Investigating Lineage Plasticity in Organoid Models of Prostate and Bladder Cancer

Transcript

• 00:00Good afternoon, my
• 00:03name is Katie Politi and I'm an
• 00:07associate professor of pathology
• 00:09and of medicine here at the Yale
• 00:12School of Medicine and Awesome Co.
• 00:16Leader of the cancer Signaling Networks
• 00:19Research program at the Yale Cancer Center,
• 00:23and it is my pleasure to introduce
• 00:27Doctor Michael Shen today.
• 00:29Doctor Shen is a professor of medicine,
• 00:32genetics and development,
• 00:34urology and systems biology at
• 00:37Columbia University Medical Center.
• 00:39He received his undergraduate
• 00:42degree from Harvard University and
• 00:44his PhD from Cambridge University.
• 00:47He then pursued his postdoctoral
• 00:50training with Phil, Doctor Phil,
• 00:53leader at Harvard Medical School,
• 00:55before becoming an independent investigator
• 00:58at Rutgers in 1994 and moved to
• 01:02Columbia University Medical Center in 2007.
• 01:05He is currently the Co leader
• 01:07of the Tumor Biology and Micro
• 01:10Environment Program at the Herbert
• 01:12Irving Comprehensive Cancer Center,
• 01:14as well as the director of Graduate
• 01:17studies in the Columbia Department
• 01:20of Genetics and Development.
• 01:22Over the past 26 years,
• 01:25Doctor Shen has investigated fundamental
• 01:28mechanisms of mammalian development
• 01:30in cancer using in vivo analysis of
• 01:34genetically engineered mouse models.
• 01:36Recently his lamp has generated novel
• 01:38culture conditions for mouse and human,
• 01:41prostate and bladder organized noise,
• 01:43which have led to the creation of a bio Bank
• 01:46of patient arrived bladder tumor organoids.
• 01:49Current work in the lab focuses on
• 01:53understanding molecular mechanisms of cell
• 01:55type differentiation in the normal as
• 01:57well as the transformed prostate epithelium.
• 02:00The epigenetic regulation of
• 02:02linear plasticity in both
• 02:04bladder and prostate cancer,
• 02:06and the role of the tumor micro environment
• 02:10in modulating treatment response.
• 02:13Doctor Sen,
• 02:14it really is a pleasure to have
• 02:16you here today and have you visit.
• 02:19I'll be at virtually from
• 02:20Columbia University,
• 02:21so thank you very much and we look
• 02:23forward to your presentation.
• 02:27Well, thank you Katie.
• 02:28It's a real pleasure to have
• 02:31this opportunity to speak to.
• 02:33This audience that yell and I
• 02:35wish this were in person, but.
• 02:38Make do as best as we can,
• 02:40so I'll go ahead and
• 02:42share my screen. Um? And.
• 02:50Hopefully. You can see my presentation.
• 02:54Is that true? Can everyone see my?
• 02:58Yes we can see it.
• 03:00Thank you. Yes excellent.
• 03:01So today I'd like to talk.
• 03:03Take the opportunity to discuss published
• 03:06work, but also a lot of work that can
• 03:09be construed as work in progress.
• 03:11Much of it focusing on the
• 03:14issue of linic plasticity.
• 03:15And we've been studying this through
• 03:18in vivo analysis in mouse models,
• 03:21but also using organoid models,
• 03:23and we've been studying this in
• 03:26both the prostate and the bladder.
• 03:30So to start with, what is plasticity?
• 03:32So if we consider that plasticity
• 03:35in the most general definition is
• 03:37the ability of a cell to change
• 03:40from one identity to another,
• 03:42we can think of this in the.
• 03:45You know, perhaps cliched Waddington model.
• 03:48As you know,
• 03:49sort of balls rolling down a Hill.
• 03:52We start with a stem cell and we
• 03:55have various differentiated cell
• 03:56types and the ability of cells too.
• 04:00Basically change their identity
• 04:02is considered to be plasticity
• 04:04and so there are different forms
• 04:07there sort of reprogramming back
• 04:09to a more progenitor state.
• 04:11There can be transdifferentiation changing
• 04:13from one identity to another etc.
• 04:16So this is a process that has been
• 04:19studied extensively in both development
• 04:22and cancer and it's important to
• 04:25think about when we talk about
• 04:27plasticity in cancer to consider that.
• 04:30In order to study plasticity,
• 04:33it's also essential to understand the
• 04:36normal pathways of differentiation
• 04:38so that one can distinguish what
• 04:41happens in the normal context from
• 04:43what might happen in a tumor context.
• 04:46So over many years we've been studying
• 04:49these issues in the prostate,
• 04:51and more recently in the bladder,
• 04:53and to start with, I just like to consider,
• 04:56you know,
• 04:57sort of a basic review of the
• 05:00prostate in the mouse.
• 05:01The prostate has a distinct anatomy there.
• 05:04It sort of has lobular structure.
• 05:06There are four different lobes in the mouse.
• 05:10There's the anterior prostate,
• 05:11the dorsal prostate, the lateral prostate,
• 05:14as well as the ventral prostate.
• 05:16However, in the human there is
• 05:19something a little bit different.
• 05:22The human prostate does not have
• 05:25a distinct lobular structure.
• 05:27Instead, it can be distinguished at
• 05:30the pathological level as having you know,
• 05:33sort of architecture of different regions.
• 05:36So there are zones that have
• 05:39been defined pathologically.
• 05:41The peripheral, central,
• 05:42and transition zones,
• 05:44so the distinct anatomy of the
• 05:47mouse and human prostate has been.
• 05:50Um?
• 05:50Of note,
• 05:51for many years and has been used sort
• 05:54of as an argument as underscoring
• 05:57the inability perhaps of the mouse
• 06:00to truly model human prostate cancer.
• 06:03Now we've known from studies over many
• 06:05years that there are many conserved
• 06:08signaling pathways in the like,
• 06:10but the relationship between the
• 06:12mouse and human prostate still
• 06:14remains somewhat mysterious,
• 06:15both in terms of normal development as
• 06:18well as the cell types can contain.
• 06:21Within the prostate. No.
• 06:23The prostate in the mouse is formed
• 06:26at late stages of fetal development
• 06:28and organic Genesis
• 06:30primarily takes place at neonatal stages.
• 06:33The prostate form,
• 06:34through a process of ductal budding
• 06:36budding from the urogenital sinus
• 06:38and initial prostate bugs are marked
• 06:40by expression of the homeo box.
• 06:42Jinan kicks 3.1 here,
• 06:44visualized by Beta Galactoside.
• 06:45ASA Valax Enoch in that we made a
• 06:48number of years ago and you can see
• 06:51even early on at the time of birth.
• 06:54Their buds that correspond to distinct
• 06:57lobes and at least initially and kicks 3.1,
• 07:00is expressed by all of the
• 07:03epithelial cells in the prostate.
• 07:05Prostate formation of course
• 07:08requires androgen signaling,
• 07:09but the requirements for androgen receptor
• 07:12are actually fairly complex and they
• 07:16involve epithelial mesenchymal interactions.
• 07:18So androgen receptor is actually
• 07:21required in the urogenital
• 07:23mesenchyme for prostate formation.
• 07:26So if you delete androgen receptor in
• 07:30the urogenital mesenchyme here in a
• 07:33TFM una testicular feminize mutant.
• 07:36And you perform a a tissue recombination
• 07:40assay as was first shown through.
• 07:43The studies of Jerry Kunia about four
• 07:46decades ago now the prostate will not form.
• 07:50However,
• 07:50if you delete androgen receptor in the
• 07:53epithelium now you will form a prostate.
• 07:56However, the prostate is not entirely normal.
• 07:59For example,
• 08:00it lacks secretory protein production.
