How do epithelial stem cells enable renewal and regeneration of organs in adults? Ophir Klein, M.D., Ph.D., explains his research in stem cell-fueled tooth renewal and intestinal stem cells.
[MUSIC] I was going to talk today about a few different areas of the lab which revolve around this theme of how epithelial stem cells enable renewal and regeneration of organs in the adult. Our lab has two main areas of focus from North to South in the GI tract. One is the craniofacial complex, primarily the oral cavity where food comes in. I’ll start by telling you about the dentition and the lining of the mouth, the oral mucosa, which is a really interesting new area that we’ve just gotten into the past few years. Then for the second part, I’ll tell you about our work in the gut and give you a little sense of how we got into that.
For the first part, I want to talk about this property that many people, including myself originally were not that familiar with, which is this idea of renewal of the dentition.
We often think of teeth as these very static structures because in humans they have minimal regenerative ability, but many other animals actually have the ability to grow their teeth continuously. If you look across the animal kingdom, there’s at least seven or eight extinct lineages that have independently or convergently evolved the ability to grow their teeth continuously, like tusks and walruses and elephants, rabbits, all their teeth grow continuously. There’s even a primate, this is a lemur for Madagascar, which has ever-growing teeth. Then like many of you will use the mouse as a genetic model for this.
I’ll start by explaining to you how the system works in mice, which is that the front teeth, the incisors grow continuously. The reason for that is because as the animal is lying on the hard material and it’s diet, it’s constantly upgrading the mineral at the tip.
That needs to be replaced continuously. That replacement is driven by stem cells that are present in the most proximal part of the tooth where teeth that have roots, like all of our teeth or the molar as the back teeth and the mice. Those don’t have stem cells, but the ever-growing teeth do have stem cells and there’s several different compartments.
There’s a mesenchymal compartment in the pulp of the tooth, which produces a Dante blasts that are similar to osteoblast. They make a soft mineral called dentin. Then the outside of the tooth has a much harder mineral called enamel, that’s produced by a very specific cell type called an ameloblast, which is probably the most polarized cell in the body. It makes this enamel, which is orders of magnitude harder than the dentin. We’ve been focused on this epithelial compartment and specifically on this region called the labial cervical loop.
There’s another cervical loop called the lingual cervical loop near the tongue. The labial one is near the lip. The labial cervical loop has been known for a long time to have proliferating cells in it. When I first started about 15 years ago, we wanted to figure out where are the stem cells in this structure and how are they regulated. This is an experiment done about ten years ago by Chunying Li, former post-doc in the lab, which is a very simple experiment, but it shows you how the system is set up.
We’re just going to focus on the epithelium. The mesenchyme is also very interesting, but I want to talk about it for the sake of time today. What Jeanine did, it was just a one-and-a-half hour or bromodeoxyuridine pulse chase to label the proliferating cells. What we saw here was that there were three main regions, epithelium, there was one in this most proximal region which is relatively quiescent, not picking up a lot of label. Then there’s these proliferating cells.
Then all the cells after this yellow arrowhead or the post-mitotic differentiating ameloblasts. When we looked at this, we realized we weren’t quite sure exactly where the stem cells were and that’s what I’ll tell you about over the next few slides.
Just to show you how the system works. If she injected BRDU and waited for 24 h instead of an hour, you can see that the labeling front has moved a couple of hundred microns. All of this movement is from cells that were labeled up in here and then advance either, we don’t know yet if it’s an active or a passive pushing process.
Then by two days they’re moving off the slide. When I was a postdoc, I started working on this and I thought this would be a great system to do various lineage tracing experiments. That’s what we did for a number of years. This is just one example that was done by a couple of foreign postdocs, Brian and Jimmy, who Brian is now at Genentech and Jimmy as a faculty member at UCLA. What they were interested in was this gene called BMI one, which is a member of the polycomb repressive complex and is expressed in both embryonic and various adult stem cells.
They saw that it was present in cells expressing that were present in this proximal part of the cervical loop. We obtained a lineage tracing tool from Mario Capecchi. Cross this with the tomato GFP reporter from Leach and low, which is red and then upon mediated recombination turns green. Before we induce the system, all the cells are red. Then within a couple of days we see some cells turning green in this proximal part where the BMI one is expressed.
This was the critical time point, was the one-week time point because as I told you, the cells only reside in this region for about a day or so before they’re pushed out.
When we see these clones of labeled premium blast present a week after the initial induction event, which is just a single dose of Tamoxifen. That’s definitive proof that we have labeled some progenitor cell. Then we take this out further again with just a single injection of tamoxifen for a month or three months, basically the entire life of the animal. We still get these waves of labeled cells coming out.
This is really the gold standard way to identify a stem cell in vivo, is to label a single time and see if it can produce progeny for the lifetime of the animal. We were really happy with this and we did a whole bunch of experiments like this for a number of years and found both using candidate approaches and doing unbiased screens. A number of different genes that were expressed here and pathways that are important.
We thought this was all exciting and great. But what we realized a few years ago was that we still didn’t really have a good answer to this question of where are the stem cells?
