Polycomb and chromatin remodeling - digital talks

On-demand webinar

Summary:

Watch the talks and panel discussion on demand from our second session of 'Epigenetics - From Bench to Clinic' held on November 9, 2020.

Hear leaders in the field present their latest research on Polycomb and chromatin remodeling. The short web-based meeting is intended to highlight exciting developments in the field and continue the sharing of knowledge within the epigenetics community.

Webinar objectives

Understand the role of polycomb in regulating of adult stem cell differentiation

Demonstrate how Polycomb group proteins influence transcriptional regulation, cell differentiation and cancer

Gain a greater understanding of the relationship between Polycomb, Inheritance and Disease.

Polycomb and chromatin remodeling

Speakers

Eneritz Agirre (Karolinska Institutet)

A primed immune epigenomic program is activated in oligodendroglia in 1 multiple sclerosis

Timestamp: 3:40- 18:50

Luciano Di Croce (Centre for Genomic Regulation)

Epigenetic regulation of adult stem cell differentiation

Timestamp: 23:44- 51:58

Kristian Helin (Memorial Sloan kettering Cancer Center)

Role of Polycomb group proteins in transcriptional regulation, differentiation and cancer

Timestamp: 52:02- 01:33:29

Danny Reinberg (New York University)

Polycomb, Inheritance and Disease

Timestamp: 01:33:30- 02:07:22

Moderator

Ali Shilatifard(Northwestern University)

Contact

Have a question? Please contact the Abcam Events Team.