• 08:03So what are the cell types that are
• 08:06found in the normal adult prostate?
• 08:09Well in both the mouse and the human
• 08:12there's an array of different cell
• 08:14types in both the epithelium and in the
• 08:18stroma that are just now really being
• 08:20able of being characterized in some detail,
• 08:23but historically we've considered
• 08:25the epithelium as containing
• 08:26three basic cell types.
• 08:28There are the luminal cells which are
• 08:31the tall columnar secretory cells.
• 08:33That produce the the prostate secretions.
• 08:36There's an underlying layer of basil cells.
• 08:39These cells are androgen receptor
• 08:41low or negative,
• 08:42and express basil cytokeratins
• 08:44unlike the luminal cells,
• 08:46which are a are high and
• 08:48expressed lumenal cytokeratins.
• 08:49And then there is a rare third type
• 08:53known as neuroendocrine cells.
• 08:56These have been very understudied and
• 08:58I'll touch upon these later in my talk.
• 09:02So this is sort of the way we thought
• 09:04about the prostate epithelium for
• 09:07many years now, and classically,
• 09:09basil cells have always considered have
• 09:11been considered to be more interesting.
• 09:13They appear to have more stemlike
• 09:16properties or does luminal cells have
• 09:18been considered to be somewhat boring,
• 09:20but I think you know for many years
• 09:23we've thought that luminal cells
• 09:25are actually the interesting cells.
• 09:27And now with the advent of single
• 09:30cell analysis,
• 09:30we see that there's considerable complexity.
• 09:33In the lumenal population.
• 09:35So what do basil cells do?
• 09:37Well, we believe that there is sort of.
• 09:41A conserved ancestral role for basil
• 09:44cells and that's in wound repair.
• 09:46So here, for example, in the prostate.
• 09:49If we damage the luminal cells by
• 09:52deletion of idcat here in inducible
• 09:55deletion of ekit here and the
• 09:58epithelial cells will Slough off.
• 10:00And undergo a notice and the basil
• 10:02cells will actually differentiate
• 10:04into luminal cells to replace
• 10:06the loss luminal cells.
• 10:08So this is shown in cartoon form here.
• 10:10So there's sort of a basil to luminal
• 10:13differentiation that takes place,
• 10:15so we think that this is a conserved
• 10:18function of basil cells in many tissues,
• 10:20and more recently,
• 10:22work from Cedric Web Con Slab has
• 10:24demonstrated that is in fact the case.
• 10:27What about luminal cells?
• 10:29Well, luminal cells, as I mentioned,
• 10:31have been considered to be somewhat boring,
• 10:34but in fact,
• 10:35in ex vivo assays at least one
• 10:37could see that there are luminal
• 10:40progenitors that are by potent.
• 10:42So if we perform lineages marking
• 10:44of luminal cells and generate
• 10:46organoids that are wholly composed
• 10:48wholly derived from luminal cells,
• 10:50then we can see that there are basil cells
• 10:53that can be formed in these organoids,
• 10:56and these basil cells are marked.
• 10:58Indicating their lumenal origin.
• 11:01So of course this is an ex vivo
• 11:05assay and ex vivo cells often
• 11:08display more plasticity than is
• 11:11found in during normal development,
• 11:15and certainly in.
• 11:17Normal context luminal cells
• 11:19are usually you'd impotent,
• 11:21but we think that normal
• 11:24developmental processes,
• 11:25among other things or,
• 11:26can be constrained by the micro
• 11:29environment and ex vivo assays such
• 11:32as organoid formation can reveal
• 11:35developmental potential that can
• 11:37be displayed in specific contexts.
• 11:40So for example,
• 11:41we think that the ability of
• 11:44luminal cells to display by potency
• 11:47to be able to generate basil
• 11:50cells is actually an ability that
• 11:53occurs early in organic Genesis.
• 11:56So this is sort of a linear
• 11:59hierarchy as is currently understood,
• 12:02in which there are by potent basil
• 12:05progenitors generating both basal and
• 12:07luminal cells during organic Genesis.
• 12:10But.
• 12:10Luminal progenitors were
• 12:12generally thought of unipotent,
• 12:13but in recent studies we found that in
• 12:17fact there is a BI potent luminal progenitor,
• 12:20so if we lineages mark luminal cells.
• 12:24At early postnatal stages
• 12:26using an inducible CK 8 driver,
• 12:30we can then mark luminal cells
• 12:33and then analyzed later.
• 12:36Luminal cells are marked,
• 12:37but also basil cells,
• 12:39and it's it's a reasonable fraction
• 12:41of basal cells that are marked,
• 12:44and this by potent progenitor,
• 12:46is fairly short lived.
• 12:47It's fairly transient,
• 12:49but it can still be detected at
• 12:51a week after birth, but again,
• 12:54the ability of this by potent progenitor
• 12:57quickly disappears thereafter.
• 12:59So interesting, Lee,
• 13:00both the by potent basil progenitor and
• 13:03this by potent luminal progenitor do
• 13:05not require androgen receptor function,
• 13:08so if we delete androgen receptor.
• 13:11In basil cells,
• 13:12we find that there's no effect on
• 13:15the formation of luminal cells,
• 13:18and similarly if we delete a
• 13:20are in the luminal cells,
• 13:22we don't see any effect on the
• 13:25generation of basil cells.
• 13:28So in cartoon form then what we
• 13:30think is going on is that there is
• 13:34a urogenital epithelial progenitor
• 13:36that gives rise to both basal
• 13:39and luminal progenitors,
• 13:41and initially the basil progenitor
• 13:43can be by potent as as well
• 13:46as the luminal progenitor.
• 13:48But that this by potency is fairly transient,
• 13:52and then in adulthood both
• 13:54luminal and basal progenitors are
• 13:56generally unipotent.
• 13:58However, this period of by potency
• 14:00is actually occurring in the
• 14:03first four weeks after birth,
• 14:05and Interestingly, this is a time
• 14:08when androgen levels are very low.
• 14:11At these pre pubertal stages.
• 14:14Now there is one other interesting
• 14:16aspect of luminal cells,
• 14:17which is that in a series of studies
• 14:20we've shown that they are favored as
• 14:23cells of origin for prostate cancer,
• 14:25so we've shown this in a number of
• 14:28different transgenic mouse models,
• 14:30as well as a hormonal carcinogenesis model.
• 14:32If we mark Basil cells in these models,
• 14:35for example in the hymec model or a tramp
• 14:38model which are well characterized,
• 14:40transgenic models of prostate cancer,
• 14:42the basal cells are marked.
• 14:44But they don't really contribute to tumors.
• 14:47However, we mark Lumenal cells now they
• 14:50readily contribute to tumor formation.
• 14:53So. What we think is going on then
• 14:56is that if we mark luminal cells,
• 14:59they will contribute tumors,
• 15:00and this argues that luminal cells are
• 15:03a cell of origin for prostate cancer.
• 15:05If we mark Basil cells,
• 15:07they don't contribute to tumors.
• 15:08However, if you explant these cells,
• 15:11they will.
• 15:11You know in this sort of ex vivo context.
• 15:14For example, in a graft undergo a
• 15:16basil to luminal differentiation,
• 15:18and now they can form tumors.
• 15:20So we consider basal cells to be a
• 15:23celeb mutation, whereas luminal cells.
• 15:25Are the true cell of origin.
• 15:27And again,
• 15:28we've analyzed this in multiple
• 15:29mouse models of prostate cancer.
• 15:31So because we've observed the
• 15:33same result every time,
• 15:35we think that luminal cells are
• 15:37generally favored as a cell of origin.
• 15:40So.