Because almost all of the markers, with the exception of this one, l rig one, which I can talk about if there are questions because it’s a little bit different. But all the other markers that we and actually other people have found spanned this fairly broad domain that included both the quiescent and the proliferating region.
We couldn’t figure out. Actually, I’ll show you on this next slide the question that we couldn’t figure out which is whether what was going on in the tooth was more similar to some of the classical models, like what you see in the blood or the hair follicle, where there’s this strict hierarchy with the somewhat more quiescent stem cells present at the top of the hierarchy, which produce committed progenitors are transit amplifying cells and then differentiate. Or whether what we saw was more similar to the paradigms that have been emerging in many epithelial systems like the gut or the esophagus that in which there don’t seem to be any long-term stem cells, but rather just proliferating stem cells that will then produce progeny that differentiate.
We’ll talk more about the gut and the lining of the mouth in the second part of the talk. This was a question that Amnon Sharir wanted to answer during his postdoc and Amnon left the lab recently to start his own group in Jerusalem. We’re still collaborating on some of these questions which are essentially, where are the stem cells and how are they regulated? This was a few years ago and Amnon embarked on a collaboration with Allon Klein in Boston with a lot of help from Pauline Marangoni, a former postdoc and now research scientists in our lab to do a single-cell RNA sequencing of the epithelium of the proximal incisor.
They dissociated, then sorted the epithelial cells and then use the platform for single-cell RNA sequencing.
I just mentioned that analysis that we used for this was called Spring and this was a program developed by Allons’ lab. What Amnon and Pauline found was that there were three main classes of cells in the epithelium, and these could be subdivided into 15 different clusters. What these three classes represented were. This first-class here represents cycling cells. This loop, when you look at the gene expression, actually show cells going in and out of the cycle.
The second class represented the pre-ameloblasts and ameloblast lineage, the cells that are going to make the enamel. We knew based on the BRDU and lineage tracing size that shows you that these cells emerge from these cells. The one that we were most interested in was this third class of cells, which we call the non-ameloblast epithelium. What these represented where all of those cells in the proximal region, which actually have many different sub-types within them. The question that we wanted to answer was, are there quiescent cells in that region that function as stem cells during homeostasis long-term stem cell, similar to what you would see for example with the bulge of the hair follicle or is all of the proliferative activity that happens in the cycling cell population responsible for the production of progeny.
We did a couple of experiments that pointed very strongly to the former hypothesis that during homeostasis, all of the activity is based on the proliferating cells. One of those was an RNA velocity experiment, which I’m showing you here. This approach for those of you who are not familiar with, it predicts the future state of the cell based on the splice to unspliced mRNA ratio. These arrows represent the trajectories that the cells undergo. The loop here of cycling cells, you can see them going in and out of the cycle.
They, as I mentioned, produce the ameloblast lineage, which we already knew. But we also see that all of the arrows from here are pointing in this direction during homeostasis.
There’s no arrows pointing back from the quiescent population into the proliferating cell. This was the first piece of evidence that we don’t see any contribution from that quiescent cell population to the proliferating cells, which suggests that they’re more of a support cell population doing something else but not contributing to the differentiating progeny. The other series of experiments that we did, and this was published a couple of years ago.
I won’t get into it in a lot of detail, but it’s summarized in this cartoon, was a lineage tracing study done with a double pulse chase of BRDU and EDU. That’s summarized here what we found, which was that we saw this very clear trajectory that we had known about from the proliferating cell populations into the differentiating ameloblast lineage. But we also saw exclusively movement from the proliferating cells back into this quiescent zone and never in the opposite direction. Together, this suggests to us very strongly that the quiescent cells do not serve as a stem cell population during homeostasis.
However, I mentioned in the title of the talk that I’m going to also talk about plasticity.
And this is a concept that we’ve gotten really interested in. I see sharing nodding, we were chatting about that a little bit yesterday. What we decided to as injure the system and then see what happens. For this, are known together with Jimmy, the former postdoc that I mentioned, developed an ex plant culture system in which they can label the cells in vivo and then pop them out and watch what they do over a period of time.
We can do that with various different tools.
In this one I’m showing you, cells are labeled with a Notch1 CreER and the same tomato GFP reporter. Notch1 labels this group of cells here called the stratum intermedium. These are cuboidal cell population that subtends the red cells, which are the ameloblast and they’re thought to provide a support function to feed them ions that they need in order to make the mineralized tissue. And what was amazing during these movies was if you look at the inset here that these cuboidal cells will send up these little projections. You can see them sending up these little fingers.
That almost seems like they’re testing what’s going on above them, but then they quickly retract them.
So I guess they check everything is okay then they pull them back. But when we give five floor uracil, which is a chemotherapeutic that kills all the proliferating cells and really wipes out the epithelium, which you can see here, then if you look in the inset again here you’ll see they’re sending these little projections and somehow we don’t understand how yet. They sense that something is wrong, and then they move up and they integrate. And from what we can tell, they essentially trans differentiate and become ameloblast to fill up that layer.
