapeutic devices?
Are there things that are occurring that allow sort of programmable matrices for,
that would maybe allow some of the parameters you're talking about to be?
Yeah, I mean, I think it's pretty, we're pretty new at being able to do it robustly,
but absolutely there's new hydrogel chemistries coming out all the time.
Two-photon locally polymerized gels in a three-dimensional way, maybe a tissue shape
than a 3D material. And a lot of the control over like some of these peg-based multi-arm gels
where you may be able to change the length of the pegs and rationally incorporate peptide ligands
into some of these stuff. So synthetic scaffolds, I think we're starting to develop cool tools.
Whether, you know, I think we're a long ways off before we can translate those into,
but maybe not too far. That's my field. We're making progress. We still, I don't think we'll
be able to fabricate a scaffold that resembles what happens when you decellularize a heart,
for instance. But I think we're learning some of the principles. And these fundamental studies,
I think, feed into that, where at least with these synthetic strategies we're developing,
we know that we can use materials principles to vary things like stiffness.
All right, let's thank Chris once again.