• 15:42We now have this view that lumenal
• 15:45cells are in fact quite interesting,
• 15:48so recently we decided to explore
• 15:50the heterogeneity of the prostate
• 15:52epithelium using single cell approaches.
• 15:55For this we had to really learn how to
• 15:59dissect the mouse prostate properly.
• 16:02You might think that's a bit of
• 16:04an over some bug, an exaggeration,
• 16:06considering we've been studying
• 16:08the prostate for over 20 years,
• 16:09but in fact it's really not
• 16:11a trivial matter of two.
• 16:13Dissect the individual mouse
• 16:14lobes all the way down to their
• 16:17junction with the urethra.
• 16:18So this is sort of a view of sort of how
• 16:21we can dissect the weight mouse lobes.
• 16:24This is actually in a green mouse,
• 16:26so you can see the lobes that are dissected.
• 16:30And then we subjected these two
• 16:33single cell RNA sequencing both
• 16:35the whole prostate as well as
• 16:38the individual lobes to analyze
• 16:41the results we collaborated with
• 16:43Roll Rabadan's lab Raul has.
• 16:47Developed you know,
• 16:49amazing algorithms that are based upon
• 16:52rather sophisticated mathematical
• 16:53approaches for analyzing single cell data.
• 16:57So in particular,
• 16:59his laboratory has developed approaches
• 17:01based upon random matrix theory,
• 17:04which demonstrate that these
• 17:06large arrays of data,
• 17:08for example as one generates using so
• 17:12single cell RNA sequencing are mostly noise,
• 17:16so.
• 17:17If you consider.
• 17:19These as giant matrices,
• 17:21well in fact the distribution of
• 17:24eigenvalues follows a sort of conserved
• 17:27mathematical distribution known as
• 17:29the Marchenko Pastur distribution,
• 17:31and the deviation for this distribution
• 17:34is actually where the signal is,
• 17:37and typically it's only
• 17:39about 2 to 3% of the data.
• 17:42So this is a hypothetical distribution.
• 17:45This is actually real data.
• 17:48This is one of our prostate datasets.
• 17:51And again, here is the signal,
• 17:54so his laboratory is developed randomly,
• 17:57an algorithm to isolate these data.
• 18:01And this proved to be very
• 18:02useful in our analysis,
• 18:04because it allowed us to identify a
• 18:06cell population that would have been
• 18:08very difficult to identify other ways.
• 18:10So when we examine an aggregated data
• 18:13set of whole prostate, we observe.
• 18:18Five different lumenal populations.
• 18:20First of all,
• 18:21we only identify one Basil population.
• 18:24So basil cells are actually
• 18:26not heterogeneous,
• 18:27but instead there's heterogeneity
• 18:28in the lumenal population.
• 18:30So, first of all,
• 18:32there's four distinct lumenal
• 18:34populations that correspond to the
• 18:36that are located distally in each
• 18:38prostate lobes alumet population
• 18:40distally in the anterior prostate
• 18:43loom D in the distal dorsal prostate
• 18:46lumel the lateral movie and eventual.
• 18:49Then there is a population that
• 18:51we call Lumpi for proxamol.
• 18:53It is similar in all four lobes and
• 18:57it is it is found more approximately.
• 19:00Finally,
• 19:01there is a population that
• 19:03we call Paraurethral.
• 19:04This population has both
• 19:06basal and luminal properties,
• 19:08and it is found in the region
• 19:11most adjacent to the urethra.
• 19:14So these are the distinct epithelial
• 19:16properties that we've identified.
• 19:18We've also identified heterogeneity
• 19:20in the stroma,
• 19:21but I won't speak to that further.
• 19:24So,
• 19:24as you can note from this
• 19:27sort of illustration,
• 19:28there is diversity along
• 19:30the proximal distal axis.
• 19:32So to give you an idea of what we
• 19:35mean by proximal and distal axis,
• 19:37here's an anterior prostate that's
• 19:39been sort of splayed out and cut in
• 19:42histological section when we refer to distal,
• 19:44we're actually referring
• 19:45to this whole region here.
• 19:47That is more than 90% of
• 19:49the volume of the prostate.
• 19:51The proximal region is just this region here.
• 19:55And these regions are quite distinct at
• 19:57the level of marker expression as well as.
• 20:00Histology,
• 20:01so here if we go along the
• 20:04proximal distal axis,
• 20:05we have the paraurethral
• 20:07region has specific markers.
• 20:08Here is the proximal region.
• 20:10It has a distinct Histology and is
• 20:13marked by specific gene expression
• 20:15of specific genes such as this one.
• 20:18PPP, one R1B.
• 20:21Then we have in the distal
• 20:23region other markers that are
• 20:25specific for distal luminal cells.
• 20:27The lume population here,
• 20:28but you will note there are also
• 20:31proxamol cells that are scattered about.
• 20:34They're not very common,
• 20:35but you can definitely find them in
• 20:38the distal region and then in between.
• 20:40There appears to be a boundary
• 20:44where these regions meet.
• 20:46If you perform electron micrography
• 20:48micrography of sort of the boundary region,
• 20:51you can see that.
• 20:54These lumit lumayan lumpi cells actually
• 20:57appear to be distinct cell types,
• 20:59so the lume cells again are
• 21:02tall columnar secretory cells.
• 21:03The loopy cells,
• 21:04on the other hand,
• 21:06have a more cuboidal appearance,
• 21:08and they don't seem to be
• 21:12particularly secretory.
• 21:13At the transcriptomic level,
• 21:15you can analyze the sort of relationships
• 21:18of these populations with each other.
• 21:22So Luis Aparicio, postdoc INR Lab,
• 21:25who who's done these computational analysis,
• 21:28has used. An approach based upon optimal
• 21:32transport theory to calculate Wasserstein
• 21:34distances between these populations,
• 21:36and then relate these
• 21:38populations to each other.
• 21:39You can see that Lumpi is sort of,
• 21:43you know, sort of at the center related
• 21:46to the distal lumenal populations,
• 21:48and then PR you and basil.
• 21:52So in order to investigate the
• 21:56function of these populations, we've.
• 22:03And so, as you might imagine
• 22:05from this sort of relationship,
• 22:07we observed that there is greater
• 22:10projector potential in the loom PPR.
• 22:13You and basil populations versus
• 22:15the distal lumenal population.
• 22:16So here's an organoid formation assay.
• 22:19You can see that the distal lumenal
• 22:22populations all have a low efficiency
• 22:25of formation of organoids lumpi as a
• 22:28much greater efficiency and PR you in
• 22:31Basel have a greater efficiency yet.
• 22:35We've also isolated these populations
• 22:37by flow cytometry and performed renal
• 22:40grafting assays so in combination
• 22:42with urogenital mesenchyme,
• 22:44these cells were formed renal grafts
• 22:46all be it with different efficiencies,
• 22:49and then we can analyze the sort of
• 22:53cell types present within these graphs.
• 22:57So using the markers that we've identified
• 22:59that is specific for each population.
• 23:02In brief,
• 23:03the loom a another distal luminal.
• 23:05Cells can make more of themselves,
• 23:07but not the loom PNP Ru populations,
• 23:10where is the loom PNP?
• 23:12Ru populations can make all of
• 23:14the other different populations,
• 23:16so this supports a sort of a projector.
• 23:21Increased progenitor potential
• 23:23for loom PNP RU populations.
• 23:26So finally one can ask what
• 23:28is the relationship between
• 23:30the mouse and human prostate?
• 23:32After all,
• 23:33anatomically they are quite different
• 23:34and histologically as well,
• 23:36so we've.
• 23:38Analyzed three independent benign
• 23:40prostatectomy specimens at the
• 23:41single cell level, and again,
• 23:43we can see that there is a
• 23:46single basil population,
• 23:48but they're different lumenal
• 23:49populations that we can relate to.