Video Transcript

00:00 - 00:11: Hi, everybody.
00:11 - 00:16: On behalf of Abcam, I'd like to welcome you to the second session in our Epigenetics from
00:16 - 00:17: Bench to Clinic event.
00:17 - 00:22: With the postponement of many epigenetic conferences because of this pandemic, we are pleased to
00:22 - 00:25: provide you with this digital platform to support the epigenetics community and help
00:25 - 00:28: you stay connected and share your amazing work.
00:28 - 00:33: Today's session will focus on Polycomb and chromatin remodeling, and we are pleased to
00:33 - 00:38: bring you a really great, exciting program, including five talks and a panel discussion.
00:38 - 00:40: This session will be chaired by Ali Shilatifard.
00:40 - 00:46: So, Ali is the chairman, Department of Biochemistry and Molecular Genetics, and director, Simpson
00:46 - 00:52: Querrey Center for Epigenetics Chicago, Illinois, and is a world-renowned biochemist and molecular
00:52 - 00:56: biologist and a respected expert in the field of transcription and epigenetics.
00:56 - 01:01: So, Ali made a seminal contribution to the field of leukemia biology by identifying the
01:01 - 01:06: first function of any of the MLL translocation partners.
01:06 - 01:10: Since then, he has continued to contribute to the field of transcription, chromatin biology,
01:10 - 01:17: and cancer biology through many discoveries, including those of the COMPASS and SEC complexes,
01:17 - 01:20: which are demonstrated to be central to the mechanisms involved in the development of
01:20 - 01:21: leukemia.
01:21 - 01:25: Currently, inhibitors being developed in Ali's laboratory towards the COMPASS family
01:25 - 01:30: and the SEC are being tested in the treatment of childhood leukemia, brain cancer, and triple
01:30 - 01:32: negative breast cancer tumors.
01:32 - 01:37: So, for his contributions to our understanding of cancer biology, Ali has been recognized
01:37 - 01:42: by the Leukemia and Lymphoma Society as the recipient of the Sword of the American Cancer
01:42 - 01:48: Society and the Amgen Award by the American Society of Biochemistry and Molecular Biology.
01:48 - 01:50: With that, I am thrilled to hand this over to Ali.
01:50 - 01:51: Great.
01:51 - 01:53: Can you hear me?
01:53 - 01:54: Yes.
01:54 - 01:55: Okay, perfect.
01:55 - 01:57: Well, good morning, everyone.
01:57 - 02:03: Welcome to the session on Polycomb and Polycomb function in gene regulation and its mutations
02:03 - 02:04: involving cancer.
02:04 - 02:10: We have a fantastic session with the world leader today talking to us about the area
02:10 - 02:14: and we hope to hear a lot about, you know, the function of the marks and how Polycomb
02:14 - 02:15: functions.
02:15 - 02:20: As you know, Polycomb was characterized in Drosophila melanogaster over 70 years ago
02:20 - 02:25: and it was shown that mutations in Polycomb resulted in issues in cell identity.
02:25 - 02:31: We all have taken Biology 101 and have seen the fly with a leg grow in its eyes and those
02:31 - 02:33: are the mutations in Polycomb.
02:33 - 02:37: Those are mutations that the eye cells think they're leg cells and they grow a leg
02:37 - 02:39: in a place they shouldn't be.
02:39 - 02:45: And I think Polycomb has turned out to be important, central in functioning with Trithorax regulation
02:45 - 02:50: of development and what we have learned through cancer genome sequencing and sequencing on
02:50 - 02:55: many different forms of cancer, that mutations in the Polycomb and Polycomb components, including
02:55 - 03:02: PRC1 and PRC2, are involved in a large number of human diseases and a series of talks today
03:02 - 03:07: are going to tell us about the biochemistry, the biology, and the genetics of Polycomb during
03:07 - 03:08: development.
03:08 - 03:13: And I hope you hear a lot about their function in the pathogenesis of the disease as well.
03:13 - 03:17: Many moons ago, beautiful work from Danny Reinberg's lab, Yi Zhang's lab, and Bob
03:17 - 03:22: Kingston's lab characterized the function of these Polycomb families, especially PRC2
03:22 - 03:26: and EZH as a histone H3K27 methyltransferase.
03:26 - 03:30: And during the past decade, we are trying to figure out as a community what is the relevance
03:30 - 03:36: of these modifications in epigenetic memory, in the regulation of gene expression, and what's
03:36 - 03:41: the difference between the context of the enzyme versus the modification in these processes.
03:41 - 03:45: And I'm sure our speakers today are going to be touching upon many of these areas in
03:45 - 03:46: here.
03:47 - 03:51: So we'll start the first session here or first talk in here with Eneritz Agirre.
03:51 - 03:57: Eneritz was a PhD student in Barcelona, and she did fantastic work during her PhD work
03:57 - 04:02: and as a postdoctoral fellow, she moved on to France and Montpellier and has studied
04:02 - 04:06: epigenetic regulation of alternative splicing.
04:06 - 04:12: She has started her own laboratory at the Curie Institute in Karolinska in Sweden, and her
04:12 - 04:16: focus is on the study of transcriptional and chromatin regulation in oligodendrocyte
04:16 - 04:20: lineage cell determination and diseases such as multiple sclerosis.
04:20 - 04:27: And she's using single-cell omics and methodologies to define the role of Polycomb in this process.
04:27 - 04:31: So without further ado, we'll go ahead and start with the first talk.
04:31 - 04:32: Okay.
04:32 - 04:35: Thanks, Ali, for the introduction.
04:35 - 04:36: Yeah.
04:36 - 04:39: Thanks for the invitation to this meeting of Abcam.
04:39 - 04:44: Yeah, as he mentioned, I'm a postdoctoral fellow at the Gonzalo Castelo Branco Group
04:44 - 04:47: at the Karolinska Institute in Stockholm, Sweden.
04:47 - 04:52: And here I'm going to present our results from studying the oligodendrocyte lineage
04:52 - 04:58: and the immune regulation modulated by the chromatin.
04:58 - 05:03: In our group, we study oligodendrocyte lineage cells, which are the glial
05:03 - 05:08: cells from the central nervous system that insulate neuronal axons, and they improve the
05:08 - 05:10: speed and efficiency of this impulse.
05:10 - 05:15: So we are studying multiple sclerosis, which is a demyelinating disease, where the myelin
05:15 - 05:21: is attacked by the immune cells, and this will distract it, and depending on the stage
05:21 - 05:26: of the progression of the disease, we'll have stages that can be relaxing with
05:26 - 05:31: our myelination from the precursor oligodendrocytes, but in other states, this is chronic and will
05:31 - 05:34: lead to failure of the system.
05:34 - 05:40: So in our group, we study these oligodendrocytes by using single-cell RNA or single-cell technologies
05:41 - 05:45: and we have already described at the RNA level the heterogeneity of the whole lineage in
05:45 - 05:49: the mouse brain and spinal cord during development.
05:49 - 05:54: We define different heterogeneous groups of oligodendrocyte cell types.
05:54 - 06:00: More recently, we're using a model, a model for multiple sclerosis, where we actually,
06:00 - 06:07: we also perform single-cell RNA sequencing in a smart-seq, and we could identify oligodendrocyte
06:07 - 06:11: subtypes specific to the disease and from the control.
06:11 - 06:17: More intensely here, we found a new function for these oligodendrocytes,
06:17 - 06:20: which would be the expression of immune genes.
06:20 - 06:27: What we saw was that disease-specific oligodendrocytes were expressing MHC class I and MHC class
06:27 - 06:34: II genes, and this we were able to validate or see in vitro in OPC cell lines that there
06:34 - 06:38: was a phagocytic function for these oligodendrocytes.
06:38 - 06:44: So we have that oligodendrocytes classically have a function of insulating axons, more
06:44 - 06:50: in the attentions, but here we found that they can also have new properties like immune properties.
06:50 - 06:55: We also have data from humans showing that this is also present and occurs in
06:55 - 07:03: the human brain, and this new function leads to a phagocytic function of these oligodendrocytes
07:03 - 07:04: and their precursors.
07:04 - 07:10: So our question now was to define how this transition from the function to immune states
07:10 - 07:15: can happen, and if this is reflected at the chromatin level, and in this case, if we are
07:15 - 07:20: able to identify specifically the oligodendrocytic states in the context of the disease.
07:20 - 07:26: So for this, we performed single-cell ATAC sequencing on an EAE model in the SOX
07:26 - 07:31: 10-CRE-EFP short mice, where we then recollected the cells at the peak of the disease, and
07:31 - 07:36: then performed 10x genomics for the single-cell ATAC to capture the chromatin accessibility of
07:36 - 07:38: these individual cells.
07:38 - 07:42: So the first thing is the clustering, and it's done from the RNA expression from the
07:42 - 07:44: single-cell RNA.
07:44 - 07:49: In this case, the clustering or identification of the cell types is based on the accessibility,
07:49 - 07:55: and for this, we define promoter regions for each of the individual cells, like the promoter
07:55 - 08:00: regions, and we calculate a gene activity value that is similar to gene expression
08:00 - 08:07: in order to cluster and get similar cells or clustering from the data.
08:07 - 08:13: As we saw at the RNA expression level, we also see that the clusters and the open chromatin
08:13 - 08:18: level are specific to the disease or the context, so there is a separation between
08:18 - 08:20: the two conditions.
08:20 - 08:25: The way to assign the cells, in order to know which specific cell type belongs, is that we use
08:25 - 08:31: our previous dataset on single-cell RNA sequencing, and we use it as a reference
08:31 - 08:37: to name and label the data in the single-cell ATAC-seq, so this would assume a correlation
08:37 - 08:39: between the expression and the chromatin accessibility.
08:39 - 08:45: This allows us to identify mature oligodendrocytes, precursor cells, and also microglia in our
08:45 - 08:46: dataset.
08:46 - 08:53: Fast checks would be to look for specific canonical markers of the lineage, that this is plotting
08:53 - 08:58: the accessibility variability over these single cells in the visualization, and if
08:58 - 09:03: we look for PTPRC, which is a marker of precursor cells, we see clearly that it is opening
09:03 - 09:04: in the precursor cells.
09:04 - 09:09: MOC is canonical for mature oligodendrocytes, opening in the mature, and then the
09:09 - 09:15: microglia markers, which also show this opening in the immune-specific cell types of the mature
09:15 - 09:19: ones, but also microglia and some of the precursors.
09:19 - 09:24: The gene enrichment analysis clearly showed that these EAE open regions are specifically
09:24 - 09:25: enriched for immune properties.
09:25 - 09:31: This is how the data looks like in a genome browser, so each line belongs to
09:31 - 09:37: a single cell, and we have the different clusters we identify, like from precursors and matures.
09:37 - 09:43: What we already see here, assuming two immune genes, is that already the promoter
09:43 - 09:48: regions of these immune genes show priming on opening the chromatin for the oligodendrocytes,
09:48 - 09:53: both for the precursors, and also as a control, we have the microglia that show this opening,
09:53 - 09:58: and in the mature ones.
09:58 - 10:02: So another thing we wanted to check is, in collaboration with Howard Chang
10:02 - 10:11: from Stanford, we wanted to use the available data from MS-associated SNPs, and
10:11 - 10:17: what we use in this case is single-cell ATAC-seq from human controlled samples, that we have
10:17 - 10:23: already defined the specific cell types, specific regulatory regions from precursors, mature
10:23 - 10:28: types, and also from microglia, and we see that we find an association and intersection
10:28 - 10:34: between these SNPs associated with MS and these oligodendrocytic open regions and microglia.
10:34 - 10:40: Moreover, when we check with our mouse data and transport it, we see that this intersection
10:40 - 10:41: is common.
10:41 - 10:46: We could suggest that these regulatory regions are associated and could have an effect on
10:46 - 10:49: the onset of the disease.
10:49 - 10:56: So we saw in our previous results that one of the causes of this pro-immune program to
10:56 - 11:00: be activated could be the induction of interferon gamma.
11:00 - 11:04: So this, in this, what we actually show is that upon treatment of interferon
11:04 - 11:11: gamma, immune genes showed an increasing expression that was also validated by RNA scope.
11:11 - 11:19: So this made us ask if this involvement of the interferon gamma-mediated activation
11:19 - 11:21: could be related to chromatin modulation.
11:21 - 11:28: So for this, we performed biotin ligase-seq and biocatalytic seq in genome Y to see all the effects
11:28 - 11:33: in the program interferon gamma treatment genome Y, in this case, in this in vitro oligodendrocyte
11:33 - 11:34: precursor test.
11:34 - 11:39: And what we can see is that we have a major upregulation upon the treatment of interferon
11:39 - 11:40: gamma.
11:40 - 11:43: We see many immune genes that are expressed.
11:43 - 11:49: And if we compare the work in the ATAC-seq, we see that there is also an opening or differential
11:49 - 11:53: accessibility more open upon the treatment, but these changes were lower.
11:53 - 11:59: Anyway, when we look at the specific immune genes, in this case from MHC class II and MHC
11:59 - 12:05: class I, we actually show that here we have the control and this is from the treatment.
12:05 - 12:10: We see that in the case of RSA, we have this higher expression and already an opening on
12:10 - 12:16: the chromatin in these immune genes, and this can be seen in this capture.
12:16 - 12:22: So how can, what could be the reason or what could be the mechanism behind this
12:22 - 12:28: upregulation of immune genes upon interferon gamma treatment?
12:28 - 12:33: For this, what we perform is sequencing using CUT&RUN, which would be similar to ChIP-seq
12:33 - 12:35: and lower cells are more specific.
12:35 - 12:41: We have H3K27me3 methylation and H3K4me3 methylation.
12:41 - 12:45: And in this case, what we actually saw was that upon treatment of interferon gamma,
12:45 - 12:53: we saw a general decrease in the K27me3 methylation with an increase in H3K4me3 methylation,
12:53 - 12:54: which are these reversed.
12:54 - 12:58: And in this case, we actually saw this increase in the immune genes that we were seeing at
12:58 - 13:00: the expression level.
13:00 - 13:06: So if we go to specific examples of these MHC class I or MHC class II genes in this locus,
13:06 - 13:11: what we see is that we have here the ATAC-seq, the opening, the chromatin, and upon treatment,
13:11 - 13:17: we have an opening of the promoters with a higher level of differential accessibility increases.
13:17 - 13:23: The H3K27 will decrease in the treatment of the interferon gamma with a major increase
13:23 - 13:27: in the signal of H3K4me3 methylation.
13:27 - 13:34: But is this H3K27me3 methylation sufficient to explain this immune gene upregulation?
13:34 - 13:41: In this case, we went to check, and what we did was to look for EZH2 inhibition.
13:41 - 13:48: That is the enzyme responsible for the deposition of H3K27me3 methylation.
13:48 - 13:55: So we performed an assay where we were inhibiting EZH2 in control and upon interferon gamma
13:55 - 13:56: treatment.
13:57 - 14:04: And here you see the data from the CUT&RUN, the signal from the H3K27me3 methylation.
14:04 - 14:11: Upon treatment, we actually can see that we have a general decrease of the H3K27me3
14:11 - 14:12: methylation.