We see that behavior not only with the stratum intermedium cells, but we also see that the cells in the proximal part that quiescent region have similar behaviors that they can undergo. So our conclusion from this was that although during homeostasis, the quiescent cells don’t have any stem cell function, they facultatively can do so.
If you think about plastic behavior as serving as a backup to enable repair from injury. I’ll just close this section with one slide because I know this talk was supposed to be about relatively recently published papers and so this was the paper that was mentioned in the title that we recently published, which we’ve been using this system now to explore various different aspects of regulation of behavior of the stem cells. And so here what we looked at was the role of splicing and epithelial homeostasis.
And I’ll just briefly say that what we found first in the tooth and then we also confirm this in the intestine was that an important splice factor called SRS F1 is required specifically for the homeostasis of the progenitor population, but not necessarily for the descendants. And the mechanism by which this SRS F1 controls epithelial proliferation is by regulating a number of different genes that are important to prevent activation of P53 and P21 and then apoptosis. And so this is just one example of some of the projects that were now undertaking to use this incisor system to understand how stem cells are regulated. With that, I’ll move on to the second cranial facial part of the talk, which for me has been a really exciting area and something that I’ve been learning about together with the lab, which is the oral mucosa.
And this is similar to the skin, but it’s on the inside of your mouth, and it doesn’t have all the same adnexa structures like hair follicles and sweat glands, it has other ones.
But what’s really amazing about it is that as you’ve probably all experienced, if you bite the inside of your mouth, that heals much more rapidly and without scarring compared to if you cut your skin. Because of these properties, it’s been used a lot in tissue engineering and transplant applications and there are, of course,many diseases and malignancies that are important in the mouth.
I want to credit Kyle Jones, a former DDS PhD student in the lab who is now a junior faculty member at UCSF, who is an oral pathologist and said, we should really look at the behavior of these epithelial cells. So what Kyle did when he began his PhD work was to write a review article saying what he was going to do. And then it was very cool.
Actually, he then went out and did it over the next few years and published his work at the end of that process. So like the skin, there’s a proliferative basal layer which then moves up and differentiates. And I should mention there are important differences between the mouse and the human that we’re still working through. And another important issue to mention is that the mouth is very highly regionalized, similar to when you think of the skin. There’s different skin, for example on the mouse, on the tail versus the back.
In the oral cavity, there are four main regions, so we’re going to talk almost exclusively about the cheek or buckle mucosa, but the palatal mucosa is different, the gingiva mucosa around the gums is different and then the tongue mucosa is also different. This is a image paper from NIH just to show you how much more quickly if you perform similar wounds in the mouth and in the skin, that the mouth heels compare it to the skin.
So, what Kyle wanted to do was first get a handle on the behavior of the oral epithelium in terms of the kinetics. And so for this, he used the H2B GFP system developed by Tumbar and Elaine Fuchs is lab many years ago in which all of the cells are labeled in green in the epithelium and then the labeling is turned off, and so those cells that are rapidly proliferating will dilute out the label and those that are proliferating more slowly will retain it. And so, what you can see here is how rapidly this label is being diluted, this actually is one of the fastest turnover rates in the adult mouse.
And then since we published a couple of years ago, I won’t get into all the details, but I’ll mention that we did a lot of modeling of the behavior of these populations, what we found was that, if you think back about that slide, I showed you about the blood versus the gut, where the blood has this classical stem cell model that’s often called the invariant asymmetry model, where every stem cell division will give rise to another stem cell plus a differentiating cell.
The model that you see present in the intestine and the esophagus is often called the population asymmetry model because their stem cells will be proliferating. Sometimes they’ll give rise to more stem cells, sometimes two daughter cells that are differentiating or sometimes a combination. We saw in the oral epithelium, just like in the gut and the esophagus, that the population asymmetry model is what holds in this tissue. So what Kyle did then was single-cell RNA sequencing to look at the structure of this population, and what he found was that similar to what I showed you in the tooth, that there was this loop of cells coming in and out of cycle which represented the proliferating cells in the basal cells.
Oh, but I should mention here, all of these cells that he’s sequenced were sorted from the basal cell layer exclusively. And so, we see the proliferating cells in the basal layer, but to us what was really interesting was that there was also a large number of differentiating cells present within the basal layer. So these were cells that we had expected to see more in the super basal layers that we’re differentiating, but they were already present in the basal layer. So what we concluded from this work was that as I mentioned, there’s this very high rate of turnover.
And the basal layer actually has a lot of differentiation going on within it.
So the cells begin this journey of differentiation much before they lift off the basal layer and how this is all controlled, we really don’t understand it, but that’s something that we’re currently looking at is how do they know the balance of proliferation and differentiation, how far to differentiate within the basal layer. But all of this initial work that Kyle did prompted us to think about is then what’s going on in the stroma that’s underlying the epithelium and how is this involved particularly in wound healing? This is a project that Jessi Cook graduate soon the lab is working on in collaboration with Mike Longacres lab at Stanford. I’m going to show you just some very preliminary data that we have and I’d be really interested in getting people’s thoughts about this. When Jessi began this project, there was relatively little known about the different mesenchymal populations in the buckle mucosa.