• 23:52The lumenal populations
• 23:53that we see in the mouse,
• 23:55so there isn't a snare population
• 23:58that seems more closely related to
• 24:00the distal lumenal populations.
• 24:02A ductal populations more proxamol,
• 24:04as well as a PR you like population.
• 24:08Again,
• 24:09using analysis of Wasserstein distances,
• 24:11this time in across species way we can
• 24:14we can definitely show this relationship.
• 24:18So the acinar cells are actually interesting.
• 24:22Lee most closely related to
• 24:24lumenal L cells in the mouse,
• 24:27the ductal cells to loom
• 24:30P&PRU 2P RU in Bloom,
• 24:32P, etc.
• 24:33So this analysis highlights loom L as
• 24:37a population of interest in the mouse.
• 24:41It perhaps is most closely related to
• 24:44the bulk of the human prostate epithelium.
• 24:48Yet the lateral lobe is understudied
• 24:50in the mouse,
• 24:52and particularly analysis of cancer models.
• 24:56So now I'd like to turn to cancer a
• 25:00little bit and think about plasticity
• 25:04in advanced prostate cancer.
• 25:07So in the current spectrum of of.
• 25:12Prostate cancer where we have.
• 25:15Treatment of with potent anti
• 25:17androgens that are very efficient at
• 25:20suppressing energon receptor function.
• 25:23Now,
• 25:23castration resistant prostate cancer is
• 25:26displaying a range of different entities.
• 25:29As sort of a spectrum that can be
• 25:33distinguished perhaps by different
• 25:35differential lineages, plasticity.
• 25:37So there is prostate cancer that is
• 25:40remains a our pathway positive it
• 25:43still expresses androgen receptor
• 25:46despite the presence of anti androgens.
• 25:49And then at the other extreme
• 25:53we have neuroendocrine prostate
• 25:55cancer which is a are negative and.
• 25:59He's most extreme forms can display
• 26:01a small cell phenotype that's
• 26:03very aggressive and lethal,
• 26:04so there's been considerable interest
• 26:06in the mechanisms of neuroendocrine
• 26:08differentiation in CR, PC, and so.
• 26:10There have been studies of CR PC that are
• 26:13trying to distinguish the different entities,
• 26:16and, for example, there is something
• 26:19considered double negative.
• 26:20That is a are negative and
• 26:22neuroendocrine negative,
• 26:23which is defined more by what it
• 26:26is not rather than what it is,
• 26:28but the relationships between these.
• 26:30Distinct entities is unclear,
• 26:33and it may be simple, maybe more complex.
• 26:39But there is widespread agreement that
• 26:41there must be a role for epigenetic
• 26:44reprogramming in this process, and so.
• 26:47A range of studies have provided evidence
• 26:50that there's increase in ezh two as PRC.
• 26:54Two activity in this spectrum of
• 26:57plasticity and recently flybot
• 26:59John Cody's lab has shown that PRC
• 27:01one activity is elevated in double
• 27:04negative prostate cancer etc.
• 27:06But this is all involved studies
• 27:09either in cell lines or in human
• 27:12prostate cancer specimens,
• 27:13so it's been difficult to sort
• 27:17of study these things.
• 27:19You know at a more detailed level
• 27:22and in terms of mechanism as well.
• 27:26So to approach this,
• 27:28we started with a mouse model that we
• 27:32had been analyzing in collaboration
• 27:35with Korea Body Schenz lab.
• 27:38So the NPP 53 mouse model uses
• 27:42inducible deletion of P-10 and P53.
• 27:45Of these animals develop a castration
• 27:49resistant prostate cancer that will.
• 27:52Display features of neuroendocrine
• 27:53differentiation and we were able to
• 27:56distinguish what we called focal
• 27:58neuroendocrine differentiation in
• 27:59which the neuroendocrine cells are
• 28:02not proliferative from overt nor
• 28:04endocrine differentiation which often
• 28:06displays a small cell phenotype.
• 28:08These are highly proliferative
• 28:10or endocrine cells.
• 28:11However, since we use lineages marking here,
• 28:15we could show that the neuroendocrine
• 28:17cells are derived from a luminal cell,
• 28:20so the initial.
• 28:23Tumor induction was from luminal
• 28:26cells and then we have.
• 28:29Alright,
• 28:30formation of CR PC neuroendocrine
• 28:33differentiation and then there is
• 28:35some type of proliferative switch
• 28:37that we don't understand that
• 28:40results in this highly proliferative
• 28:43neuroendocrine prostate cancer.
• 28:44So because we use linear tracing here,
• 28:48we provided evidence that in fact
• 28:51this was a transdifferentiation of
• 28:53luminal cells to neuron can cells.
• 28:56So when we think about transdifferentiation,
• 28:59there's sort of a fundamental
• 29:01question both in developmental
• 29:03as well as cancer context,
• 29:05which is what is really going
• 29:07on in terms of the pathways that
• 29:10result in transdifferentiation.
• 29:12Well,
• 29:13it's possible that this change in
• 29:15identity occurs through a transition that
• 29:18happens normally during development.
• 29:21Alternatively,
• 29:21it's possible that this change
• 29:24in identity occurs through a
• 29:26transition that is wholly or at
• 29:28least partially novel that doesn't
• 29:30really occur in normal development,
• 29:33so there could be a hijacking of
• 29:36an alternative pathway or or some
• 29:39other pathway or process that
• 29:41doesn't occur in normal context.
• 29:44To understand this,
• 29:45of course it is first of all important
• 29:49to discover how neuroendocrine
• 29:52cells differentiate normally.
• 29:54So.
• 29:55It's remarkable that there's very
• 29:57little in the published literature about.
• 30:00Origin in fact,
• 30:02features of neuroendocrine cells
• 30:03in the normal prostate,
• 30:05enlarged likely because they are quite
• 30:08rare and what is also interesting
• 30:11is that there are different models
• 30:13have been put forth for their origin.
• 30:17So one model says that they
• 30:19actually are of epithelial origin
• 30:21and arise from Basel progenitors.
• 30:24Another model is that they arise from neural
• 30:27Crest and so Cedric Pompons lab published.
• 30:31That they came from basil
• 30:33cells and a more recent paper.
• 30:34From that they're from their old Crest
• 30:37in both these papers use linear tracing.
• 30:39In fact, we believe that both of
• 30:41these papers are incorrect and most
• 30:44likely they arise from an early
• 30:46urogenital epithelial progenitor.
• 30:48So neuroendocrine cells are very rare.
• 30:50But what is make some particularly
• 30:53in the mouse, prostate is they're
• 30:55highly asymmetrically distributed,
• 30:57so they are mostly found in the proximal
• 31:00region, which I showed you earlier.
• 31:02So nearly all the owner can
• 31:05cells are proxamol.
• 31:06They're very rare distally.
• 31:09And remarkably, neuroendocrine cells,
• 31:11despite their rarity,
• 31:12are heterogeneous.
• 31:13So about 80% of neuron can cells
• 31:17have adluminal like phenotype.
• 31:19They actually express androgen receptor
• 31:21remarkably and they express lumenal
• 31:24cytokeratins and then another 20% of
• 31:26your endocrine cells are basil like
• 31:29they expressed basal cytokeratins and P.
• 31:3263.
• 31:34They can be detected very early
• 31:37in organic Genesis at burn.
• 31:46Many, most perhaps all Durand Prince cells
• 31:49are actually formed at prepubertal stages,
• 31:52and since neuroendocrine cells I didn't
• 31:56mention this on previous slide are.
• 31:59But never divide there.
• 32:00They appear to be post mitotic.
• 32:03We believe that they are made and
• 32:06are not subsequently generated.