14:12 - 14:17: When we compare to this control and upon the treatment for the RNA-seq at a spatial level,
14:17 - 14:18: this was increased.
14:18 - 14:26: Also, when looking at specific QRT-PCR of specific immune genes upon the treatment
14:26 - 14:32: of EZH2 inhibition, we saw a general increase in this interferon gamma treatment, which
14:32 - 14:38: will also mean the group of genes that are pre-related genes with the inhibition of EZH2.
14:38 - 14:46: EZH2 saw an immune gene authority component that was significantly enriched, which suggests
14:46 - 14:53: that H3K27me3 methylation removal, together with the treatment of interferon gamma,
14:53 - 14:58: could promote this immune gene program in our oligodendrocyte precursor cells.
14:58 - 15:04: So, we have a transcriptional regulation at the chromatin level of the immune role for
15:04 - 15:09: the oligodendroglia in disease, that this immune role or this function.
15:09 - 15:13: The oligodendroglia can express these immune genes in the context of disease.
15:13 - 15:18: We also show that the immune genes show this primary susceptibility of the oligodendroglia
15:18 - 15:24: in the promoter regions that can be validated at the single-cell level and also using biotin sequencing.
15:24 - 15:30: We have a subset of MS-susceptible SNPs that are located in this oligodendrocyte-specific
15:30 - 15:36: operant chromatin region, suggesting that it could affect the onset of the disease.
15:36 - 15:42: And this interferon gamma plays a role in this transition from the classical
15:42 - 15:46: function of the oligodendrocyte to immune states in the context of disease.
15:46 - 15:52: And this interferon gamma induces the decrease of H3K27 demethylation and this immune gene
15:52 - 15:56: promoters for oligodendrocytes or oligodendroglia sets.
15:56 - 16:03: So, yeah, we would like to thank Goncalo Castelo-Branco, all the group, this work being
16:03 - 16:08: done together with Mandy Meijer, who has worked on the experimental part, and our collaborators
16:08 - 16:13: from Istanbul, Howard Cheng, and all the data available through bioRxiv.
16:13 - 16:14: So, yeah.
16:14 - 16:15: Thanks a lot.
16:15 - 16:18: And if you have any questions, yeah.
16:18 - 16:19: All right.
16:19 - 16:22: Well, thank you so much for that great talk.
16:22 - 16:24: We'll start with a couple of questions.
16:24 - 16:30: One question that came in online is that how can you translate these findings into a model
16:30 - 16:33: in humans, into a mouse in the human?
16:33 - 16:38: So the disease you see in humans, how can you translate into mouse and vice versa?
16:38 - 16:39: Yeah.
16:39 - 16:44: Actually, now currently we have also data from human samples.
16:44 - 16:50: So we are also performing single-cell RNA-seq from our CARBA datasets or brain banks.
16:50 - 16:57: So we already verified that we have something that these immune genes are also expressed in
16:57 - 16:58: the human.
16:58 - 17:05: And a way to kind of translate it would be like doing another different phase more based
17:05 - 17:06: on the human data.
17:06 - 17:07: Yeah.
17:07 - 17:08: Okay.
17:08 - 17:14: And the second one is, is this immune function of the oligodendrocyte present in other diseases apart
17:14 - 17:15: from multiple sclerosis?
17:15 - 17:18: Is it a specific disease or a general function?
17:18 - 17:19: Yeah.
17:19 - 17:27: We actually saw that it could be any of the genetic diseases that have an important
17:27 - 17:28: immune component.
17:28 - 17:34: And it's the case, for instance, of Alzheimer's or even during aging, this immune component
17:34 - 17:35: is increased.
17:36 - 17:43: So this is also evidence that will affect the oligodendrocytes in this kind of neurodegeneration.
17:43 - 17:44: Yeah.
17:44 - 17:50: And are you able to use some of this CRISPR screening system to ask for the suppressors
17:50 - 17:52: or enhancers of the phenotype?
17:52 - 17:53: Yeah.
17:53 - 17:54: Actually, yes.
17:54 - 17:57: We have some, we'll be like first testing candidates.
17:57 - 18:02: We have some transcription factors or specific modifiers that we could choose.
18:02 - 18:08: And the next step will be followed by Cas9 assays to really mutate and see how this
18:08 - 18:09: is modulated.
18:09 - 18:15: We have another question coming from Andrea Punti who's asking, do you know if the oligodendrocytes
18:15 - 18:22: increase their antigen presentation ability on EZH2 inhibitors?
18:22 - 18:28: We didn't, yeah, because we've been testing, we tested them in vitro.
18:28 - 18:33: Actually, we know that they're actually presenting and this is the function of phagocytosis.
18:33 - 18:39: So I would say yes, but well, we need to still, because we are ongoing with all the assays
18:39 - 18:42: from the Interferon Gamma and also the inhibitions.
18:42 - 18:47: We have some candidates, but not still, but I would say yes.
18:47 - 18:48: All right.
18:48 - 18:50: Well, there are no more questions.
18:50 - 18:52: We're going to move on to the next speaker.
18:52 - 18:53: Right.
18:53 - 18:58: So some really great talks and a really good discussion.
18:58 - 19:01: I'm just going to use this opportunity to thank you all on behalf of Abcam for joining
19:01 - 19:02: the session.
19:02 - 19:07: For those of you who've just joined, my name is Vinit Bharara and I'd like to take you
19:07 - 19:13: through some of the work we're doing here at Abcam to support the work in the Polycomb
19:13 - 19:16: and chromatin remodeling field.
19:16 - 19:21: And at the moment, epigenetics is one of our core focus areas.
19:21 - 19:28: We like to ensure that the important role it plays is given due attention by us
19:28 - 19:34: and on a more general level as well, we make sure we produce really well-validated antibodies
19:34 - 19:40: to Polycomb, to the Polycomb complex, and the different enzymes involved and accessory components.
19:40 - 19:47: So we make sure at this stage, we do a lot of validation such as knockout validation
19:47 - 19:51: and we make sure that the antibodies are recombinant so you can get reproducibility and that
19:51 - 19:52: selectivity.
19:52 - 19:58: We are also constantly innovating and we seek out opportunities to work with researchers
19:58 - 19:59: like yourselves.
20:00 - 20:04: and collaborate in emerging areas of epigenetics.
20:04 - 20:05: In the next few slides, I'll take you
20:05 - 20:07: through some of the specific activities,
20:07 - 20:11: some of the content we produced and the efforts
20:11 - 20:14: we do to provide support to the epigenetics community.
20:15 - 20:18: So we've created something like a hub page
20:18 - 20:21: which can help you navigate through your epigenetic content
20:21 - 20:23: and you can quickly access the most relevant materials
20:23 - 20:26: and that's here that you can see on your screen.
20:26 - 20:30: You can get to this page just by scanning the QR code
20:30 - 20:31: and you can find out more there.
20:31 - 20:34: And it covers a range of topics, right, from DNA and RNA
20:34 - 20:38: and chromatin remodeling, histone-mediated epigenetics
20:38 - 20:39: and so on.
20:39 - 20:42: So another thing I wanted to mention
20:42 - 20:44: is that we are quite proud of the guides
20:44 - 20:47: and the tools that we put together for researchers.
20:47 - 20:51: This is a great example here of our epigenetics application guide
20:52 - 20:55: which is not just for people who are just starting off in the field,
20:55 - 20:59: but even for some of you who do a fair amount of epigenetics.
20:59 - 21:01: There might be tools and tips and tricks in here
21:01 - 21:03: that you could use to start your next assay
21:03 - 21:07: or find out the answer to your next question.
21:07 - 21:09: And it can help you get started
21:09 - 21:12: and also refine your skills in those techniques.
21:12 - 21:15: And this is a guide written by epigeneticists for epigeneticists.
21:15 - 21:18: So you'll find it quite helpful if you haven't seen it already.
21:18 - 21:19: Most recently, for example,
21:19 - 21:21: we've introduced CUT&RUN and CUT&TAG
21:21 - 21:23: and it's a great chapter to get you started on these techniques.
21:23 - 21:25: if you haven't already.
21:26 - 21:28: Coming to the kinds of products we make,
21:28 - 21:31: as I said, we constantly innovate,
21:31 - 21:35: try to develop reagents in the most innovative and emerging areas.
21:36 - 21:40: Here's an example of coverage that we try to have
21:40 - 21:43: for the histone H3 mutant antibodies,
21:43 - 21:47: especially those of you working in glioma, AML, and sarcoma
21:47 - 21:48: might find this quite useful
21:48 - 21:51: where these mutants are seen in such cancers.
21:52 - 21:59: So the mutants are the H3K27M, K36M, and the G34W/V/R.
21:59 - 22:02: All these antibodies that we have to these mutants
22:02 - 22:05: are recombinant monoclonals and validated in IHC
22:05 - 22:06: and mutant tissues.
22:06 - 22:09: So they work, and IHC is one of the key techniques.
22:09 - 22:12: Of course, some of them work in SIP, etc.,
22:12 - 22:14: but a lot of people who work in these mutations
22:14 - 22:17: look at IHC as a key application.
22:18 - 22:20: So now these, again, as I said,
22:20 - 22:25: can be used in classifying and characterizing different cancers,
22:25 - 22:28: both the low and high-grade types of cancers.
22:28 - 22:29: So if you want to find out more, again,
22:29 - 22:31: you can scan the QR code and get straight to the page
22:31 - 22:33: or you can search for these on our website.
22:34 - 22:37: Right. Coming to the world of Polycomb,
22:37 - 22:38: just an example here,
22:38 - 22:42: we constantly release new and new antibodies
22:42 - 22:43: to the Polycomb field.
22:43 - 22:48: Here's an example of KMT2D or MLL2 that we recently launched.
22:49 - 22:51: And as you can see, some others, RBBP4,
22:51 - 22:54: P7, some are knockout validated,
22:54 - 22:56: which means we've knockout tested it in a knockout cell line.
22:56 - 22:59: So you can see, as you can see, the western blot
22:59 - 23:00: is a band that's missing.
23:00 - 23:03: That's a knockout cell line, so it's specific to that target.
23:04 - 23:08: And yeah, watch out on our website or look for a target.
23:08 - 23:11: We constantly release new antibodies.
23:11 - 23:13: There's also a page that you can go to,
23:13 - 23:16: which you'll see coming up in your chat box as well.
23:16 - 23:17: This is a new products page. It's something new
23:17 - 23:20: that we've done at Abcam, and you can see
23:20 - 23:22: that these are some of the new products
23:22 - 23:26: we've made in the Polycomb field, like CHD4 and the BAF170.
23:27 - 23:30: And you can stay up to date with these new products
23:31 - 23:33: that we keep updating on this page.
23:33 - 23:36: So if there's anything or if you want to collaborate with us
23:36 - 23:38: or you want to test some of these in your lab
23:38 - 23:40: or you would like some samples, just email us
23:40 - 23:42: at epigenomics@abcam.com,
23:42 - 23:44: and we'll get back to you directly.
23:44 - 23:47: All right, so Luciano is a senior group leader
23:47 - 23:49: at the Center for Genomic Regulation,
23:49 - 23:54: CRG, in Barcelona, and he's also an elected member of EMBO,
23:54 - 23:58: and so it's an honor to have him here for giving his talk.
23:58 - 24:03: He started his work studies as a PhD student.
24:03 - 24:06: He's Italian, so although he spent most of his time
24:06 - 24:08: in Spain, he's still Italian,
24:08 - 24:10: and he got his degree from Rome,
24:10 - 24:12: La Sapienza, the university,
24:12 - 24:16: and his research interest for the past 20 years,
24:16 - 24:18: I met Luciano almost 20 years ago.
24:19 - 24:21: For the past 20 years, he has been focusing
24:21 - 24:24: on understanding how epigenetic modification of chromatin
24:25 - 24:28: and changes in the process establish in place
24:28 - 24:30: and how they can affect gene expression,
24:31 - 24:33: chromatin structure, and more recently,
24:33 - 24:36: the three-dimensional structure of chromatin in the process,
24:36 - 24:38: and a disease of choice in his lab
24:38 - 24:41: that has been studying the role of chromatin modification
24:41 - 24:42: and chromatin structure is leukemia,
24:43 - 24:45: and he has been looking at leukemia
24:45 - 24:47: as a cancer model to understand the process
24:47 - 24:50: and how chromatin and chromatin modification,
24:50 - 24:52: both on the Polycomb and on the Trithorax
24:52 - 24:54: functions in this process.
24:54 - 24:58: He wrote a very beautiful review of the Polycomb and Trithorax
24:58 - 25:02: with Cavalli about a year and a half ago in Cell,
25:02 - 25:05: which really nicely covers the dynamic
25:05 - 25:08: of the Polycomb and Trithorax for almost 70 years
25:08 - 25:10: of the studies of genetics and biochemistry
25:10 - 25:12: that have been in that process.
25:12 - 25:15: So, his training was done in Germany,
25:15 - 25:17: but I don't think he speaks a lick of German.
25:17 - 25:20: He was in Marburg for many years as a postdoctoral fellow,
25:20 - 25:24: and then moved back to Milan to do a second postdoc study,
25:24 - 25:27: and from Milan, he moved to CRG in Barcelona,
25:27 - 25:30: where he's been since and is an ACRE professor
25:30 - 25:34: and highly cited and very collaborative
25:34 - 25:35: and a wonderful individual.
25:35 - 25:37: So, with that, Professor DiCroce.
25:39 - 25:41: Thanks a lot, Ali.
25:41 - 25:44: Beautiful introduction. Can you hear me?
25:44 - 25:48: Yeah. So, thanks for the invitation
25:48 - 25:51: to share our results with you today.
25:51 - 25:54: And today, I will talk about the role of,
25:54 - 25:56: just moving this window on the other side.
25:59 - 26:03: The role of genetic regulation in stem cell differentiation,
26:03 - 26:06: particularly I will focus on the role of the PRC2 complex
26:06 - 26:08: in metabolic stem cells.
26:08 - 26:12: It was mentioned by Ali, Polycomb and Trithorax
26:12 - 26:15: were initially identified in Drosophila
26:15 - 26:16: more than 70 years ago,
26:16 - 26:18: but these two multiprotein complexes,
26:18 - 26:21: they are conserved from Drosophila to mammals.
26:21 - 26:23: Not only the function, but also the structure,
26:23 - 26:26: the composition, and the target genes are conserved.
26:26 - 26:27: Of course, in the human genome,
26:27 - 26:30: there are many more target genes for Polycomb and Trithorax
26:30 - 26:33: because they are given to this function of our genome.
26:34 - 26:36: Now, an additional difference is that in mammalian cells,
26:36 - 26:39: the Trithorax complex is called COMPASS MLL complexes.
26:40 - 26:43: And as I said, these two multiprotein complexes
26:43 - 26:46: are important for cell differentiation
26:46 - 26:49: and animal development, but also, as was mentioned by Ali,
26:49 - 26:53: they are often misregulated in several human pathologies,
26:53 - 26:55: including in human cancers.
26:56 - 26:58: And Polycomb comes in two flavors,
26:58 - 27:01: the PRC1 complex and the PRC2 complex.
27:01 - 27:06: Mechanistically, this is the classical modus operandi
27:06 - 27:08: or canonical modus operandi of the Polycomb
27:08 - 27:10: repressive complex; the PRC2 complex
27:10 - 27:13: leads to the deposition of the trimethyl mark
27:13 - 27:15: on lysine 27 of histone H3,
27:15 - 27:17: which then becomes a docking site
27:17 - 27:19: for the other Polycomb complex,
27:20 - 27:21: the canonical PRC1 complex,
27:21 - 27:24: which, through the CBX protein, can recognize this mark
27:24 - 27:28: and bind to the place where PRC2 deposits the mark.
27:28 - 27:30: Now, the presence of both PRC1 and PRC2
27:30 - 27:33: at a promoter leads to gene silencing,
27:33 - 27:37: which leads to maintaining the repressive state of the promoter,
27:37 - 27:39: but also leads to chromatin compaction,
27:39 - 27:41: repulsion of this RNA polymerase machinery,