We know that the mesenchymal fibroblasts do a number of different things: they perform functions like production of extracellular matrix, they’re important in signaling tissue maintenance, but Jessi really wanted to resolve the identities and functions of these fibroblasts and then understand how can we explain these differences between wound healing that we see in the skin and the mouth. So together with Michelle Griffin, a postdoc in Mike’s lab, what Jessi did was to perform these wounding essays in both the mouth and the facial skin, and then initially do single-cell RNA sequencing to get a handle on these different populations, and so she got a nice dataset out of this. Here what I’m showing you is a combination of both the buckle and the skin fibroblasts. We have 10 populations, but when we look at which of these populations represents what the origins of the cells are, what we see is actually there’s a fair amount of differentiation between the mouth and the skin. The blue clusters are from the buckle mucosa and the red from the skin.
There’s some overlap, but also some very distinct clusters. Then Jessi and Michelle began to look deeply into the sub-clusters here, and they first did some gene ontology and they found that there were different terms associated with different clusters and they’re marching through all these. But the one that they’re particularly interested in is in this growth and regeneration because they want to understand what’s controlling the healing. They focused in on three different pathways. I’ll just tell you about one of them now, which is the one that’s given us the most interesting functional results so far, which is this Gas6 and Axl pathway.
Axl is the receptor tyrosine kinase for the Gas6 ligand. What we see is that there are important differences in expression in the mouth and in the skin. We see that it’s expressed at higher levels in the unwounded buccal mucosa, both in the mesenchyme and in the epithelium, compared to the skin, and so this led us to then just go ahead and see is there any functional relevance to this. Right now we’re just doing the antagonists and agonists experiments and we’re beginning to work on the genetics. There’s a small molecule which blocks the activity of axle, called bemcentinib, and when we give it to the unwounded buccal mucosa, you can see that it’s already leading to a significant atrophy of the epithelium compared to the control, and then when we give it during the wound repair process, this is four days after the wounding, you can see there’s already nice healing in the control, whereas in the one that’s been treated with bemcentinib, there’s poor healing.
Then when we took the gas x protein, the agonists, and we put it on skin after wound healing, we see that there’s improvement compared to the control and the Gas6 treated one.
We’re thinking that perhaps this could have some therapeutic applications for wound healing in the skin, and so we’re beginning to look into this further. As I mentioned, this is just very preliminary work going on in the lab right now, but Jessi and Michelle are spending a lot of time going through all of these different subpopulations. In terms of the growth and regeneration cluster, we have a couple of other pathways that we’re beginning to look at, and mechanistically, we want to understand better what the downstream targets of Gas6 and Axl are, and so as I mentioned beginning some of these genetic studies.
Jessi put this together, this represents what the different clusters are involved in.
We have ones that are involved in production of the matrix, very signaling and tissue maintenance processes. For the last part of the talk, I’ll move a little further down the GI tract into the intestine, and I’ll start by just a minute or two of Introduction to the system. I’m sure all of you in a stem cell crowd are familiar, but just to remind you that the intestinal epithelium is one of, if not the most highly regenerative tissue in the mammalian body, it turns over every several days in both mouse and human, and in the small intestine, there are two main compartments there, the villi, which are the business end where the absorbed absorption of nutrients and secretion of hormones and mucus happens, and then there are the crypts, which are where the progenitor cells are housed.
If you slice through here, you get a view like in this cartoon where the stem cells and their descendants are housed in the crypt and then they move in this conveyor belt-like fashion, similar to what I showed you in the incisor over the course of a few days, and then they’re sloughed off the tip of the villus through a process called anoikis. For many decades, there was a lot of controversy about the identity of the stem cells in the crypt base, and about 15 years ago, an important experiment done by Hans Clevers Lab showed that a gene called Lgr5, which is a part of the wind signaling pathway, is expressed in these green cells here called crypt base columnar cells, which were at the time one of the two candidate populations for the stem cells, and work over the past many years has demonstrated fairly definitively that these crypt base columnar cells are the homeostatic stem cells.
We got into this somewhat serendipitously right around when I was finishing my post-doc through a close collaboration with Fred de Salvajes Lab, who’s been a mentor and collaborator to me for a couple of decades now, and Fred’s lab had recently made a mouse, this was right around when Lgr5 was reported, and the hypothesis was that eliminating the Lgr5 cells would have catastrophic effects for the gut because they were thought to be so important for the turnover, and so the mouse was a diphtheria toxin receptor knocked into the Lgr5 locus together with the GFP reporter, and Fred contacted us and said, ” Hey, you know, we’re getting some interesting results.