• 32:08By lineages tracing, we believe
• 32:10that they have an epithelial origin,
• 32:13so using an NCX Cree driver we can
• 32:16mark most of the prostate epithelial
• 32:19cells and in fact the vast majority of
• 32:22neuroendocrine cells are marked by NCX Creek,
• 32:26so we believe they have a
• 32:29prostate epithelial origin.
• 32:30So there's more than 95% of the
• 32:34neuroendocrine cells are marked
• 32:36in this experiment.
• 32:38Can finally neuroendocrine cells you know?
• 32:43Likely arise,
• 32:44as has been shown in the lung
• 32:47through a pro neural pathway,
• 32:49in which sort of the master regulator
• 32:53at the top of of this of the sort of
• 32:58transcription factor hierarchy is ASE L1.
• 33:00So if we delete ACL one in
• 33:04the mouse prostate,
• 33:05we can actually recover mice that
• 33:07have prostates that completely
• 33:09lack your endocrine cells.
• 33:11And yet the prostate appears to be normal
• 33:14and there is a normal proximal region.
• 33:17So we do have a rare escaper cells in
• 33:21the Periorbital region which are likely
• 33:25due to incomplete deletion by index 3.1.
• 33:29So our current model for the
• 33:31origin of your endocrine cells is
• 33:34that they likely arise from your
• 33:36original epithelial progenitor,
• 33:38although we haven't excluded
• 33:39the possibility they arise from
• 33:41an early lumenal progenitor.
• 33:44But in either case progenitor activity
• 33:46coincides with the developmental stages
• 33:49in which androgen levels are very low,
• 33:51and we're currently studying.
• 33:54The molecular properties of
• 33:56normal neuroendocrine cells.
• 33:58To understand you know in more
• 34:02detail their regulation.
• 34:04Moving on to cancer,
• 34:06we've used the NP 53 mouse model
• 34:10to generate organoid lines that
• 34:13displayed neuroendocrine phenotypes so.
• 34:17This is work from a talented
• 34:19postdoc in my lab, Jolly.
• 34:21She has used NP 53 tumors and established
• 34:24a large number of organ would lines,
• 34:27some of which have neuroendocrine features.
• 34:29As you can see here,
• 34:31this is the sort of primary tumor,
• 34:34and you'll note that it's heterogeneous.
• 34:36These are the organoids
• 34:38are established there,
• 34:39green because of the linear smirking.
• 34:42Here's a different line that you can
• 34:44see the region of small cell Histology.
• 34:47And these organoids are heterogeneous.
• 34:49You can see that they have a
• 34:53neuroendocrine region as well
• 34:54as a non ***** can region that
• 34:57is mesenchymal in nature.
• 34:58So this can be more clearly
• 35:01revealed by marker analysis.
• 35:02So here's a line that's
• 35:04relatively homogeneous.
• 35:05It expresses Synaptophysin and Chromogranin
• 35:07a so it has a neuroendocrine phenotype.
• 35:11Here's a different line that is more
• 35:14heterogeneous there that has sort of
• 35:16mixed expression of neuroendocrine
• 35:18markers as well as androgen receptors.
• 35:21To some extent,
• 35:22it actually has a double positive
• 35:25or African phenotype.
• 35:27This is a heterogeneous line that
• 35:30I showed you earlier,
• 35:31so it is it has intermixing of
• 35:34neuroendocrine cells and non your
• 35:36endocrine cells that express
• 35:38androgen receptor.
• 35:39Here is a different heterogeneous line
• 35:41again with a similar intermingling.
• 35:44So these new rendering lines,
• 35:47whether homogeneous or heterogeneous
• 35:49or highly stable during passaging,
• 35:51they can be passaged for
• 35:53more than 20 passages,
• 35:55and the heterogeneous lines will
• 35:58maintain their heterogeneity.
• 36:00So we can analyze the heterogeneity
• 36:02using single cell RNA sequencing and
• 36:04so this is the first line I showed you,
• 36:07and Interestingly the clusters
• 36:08are sort of grouped together,
• 36:10so we have an ARPU C cluster on any PC
• 36:14cluster and and and and a grouping of DN PC.
• 36:18And this is this is the
• 36:20other heterogeneous line,
• 36:22and we see a similar arrangement of clusters.
• 36:26So the heterogeneity of these organize
• 36:29is striking because it suggests that
• 36:32we've been we're able to capture
• 36:34much of the spectrum of CR PC within
• 36:39organoids established organoid lines.
• 36:41So what can we do with these organoid lines?
• 36:44Well, we can do a number of things.
• 36:47One thing is we can examine, you know,
• 36:50sort of the epigenetic marks that are
• 36:52displayed in these organoid lines.
• 36:54So for example,
• 36:55we've been pursuing cut and tag
• 36:58analysis here,
• 36:58looking at H3K27 trimethyl.
• 37:00So this is the mark deposited by PRC 2.
• 37:04And so I'll just show you just
• 37:07little this little tidbit here.
• 37:10What's interesting here is that
• 37:12actually the non neuroendocrine
• 37:14lines appear to be have a somewhat
• 37:16higher level of H3K27 trimethyl
• 37:18than the neuroendocrine lines.
• 37:20So that's something interesting that
• 37:23we are currently following up on.
• 37:26We've also been collaborating
• 37:28with Andrea Califano's laboratory.
• 37:31Which has developed a set of
• 37:34computational systems approaches
• 37:35to identify master regulators
• 37:38that drive biological processes,
• 37:40and so one of the sort of.
• 37:46Analytical approaches that they've
• 37:47developed is known as Meta Viper,
• 37:50where we can take single cell
• 37:52RNA sequencing data and analyze
• 37:55this to infer protein activity
• 37:57at the single cell level.
• 37:59So Alessandro Vasi Evo in Andre's lab
• 38:02postdoc in Andre's lab has done this,
• 38:05and again using the same organoid
• 38:07line as I showed you earlier
• 38:10here by protein inference,
• 38:12we can see again clustering of ARPU.
• 38:15See any PC and NPC.
• 38:18The the RPC cluster is elevated.
• 38:24Using an androgen receptor signal is
• 38:27is enriched for androgen receptor
• 38:30signature the any PC cluster is
• 38:32enriched for a neuroendocrine signature
• 38:34and when we can predict master
• 38:38regulators using this sort of approach.
• 38:41Notably one of the newer endocrine master
• 38:44regulators that's predicted is ACL one.
• 38:48So this is a way that we are.
• 38:51This is a method that we're employing
• 38:54to identify candidate intrinsic drivers
• 38:56of neuroendocrine differentiation
• 38:57that we're currently seeking to
• 39:00validate in functional assets.
• 39:02So.
• 39:02One of the sort of interesting
• 39:05questions we can ask is does trans
• 39:08differentiation occur at some level
• 39:11in organoid cultures and we have
• 39:13some evidence that that it might
• 39:16one way that we've been looking at
• 39:19this is using single cell a tax seek
• 39:22to examine chromatin Accessibility,
• 39:24and we can see that this is the
• 39:28same line again by single cell,
• 39:30a taxi there, seven clusters,
• 39:33and these clusters are.
• 39:34Have open chromatin at chromogranin.
• 39:36A neuron can marker these in a R and what
• 39:40you can see is that there is one cluster
• 39:43here that has open chromatin for both.
• 39:46Chrome Grande are you can
• 39:48see this more readily.
• 39:51Looking at the genomic locus so this
• 39:54cluster seven has accessible chromatin
• 39:57at both chromogranin, A&AR and so.
• 40:01This cluster, we believe,
• 40:03corresponds to a potential transitional
• 40:05population in the process of
• 40:08neuroendocrine differentiation.
• 40:10We've also been trying
• 40:12to assay this directly,
• 40:13so this is very preliminary data,
• 40:16but we can isolate non your
• 40:18endocrine cells by flow cytometry.