27:41 - 27:44: and other mechanisms
27:44 - 27:46: which have been recently able to detect.
27:47 - 27:50: Now, as I mentioned, this is the canonical way,
27:50 - 27:53: but if you look at the canonical modus operandi,
27:53 - 27:57: the PRC1 complex follows the steps of PRC2 complex,
27:57 - 27:59: given the fact that it recognizes the modification
27:59 - 28:00: which is deposited by the PRC2 complex.
28:00 - 28:03: Then the question is, what is targeting the PRC2 complex
28:03 - 28:05: to specific places of the genome?
28:06 - 28:10: What we know from many published recent data,
28:10 - 28:13: is that the PRC2 complex can interact with RNA,
28:13 - 28:17: which can prevent this association to chromatin,
28:17 - 28:20: this specifically occurs at active genes,
28:20 - 28:22: or vice versa, can also facilitate
28:23 - 28:24: the engagement of the PRC2 complex
28:24 - 28:27: to specific regions of the genome.
28:27 - 28:31: And that's the function of many long non-coding RNAs.
28:31 - 28:36: But the PRC2 components can also interact with specific factors,
28:36 - 28:38: like for example, PCL protein,
28:38 - 28:40: which will probably provide recognition
28:40 - 28:42: of this modification, but also the affinity for DNA,
28:42 - 28:47: and particularly for stretches of unmethylated CpG dinucleotides.
28:47 - 28:49: So the binding of PCL protein
28:49 - 28:51: to core components of the PRC2 complex
28:51 - 28:55: leads to the recruitment of the PRC2 complex to CpG islands.
28:55 - 28:57: But PCL proteins are not the only proteins
28:57 - 28:59: which can interact with PRC2;
28:59 - 29:01: there are other proteins that have been characterized,
29:01 - 29:04: and another classical example is JARID2,
29:04 - 29:08: which brings the PRC2 complex to different places in the genome,
29:08 - 29:10: where there is actually interesting close talk
29:10 - 29:12: between PRC1 and PRC2,
29:12 - 29:14: which is mediated by H2A monoubiquitination,
29:14 - 29:17: which I don't have time to discuss here.
29:17 - 29:19: Basically, the data from our lab and many other labs
29:19 - 29:22: lead to the identification and classification
29:22 - 29:25: of the PRC2 complex in two main variations,
29:25 - 29:30: the PRC2.2 complex, which contains JARID2 and APP2,
29:30 - 29:33: and the PRC2, sorry, PRC2.1,
29:33 - 29:35: which contains PCL proteins
29:35 - 29:41: together with the EPOP or Polycomb proteins.
29:41 - 29:46: Now, what's the balance between these two complexes
29:46 - 29:48: and whether they have different characteristics?
29:48 - 29:50: This is not clear at the moment.
29:50 - 29:52: And for example, what we know is that the PCL proteins,
29:52 - 29:56: as I mentioned before, can target the PRC2 specific places.
29:56 - 29:58: But what are the PCL family members?
29:58 - 30:01: Let me give you just a few additional pieces of information here.
30:01 - 30:05: The PCL family of proteins is composed of three members,
30:05 - 30:08: PCL1, PCL2, and PCL3,
30:08 - 30:10: also called, this is the formal name,
30:10 - 30:14: PHF1, MTF2, and PHF19.
30:14 - 30:19: We recently elucidated the role of PHF19,
30:19 - 30:20: also called PCL3.
30:20 - 30:22: I'm going to talk about this a bit more in detail.
30:22 - 30:24: But as you can see, all of them contain
30:24 - 30:27: these very interesting set of chromatin-related domains,
30:27 - 30:30: trithorax domain, PHD domain, actually two of them,
30:30 - 30:31: and a plant homeodomain.
30:31 - 30:34: PCL3 is particularly interesting
30:34 - 30:38: because it also has a short isoform,
30:38 - 30:40: which is called PHF19 short isoform,
30:40 - 30:42: which lacks some of the key domains
30:42 - 30:44: indicated which are missing here.
30:45 - 30:48: Those domains are important for the function,
30:48 - 30:50: actually, for the binding to the core component of PRC2,
30:50 - 30:52: which means this protein does not interact
30:52 - 30:54: with the core component of the PRC2 complex.
30:54 - 30:56: Now, the interesting thing is that the ratio
30:56 - 30:58: between the long and the short isoform,
30:58 - 31:00: this ratio leads to isoform,
31:00 - 31:03: dramatically change between normal cells and cancer cells.
31:03 - 31:05: So what happens if we break this balance
31:05 - 31:06: between these two complexes?
31:06 - 31:09: For example, if we remove PCL proteins,
31:09 - 31:12: would other PCL family members surrogate for the function
31:12 - 31:15: and then keep bringing the PRC2
31:15 - 31:17: to the same places across the genome?
31:17 - 31:20: Or maybe the lack of PCL3, like in this case,
31:20 - 31:22: will lead to destabilization
31:22 - 31:23: of this variation of the PRC2 complex.
31:23 - 31:25: The only one left in the nucleus
31:25 - 31:28: would be the PRC2.2 complex.
31:29 - 31:33: So we have, as I said, we have worked on PHF19,
31:33 - 31:34: also called PCL3.
31:34 - 31:36: In the past year, we have identified this protein
31:36 - 31:38: which controlled back many years ago,
31:38 - 31:40: which controlled the occupancy of PRC2,
31:40 - 31:43: and also it's relevant, important
31:43 - 31:44: for the proper differentiation
31:44 - 31:47: of mouse embryonic stem cells.
31:47 - 31:49: So cells which lack PHF19
31:49 - 31:51: are actually impaired in differentiation.
31:51 - 31:53: Now, what we and other groups recently realized
31:53 - 31:55: is that PHF19, as I mentioned before,
31:55 - 31:59: is misexpressed in several human tumors.
31:59 - 32:02: And for example, what we recently published is that
32:02 - 32:07: the role of PHF19 is very critical
32:07 - 32:10: in a prostate cancer model system,
32:10 - 32:13: and mechanistically what we found
32:13 - 32:14: is that deletion of PHF19,
32:14 - 32:15: I'll talk about the long guys
32:15 - 32:18: from the one which interacts with PRC2 complex,
32:18 - 32:22: deletion of PHF19 will be respected
32:22 - 32:23: was that upon deletion,
32:23 - 32:26: we would have seen a decrease
32:26 - 32:28: of recruitment of PRC2 complex.
32:28 - 32:30: But actually what we observe is the opposite.
32:30 - 32:33: You can see here, upon deletion of PHF19,
32:33 - 32:35: there's more PRC2, as indicated here,
32:35 - 32:38: and an increased deposition of the repressive mark.
32:38 - 32:40: And this actually is, of course,
32:40 - 32:41: not only to see big faults at the promoter,
32:41 - 32:43: but it also, of course, genome-wide,
32:43 - 32:45: and I have time to explain this one here.
32:46 - 32:49: And then trying to look for a potential explanation for this.
32:49 - 32:52: What we found is that JARID2 and MTF2,
32:52 - 32:54: which are the other family members,
32:54 - 32:56: actually upon deletion of PHF19,
32:56 - 32:57: their occupancy increases,
32:57 - 33:01: which means that upon deletion of one specific PCL protein,
33:01 - 33:04: there is a kind of reorganization of PRC2 variations,
33:04 - 33:05: which have different characteristics.
33:05 - 33:07: And particularly in this case,
33:07 - 33:09: they will lead to a reduced proliferation
33:09 - 33:12: and increase in metastatic potential of these cells.
33:14 - 33:14: So this is in cancer,
33:14 - 33:18: but what is actually the true physiological role
33:18 - 33:20: of PHF19 in animal development?
33:20 - 33:23: And most of the data which have been mainly,
33:24 - 33:25: most of the model systems which have been used
33:25 - 33:27: for studying PRC2 complex,
33:27 - 33:30: PRC1 and PRC2 complexes are in embryonic stem cells.
33:30 - 33:32: But what's the role of Polycomb in adult stem cells?
33:32 - 33:36: This is quite unclear in the field.
33:36 - 33:36: So what we did,
33:36 - 33:39: we generated a mouse conditional knockout
33:39 - 33:42: for PHF19, PCL3.
33:42 - 33:44: As was mentioned before by Ali,
33:44 - 33:46: this animal developed homeotic transformations.
33:46 - 33:50: It does not as dramatic as the amniotic phenotype
33:50 - 33:52: that Ali was mentioning before,
33:52 - 33:55: but actually there are several skeletal axis malformations.
33:55 - 33:57: For example, in the presence of,
33:57 - 34:01: at the location of the vertebral C7,
34:01 - 34:02: there is an extra rib
34:02 - 34:05: and this is a narrow course on control animals.
34:05 - 34:06: Let's say wild-type animals.
34:06 - 34:09: Also there is inversion of different vertebrae.
34:09 - 34:14: But as mentioned, this was expected being Polycomb,
34:14 - 34:17: other deletion of the Polycomb protein
34:17 - 34:20: triggers similar homeotic transformations.
34:20 - 34:25: What we did not expect is that these animals during,
34:25 - 34:27: that actually when we look at the blood system,
34:27 - 34:31: they have a reduced number of blood cells,
34:31 - 34:32: as mentioned here,
34:32 - 34:34: control animals versus knockout animals.
34:34 - 34:35: And when these animals become old,
34:35 - 34:37: actually they develop splenomegaly.
34:37 - 34:41: So these two things together suggest actually
34:41 - 34:44: that there might be a defect in the blood system
34:44 - 34:45: that reduces cellularity
34:45 - 34:47: and the presence of splenomegaly in old animals
34:47 - 34:50: indicates that there is most likely a defect
34:50 - 34:51: in the blood system.
34:51 - 34:53: So, and what we did to the gold standard technique
34:53 - 34:56: for studying the defect in the hematopoietic system.
34:56 - 34:58: is by performing CDER transplantation.
34:58 - 35:00: Let me guide you to this.
35:00 - 35:02: At the beginning, the experiment starts
35:02 - 35:07: with mouse irradiation by irradiated animals.
35:07 - 35:09: We completely kill all the blood cells,
35:09 - 35:10: which means the animal dies
35:10 - 35:13: because there is no blood left in the vein.
35:13 - 35:15: Unless we transplant the bone marrow
35:15 - 35:17: coming from either control animals,
35:17 - 35:21: but also now we can compare wild-type bone marrow
35:21 - 35:24: and compare it with bone marrow coming from animals
35:24 - 35:27: that have been deleted for Fischer-Frontier.
35:27 - 35:29: So we can transplant this bone marrow
35:29 - 35:33: in, of course, in parallel in irradiated animals.
35:33 - 35:36: And what we observe actually, there's no difference,
35:36 - 35:37: but you can push the system
35:37 - 35:39: where they transplanted this again
35:39 - 35:41: from the first recipient to the second recipient.
35:41 - 35:43: And again, we don't observe any differences here,
35:43 - 35:46: but then we perform the third transplantation
35:46 - 35:49: and then all the animals that received the bone marrow
35:49 - 35:52: coming from Fischer-Frontier knockout mass model,
35:52 - 35:53: they all fail and die.
35:54 - 35:57: Which means firstly, there is a defect in the durability
35:57 - 36:01: and second, since the defect occurs
36:01 - 36:05: after the third transplantation indicates the defect
36:05 - 36:08: is at the level of the long-term hematopoietic stem cells.
36:08 - 36:11: Let me put you in the context of this.
36:11 - 36:13: So if the defect would have arisen
36:13 - 36:14: after the first transplantation,
36:14 - 36:15: most likely the defect would have been
36:15 - 36:18: a more differentiated cell type in our blood system.
36:18 - 36:22: If the defect would have arisen after the second transplantation,
36:22 - 36:25: the defect would have been, let's say, ascribed
36:25 - 36:27: to more committed precursors,
36:27 - 36:29: but as we've seen the defect occurs
36:29 - 36:31: after the third transplantation,
36:31 - 36:35: the defect occurs in this very rare cell population.
36:35 - 36:38: So this forced us to study a bit more in detail
36:38 - 36:40: the hematopoietic stem cells.
36:40 - 36:41: And that's what we did.
36:41 - 36:46: We isolated hematopoietic stem cells using the SLAM marker.
36:46 - 36:47: There are very few cells for each animal
36:47 - 36:50: and we perform both bulk transcriptomics
36:50 - 36:51: and single-cell transcriptomics.
36:51 - 36:53: And what we found, I can summarize here
36:54 - 36:55: with just one slide,
36:56 - 37:00: is that deletion of HFN team leads
37:00 - 37:03: to reduced ribosomal biogenesis
37:03 - 37:05: and reduced metabolic function.
37:05 - 37:09: Also, we observed there is reduced oxidative phosphorylation
37:09 - 37:11: and a reduced protein function.
37:11 - 37:13: All of these three things together
37:13 - 37:15: suggest that upon deletion of HFN team,
37:15 - 37:19: the hematopoietic stem cells become more quiescent.
37:19 - 37:22: And also what we observed,
37:22 - 37:24: I will come back to this at the end of my talk,
37:24 - 37:27: is that when we look at the transcriptomics,
37:27 - 37:32: they pretty much resemble the signature of a leukemic stem cell.
37:34 - 37:35: Now, this data, as I said,
37:35 - 37:38: the transcriptomic data indicate the cells are more quiescent.
37:38 - 37:39: Can we prove it?
37:39 - 37:43: What we did, we did Ki67 staining.
37:43 - 37:46: You can see there is reduced staining in knockout cells.
37:46 - 37:48: And also we performed BRDU incorporation.
37:48 - 37:51: And again, knockout cells incorporate less,
37:52 - 37:53: reduced amount of BRDU.
37:53 - 37:54: Again, corroborating the fact
37:54 - 37:56: that these cells are more quiescent.
37:56 - 37:58: So they are more quiescent,
37:58 - 38:02: but can those cells undergo some differentiation?
38:02 - 38:04: And the way in which we assess this question
38:04 - 38:08: is by sorting a single cell in a single well here;
38:08 - 38:10: we performed this experiment multiple times.
38:10 - 38:11: And then from the single cell,
38:11 - 38:14: we let the cells grow and proliferate.
38:14 - 38:16: And we count the number of cells
38:16 - 38:18: and how many times the cells divide
38:18 - 38:20: for the first four days and after.
38:20 - 38:22: So we have the kinetics of the growth.
38:22 - 38:24: And then at day nine,
38:24 - 38:27: we count the size of the whole colony.
38:27 - 38:29: And after 14 days, we characterize the colony,
38:29 - 38:30: looking at the markers
38:30 - 38:33: for whether cells have been differentiated or not.
38:34 - 38:37: Here summarized, again, the data in a single slide,
38:37 - 38:42: what we found that when you see the cell here
38:42 - 38:45: and you let it wait for 14 days,
38:45 - 38:46: most of the cells differentiate.
38:46 - 38:49: There is basically no hematopoietic stem cell.
38:49 - 38:52: There is no stem cell left, only 2%.
38:52 - 38:56: But when you perform an experiment with cells
38:56 - 38:58: that have been isolated from knockout animals,
38:58 - 39:01: we see the number increases by at least 10 times
39:01 - 39:02: indicated here.
39:02 - 39:07: So definitely, the cells are unable
39:07 - 39:09: to undergo full differentiation
39:09 - 39:11: and remain more in an undifferentiated state
39:11 - 39:13: as this data suggests.
39:14 - 39:19: So this pushed us to go back and look a bit more in detail
39:20 - 39:22: in the transcriptome analysis.
39:22 - 39:25: And what we have found is that among the upregulated genes,
39:25 - 39:27: something that we missed at the very beginning
39:27 - 39:28: is that there is an upregulation
39:28 - 39:30: of the retinoic acid pathway.
39:30 - 39:33: While among the genes that become downregulated,
39:33 - 39:35: the pathways become downregulated,
39:35 - 39:36: again, with what I mentioned before,
39:36 - 39:41: mitochondrial transport, oxidative phosphorylation.
39:41 - 39:46: So this data here, somehow calls to our attention,
39:46 - 39:49: a paper that we read a few years back
39:49 - 39:50: is a paper that was published
39:50 - 39:53: by Andreas Trumpp and Nina Cabezas.
39:53 - 39:55: In this paper, basically what they did,
39:55 - 39:59: they took the long-term hematopoietic stem cell population
39:59 - 40:00: and so,
40:00 - 40:03: we divided this population actually into two populations,
40:03 - 40:06: which are called dormant long-term hematopoietic stem cells