You guys want to look at this together,” and this was really great initial foray, and then this is what led us to continue to work on the gut over the past several years. This here is just showing you the reporter in green, in low magnification and high magnification views in red is a marker of proliferation, and what you can see is that both the stem cells themselves, as well as their transit amplifying descendants are proliferating, so the crypt base columnar cells turnover fairly rapidly. As I said, we had this hypothesis that the elimination of these stem cells by giving the toxin was going to have a catastrophic effect on the gut, and we were really surprised when we gave this to the mice and within a day we saw the stem cells were gone, the Lgr5 expressing cells were gone, but over at least two rounds of renewal 10 days, histologically and by many other measures the gut look totally fine.
This was really perplexing at first, and there was another finding that I found really exciting, which was that when we withdrew the toxin, the Lgr5-expressing cells rapidly reappeared.
Here we’re giving the toxin for several days, you can see there’s no green cells present, and then within a couple of days of withdrawing the toxin, the Lgr5 expressing cells come back, and then within a few days they’ve repopulated the base of the crypt. The gut likes to have these cells there, but it can do fine without them, and I’ll just tell you the punchline after many experiments by many labs is that the reason that the gut can tolerate the absence of these cells is that there’s a lot of plasticity present, and so cells of many other lineages that have begun to commit can actually de-differentiate and repopulate the base of the crypt.
Now there’s still a lot that we don’t know about this process, but that’s the short version. These experiments got me and the folks in the lab really excited about trying to understand the plasticity of the gut, and then thinking in the long term about if we can understand how the stem cells respond to injury and inflammation, can we then harness some of this regenerative ability to think about treating disease? I’ll give you one short example of a story related to this plasticity that we published a couple years ago, and then for the last part of the talk, I’m going to tell you about some unpublished work which again would be interesting getting people’s feedback on.
This story began with this question of how infection with worms affects renewal of the epithelium, and we got interested in this through a close collaboration with Rich Loxley’s lab at UCSF who has been studying these worms for many years, and what we learned from Rich was that vertebrates have co-evolved, many animals have co-evolved with these vertebrates for all of our history, and they still infect billions of people in the developing world and virtually all animals in the wild, and while they’re generally not fatal, they can cause some morbidity, but often they’re non-symptomatic.
They’ve been largely eradicated in the developed world, but because the immune responses are the result of this co-evolution between the parasite and the host, there is an idea that there may be some negative consequences to having eliminated the worms, including that being exposed to them early in life as part of this hygiene hypothesis may protect against some immune-mediated diseases. Actually, some patients with conditions like inflammatory bowel disease will actually self-administer helmets in an attempt to shift their immune balance. The system that we used in collaboration with Rich was a natural mouse helminth called Heligmosomoides polygyrus, which as part of its life cycle, this represents the crypts here and the mesenchyme, the larva will burrow through the epithelium and then live in the mesenchyme for about weeks in a structure called a granuloma, which is full of immune cells and then it will emerge and live as an adult wrapped around the villi in the gut.
This is an image just to show you how rich the granuloma is in these immune cells.
Here you can see in this histological section, the worm sneaking in and out of the plane of the board. From a stem cell biology perspective, what to us was interesting was that it was known for a long time that the crypts directly above the granuloma are hyper proliferative and so this suggested to us that something either made by the worm or made by the immune cells, or perhaps based on disruption of the mesenchyme was causing hyper activation of a stem cell program leading to hyper proliferation. This project was led by Ysbrand, a former student in our lab and Adam, a former post-doc in Rich’s lab. What they did was just infect those Lgr5-GFP reporter that I showed you in the previous slide with the worms and they got this really surprising result, which was the opposite of our hypothesis. Which was that compared to the control in which all of the crypts are full of these green Lgr5 cells, the cells that are right around the granuloma and infected mouse actually turned off Lgr5 compared to those crypts at a distance.
But yet we know that they’re hyper proliferative. We thought maybe something is screwy with the reporter so we looked at a surrogate marker which is called OLMF4, Olfactomedin 4, which is a target or the notch pathway. This is by NC2 and we see a similar thing which is that the Crips right above the granuloma, have turned off expression of OLMF4. This was interesting. What Ysbrand decided to do is to punch out hundreds of these little granuloma and then sort out the crypt base epithelium and do a bulk RNA sequencing experiment to compare epithelium at a distance from the granuloma with the crypt base epithelium around the granuloma.
He got this really nice dataset. There were a couple of important findings from this. Here’s the granuloma associated samples and here’s the samples that are from the non-granuloma associate epithelium.
These are the genes that are down regulated in the granuloma associated epithelium. What he saw was that it wasn’t just Lgr5 and OLMF4 that are down regulated, but these green genes represent the entire suite that constitutes the transcriptional signature of the stem cells, so they’re all turned off.
We also see that there’s a lot of genes that are up-regulated and we’re still actually in the process of working through them, but one that we focused on was this gene Ly6a or Sca-1, which is expressed in a lot of adult stem cell populations but hadn’t really been studied very much in the intestine and there’s a nice antibody for this.