• 40:20Mark them with expression of RFP and
• 40:23then culture than honor in cells,
• 40:26and neurons can sell separately for
• 40:28several passages and they maintain
• 40:31their non ***** couldn't border and can
• 40:34phenotypes if we coculture them together.
• 40:36However,
• 40:37we now see their rare cells
• 40:39that that our RFP positive.
• 40:41But now express neuron could
• 40:43markers such as synaptophysin or
• 40:46Chromogranin A and interesting Lee.
• 40:48They maintain the expression of the Menton.
• 40:51So these appear to be a transitional cell,
• 40:55since phenotype seemingly corresponds
• 40:57to the what the a taxi might predict.
• 41:02OK,
• 41:02so these are some of the approaches
• 41:04that we've been employing to study
• 41:06language plasticity in prostate cancer.
• 41:09In the remaining 10 minutes or so,
• 41:11I'd like to switch over to bladder
• 41:14cancer and explain how we've
• 41:16been using organize to study
• 41:18plasticity in bladder cancer.
• 41:20So bladder cancer.
• 41:24Is of course a major health problem.
• 41:27It's quite understudied.
• 41:28The normal bladder contains again
• 41:31sort of three epithelial cell types,
• 41:34as it were basil cells,
• 41:36intermediate cells,
• 41:37and umbrella cells,
• 41:39and bladder cancer can be roughly
• 41:41divided into non muscle invasive
• 41:44disease and muscle invasive disease.
• 41:47Historically,
• 41:47these have been considered to
• 41:50be almost distinct entities,
• 41:52and it's unclear what the
• 41:54relationship actually is so.
• 41:56There are two forms of non
• 41:59muscle invasive bladder cancer,
• 42:01papillary and carcinoma insight,
• 42:03two and carcinoma insight two has
• 42:06been considered to be sort of the
• 42:08precursor to muscle invasive disease.
• 42:11However,
• 42:11there is also sorry there is also some
• 42:15evidence that papillary disease can
• 42:17progress to muscle invasive disease,
• 42:20so we've been interested in studying
• 42:23progression of bladder cancer and.
• 42:26To pursue this,
• 42:27we've actually established patient
• 42:29derived bladder tumor organoids,
• 42:31and these have been established through
• 42:33collaboration with urologists who
• 42:35perform transurethral resection's.
• 42:36So what they do is they sort
• 42:39of go in and extract,
• 42:42sort of like the tops of
• 42:44these of these tumors here.
• 42:47This is what they actually
• 42:49view through the cystoscope.
• 42:51This might look a little uncomfortable
• 42:53for men in the audience, but.
• 42:56This is how it's done and we
• 42:58take these samples and we can
• 43:01establish organoid lines in culture,
• 43:04which we can seriously passage.
• 43:06These organoids can also be grafted.
• 43:08Orthotopic Lee.
• 43:09This orthotopic grafting it uses
• 43:11ultrasound guided implantation into
• 43:13the bladder wall which is a very
• 43:16efficient process so we can readily
• 43:18interconvert organoids into Xena graphs.
• 43:20We can also take the Xeno grafts and convert
• 43:23them back to organoids, so all of these.
• 43:27Together with the parental tumor can be
• 43:31analyzed by sequencing or histopathology etc.
• 43:34So this just shows you some of the organoid
• 43:38lines that we established and what is,
• 43:41I think, evident is that the organoids
• 43:43and the Xeno grafts retained the
• 43:46characteristic characteristic
• 43:47Histology of the parental tumor.
• 43:50We also can establish organoid lines that
• 43:52have less common histological variants,
• 43:55such as squamous cell carcinoma.
• 43:57In recent unpublished work,
• 43:59we've now increased our bio bank to
• 44:02approximately 50 organoid lines.
• 44:05We've been able to establish organized
• 44:08lines from cystectomy samples,
• 44:09as well as from transferring for receptions,
• 44:13and have also a stab Liszt several
• 44:17lines that contain variant histologies.
• 44:20So in collaboration with David
• 44:22Solids Group at Memorial Sloan,
• 44:24Kettering Owen,
• 44:25the previous pathology was in
• 44:28collaborate with collaboration
• 44:29with him at all Media Memorial.
• 44:31We've analyzed these organoid lines
• 44:33molecularly using the targeted sequencing
• 44:36platform at Memorial MSK impact.
• 44:37We sequenced the organoids parental
• 44:40tumor and normal bloods and we can
• 44:43show generally that the organoids.
• 44:45Display mutational profiles that are
• 44:47concordant with that of the parental tumor.
• 44:50We can examine.
• 44:52Sort of the mutational profiles
• 44:54of these organoid lines,
• 44:56which really recapitulates sort of
• 44:58the distribution of of mutations
• 45:00in human bladder cancer.
• 45:02So we can see that among the common
• 45:05mutations we see mutations in a
• 45:07lot of epigenetic regulators which
• 45:10are frequently mutated in bladder
• 45:12cancer such as KDM 6A KMT 2C and 2D.
• 45:15As well as error 1A.
• 45:20Interestingly, we also see were also able
• 45:23to capture mutations that are relatively
• 45:25rare but interesting in bladder cancer,
• 45:28such as mutations and ERB B2 and of
• 45:31note we have very few nations in RB,
• 45:35so many bladder cancer cell lines were
• 45:38established from metastatic bladder
• 45:39cancer and contain RB mutations,
• 45:41whereas are organoids.
• 45:43Are generally established from non
• 45:46muscle invasive bladder cancer
• 45:48or earlier stages of muscle.
• 45:51Invasive disease and lack RB mutations.
• 45:55So what can we do with these organoid lines?
• 45:59One thing we can do is we can
• 46:02examine their drug response,
• 46:04and of particular interest were able
• 46:06to establish organoid lines from
• 46:08patients in a longitudinal fashion.
• 46:10So patients will often undergo
• 46:12transurethral resection, be treated,
• 46:14and then sometime thereafter,
• 46:16their tumors will unfortunately recur.
• 46:18And then we have an opportunity to
• 46:20establish another organoid line.
• 46:22So here's an example of a patient.
• 46:25And a pair of organoid lines where
• 46:28patient was not otherwise treated.
• 46:30So the tumor was removed,
• 46:32but the patient was not otherwise treated,
• 46:35and we can see in in terms of
• 46:37response to a range of different
• 46:40drugs that Interestingly,
• 46:41the organoid lines display nearly
• 46:43overlapping drug response profiles.
• 46:45However,
• 46:46in this case the different patient
• 46:48was treated with Mitomycin C and BCG.
• 46:52The tumor relapsed after over a year,
• 46:55and now the recurrent organoid line is much
• 46:59more resistant to a range of different drugs.
• 47:02However,
• 47:03displays similar responses to other drugs,
• 47:06so.
• 47:07It's of interest to us to understand how.
• 47:12Drug response has altered the
• 47:14properties of these organoids.
• 47:16So one thing we've started to do
• 47:19is to perform single cell analysis
• 47:21here of this recurrent pair,
• 47:24and Interestingly,
• 47:25the recurrent organoid line is
• 47:27actually much more heterogeneous
• 47:29than the sort of the parental of the
• 47:32order online from the parental tumor.
• 47:34And what is interesting is if you
• 47:37re aggregate these together now
• 47:40you can identify.
• 47:41A cluster that is actually in common
• 47:46between both the between the 1st
• 47:50and the 2nd organoid lines so.
• 47:54This cluster, we believe,
• 47:56corresponds to a transitional
• 47:58population and we can identify
• 48:01markers that are specific for
• 48:03this transitional population,
• 48:05so our hope here is that we can use
• 48:08utilized this pair of organoid lines
• 48:12and other examples of recurrent
• 48:15disease that we have to sort
• 48:18of replay in organoid culture.