40:06 - 40:10: and activated long-term hematopoietic stem cells.
40:10 - 40:13: The difference in these two populations here,
40:13 - 40:15: the different characteristics are that
40:15 - 40:19: the dormant cells have a very high retinoic acid pathway,
40:19 - 40:21: and low biosynthetic pathway,
40:21 - 40:24: low metabolic activity, and low ROS production,
40:24 - 40:26: which is actually very much similar to
40:26 - 40:28: this effect that we observe
40:28 - 40:30: upon PHF-13 deletion.
40:30 - 40:31: Then that poses the question,
40:31 - 40:34: is PHF-13 actually controlling
40:34 - 40:36: the switch between dormant to activated cells?
40:36 - 40:38: Those cells are the ones that then give rise to
40:38 - 40:40: the entry in the differentiation path,
40:40 - 40:42: they give rise to all the blood cells.
40:42 - 40:46: Is PHF-13 controlling the activation of these cells?
40:46 - 40:48: In other words, is the balance between
40:48 - 40:54: PRC2.1 and PRC2.2 or PRC2 variations
40:54 - 40:56: actually relevant for controlling
40:56 - 40:59: dormancy versus activation of
40:59 - 41:02: differentiation in the hematopoietic system?
41:03 - 41:06: This is the question we want to address,
41:06 - 41:08: how PHF-13 regulates
41:08 - 41:10: hematopoietic stem cell quiescence
41:10 - 41:13: and what is the mechanism behind this?
41:13 - 41:16: Basically, what we expected is
41:16 - 41:18: that different variations of the PRC2 complex,
41:18 - 41:21: although they share several target genes,
41:21 - 41:23: they also have their own specific target genes.
41:23 - 41:26: If we consider a potential set of
41:26 - 41:29: PCL3 PHF-13 target genes upon deletion of PHF-13,
41:29 - 41:31: we expect those genes to lose
41:31 - 41:35: K27 trimethylation and to become activated.
41:35 - 41:38: Globally, we would expect to have a reduced
41:38 - 41:41: K27 trimethylation level in
41:41 - 41:44: the cells and activation of a specific set of genes.
41:44 - 41:47: Actually, what we obtained is exactly the opposite.
41:47 - 41:50: What we notice is that upon deletion of PHF-13,
41:50 - 41:52: there is a dramatic increase
41:52 - 41:54: of K27 trimethylation deposition.
41:54 - 41:56: That was quite surprising,
41:56 - 41:59: and then we dug a bit more in detail.
41:59 - 42:01: What we asked ourselves is where
42:01 - 42:05: this K27 trimethylation accumulated.
42:05 - 42:08: Clearly, it's not being accumulated at
42:08 - 42:11: genes which control stem cell characteristics,
42:11 - 42:13: so there is no difference between control and
42:13 - 42:15: the level of K27 in
42:15 - 42:18: those sets of genes in control cells and knockout cells.
42:18 - 42:20: But then when we look at the cells which
42:20 - 42:21: control differentiation of
42:21 - 42:22: transcription factors that
42:22 - 42:24: implicate in blood differentiation,
42:24 - 42:26: you can see that significantly,
42:26 - 42:29: the K27 trimethylation accumulates
42:29 - 42:31: specifically at those promoters.
42:31 - 42:34: For example here, you can see
42:34 - 42:36: these CBP-alpha master genes
42:36 - 42:37: in multiple differentiation,
42:37 - 42:39: control cells, very low level of K27,
42:39 - 42:42: increased level in PHF-13 knockout,
42:42 - 42:44: the same is true for PAX5 and
42:44 - 42:45: DAP, all genes which are
42:45 - 42:47: relevant for hematopoietic differentiation.
42:47 - 42:49: You can see this accumulation of K27
42:49 - 42:51: goes hand-in-hand with a reduction in
42:51 - 42:54: active signals suggesting that
42:54 - 42:57: those genes become locked in a repressive state.
42:57 - 42:59: Basically, we expected
42:59 - 43:01: those genes to become upregulated,
43:01 - 43:02: but actually what we realized is that it's
43:02 - 43:04: a compensation from the other variations
43:04 - 43:05: of the PRC2 complex.
43:05 - 43:08: Interestingly, some of these variations are capable of
43:08 - 43:14: compensating for preventing activation of some of the genes,
43:14 - 43:15: but not for all of them,
43:15 - 43:22: and what is specified is still unclear.
43:22 - 43:24: If you remember, this is the last data
43:24 - 43:25: I want to share with you,
43:25 - 43:28: that when we transplant the bone marrow,
43:28 - 43:31: we observe a defect after the third transplantation,
43:31 - 43:33: and this was done by using
43:33 - 43:35: bone marrow obtained from young animals.
43:35 - 43:39: But if we take bone marrow coming from old animals,
43:39 - 43:41: we already saw that this animal develops
43:41 - 43:42: phenotypes instead of when they
43:42 - 43:44: become old after 80 weeks,
43:44 - 43:46: actually, a few weeks after transplantation,
43:46 - 43:48: it already presents phenotypes.
43:48 - 43:51: Not only that, but then if we look at the blood system,
43:51 - 43:53: the old blood system failed.
43:53 - 43:56: You see the bone become completely pale,
43:56 - 43:58: which means there are no erythrocytes
43:58 - 44:00: in this animal model here.
44:00 - 44:02: Actually, we could not even
44:02 - 44:04: bring it to the third transplantation
44:04 - 44:05: because all these animals would die.
44:05 - 44:08: This is definitely something that is linked to
44:08 - 44:10: the exhaustion of the characteristics
44:10 - 44:13: of the hematopoietic stem cells.
44:13 - 44:15: Basically, I'll finish here
44:15 - 44:18: and sum up what I've been mentioning,
44:18 - 44:21: tell you with this final cartoon.
44:21 - 44:23: We know that PHF-13 and PRC2 complex
44:23 - 44:25: seed genes which are
44:25 - 44:29: important for developmental differentiation.
44:29 - 44:31: Developmental genes are important for differentiation.
44:31 - 44:34: Those are typically genes which are
44:34 - 44:37: varied which present both active and inactive marks.
44:37 - 44:39: Now, deletion of PHF-13 leads to
44:39 - 44:41: the reassembly of
44:41 - 44:43: novel variations of the PRC2 complex which
44:43 - 44:46: contain PRC1, PRC2, and JARID2.
44:46 - 44:48: In this complex, they might have
44:48 - 44:50: different characteristics because when we found that
44:50 - 44:51: the variability is lost and there is
44:51 - 44:54: accumulation of K27 trimethylation.
44:54 - 44:56: Those genes, instead of being variegated,
44:56 - 44:59: which means instead of being able to go either
44:59 - 45:03: being activated or repressed, they stay repressed.
45:03 - 45:06: When the differentiation signal arrives,
45:06 - 45:08: these genes cannot be activated.
45:08 - 45:11: More generally, this upon deletion of PHF-13,
45:11 - 45:12: hematopoietic stem cells are more
45:12 - 45:16: quiescent, they accumulate K27 trimethylation,
45:16 - 45:17: specifically genes which
45:17 - 45:19: control hematopoietic differentiation.
45:19 - 45:22: This is true not only at the promoter level,
45:22 - 45:24: but also when we look at the non-coding regions which
45:24 - 45:25: control these genes, we also see
45:25 - 45:30: similar characteristics, similar phenotype.
45:30 - 45:31: Also, as I mentioned,
45:31 - 45:34: we observe a leukemic stem cell
45:34 - 45:36: signature upon deletion of PHF-13.
45:36 - 45:38: Here, just one data point in this direction,
45:38 - 45:42: when we isolate cells from the spleen,
45:42 - 45:43: we see the appearance of
45:43 - 45:44: this aberrant population.
45:44 - 45:46: Here, we have knockout animals
45:46 - 45:48: which are not present in control animals.
45:48 - 45:51: This aberrant population is what we believe is
45:51 - 45:53: a pre-leukemic state and this might be
45:53 - 45:57: linked to asymmetric versus asymmetric deletion.
45:57 - 46:00: I'll finish here and just mention that
46:00 - 46:03: the work was performed in Barcelona,
46:03 - 46:04: in a CRG which is located,
46:04 - 46:07: you can see in this building here.
46:07 - 46:09: In a different way,
46:09 - 46:11: all the lab members have
46:11 - 46:13: contributed to the work I've been presenting here,
46:13 - 46:16: but particularly the work on PHF-13 in
46:16 - 46:18: the hematopoietic system was performed
46:18 - 46:22: by Arantxa, Isabel, and Pedro in our lab.
46:22 - 46:24: I would like to thank our collaborators and sorry,
46:24 - 46:27: I forgot to mention these are the people in the lab.
46:27 - 46:30: Thanks and happy to take some questions.
46:35 - 46:37: Well, thank you, Luciano.
46:37 - 46:40: We have time for a few questions.
46:40 - 46:43: The first question is that,
46:43 - 46:47: is there any reduction in the PRC1 occupancy
46:47 - 46:53: at differentiation genes in PHF-19 knockout cells?
46:57 - 47:00: That's something that we don't know.
47:00 - 47:01: It's a topic that we've been
47:01 - 47:03: discussing often in our lab meetings.
47:04 - 47:08: We do not have the time to perform this experiment yet,
47:08 - 47:10: but this experiment had been planned and
47:10 - 47:13: we will know in the next months.
47:14 - 47:16: Ibrahim Natu is asking,
47:16 - 47:20: what is the determining factor for the presence of
47:20 - 47:26: PHF-19 or JARID2 within the same complex?
47:27 - 47:32: Actually, they are never in the same complex.
47:32 - 47:35: They actually fight for the recognition of
47:35 - 47:38: the PRC2 core component
47:38 - 47:40: of the same region of the PRC2 core component.
47:40 - 47:42: Either you incorporate one of
47:42 - 47:45: the PCL family members or you incorporate JARID2.
47:45 - 47:49: They are actually the two proteins and mutually exclusive.
47:49 - 47:52: You have the JARID2-PRC2 variation
47:52 - 47:54: or the PCL-PRC2 variations.
47:54 - 47:57: They cannot coexist.
47:57 - 48:00: This is clear from structural studies actually.
48:00 - 48:02: We don't have time to discuss here,
48:02 - 48:05: but this is based on structural studies.
48:05 - 48:07: Vastava Parakash is asking,
48:07 - 48:10: are there any sites that are specific for
48:10 - 48:16: PHF-19 which cannot be compensated upon PHF-19 loss?
48:16 - 48:18: Yeah. Actually, that's something that we know.
48:18 - 48:20: There are several genes which
48:20 - 48:22: control the sameness characteristic in
48:22 - 48:26: cell cycle and which actually become upregulated.
48:26 - 48:31: The linked question to this is,
48:31 - 48:36: why are those promoters,
48:36 - 48:39: either PCL family member or JARID2 cannot compensate?
48:39 - 48:41: That's something that what we are studying is
48:41 - 48:44: a specific targeting of what together with
48:44 - 48:48: the different variation of PCL-PRC2 complex,
48:48 - 48:50: there is an additional factor which can justify
48:50 - 48:53: the specific and unique targeting to a specific locus.
48:53 - 48:56: That's still unclear at the moment.
48:56 - 48:58: Peng Lu is asking, what happens to
48:58 - 49:02: the occupancy of RNA polymerase
49:02 - 49:05: II upon the loss of PHF-19?
49:06 - 49:09: We have studied that in the past,
49:09 - 49:13: and particularly with respect to the fact that
49:13 - 49:18: actually together with PCL protein and EPOP,
49:18 - 49:21: the other two factors are part of the same PRC2 variation,
49:21 - 49:24: which is called elongin B, elongin C.
49:24 - 49:26: As you know, because also you have been
49:26 - 49:28: working on these two proteins in the past,
49:28 - 49:30: elongin B, C have been associated
49:30 - 49:32: with elongation of polymerase.
49:32 - 49:35: Now, we have done several experiments,
49:35 - 49:37: actually a few of these experiments
49:37 - 49:39: done in collaboration with the duality.
49:39 - 49:42: We could not monitor any change in
49:42 - 49:45: the occupancy of any of the variations of the polymerase.
49:45 - 49:47: The link between elongin B,
49:47 - 49:49: C and polymerase or
49:49 - 49:52: elongation of polymerase is still unclear.
49:53 - 49:54: There's been a recent paper.
49:54 - 49:57: Studies that came out of Cech's lab
49:57 - 49:59: where he suggests that actually
49:59 - 50:02: the mere process of transcription,
50:02 - 50:04: this was opposite to what Christian's lab has shown,
50:04 - 50:06: that the transcription itself is required
50:06 - 50:09: for the recruitment of Polycomb.
50:09 - 50:13: I think there's a dichotomy in here
50:13 - 50:17: on the role of transcription and Polycomb recruitment.
50:17 - 50:19: The traditional model is that what you, Danny,
50:19 - 50:21: and Christian have shown is that
50:21 - 50:24: if you basically get rid of transcription,
50:24 - 50:25: Polycomb comes in,
50:25 - 50:28: activates transcription, Polycomb pops off.
50:28 - 50:30: But does the transcription
50:30 - 50:32: itself drive recruitment of Polycomb?
50:32 - 50:34: I think maybe this one of
50:34 - 50:36: these questions is trying to drive into.
50:36 - 50:39: Maybe there's been specificity on that.
50:39 - 50:41: I think we can discuss at the end.
50:41 - 50:42: We have a little time for
50:42 - 50:44: discussing this more general aspect.
50:44 - 50:46: But actually, one of the recent papers from
50:46 - 50:49: Cech also indicated the presence of an RNA
50:50 - 50:52: that actually can prevent the binding
50:52 - 50:56: of a Polycomb directly.
50:56 - 50:57: That is transcription itself
50:57 - 50:59: affects Polycomb recruitment.
50:59 - 51:02: RNA is a by-product of transcription.
51:02 - 51:05: I'm going to do two last questions here.
51:05 - 51:08: Lisa Marula is going to ask,
51:08 - 51:12: does the highest PRC1 behave in this process?
51:12 - 51:14: This was linked to the previous question.
51:14 - 51:16: It's very similar to the previous question.
51:16 - 51:19: We are investigating that. We don't know yet.
51:19 - 51:22: Andrea Punti wants to know whether
51:22 - 51:25: the effect you observed in the PHF19
51:25 - 51:29: null stem cells can be
51:29 - 51:34: recapitulated by the EPOP or Poly I or II loss.
51:34 - 51:38: Actually, that's something that I don't know about Poly,
51:38 - 51:40: but we have recently generated
51:40 - 51:42: the EPOP knockout mouse model,
51:42 - 51:44: and we will actually look into
51:44 - 51:46: the blood system in the metabolism cell
51:46 - 51:49: and see whether they have different characteristics or not.
51:49 - 51:51: There are a few more questions in there,
51:51 - 51:52: but for the sake of time,
51:52 - 51:53: we're going to move on to our next speaker.
51:53 - 51:55: You can go online and answer
51:55 - 51:56: some of these questions, please.
51:56 - 51:57: I will. Thanks.
51:57 - 51:59: Thank you very much.
51:59 - 52:03: We're going to move on to our penultimate speaker,
52:03 - 52:04: Professor Kristian Helin.
52:04 - 52:07: I'll say Luciano was the most handsome epigeneticist.
52:07 - 52:09: Kristian is probably the best fit.
52:09 - 52:11: If you're at a meeting, he would be sending you to
52:11 - 52:13: a challenge with a volleyball,
52:13 - 52:14: soccer, or zooming challenge.
52:15 - 52:16: But hopefully you're at home,
52:16 - 52:18: so all of our hearts are in great shape.
52:18 - 52:21: So Kristian is a professor of cell biology,
52:21 - 52:23: is an endowed professor,
52:23 - 52:25: a Dean Hoop Professor of Cell Biology,
52:25 - 52:28: and also chairman of the Department of Cell Biology,
52:28 - 52:31: and the director of the Center for Epigenetic Research
52:31 - 52:35: at the Memorial Sloan Kettering in New York City.
52:35 - 52:38: He's also a professor at the University of Copenhagen,
52:38 - 52:40: where he was previously, moving
52:40 - 52:42: to the United States at Sloan Kettering.
52:42 - 52:45: He received his MS in chemical engineering
52:45 - 52:48: from the Technical University of Denmark,
52:48 - 52:50: and PhD from Copenhagen.
52:52 - 52:55: Post-PhD, Kristian moved on to Harvard,
52:55 - 52:56: where he was a postdoctoral fellow,
52:56 - 52:58: studying the E2F transcription factor
52:58 - 53:00: in cancer pathogenesis,
53:00 - 53:02: a work he has continued on,
53:02 - 53:04: and trying to understand the role of many of these factors
53:04 - 53:06: in cancer pathogenesis,
53:06 - 53:10: and how epigenetics also impinges on this process.