I sprained an atom stained the infected tissue, and as we predicted from the RNA sequencing, we see that those crypts above the granuloma that we knew it turned off Lgr5 express high levels of Sca-1, whereas those crypts at a distance that maintain Lgr5 don’t express Sca-1. Now that we have the antibody and the in vivo reporter, we can then sort these cells. As you’re probably all aware, if you sort Lgr5 expressing intestinal stem cells and put them in major gel, they’ll form structures called the organoids. We put the Sca-1 expressing cells and the Lgr5 expressing cells in culture, and indeed the Lgr5 positive, Sca-1 negative cells formed organoids just as we would expect from wild-type mice.
But the Sca-1 positive cells form these very unusual cystic structures called spheroids. We weren’t sure what these were initially, and then we looked in the literature and what we saw was that if you take fetal epithelium from the mouse and put it in [inaudible] gel, it will form spheroids and there’s actually a beautiful inverse relationship between the age of the embryo and postnatal mouse and its ability to form spheroids.
The older it gets, the more it will form organoids and the less it will form spheroids so by the time it’s fully developed, it forms only organoids and not spheroids. We then looked both at the gene expression patterns from the spheroids as well as back in-vivo dataset, and what we saw was that in both these cases there was a significant up-regulation of genes that are expressed in the fetal gut.
Our proposal was that in addition to the homeostatic renewal, which happens from Lgr5 expressing cells, and what I mentioned before, that can happen when we ablate the Lgr5 expressing cells through this de-differentiation of committed progenitors, that there’s also this process that we call developmental reprogramming, in which the entire transcriptional signature of the crypt shifts over from what we see in the adult, something that looks more like the embryo and this somehow enables better repair or better ability to deal with injury, and then we see that this gene expression program reverts back to the adult program after the injury is healed.
As I mentioned, I want to close with unpublished story that we’re in the process of working on right now. I want to mention this is all the work of Rachel’s Zwick, who’s a really terrific post-doc and came up with this idea of thinking about how does the gut vary along the length of the small intestine? What Rachel is doing is actually revisiting ideas that haven’t been thought about in a couple of millennia. We all learn in medical and graduate school that the small intestine is divided into three main regions, the duodenum, jejunum, and ileum, but there really are not clear anatomic landmarks between these and there’s not a fixed transition point.
Initially what Rachel wanted to do was ask whether stem cells maintain the regional specialization that we know happens along the length of the gut.
But then she realized she needed to actually back up and first define the molecular patterns of regional specialization. In order to do that, she came up with a very clever approach, which is to take the small intestine, this is now using these, again Lgr5-GFP reporter mice and cut it. It’s about 37 cm long, so she cut it into 30 equal one cm segments and then in collaboration with Chris McGinnis, at Gartner’s Lab at UCSF using their multi-sig barcoding approach, barcoded each of these 30 segments so we could then pull them to sequence and know the origin and anatomically along the length of the guide of each of the cells. She applies a sample of barcode, sorts them. There’s actually two different populations that she looks at.
One is just the stem cells and the second is all of the epithelial cells, including the stem cells so what I’m going to show you is actually a combination of these. It looks like it’s enriched in the stem cells just because we put the two together here, but we also look at them individually.
Then we use the [inaudible] sequencing platform. For those of you who are looking at this, you’ll notice in this single cell that it looks like there’s a lot of stem cells, and again, that’s because we’ve put the two populations together here and so we see all of the expected cell types. This is the absorptive lineage, the enterocyte lineage, and then here are the secretory cells, and now we can, rather than a coating them based on their identity, we can color them based on their anatomic origin, and what we see is that the epithelial cells and particularly the enterocytes show this very high degree of zonation.
But not all of the cells do so for example the tuft cells, you don’t see that there’s donated the same way that the enterocytes are in which there’s this very clear demarcation between the proximal to distal origin of the cells, and I’ll get back to the stem cells in a second. Rachel then worked with Dario epithelia bioinformatician, who works closely with our lab, to actually calculate the positions of the molecular boundaries and here what she saw, and to me this is one of the most important findings from our study, was that although we often think of the gut being divided into these three classical regions, again defined by the Greeks millennia ago. What we actually see is that there are five main domains, and what is remarkable about the way that this algorithm works is that the proximal distal origin of the cells.
You see these numbers here. This was calculated by the algorithm and we didn’t feed the information and it’s almost perfect.
There’s a couple of places like here where they’re switched a little bit. But the proximal region localization of the cells is very important component of their bioinformatic identity. So we saw these five domains, and Rachel has labeled them A through E, and they do not correspond perfectly with the classical anatomical domains. There are some regions of where they are more similar in somewhere they’re not, and so she then went in and wanted to look more carefully at the regional segregation of these domain defining genes, and the way that she does this is using an approach called a Swiss roll in which the gut is coiled up and the proximal part is on the outside and the distal part is on the inner, inner part of the Swiss roll.
I’ll just give you one example of what this looks like.