• 48:20The events that take place during
• 48:23the emergence of treatment.
• 48:25Resistance.
• 48:28So finally I'd like to address the
• 48:31issue of tumor progression in organoid,
• 48:34so non muscle invasive disease you know
• 48:37can be classified into two lumenal
• 48:39categories as well as a basil category.
• 48:42Muscle invasive disease is more complex
• 48:44and again as I mentioned earlier,
• 48:47the relationship between these
• 48:49entities has been somewhat unclear.
• 48:51However, there is a question of whether
• 48:53you know if we have progression,
• 48:56whether there might be a switch
• 48:59and subtype specifically.
• 49:00From sort of Class 2 luminal tumors to Basel
• 49:03to a basal squamous muscle invasive tumor.
• 49:06So we believe that we can
• 49:09recapitulate this in organoid culture.
• 49:11So many of our organoid lines are
• 49:14phenotypic least able and culture.
• 49:17If you look at different markers there
• 49:20stable and organoids as organoid Xena,
• 49:22graphs ansina graph derived
• 49:25organoids however.
• 49:26Sort of a little over a
• 49:29majority of the organoids from.
• 49:30Non muscle invasive tumors displayed
• 49:33this sort of phenotypic plasticity.
• 49:35This is a they start with the luminal
• 49:38phenotype here and they become
• 49:40basil during organoid culture.
• 49:42But notably this phenotype can be
• 49:44largely reversed by xenografting,
• 49:46so we believe this there's an effect
• 49:48of the tumor micro environment
• 49:50that can repress this basil to
• 49:53luminal differentiation.
• 49:54In fact can reverse it but if we
• 49:58remove the tumors again from the.
• 50:00In a graph and culture miss organoids,
• 50:03they will again undergo the
• 50:06space little differentiation.
• 50:07So John Kristen Terrific Post
• 50:09Document Lab has pursued a small
• 50:11molecule screen to examine organoid
• 50:14lines that display plasticity to
• 50:16see if it's possible to revert it,
• 50:19and he identified one big hit here GSK.
• 50:22Else one, which is a.
• 50:25KTM one inhibitor and it is able to
• 50:30partially revert this plasticity,
• 50:32so the.
• 50:43I'm not showing this here,
• 50:45but if he knocks out KTM one day,
• 50:47he can also see this effect. So.
• 50:52He is currently investigating
• 50:54the mechanisms by which KTM
• 50:561A regulates this transition.
• 51:01In parallel, he's also been pursuing a taxi.
• 51:04Can Alesis to examine sort of the
• 51:06epigenetic States and eventually
• 51:08the epigenetic marks that are
• 51:10associated with this plasticity.
• 51:12So if you examine, for example,
• 51:14these three organoid lines,
• 51:16this is adluminal line.
• 51:17This is a basil line and this
• 51:20is what we call a plastic line.
• 51:23The plastic line looks like a basil line,
• 51:26although it started with a luminal phenotype.
• 51:29So we've performed a taxi.
• 51:31Can Alesis and we can cluster
• 51:33these so you see that the
• 51:36lumenal lines clustered together.
• 51:38The basil lines clustered together
• 51:40and these are the plastic lines which
• 51:42in the first principle component
• 51:45cluster together with basil.
• 51:46But they're separated along
• 51:48the second principle component.
• 51:52So now if we look at, you know,
• 51:55sort of genomic tracks, you can see that
• 51:59at basil markers such as keratin 14,
• 52:02that the basil lines of course
• 52:04have open chromatin and the plastic
• 52:07lines also have open chromatin,
• 52:09which of course is to be expected
• 52:12because they have a basal phenotype.
• 52:14But what's interesting
• 52:16is that lumenal markers.
• 52:19Such as gotta three.
• 52:21We we see that the plastic organoid lines
• 52:24seem to have partially open chromatin,
• 52:27so they retain an epigenetic
• 52:30memory of their lumenal origin.
• 52:33So we are currently pursuing
• 52:35studies to determine whether we can
• 52:37actually detect this epigenetic
• 52:39memory in human prostate tumors,
• 52:42which might indicate.
• 52:44A specific pacifically base of the
• 52:47basal squamous category to determine
• 52:50whether they may have in fact
• 52:53originated from more lumenal tumor.
• 52:59Finally, we're also pursuing motif
• 53:01discovery approaches to identify
• 53:04candidate transcription factors that
• 53:06might be driving this transition,
• 53:09and we're coupling this with
• 53:12other computational approaches.
• 53:14We think that this really is modeling
• 53:18something that's happening during, you know.
• 53:21Sort of transition to muscle
• 53:24invasive disease because.
• 53:25As has been noted in breast
• 53:28cancer for Andy Walt,
• 53:29what we've found is at the at the
• 53:32invasive front of these tumors.
• 53:34Now we see the expression of a basal marker
• 53:37cytokeratin 14 at the invasive front,
• 53:40but this is not expressed within
• 53:43the tumor body.
• 53:45So in closing,
• 53:46then I'd like to underscore that we think
• 53:49that these organoids represent a model
• 53:51system for studying tumor plasticity.
• 53:54So we can identify transitional
• 53:56populations in patients in order rates
• 53:58from patients with recurrent disease,
• 54:01and we think that the sort of plasticity
• 54:04we're studying culture can reflect
• 54:06processings of disease progression in vivo,
• 54:09and we're using computational
• 54:10systems approach is to identify
• 54:13the drivers of plasticity.
• 54:15And more generally,
• 54:16we believe that organoid models.
• 54:19Are incredibly useful because they
• 54:22will allow mechanistic studies
• 54:24of complex questions in cancer
• 54:27biology that may be inaccessible,
• 54:29used using other approaches.
• 54:33Of course,
• 54:34the work that I've described
• 54:36involved large team of terrific
• 54:39scientists really did all the work.
• 54:41So notably,
• 54:42Laura Crowley is a graduate student.
• 54:45My lap who led the single cell
• 54:47analysis of the prostate epithelium
• 54:50together with Francesco Comma,
• 54:52Boolean, former postdoc Moshe Botton,
• 54:54now at George Washington University.
• 54:57The work on normal neuroendocrine
• 54:59differentiation was performed
• 55:01by graduate student gave alone.
• 55:04And the prostate ***** organized by
• 55:07Jolly a terrific postdoc in the lab.
• 55:11Then former lab members soup soup
• 55:14Lee started the bladder organized
• 55:16project and I mentioned John Kristen
• 55:19is a current post Doc who's played a
• 55:23major role in continuing this project.
• 55:25We've had terrific collaborators.
• 55:27Roll rabadan Andrea Califano for
• 55:29computational systems biology.
• 55:31Jim Mckiernan,
• 55:31whole team of talented urologists
• 55:34who provided samples.
• 55:35We've collaborated with Korea Body Shen,
• 55:38in analysis of Mouse models,
• 55:40which how Lewin, epigenetic analysis.
• 55:42Need help search for pathology as
• 55:45as well as Max load at Cornell.
• 55:49For pathological analysis and
• 55:51a great group of collaborators
• 55:53at Memorial Sloan Kettering,
• 55:56David Solid Hickman to Marty
• 55:58and Barry Taylor.
• 56:00So thank you very much and I'll
• 56:02be happy to take any questions.
• 56:05Thank you very much, Michael.
• 56:07That was a wonderful talk and really
• 56:10amazing to see all of these models
• 56:12and everything that can be done
• 56:14with all of these different models.
• 56:17I would like to remind the audience
• 56:20that they can type questions in
• 56:22the chat and I will read them
• 56:25out and ask them to Michael.
• 56:27I will I will get started
• 56:30with a question Michael.