53:10 - 53:12: Previously, he was a founding member
53:13 - 53:16: of the Department of Environmental Oncology
53:16 - 53:19: at the European Institute of Oncology in Milano,
53:19 - 53:21: and he was recruited back to Copenhagen
53:21 - 53:25: as a founding director of the biotech research innovation
53:25 - 53:27: at the University of Copenhagen,
53:27 - 53:29: where he has still an appointment
53:29 - 53:32: prior to joining the Memorial Sloan Kettering.
53:32 - 53:34: Christian has done beautiful work
53:34 - 53:36: during the past 20 years
53:36 - 53:38: in the identification of many of these epigenetic factors,
53:38 - 53:41: the methylases and demethylases,
53:41 - 53:42: and his laboratory has focused
53:42 - 53:46: on the role of these factors on cancer pathogenesis,
53:46 - 53:48: and he's going to expand upon the work
53:48 - 53:50: that he has done for us today.
53:50 - 53:51: With that, welcome, Kristian.
53:53 - 53:54: Thank you.
53:54 - 53:56: Thank you very much, Ali.
53:56 - 53:57: Well, you're already at the beach.
53:57 - 53:58: Look at that.
53:58 - 54:00: That's what I was thinking.
54:00 - 54:04: I mean, first, I would like to thank the Abcam team
54:04 - 54:06: and Tony for inviting me here,
54:06 - 54:08: and second, I have to say,
54:08 - 54:09: well, the picture a little bit
54:09 - 54:13: about an Abcam meeting in November
54:13 - 54:16: should actually be in a beach location
54:16 - 54:17: somewhere in the Caribbean,
54:17 - 54:20: but let's hope that this is going to happen in the future
54:20 - 54:24: when some of the vaccines are going to be available.
54:24 - 54:29: So let me take off this and go to the presentation.
54:38 - 54:39: There you go.
54:39 - 54:44: Sorry for that.
54:51 - 54:52: There you go.
54:52 - 54:53: Everybody can see the screen?
54:54 - 54:56: Somebody says yes?
54:57 - 54:58: Yes.
54:58 - 55:00: Yeah, go on.
55:00 - 55:01: Great, thank you.
55:01 - 55:04: So what I'm going to talk to you about today
55:04 - 55:06: is, of course, also Polycomb group proteins,
55:06 - 55:10: and it's going to be a long session on polycombs.
55:10 - 55:12: I'll try to take a little different tact
55:12 - 55:16: than Luciano just did and some of the other speakers.
55:16 - 55:18: So I'm going to give two parts of my talk.
55:18 - 55:20: One is going to basically understanding
55:20 - 55:22: how PRC2 is recruited to chromatin,
55:22 - 55:25: which we heard about in plants before,
55:25 - 55:27: and the second is taking another approach
55:27 - 55:30: to actually try to understand the biological functions
55:30 - 55:34: of the K27 methylation and potential auto-acetylation
55:34 - 55:39: in regulating normal transcription and also differentiation.
55:41 - 55:43: So first, a few disclosures.
55:43 - 55:44: I'm a consultant for Noble Holdings,
55:44 - 55:47: Hannibal Health Innovation, and Terra Bioscience.
55:49 - 55:52: So you just heard about the Polycomb group proteins.
55:52 - 55:54: There are different ways you can present them.
55:54 - 55:55: I'm normally starting
55:55 - 55:57: with the Polycomb repressive complex two
55:57 - 56:00: because this is where my lab has done most work
56:00 - 56:02: over the years, and we also see that
56:02 - 56:07: as one of the keys actually to understand
56:07 - 56:11: how this repressive system is regulating transcription.
56:11 - 56:14: The PRC2 complex is conserved during evolution.
56:14 - 56:16: Actually, the crystal structures came out
56:16 - 56:19: from yeast, one of the first,
56:19 - 56:20: so it's so much conserved during evolution.
56:20 - 56:25: It's also in some unicellular organisms too.
56:26 - 56:27: PRC1, we have to remember,
56:27 - 56:29: is not conserved to the same degree.
56:29 - 56:33: So PRC2 has these core members, EED, SUZ12,
56:33 - 56:35: and then a catalytic subunit,
56:35 - 56:38: which is either EZH1 or EZH2.
56:38 - 56:41: It's the only known K27 methyltransferase complex
56:41 - 56:42: in mammalian cells.
56:42 - 56:45: In plants, we actually know different monomethylases,
56:45 - 56:49: but we do not have those in mammalian cells.
56:49 - 56:51: We know that these are all essential
56:51 - 56:52: for normal development.
56:52 - 56:54: If you knock out these core subunits,
56:54 - 56:57: EED, SUZ12 or EZH2,
56:57 - 56:59: you see problems in gastrulation
56:59 - 57:01: in a mouse that is between E7,
57:01 - 57:04: around E7.5 in mouse development.
57:04 - 57:07: We know that the complex is important
57:07 - 57:09: for the maintenance of transcriptional repression.
57:09 - 57:11: We don't think that it is actually important
57:11 - 57:12: for setting up transcriptional repression,
57:12 - 57:16: but maintaining it in a transcriptional pattern,
57:16 - 57:19: which can all be overcome by signaling
57:19 - 57:21: and by transcription factors.
57:21 - 57:23: One thing which is important to know
57:23 - 57:25: is that the PRC2 complex
57:25 - 57:27: actually does mono-, di-, and trimethylation,
57:27 - 57:29: and mono-, di-, and trimethylation are very widespread
57:29 - 57:30: in the genome.
57:30 - 57:35: So 5% to 10% of H3 has monomethylation,
57:35 - 57:37: up to 50%, 60% have dimethylation,
57:37 - 57:42: and around 7% to 10% have K27 trimethylation.
57:42 - 57:45: And when we talk about the Polycomb group proteins,
57:45 - 57:48: most people are only thinking about the trimethylation,
57:48 - 57:50: but it's important to remember,
57:50 - 57:51: in particular, when you think about recruitment,
57:51 - 57:54: but potentially also the biological function
57:54 - 57:56: that Polycomb group proteins are actually,
57:56 - 57:58: or PRC2 is doing mono-, di-, and trimethylation.
58:00 - 58:03: So PRC2, of course, has been linked to cancer.
58:03 - 58:05: This is the way we also actually started working on
58:05 - 58:07: in my lab, the link to cancer
58:07 - 58:11: and the link to the E2F transcription factor pathway.
58:11 - 58:16: It's been shown that EZH2 has hyperactive mutations
58:16 - 58:18: in lymphoma and melanomas.
58:18 - 58:21: It's been shown that there are specific dependencies
58:21 - 58:24: in tumors where there are mutations
58:24 - 58:26: in SWI/SNF components.
58:26 - 58:28: And it's also been shown in some cases
58:28 - 58:30: that you will see gene amplification,
58:30 - 58:33: particularly EZH2 in some cancers.
58:33 - 58:35: So this, of course, has led some companies
58:35 - 58:38: to develop inhibitors to EZH2 and EED,
58:38 - 58:40: and now five of them are in clinical trials,
58:40 - 58:43: and a diversity of clinical trials now.
58:43 - 58:46: An EZH2 inhibitor has actually been approved recently
58:46 - 58:49: for patients with relapsed or refractory lymphomas,
58:49 - 58:51: and for patients aged 16 and older
58:51 - 58:53: with metastatic and locally advanced
58:53 - 58:55: epithelial sarcomas.
58:55 - 58:58: Those are the ones with SWI/SNF mutations.
58:58 - 59:01: What it actually also gives us is actually some tools
59:01 - 59:03: to target other types of tumors,
59:03 - 59:05: mainly in combination therapies,
59:05 - 59:08: where the hope is that the targeting of PRC2
59:08 - 59:10: will lead to the deep repression of other genes,
59:10 - 59:11: which are important for targeting
59:11 - 59:14: by other types of treatments.
59:14 - 59:16: And so this is the hope we are going to see in the future
59:16 - 59:18: that we will see the development of that.
59:18 - 59:21: Recently, we and also Alicia Lazaroff's lab
59:21 - 59:23: published that DIPG tumors,
59:23 - 59:26: which have these K27M mutations,
59:26 - 59:28: are expressing K27M mutant,
59:28 - 59:31: are very sensitive to EZH2 inhibitors as well,
59:31 - 59:33: and we hope that there will be clinical trials
59:33 - 59:35: on these types of tumors as well.
59:37 - 59:39: So one of the questions we've been working on
59:39 - 59:43: over the years is how the Polycomb repressive complex two,
59:43 - 59:46: it's actually the binding to chromatin is regulated.
59:46 - 59:48: And I will tell you about today is mainly the work
59:48 - 59:51: by two postdocs and a graduate student in my lab.
59:52 - 59:54: So Jonas Heufeld and Lynn Hildreth.
59:56 - 60:00: So just going back to how these K27M mutations
60:00 - 60:05: and patterns, they actually look in the genome. So this is a screenshot of three different
60:05 - 60:09: genes in mouse embryonic stem cells. And what you see here is that SUZ12, which is one
60:09 - 60:16: of the core components of PRC2, binds really nicely to the PRDM12 gene. And you see that
60:16 - 60:22: co-localizing with H3K27 trimethylation you see on the top here. What you also see actually
60:22 - 60:30: there are some spurious, or whatever you call them, at least reasonable levels of K27 trimethylation
60:30 - 60:35: also outside the regions where you actually see stable PRC2 binding, which are overlapping
60:35 - 60:40: by and large with, you know, with K27 dimethylation. These are regions you find throughout the
60:40 - 60:45: genome and about 50% of K27 trimethylation you're actually finding outside where you
60:45 - 60:51: see detectable PRC2 binding, suggesting that perhaps the model we have for PRC2 is not
60:51 - 60:57: reasonably correct, that it had to bind there for a long time to actually see trimethylation.
60:57 - 61:00: What you also see are, you know, very high levels of dimethylation throughout the genome
61:00 - 61:04: and also monomethylation. And monomethylation you particularly find on genes which are highly
61:04 - 61:10: expressed. We don't really know the function of those, but we know that PRC2 is actually
61:10 - 61:14: the complex which is actually mediating those.
61:14 - 61:20: So Polycomb binding to target genes, I'll try to tell you a little about what we think
61:20 - 61:24: we know and also what we are actually currently discussing, and I guess we'll have that discussion
61:24 - 61:30: in the end of this session as well. So I'm talking mammalian cells here. I try to relate
61:30 - 61:36: it to what the talks in perhaps Drosophila, but also what we just heard about in plants.
61:36 - 61:40: So in mammalian cells, we have 3,000 to 5,000 Polycomb-targeted genes in each cell. And
61:40 - 61:45: of course, they differ from one cell to another because these are genes which should be repressed.
61:46 - 61:52: PRC2 but also PRC1 are enriched at CpG islands of silenced genes. So we don't have PIEs,
61:52 - 61:58: if you want. In mammalian cells, we have CpG islands, which are the PIEs, but they don't
61:58 - 62:03: have really defined sites. They have these CpGs. And when they are non-methylated, PRC2
62:03 - 62:06: can bind there. Also, PRC1, by the way.
62:06 - 62:14: So PRC2 binding co-localizes both with K27 trimethylation and H2A lysine 119 monomethylation.
62:14 - 62:20: We heard about before, which is mediated by Ring1A and Ring1B catalysis from the
62:20 - 62:28: PRC1 complex. KDM2B of the variant PRC1 complex can bind directly to CpGs. It has this CXC
62:28 - 62:36: motif that can direct it directly to CpGs. And we also know that PCL1, 2, and 3 of PRC2
62:36 - 62:41: have been suggested to bind directly to CpG islands. Our data suggests your patient also
62:41 - 62:45: needs SUZ12 to see that binding.
62:45 - 62:49: And this was the discussion we had before, where basically a binding of PSG2 occurs in
62:49 - 62:55: response to transcriptional silencing, as we also showed in plants. First, you basically
62:55 - 63:00: knock down, basically mediate the transcriptional repression of a gene by transcription factors,
63:00 - 63:04: perhaps by age negativity. And not before you see that gene repression, you actually
63:04 - 63:11: see PSG2 binding. We know that PSG2 is not required for the binding of a lot of genes
63:11 - 63:16: when you exit out of embryonic stem cells during differentiation, but you do require
63:16 - 63:21: actually PSG2 to maintain the transcription repression, which we showed a number of years
63:21 - 63:22: ago.
63:22 - 63:27: RNA has been shown both to promote and inhibit binding of PSG2 to chromatin, and I think
63:27 - 63:34: the jury is still out there exactly how RNA is actually regulating PSG2 activity on genes.
63:35 - 63:39: It seems to be both required for the recruitment in some cases, but definitely also it's required
63:39 - 63:43: to degrade it to see full repression.
63:43 - 63:48: So there are some models we're discussing at the moment. If you go back to what Luciano
63:48 - 63:52: was just telling you, there's the hierarchical model where PSG2 is recruited first through
63:52 - 63:59: to the CpG islands, at least to the trimethylation of K27, and that recruits basically the PSG1
63:59 - 64:04: complex through what is called the canonical PSG1 complex through the chromatogram of
64:04 - 64:08: Cpx proteins, and that leads to H2A monoubiquitylation.
64:08 - 64:12: We know for a number of years that this is not the whole mechanism how it works, because
64:12 - 64:17: we know that when we knock out PSG2, we still see no effect on basically the global levels
64:17 - 64:23: of H2A monoubiquitylation in the cells, and we also know that PSG1 can be recruited through
64:23 - 64:26: the variant PSG1, which has this Kdm2B.
64:26 - 64:31: But what has been suggested in recent years, in particular by Rob Close's laboratory, is
64:31 - 64:35: that there's an alternative model for how you can actually see this recruitment, where
64:35 - 64:42: PSG1 through RYBPP or through Kdm2B in that complex as well is coming in first, and that
64:42 - 64:48: leads to the recruitment of the PSG2 complex. In particular, here it's been suggested that
64:48 - 64:56: the R2 of the PSG2 complex can bind to monoubiquitylated H2A ubiquitin, and that can lead to the recruitment
64:56 - 65:01: of PSG2. So, we are discussing that a lot. I think in one camp they say, well, this is
65:01 - 65:04: the way it works. I think we are more in the camp where they actually interact together
65:04 - 65:09: and there's interdependent recruitment of the two complexes.
65:09 - 65:13: If you go a little closer on PSG2, you can actually look here at the core domain, which
65:13 - 65:20: we try to basically depict using the crystal structure data together with the EM data and
65:20 - 65:26: put together a whole complex, which is shown here. So, SUZ12 is shown in yellow. SUZ12
65:26 - 65:31: has a C-terminal, also called the VEFS domain, and this is required for actually formation
65:31 - 65:37: of a complex where EED and EZH2 are part of it, and that forms also the catalytic activity.
65:37 - 65:43: Then it has this N-terminal extension, which is not touching EED or EZH2 and is required
65:43 - 65:48: for binding to other proteins involved in recruitment. What I'm showing over here to
65:48 - 65:55: the right is actually what happens when you express various versions of SUZ12. If you
65:55 - 65:59: have the VEFS domain, which is basically here, this domain, which is involved in catalytic
65:59 - 66:04: activity, we say here first a knockout, you knock out SUZ12, you lose SUZ12, but you
66:04 - 66:10: also lose EZH2 and EED because it's required for stability, and you lose mono-, di- and trimethylation.
66:10 - 66:16: If you express the VEFS domain only, you actually retain the expression of EZH2, EED, and you
66:16 - 66:21: regain mono-, di- and trimethylation, but if you take the N-terminus, which is shown here,
66:21 - 66:26: it's not enough for stability, and you actually lose mono-, di- and trimethylation. So basically,
66:26 - 66:33: the catalytic activity of the PRC2 complex is actually required here, this C-terminus
66:33 - 66:40: of SUZ12. On the other hand, you can take the N-terminus of SUZ12, which is shown here,
66:40 - 66:45: and you can look for now for the binding of SUZ12 throughout the genome. So one thing
66:45 - 66:49: we can show over here is that this is the wild-type mouse embryonic stem cells. If you