But she’s done this now for many different genes, and so she sees this really beautiful segregation of these domain defining genes. This is one example, these are two fatty acid binding proteins. One of them is expressed more proximately, one more distally. But these two are actually fairly well studied, but she can now go in with many different genes that have not been studied as much in the context of the gut and look at their regional localization, and in terms of functionally, what these different domains do, they seem to be primarily defined based on the metabolism of what they absorb, and so when she then takes these enterocytes and then re clusters them, looking at just the absorptive cells by themselves into the five domains, they break up very nicely, and what Rachel has done is now identified many different pathways that are expressed in each of these domains, and is beginning to look at how both the metabolism affects the patterning and vice versa, and I think this is really a treasure of information and Rachel is going to hopefully go on the job market sooner, will then be able to use a lot of this in her own lab, and so she’s doing things like giving them different diets and seeing how that affects the patterning.
So this work was really interesting for us from the mouse.
But we also wanted to know whether this carried into other species, and so we were fortunate to collaborate with J. Gardener, a transplant surgeon at UCSF, who enables us to obtain organs from donors that are not being used for transplant, and so Rachel gut, almost every in the lab to work with her as this human intestine was brought into the lab, and this is from a healthy donor whose or other organs are being used for transplant, and here of course, the gut is many tens of feet long, so we can’t use the entire segment, but she again divided it up into three pieces and then just use the first bit of each of those pieces, and of course also the human was not labeled with an algebra five GFP. So we can hear, just sort out the crypt base epithelium based on expression of CD44, and so she did that and she got a very nice dataset. All of the cells together. You can see the progenitor cells here and then the enterocytes as they differentiate, and these were barcoded in the same way as what I showed you with the mouse.
So similar to the mouse, there’s a very high degree of nation in the human, and she could then make a similar heatmap to the one that I showed you for the mouse. But now with the human data, and again, we see that there are five metabolic domains, five molecular domains, just like what we see in the mouse. Some of the boundaries are actually identical in terms of the segments, and I should mention here there’s only 15 numbers on the bottom because it turned out that bioinformatics, it was easier to do pairwise comparison of segments so each segment actually represents two in the mouse, but otherwise it’s the same.
So what we see is that the most proximal and the distal boundaries mirror what we see in the mouse, but then the boundaries in the middle actually are shifted over, and so in a way actually, what’s to me more surprising is how similar both the domains and their boundaries are. There are some differences that are present, but considering how different the human and the mouse are in terms of our diet and our lifestyle, that the patterning of the gut is overall quite similar between these two species.
The last thing that Rachel wanted to do is understand whether these functional domains are maintained by similarly regionalize stem cells, and that was actually her initial question when she began the project. When she then looks at the heatmap for just the stem cells, what she sees is that here, there are actually only three subpopulations, and how these three subpopulations feed into the metabolic domains is something that we don’t understand. There must be some information which is inherently present in the stem cells, but also additional information which is provided by the external environment in some way so these three different domains here, and similarly that there are subtle differences between the mouse and the human in terms of where the stem cell domains are present.
Then what Rachel is now working through, and I’ll just give you one example of this on this last slide is to see whether we can actually demonstrate that these regional stem cells control the specialization of enterocytes, the differentiation of the downstream cells. So she and Dario performed in-silico predictions of upstream transcription factors that would be regulating gene expression in the progeny.
She’s identified a number of candidate patterning factors and it’s using CRISPR gene editing to perturb this and regional organoids, and this is just one example. This is a gene called C dx1, and so she’s deleted. See dx1, which is highly expressed in the distal or organize the domain E derived organoids, and what she sees is that this leads to a shift to gene expression program that looks more like what we see in the more proximal organoids.
Now of course there’s going be, it’s gonna be probably multi-factorial as many, many different pathways and transcription factors involved in this. So this is just a first step in understanding how identity is controlled.
So to summarize this last part, what Rachel is found is this regionalization of both the enterocytes and the stem cells and is beginning to look at genetic control of this process, and she’s identified these five different domains that are present in both mouse and human. There is evidence for regional population of the stem cells, and now, as I mentioned, we’re trying to figure out how the regionality of the stem cells directs differentiation of the enterocytes and downstream how that affects the function of the gut including metabolism. So I’ll stop there. I mentioned the folks in the lab who did the work and all of our collaborators who have helped us and also want to acknowledge our funding agencies, and thank you for your attention.
I’m happy to take questions if there’s time.
[APPLAUSE] Question. The first one is on your last story. There’s stem cell regionalization into three, could that also be metabolism driven? What do you think is driving that, and in terms of that regionalization temporal format, is it a prenatal event when metabolism is not so much of a queue or postnatal? Those are both great questions.
So I’ll say for the second one about how developmental the origin of the regionalization is.
So that’s something that we’re working on. We have similar datasets from relatively early and gut development through early postnatal life, and so I don’t have a clear answer yet. My guess is that well, or I should say, I think we have some evidence that there’s some of the patterning that happens without any external cues. You can imagine that either there’s a reinforcement of this patterning by metabolism.
That’s a refinement of it. That’s something that we now want to figure out is how much of this is hard wired and how much of it is is externally controlled. Depending on how you define hardwired, there is evidence, not just from our group but from others. That for example if you make organoids from proximal versus distal parts of the gut and propagate them for a long period of time, they’ll keep their regional identity. So it’s not entirely just dependent upon seeing nutrients.