• 56:32One of the things that you see in
• 56:35the organoids and I was thinking in
• 56:38particular about the prostate cancer
• 56:41ones is there's a the heterogeneity
• 56:44and I was just wondering whether
• 56:47you see shifts in the heterogeneity
• 56:50or of the organoids themselves when
• 56:53you use different types of media.
• 56:56So can you see shifts based on
• 56:59how you grow them?
• 57:01And then I was also wondering whether
• 57:04you have applied certain types of
• 57:07therapies to those prostate cancer
• 57:09organoids and whether you see changes
• 57:12in that heterogeneity and shifts when
• 57:15you when you use different different
• 57:18types of treatments.
• 57:19OK, so Katie, that's a great question.
• 57:22Well questions so first
• 57:24question regarding media so.
• 57:26One of the things that I forgot
• 57:29to mention is that we use our own
• 57:33sort of homegrown media for all
• 57:35of these organoid experiments.
• 57:38This is a complex medium containing
• 57:42hepatocyte media and serum.
• 57:44You know we.
• 57:47Develop this years ago to grow
• 57:50mouse prostate organoids so it's
• 57:53quite different from the ENR
• 57:56based media that many groups
• 57:59use to pursue organoid assays.
• 58:02There are although.
• 58:05You might imagine that some of the
• 58:08growth factors involved are in common.
• 58:10There are undoubtedly differences
• 58:12between the media compositions in
• 58:14terms of what's actually going on.
• 58:17We know that, anecdotally,
• 58:19for bladder tumor organoids that
• 58:21it is probably easier to establish
• 58:24patient derived organoids using
• 58:26our media than in our based media.
• 58:31And we also know that you can transition
• 58:34organoid lines from one media to the other,
• 58:37but it may not be that straightforward.
• 58:40So we have some experience with DNR
• 58:43based media, but all the analysis
• 58:46that I've showed you today were done
• 58:49in our sort of homegrown media.
• 58:52So do we observe shifts in
• 58:55composition in different media?
• 58:58We have not really examined that.
• 59:02In part because it can be different,
• 59:05it can be difficult to transition
• 59:07organ lines from one medium to another.
• 59:10Be a composition to another.
• 59:13In terms of drug treatment,
• 59:15we have only started to do this with
• 59:19respect to the prostate organoids the.
• 59:22Bladder organizer, something that
• 59:24we've been examining in more detail.
• 59:28We've been particularly interested in
• 59:30mechanisms of cisplatin resistance,
• 59:32which, of course is of considerable
• 59:35translational interest.
• 59:37So those are studies that are that are,
• 59:41you know,
• 59:42currently being pursued to examine
• 59:45how cisplatin treatment alters.
• 59:48Or the phenotype,
• 59:49and perhaps the heterogeneity
• 59:51of the organized,
• 59:52but I don't have any results yet.
• 59:55To show you.
• 59:57Great, thank you.
• 59:58There's a question from Gefsky.
• 01:00:01Are have you been able to analyze human
• 01:00:04lumenal bladder tumors for plasticity,
• 01:00:07markers and correlate those
• 01:00:09results with subsequent development
• 01:00:10of muscle invasive tumors?
• 01:00:14So that's a great question.
• 01:00:16Obviously we want to extend what
• 01:00:18we've been doing in organoid
• 01:00:20culture to human specimens, so.
• 01:00:26I have to confess that this work that we've
• 01:00:28only recently gotten started, so we don't.
• 01:00:33Part of the problem is actually
• 01:00:35having a cohort of patients
• 01:00:37that's suitable for this so.
• 01:00:40Yeah, we are now in the process
• 01:00:43of trying to assemble patient
• 01:00:45cohorts where we can actually.
• 01:00:49Have samples are sort of launch eternal
• 01:00:54from patients who have progressed,
• 01:00:57say from high grade non muscle invasive
• 01:01:00disease to muscle invasive disease.
• 01:01:03Assembling these cohorts is very nontrivial.
• 01:01:06Fortunately we are part
• 01:01:09of a large collaboration.
• 01:01:12Led by Corey Body Shan,
• 01:01:14together with collaborators at Memorial
• 01:01:17Sloan Kettering and at Johns Hopkins,
• 01:01:20so collaborates Atmore Memorial
• 01:01:22include David Solid and colleagues,
• 01:01:25as well as you know,
• 01:01:28people like Jonathan Rosenberg,
• 01:01:30DeAndre Jordan, Barry Wagner,
• 01:01:32Bernie Buckner, and at Johns Hopkins.
• 01:01:35Led by David Mcconkey,
• 01:01:37Nojan and others to try
• 01:01:40to gather together the.
• 01:01:42Cohorts that are essential to
• 01:01:44address this type of question
• 01:01:46because they don't currently exist,
• 01:01:48and these types of samples are rare.
• 01:01:53Thank you we have another question
• 01:01:55from Mike Hurwitz who says great talk
• 01:01:58within the different luminal subtypes.
• 01:02:01Do you think some are more
• 01:02:03likely to develop into cancer?
• 01:02:06Any correlate in human prostates?
• 01:02:09OK, so this is a question about
• 01:02:11prostate and of course we're very
• 01:02:13interested in cell of origin.
• 01:02:15You know, we've always been sort of
• 01:02:17dissatisfied with our previous analysis
• 01:02:19of Cell of Origin because you know you
• 01:02:22have luminal cells and basil cells.
• 01:02:24There's only so much you can say,
• 01:02:26but we we think that there's still
• 01:02:29something to explore here because.
• 01:02:32This is well known phenomenon in which.
• 01:02:38You know 85 to 90% of prostate
• 01:02:41cancer patients.
• 01:02:45With sort of intermediate risk disease,
• 01:02:48you know a will actually have indolent.
• 01:02:53Prostate cancer, whereas the remaining
• 01:02:5610 to 15% of patients actually
• 01:02:59have aggressive disease and it's
• 01:03:01difficult to distinguish between the
• 01:03:04indolent and aggressive tumors and
• 01:03:06despite a lot of molecular analysis,
• 01:03:10they haven't tremendously improved over.
• 01:03:12Just simple histological police and grading.
• 01:03:15So we think that it remains
• 01:03:18possible that cell of origin could
• 01:03:21explain at least partially.
• 01:03:24The difference between indolent
• 01:03:26and aggressive disease.
• 01:03:27And so that's something we're very
• 01:03:30interested in pursuing. The.
• 01:03:31We we know already that you know from
• 01:03:35the literature that both proximal
• 01:03:38and distal luminal cells can be
• 01:03:42cells of origin in mouse models.
• 01:03:46But that does not necessarily answer
• 01:03:49the question because you know they
• 01:03:53may be different in terms of their
• 01:03:56phenotype or response to treatment or.
• 01:04:00Ultimate outcomes so you know we're in the
• 01:04:04process of pursuing these types of studies.
• 01:04:08It's not really clear what's
• 01:04:11going on in the human prostate.
• 01:04:15And again,
• 01:04:16I think we're just scratching the
• 01:04:17surface in terms of understanding
• 01:04:19the relationship between the
• 01:04:21lumenal populations in the mouse
• 01:04:23and the populations of the human.
• 01:04:25There's a lot more work that
• 01:04:27needs to be done there.
• 01:04:31Well, thank you very much Michael for
• 01:04:34this visit for this fascinating talk.
• 01:04:36I I know I it made me think of a
• 01:04:39lot of things or some parallels
• 01:04:41in in the world of lung cancer.
• 01:04:44So it was really great to think
• 01:04:47about this this so thank you very
• 01:04:49much for visiting us today and
• 01:04:51thank you everybody also who joined
• 01:04:54and have a wonderful afternoon.
• 01:04:56Well, thank you Katie. Thank you every.