66:49 - 66:55: look at SUZ12 binding, you'll see here it's actually binding here to CpG islands, we know.
66:55 - 67:00: When you take a knockout of EZH2 and EZH2, there's no trimethylation activity in these
67:00 - 67:03: cells where you have EED. When you knock out that, there's no trimethylation activity in
67:03 - 67:09: these cells. You actually retain the binding of SUZ12 in these cells, and of course, what
67:09 - 67:13: you see here is that when you knock out SUZ12, there's no binding. So basically, that shows
67:13 - 67:19: that SUZ12 can bind to the chromatin in the absence, totally in the absence of the other
67:19 - 67:26: subunits, and it's also, you don't require the trimethylation of K27 to see this binding.
67:27 - 67:32: So now focusing on the N-terminus only, here we'll take various mutants, which we showed
67:32 - 67:37: you before, and now if you only express what is known as the delta-VEFS here or the N-terminus,
67:37 - 67:43: what you see here, you see good binding to CpG islands as well. So that part of SUZ12
67:44 - 67:50: is required for specific DNA binding, and we also know that it's essential for specific binding
67:50 - 67:55: because when you delete it, you don't see the binding anymore. However, what was important,
67:55 - 67:59: that this domain, actually, this VEFS domain was sufficient to give you global
67:59 - 68:06: PSG2 activity in the cell. So you can have non-distributed PSG2 binding throughout the
68:06 - 68:11: genome, which is sufficient to catalyze the global levels of K27 trimethylation,
68:11 - 68:18: but not specific on CpG islands. So we also heard about this before,
68:18 - 68:22: that PSG2 is organized in two distinct subcomplexes, and particularly what we know
68:22 - 68:28: is that the N-terminus of SUZ12 is actually binding to these subunits, which are part of
68:28 - 68:35: two different PSG2 complexes. One is called PSG2.1, and the other one is called PSG2.2,
68:35 - 68:42: and these are two distinct complexes. One of them contains all-POLY or PCL1, 2, and 3,
68:42 - 68:48: and the other one, JARID2 and AEBP2. We know the crystal structures of that as well. We know that
68:48 - 68:54: there is competition between these complexes, so as Luciano said, they cannot bind to the N-terminus
68:54 - 69:00: of SUZ12 at the same time. So what we were interested in, actually, is trying to understand
69:00 - 69:05: how these complexes contributed to the ability of the N-terminus of SUZ12 to bind to this
69:05 - 69:11: specific site. And the way we did that was by generating embryonic stem cells where we
69:12 - 69:18: consecutively knocked out, first individually and then all of them together, the various subunits
69:18 - 69:25: which can bind to the N-terminus of SUZ12. So just to give you examples here, you have AEBP2,
69:25 - 69:31: EPOP, JARID2, PCL2, or PCL1, 2, and 3 knockout embryonic stem cells. You see expressed the same
69:31 - 69:35: levels of SUZ12. When you knock out AEBP2, you see the expression of the other subunits,
69:35 - 69:41: and you retained all the methylation patterns. So AEBP2 is not required for the activity.
69:41 - 69:46: Similarly, EPOP, you'll see here, you don't require that. JARID2, you don't see a drop in
69:46 - 69:53: the global levels. When you knock out PCL2, you actually start seeing some PCL2 knocked out,
69:53 - 69:57: and then also here, PCL2 knocked out. When you look now at the binding of SUZ12,
69:57 - 70:02: you completely lose binding of SUZ12 when you knock out SUZ12, and you see diminished binding
70:02 - 70:08: in PCL2 and PCL1, 2, and 3 knockout. Importantly, you don't see any effect whatsoever in the JARID2
70:08 - 70:14: when you look at this gene, in fact, and you don't see anything on K27 trimethylation.
70:14 - 70:20: And what you also have to see here is that even though you seem to lose most of the SUZ12 binding,
70:20 - 70:25: you actually do retain most of the K27 trimethylation on these specific types.
70:26 - 70:32: If you do that over the genome, this is SUZ12 binding, and this is basically K27 trimethylation,
70:32 - 70:38: and even in the presence of PCL1, 2, and 3 binding, absence of PCL1, 2, and 3, you see a
70:38 - 70:45: reduction of K27 trimethylation, but you still see abundant K27 trimethylation in these cells.
70:47 - 70:51: So it seems like you have to knock out more of those core subunits. We have to try to distinguish
70:51 - 70:55: which of them are required, so we did a systematic knockout of these
70:55 - 71:02: various subunits. Here we have PCL1, 2, 3 knockout together with EPOP. We have JARID2,
71:02 - 71:06: AEBP2, and so on and so forth. And here we have all six subunits knocked out.
71:07 - 71:12: When you look at those, not going to go fast through them, you knock out SUZ12 binding,
71:12 - 71:16: and in the present when you knock out all six of them, you actually see that now you're actually
71:16 - 71:21: knocking out all K27 trimethylation activity in these cells. So it seems like that the end term
71:21 - 71:29: together with those six subunits is actually giving you the specificity of CpG island binding
71:29 - 71:35: of the PRC2 complex, and therefore the localization of H3K27 trimethylation.
71:38 - 71:43: So taking that into a broader perspective, what we've shown is that there's a parallel and
71:43 - 71:49: interdependent recruitment of PRC2 complexes to maintain repression of target genes.
71:50 - 71:55: So we start over here with PRC2. We have PRC2, which actually leads to the methylation of
71:56 - 72:02: monomethylation and trimethylation. The complex can be recruited to the CpG islands through the PRC2.1
72:02 - 72:08: complex, and potentially it can also be recruited through H2A monoubiquitylation through the binding of
72:08 - 72:13: JARID2 here as well. And monoubiquitylation, of course, is mediated by the PRC1 complex.
72:13 - 72:18: What we know in our cells, and also studies from other laboratories, is that when you knock out the
72:18 - 72:26: PRC2 complex here, you don't see a dramatic decrease in K27 trimethylation in embryonic stem
72:26 - 72:33: cells. So the idea that PRC1 is that setting up the PRC2 complex for mediating K27 trimethylation
72:33 - 72:38: is probably not the whole story, but it seems that the two complexes have to work together
72:38 - 72:43: before that happens. We know that canonical PRC1 can actually be recruited through the binding to
72:43 - 72:49: K27 trimethylation, which is shown here, and then we have this RYBP-YAF-KDM2 complex,
72:49 - 72:54: the variant one, which can be recruited into the CpG islands. So there seems to be a systematic
72:54 - 72:59: recruitment of this complex to actually maintain transcription repression. We also know that this
72:59 - 73:05: transcription repression is transient. It's happening all over the genome, and it's particularly
73:05 - 73:12: focalized on the CpG islands, and one of the particular things here is that these CpG islands
73:12 - 73:15: are devoid of, when they are in this state, they are devoid of
73:16 - 73:20: nucleosome occupancy, so it may be a spatial issue as well.
73:23 - 73:28: So let me go to the last part of my talk, which is about the biological effects of K27
73:28 - 73:34: methylation and acetylation, and one of the questions that we ask for a number of years is
73:34 - 73:39: how important is methylation of K27 methylation for the effect that polycomb group proteins have
73:39 - 73:45: on gene expression and cellular processes, and the work I'm presenting here will be the work of
73:45 - 73:50: Faizaan Mohammad, who recently left the lab, and then Aditya Sankar, who took over the project when
73:51 - 73:59: Faisan left, and the reason why we're asking that question, of course, is that there is an
73:59 - 74:04: increasing number of proteins that are methylated on lysines, and they have been identified through
74:04 - 74:09: the years, and here are just some reviews. We're actually discussing the function of
74:10 - 74:17: lysine methylation of non-histone target of other proteins, and the big question is, of course,
74:17 - 74:21: how important are these non-histone protein targets for the biological effects of the lysine
74:21 - 74:27: methyltransferases, or in other words, are the histones the key targets of the lysine methyltransferases?
74:28 - 74:34: If you look at PRC2, we know several non-histone proteins that are methylated by PRC2, for example,
74:34 - 74:42: GATA4, Elongin A, ROR-alpha, and JARID2, and the question is how important is H3K27 methylation
74:42 - 74:50: for PRC2 function? So one of the questions is how to study the importance of K27 methylation,
74:50 - 74:57: and you can, of course, do that in certain mutations in H3, lysine 27, you mutate it to
74:57 - 75:03: another amino acid, so in this way it cannot be methylated anymore, and that has been proven
75:03 - 75:08: feasible in Drosophila, but very, very difficult to do in mammalian cells, and one reason for that
75:08 - 75:14: is that in Drosophila, there's a single cluster of canonical H3, which has been studied now,
75:14 - 75:20: which can be deleted and then be rescued by a single back, whereas in mouse, we have 12 genes
75:20 - 75:27: for H3.1 and H3.2, and we have two genes for H3.3, so in mouse, you will have to mutate 28 alleles
75:27 - 75:33: to actually see that, or you have to replace several different areas on the genome, and it's
75:33 - 75:38: similar, very similar in human cells, and of course, there's one caveat by doing that, that
75:38 - 75:45: you see effects not only on lysine 27 methylation, but also on acetylation, so despite that,
75:45 - 75:51: there are some published data. In Drosophila, studies have shown that substitution of
75:51 - 76:00: K27 with H3K27R or H3K27A in canonical, that means the replication-dependent H3 leads to
76:00 - 76:07: a polycomb phenotype, which was not really carefully, let's say, studied if you look at
76:07 - 76:12: gene expression, but it led to the de-repression of some polycomb-type genes, and these are studies
76:12 - 76:22: from Jörg Müller’s lab and also from a lab at UNC. In Drosophila, substitution of replication
76:22 - 76:30: independent H3.3K27 with H3.3K27R was actually also found to lead to increased
76:30 - 76:35: expression of polycomb-type genes, also suggesting that H3.3 has a role in repression of polycomb-type
76:35 - 76:42: genes, and this has actually also been published previously by Gabrieli's lab, that has potential
76:42 - 76:48: role here. So, if you look at mammalian cells, the generation of H3.3K27R in mouse embryonic stem
76:48 - 76:55: cells shows that K27 methylation is not required for enhanced activity, which was recently published
76:55 - 77:02: by Bing Xu’s lab. They did not study the role of replication-dependent H3.3.
77:02 - 77:10: So, what did we do? Well, we were, or Faizaan was kind of, you know, bold enough to take a strategy
77:10 - 77:18: to generate H3.1 and H3.2K27R mouse embryonic stem cells, and what he did was to, he took
77:18 - 77:23: advantage, or he took advantage of a new adenine base editor, which was published now three years
77:23 - 77:31: ago by the new lab at MIT, and that editor can actually convert adenines, so change them to
77:31 - 77:37: guanines, and when you do that, it does that seven to ten nucleotides from the PAM sequence.
77:38 - 77:44: So, importantly, if you look at the H3.1 and H3.2, they are completely conserved in the nucleotide
77:44 - 77:48: sequence, so the hope is that you can go in with a single guide RNA and then mutate them all in
77:48 - 77:56: one go, and that's in fact what Faizaan did. He mutated them all, not in one go, but almost in one
77:56 - 78:02: go because he found a clever way of identifying them by doing PCR, and what is shown over here
78:02 - 78:10: are mouse embryonic stem cells where all the H3.1 and H3.2 lysines have been changed to an adenine,
78:10 - 78:16: and when you characterize them, you see almost no monomethylation, dimethylation, or trimethylation,
78:16 - 78:20: at least you don't detect it by Western blot. You don't see an effect on monomethylation,
78:20 - 78:25: which is consistent with the data of knockout PRC2. You don't see any effect. You don't see
78:25 - 78:32: an expression on H3.3 either. You don't see effect on K36 trimethylation or K4 trimethylation either,
78:32 - 78:37: but you see a relatively big drop in K27 methylation, and this is most likely the
78:37 - 78:42: K27 methylation, which is outside the enhancer regions, which is quite substantial as well.
78:43 - 78:53: So Faizaan, and now also joined by Aditya, wanted also to knock out H3.3, K27, and the way they did
78:53 - 79:02: that was first to knock out H3.3 in the cells, in these canonical K27R cells, and then basically put
79:02 - 79:09: back in K27 H3.3 wild type or a mutant H3.3 in this way. They had a system where they were sure
79:10 - 79:16: they were looking at the same cell populations. So when they did that and characterized these
79:16 - 79:20: cells that we've shown over here, you basically lose mono-, di- and trimethylation, and you have very
79:20 - 79:27: minimal K27 methylation. Of course, you shouldn't have any, so it shows the specificity of the
79:27 - 79:32: antibodies. If you want, these are two different monoclonal antibodies to K27 methylation.
79:33 - 79:39: I just want to show here, as a comparison, when you knock out SUZ12, you don't see exactly the
79:39 - 79:43: same. While you do, of course, lose mono-, di- and trimethylation, but you see a huge increase in
79:43 - 79:51: K27 methylation. Basically, in some way, perhaps suggesting that K27 or the activity of K27
79:51 - 79:56: methylation may actually be a contributor to some of the phenotypes you're looking,
79:56 - 79:59: that you need this K27 methylation mark actually leads to
80:00 - 80:05: the increased expression of some genes. I'll come back to that later. If you now look at
80:05 - 80:11: embryoid body formation, which is shown here, ES cells undergoing differentiation. When
80:11 - 80:15: you do that with a SUZ12 knockout, you see that they don't look like normal embryoid
80:15 - 80:21: bodies. These are the wild-type controls. Whereas if you knock out canonical K27, you see something
80:21 - 80:27: which is very similar, at least not as bad as you see the SUZ12 knockout. But if you
80:27 - 80:34: knock out all K27, H3K27, possibility to methylate those, you see something which
80:34 - 80:41: is very, very similar at the phenotypic level as you knock out SUZ12. It seems like K27
80:41 - 80:49: is required for embryoid body formation, giving you a polycomb phenotype.
80:49 - 80:54: What happens to transcription during this process? Well, what you can look at here are
80:54 - 80:58: something which is, I'll just orient you a little here. These are the wild-type cells,
80:58 - 81:04: and these are basically the SUZ12 knockout cells, so PRC2-. These first two here are canonical,
81:04 - 81:09: and the next two are the pan-mutant. If you look at the pluripotency markers, basically
81:09 - 81:15: when you see a reduction of those in, let's say, the canonical, whereas you don't see
81:15 - 81:19: this reduction in knockout SUZ12 cells, and you don't see that when you knock out
81:19 - 81:26: all K27, you substitute those. You see it's very similar on pluripotency and naive genes,
81:26 - 81:30: whereas you do see some differences between them when you look at the three embryonic
81:30 - 81:38: germ layers, whether you knock out basically pan or only the canonical one. So it seems
81:38 - 81:48: that they have different, let's say, roles in basically downregulating those genes during
81:48 - 81:49: differentiation.
81:49 - 81:54: But the point is that you see a failure to downregulate pluripotent and naive stem cell
81:54 - 81:59: markers in the pan-mutant, and you also see a failure to induce efficient expression of