Presumably there’s also epigenetic locks on the end. I think it’d be really interesting.
In terms of your question about whether the stem cells themselves are informed by nutrients. You could easily imagine a way that would happen because as the food is absorbed by the lymphatics into the bloodstream, it’s going to be traveling presumably down around that region. I think it’s very possible that happens.
We don’t have any direct evidence for that yet as far as I know, but that’s what we’re thinking about with some of these diet experiments. Can I just ask you to quickly speculate the two modules of the stem cell, was the one on top in the intestine fold and give rise to multiple cell types versus small regional limited span, what drives that? You are at the very unique position because you studied both types. Actually, I would say as of right now, all of the stem cells that we’re studying, none of them fulfill the invariant asymmetry, the classic model. We thought that the two did for long time, because we modeled it, not just us, but it’s a small field that other labs are working on is also model this work after the hair follicle.
The tooth is actually considered an ectodermal appendage that developmentally is very similar to the hair follicle. We just assume that it would be like what you see in the bulge, but it’s actually very different in the tooth. I think this is a really interesting question. I can speculate on that there are different reasons that you might want. Classically, people have talked about things like certain stem cells.
You might want to protect their genomes and have them divide less.
I think the evidence around things like the immortal strand hypothesis is not that’s fallen a little bit, I think, by the wayside. But in general, the idea that you want to proliferate less in certain regions or that you need to proliferate less for example in the muscle, you don’t need a lot of turnover unless you have a big injury. Whereas epithelia that have to turn over very rapidly are going to want to have a just a continuous proliferation. But yeah, to me it’s something that could be an exciting project for somebody to compare.
If you can think of related systems in which you see these different proliferation behaviors, how that happens. You mentioned so many interesting things. I just wanted to ask you about this, [inaudible] axial interaction you mentioned.
Because I know another epithelial junction mesenchymal sites in the body plays a really important role. For instance, if that’s missing, then the photoreceptor the RPE fails to digest the pinocytosis the outer [OVERLAPPING] I’m just wondering what do you think is happening there?
The signal seems to be coming from the mesenchymal fibroblast to the epithelial cells to tell them to do what? Well, actually right now we’re figuring it out, because both the signal and the receptor are expressed in both compartments to a certain degree.
That’s why we’re getting the mutants in to begin to look tissue in a tissue specific fashion, where they’re important and what their roles are. The easiest guess, but I don’t want to promise. The easiest guess, basically, they’re telling these cells to either proliferate or to secrete ECM, plug up the wound, and begin to fill it in with a scar.
Based on the inhibitor or an agonist data that suggests that that’s what’s going on. But again, these are very preliminary, so we haven’t even done the simple things that I’m sure you’re thinking about.
Like have you looked at how the drugs are affecting proliferation, and we looked only a little bit of gene expression. All those things are things that we need to do. Maybe what I’ll do is I’ll take one or two coming in Zoom.
You may have alluded to this answer already, but the question is from the earlier part of your talk where you talked about the incisor and mentioned that it’s actively cycling cells but seemed to maintain normal homeostasis. The question I was wondering, what about under conditions of injury will the quiescent cells maybe then jump back into cycle. Yeah, so that’s what I very briefly mentioned, that we see not only those stratum intermedium cells that I showed in the movie, but also we see behaviors in the most proximal part where there’s very quiescent cells that can. Those cells will very rarely proliferate under homeostasis. The image that I showed of the one-hour chase there, you see one proliferating cell there, you’ll see more of that happening.
One thing that I’m hoping somebody will be interested in studying in the future is, there’s all these different subpopulations there that we identified by the single-cell RNA seek, and we don’t really know very much about them. I’m presuming that some of them are going to have different proliferative behaviors in response to these injuries, and we actually have a nice model now with the exponent culture system, and so we do see cells proliferate more, at this point we don’t really know anything about which ones are doing it and what their behaviors are. One last question in the Zoom was asking you to speculate a bit more clinically particularly some of the last work that you did with the human gut and the regionalization. Does this give you any insight into how to approach clinical conditions like short gut or neck or even better transplantation strategies? I think that’s a great question and I know that Rachel, the postdoc did the work, is thinking a lot about that.
We’re just still at a very early stage as a field of thinking about how to use stem cells in the gut for repair. But you can imagine either by understanding how this regionalization is controlled, that that would enable you to perhaps target therapies better to certain regions of the gut that in which you see various conditions being more active even at an individual patient level. Or that as we move into thinking about using stem cells and organoids for repair, that understanding their identity will be helpful. Because you don’t know if putting an organ or tissue engineered construct that’s similar to one part of the gut into another, is that going to remodel and how much and can you control that? I think all those questions are going to be interesting to think about therapeutically.
I think we’ve exhausted all the questions on Zoom. Thank you very much.
With that, let’s give a round of applause and thank you for launching the new academic year. [APPLAUSE] [MUSIC].