81:59 - 82:04: genes involved in the formation of the three different embryonic germ layers. And of course, one
82:04 - 82:08: of the reasons could be here is that you lack K27 acetylation.
82:10 - 82:15: So the question we ask ourselves is the failure to induce differentiation due to changes in
82:15 - 82:22: K27 acetylation of those. In other words, is K27 acetylation there for maintaining
82:22 - 82:28: gene repression, or is K27 acetylation by activating genes required for differentiation?
82:28 - 82:33: So data in the mouse embryonic stem cells and Drosophila suggest that H3.3 K27 acetylation
82:33 - 82:41: is not required for gene activation. So how about K27 acetylation in general?
82:41 - 82:47: So what we did was actually use a system. So first here to look at what happens in the
82:47 - 82:52: mouse embryonic stem cells, what happens to gene expression in general. And so these are
82:52 - 82:57: the gene expression you see when you knock out SUZ12 in mouse embryonic stem cells, which
82:57 - 83:01: are growing. So you see upregulation in a large number of genes, and you see the same
83:01 - 83:06: in the canonical and the pan-mutants. You see the upregulation. You see more here in
83:06 - 83:09: the pan-mutant than you see in the canonical mutant.
83:09 - 83:13: And what we also show over here is that there's a highly significant overlap of gene expression
83:13 - 83:19: profiles of cells lacking SUZ12 and those expressing the K27R mutant. And we see strongly
83:19 - 83:24: upregulated genes in this steady state situation, if you want, in the mutants which do not have
83:24 - 83:31: K27 methylation or K27 acetylation. So it seems like at least in embryonic stem cells,
83:31 - 83:36: you do not require K27 acetylation to see the upregulation.
83:36 - 83:45: The question is, what happens when you now differentiate the cells? So what happens when
83:45 - 83:50: you really activate genes during a differentiation process? And what we did there was actually
83:50 - 83:56: take mouse embryonic stem cells and then differentiating them into epiblast-like cells. This is early
83:56 - 84:01: developmental stage, and we can do that. And we know that when we knock out the polycomb
84:01 - 84:06: repressive complex in these cells, they actually allow to differentiate in contrast to what
84:06 - 84:12: you see in the embryonic stem cells. So we know that this is not related to the methylation
84:12 - 84:18: of K27, but it could be that acetylation is required. So when we did that, we see absolutely
84:18 - 84:23: no effect of knocking out, of changing K27 in those cells. So we get a phenotype which
84:23 - 84:31: is very, very similar to when we knock out basically PRC2, namely nothing. But what we
84:31 - 84:36: do see is that we see the upregulation of exactly the same genes, and we see this downregulation
84:36 - 84:41: of the same genes during that differentiation process, suggesting that K27 acetylation
84:41 - 84:46: is not required for the induction of gene expression whatsoever in any of the cases
84:46 - 84:54: that we studied so far. So with that, I would like to end my talk by summarizing this last
84:54 - 84:59: part of my talk about PRC2 and K27 methylation. So what we have provided is the first study
84:59 - 85:07: of a pan-mutant K27 in any organism, and that pan-mutant K27, canonical mutant K27R,
85:07 - 85:11: and PRC2 knockout have very similar effects in gene expression in proliferating mouse
85:11 - 85:19: ES cells and epiblast-like cells. However, pan-mutant K27, but not canonical mutant K27
85:19 - 85:24: only replicates the polycomb knockout phenotype and transcription effects in the formation
85:25 - 85:30: of embryoid bodies. And what I think is also important, even though that's a nice mark
85:30 - 85:36: perhaps for enhancers and also for transcription, is that K27 acetylation is not required for
85:36 - 85:41: ongoing or activated transcription in self-renewing embryonic stem cells and during differentiation.
85:42 - 85:48: The historical role of K27 methylation is to maintain gene expression for a mechanism
85:48 - 85:52: that does not require the prevention of K27 acetylation, and the reason why I'm saying that
85:53 - 85:58: is that there have been studies suggesting that the major function of K27 methylation is
85:58 - 86:01: potentially actually to prevent acetylation, that is gene activation.
86:02 - 86:06: And then finally, we have demonstrated the feasibility of performing site-directed
86:06 - 86:11: mutagenesis on all histone genes in mammalian cells, thereby opening up for studies addressing
86:11 - 86:20: causal effects of specific histone post-translational modifications,
86:20 - 86:24: where we don't look at the protein itself, but we actually are looking at
86:25 - 86:30: the substrate we are affecting by mutating that and how it actually conveys a phenotype.
86:31 - 86:36: So with that, I'd like to end my talk by again acknowledging Jonas, Anne,
86:36 - 86:41: Lin, and also Tulin for the polycomb studies initially in the first part of my talk,
86:41 - 86:47: and Aditya and Faizaan for the second part, and our collaborators on proteomics,
86:47 - 86:51: in particular Ole Norberg-Jensen and Andrey Tvarsdovskiy from the University of Southern Denmark
86:51 - 86:55: and Simone Sidoli from the Albert, who is now at the Albert Einstein.
86:55 - 86:57: And with that, I'd like to take questions. Thank you.
87:01 - 87:05: All right. Well, thank you, Kristian, for the very stimulating talk.
87:05 - 87:11: We'll start with a few questions. The first two are from Simele Skilci, and she's asking,
87:11 - 87:19: is there competition among the PRC2 and PRC1 complexes? Although they are not recognized in
87:19 - 87:24: the same modification, she wants to know if they are building on the same place on the genome.
87:27 - 87:34: Well, if there is competition, I don't think we have any data on competition between two complexes.
87:35 - 87:40: I think what we have to think about is that these complexes are associated with chromatin very
87:40 - 87:46: transiently, and they may not sit there at the same time, but I don't think they compete with
87:46 - 87:51: each other for the binding. At least, we do not have any data suggesting that, but I don't think
87:51 - 87:57: they are allowed to sit there at the same time. I think there's a… I like… there's a very nice
87:57 - 88:03: review published in Cell recently by Tom Misteli discussing chromatin structure and the default of
88:03 - 88:09: chromatin in general, and I think for people also studying post-translational modifications
88:09 - 88:12: of chromatin, I think it's a very valuable review, which I'd like
88:13 - 88:18: to recommend very highly, and suggesting quite a lot about the transient
88:20 - 88:24: nature of many of these modifications, and particularly the binding to chromatin.
88:25 - 88:31: So, also, Simele wants to know, what is the role of the catalytic activity of PRC2?
88:31 - 88:35: If you use an inhibitor of EZH2, what happens to reprogramming?
88:36 - 88:44: Well, so, an inhibitor of PRC2 are actually reflecting very much many of the experiments
88:44 - 88:50: I showed you now, and it's reversible, at least in the studies we've done so far. One of the papers
88:50 - 88:56: is…I also discussed one published in 2018 in Nature Structural Molecular Biology. We have
88:56 - 89:02: extensively used inhibitors and showed the reversibility of PRC2 repression, losing that,
89:02 - 89:09: and as soon as we wash out the inhibitor, we do get gene expression again very quickly.
89:11 - 89:17: All right. Paul wants to know, what happens to phosphorylation of a histone H3 and serine 28
89:18 - 89:20: in a K27 replacement lines?
89:25 - 89:29: What happens to the phosphorylation of serine 28 in a…
89:30 - 89:30: H3.
89:31 - 89:33: In a K27 replacement lines?
89:35 - 89:41: Nothing. What we observed, we see that serine 28 is phosphorylated. We've actually checked that.
89:41 - 89:49: So, serine 28 is phosphorylated. It's also phosphorylated by consummate treatment. I'm
89:49 - 89:53: not really sure why the question is asked, but I can answer the answer to the question.
89:53 - 89:56: I'm…don't shoot the messenger. I'm just reading.
89:56 - 90:01: The next question, I think the answer is very obvious as well. Ping Lu wants to know what
90:01 - 90:06: happens to trimethylation when you get rid of EED. Given that EED is central to formation of
90:06 - 90:12: the complex, the answer is very evident. Yes. So, we knock out… so, this has been
90:12 - 90:18: shown by many just to recap that. You knock out EED, you knock out SUZ12, you completely lose
90:18 - 90:25: K27 methylation. If you knock out EZH2, you retain some methylation. If you knock out EZH2 and EZH1,
90:25 - 90:32: you completely ablate all methylation. So, an anonymous reader or attendee wants
90:32 - 90:40: to know that it could be that H3K27 acetylation is not required for ES renewal and differentiation.
90:40 - 90:46: And I think the answer from some of the work that we publish and Joanna Wysocka did, when you get
90:46 - 90:51: rid of monomethylation and enhancers, you abolish acetylation and enhancers and the cells are still
90:51 - 90:56: pluripotent. So, I assume you guys see similar things with the loss of acetylation in your system.
90:57 - 91:03: Well, pluripotent is…I would call them self-renewing. If you think about pluripotent,
91:03 - 91:06: that they're expressing the pluripotent transcription factors, yes. They're doing fine.
91:06 - 91:09: And the renewal state of the ES, you see…
91:12 - 91:15: Sorry? For ES renewal function.
91:15 - 91:17: Yes, no problem. No problem.
91:18 - 91:22: But we do…we are not able to use the system to understand whether K27 acetylation
91:23 - 91:28: is required for, you know, generating formation of the free embryonic germ layers,
91:29 - 91:31: which is what I call pluripotent.
91:32 - 91:37: I mean, I might add to that that we showed in 2017 work by Ryan Rickle in my lab that when
91:37 - 91:42: you get rid of the monomethylation-specific line enhancer, you abolish acetylation and
91:42 - 91:47: enhancer. And we're able to generate flies that are kicking around, looking just as normal,
91:47 - 91:52: and they have no monomethylation, no acetylation on their enhancer. So, that says that pluripotency
91:52 - 91:57: and self-renewal and differentiation does not seem to be affected, at least in *Drosophila melanogaster*.
91:58 - 92:04: I think the important thing with our study is that we completely abrogate K27 acetylation
92:04 - 92:10: as well. And as you know, that K27 acetylation is not only associated with promoters or with
92:10 - 92:11: enhancers, but also promoters.
92:12 - 92:12: Exactly.
92:13 - 92:14: But…
92:14 - 92:16: The motor function, we did not look at. I guess we should.
92:17 - 92:22: And the other thing is that, of course, it doesn't say that P300-CBP are not required,
92:22 - 92:26: but because it's known that P300-CBP are doing many, many other things than K27 acetylation.
92:27 - 92:32: All right. I'm going to ask the last question in here by Andrea Punti, who asks,
92:33 - 92:40: in your work, the TKO of the PCL gene in a mouse's embryonic stem cells doesn't affect
92:40 - 92:47: EPOP level. However, in a similar work from the Bracken group, they do see a massive reduction in
92:47 - 92:54: EPOP in the TKO of the PCL mouse embryonic stem cells. And how do you define this discrepancy?
92:56 - 92:56: I don't.
92:59 - 93:05: We haven't done the studies that Adrian's lab has done. We did publish it back-to-back, though,
93:06 - 93:11: and I actually did not notice that there was this discrepancy. I'll have to talk to Adrian about that.
93:11 - 93:15: Okay. So, there are a lot more questions online for you. Very stimulating talk. We won't be able
93:15 - 93:20: to go through everything. Maybe at the discussion session at the end, after Danny's talk, we all
93:20 - 93:24: can come back and address some of these things. But I highly encourage you to go online and answer
93:24 - 93:28: some of these very interesting questions. Thank you, Kristian.
93:28 - 93:29: You're welcome. Thank you.
93:29 - 93:36: All right. So, we're going to move to the final talk of the session by Professor Danny Reinberg.
93:36 - 93:40: Is Danny around? I don't see his name online. Yes, I am around.
93:40 - 93:48: All right. Welcome, Dr. Reinberg. All right. Danny is, if I introduce Luciano as the most handsome
93:48 - 93:53: and Kristian as the fittest, I define Danny as one of the kindest and caring colleagues in
93:53 - 94:01: the field. I've known Danny for over 30 years now, and I think his lab has contributed tremendously
94:01 - 94:06: to our understanding of transcription, transcription machinery, and now epigenetic
94:06 - 94:12: factors and regulation in this process, and he's going to tell us about. But as far as,
94:12 - 94:17: I guess, his caring part goes, I see Danny on many meetings, and he always asks the most astute
94:17 - 94:22: questions right on the heart of the mechanism of someone who's given a talk. And I think
94:22 - 94:26: he does that because he really cares about the field and understanding and trying to set a
94:26 - 94:33: mechanistic insight to what we have learned. He was a graduate student in Horowitz's lab
94:34 - 94:40: in Einstein before Horowitz moved to Memorial Sloan-Kettering, and then he was a postdoctoral
94:40 - 94:46: fellow in Bob Rader's laboratory, and this is around the time where there's a race in the world
94:46 - 94:51: on identifying all the basal factors that are involved in the initiation of transcription, and Danny
94:51 - 94:57: did a fantastic job in defining how many of these TF2A, B, C, D, E, F, Gs, and reconstituting
94:57 - 95:03: transcription and moving forward. And his lab around the late 1990s got very much interested in
95:03 - 95:07: the role of transcription through chromatin. I think it was the first approach, and that came
95:07 - 95:13: up with the work of George Orphanides, where he identified this factor named FACT, facilitate
95:13 - 95:17: transcription through chromatin, and he got very much interested in the chromatin function in the
95:17 - 95:22: process and beautiful work that he did when Yizhang was a postdoc in his lab, defining the
95:22 - 95:27: polycomb and moving forward from that process. And since then, he has done a tremendous amount of work
95:27 - 95:33: trying to understand the role of epigenetic factors, histone modification, and epigenetic
95:33 - 95:38: memory, and I think he's going to tell us about some of his work on which mark is really actually
95:38 - 95:44: involving memory and function in that process. For his fantastic work, he's very well recognized
95:44 - 95:49: for the work he has done. He's an endowed professor at New York University in the
95:49 - 95:54: Department of Biochemistry there, and he is a member of the National Academy of Sciences,
95:54 - 95:58: member of the National Academy of Medicine, and many other honors and accolades. So,
95:58 - 96:02: Danny, great to have you here today, and we look forward to your talk.
96:05 - 96:06: I don't see my slides.
96:10 - 96:11: We don't see your slides either.
96:12 - 96:13: Well, they are there.
96:15 - 96:18: He punched that green button on the center. It says share screen.
96:18 - 96:21: I am, I am, I am, and then I don't see them. Oh, there.
96:22 - 96:22: There you go.
96:22 - 96:30: All right, he didn't know. Well, I will start by thanking Tony and the Abcam group for giving me the
96:32 - 96:39: ability to present our work in here and put our studies within the context of what Adi described
96:40 - 96:47: very well, which is mechanism. What we're talking in here is, or the meeting is centered from
96:47 - 96:57: epigenetics or epigenetics from bench to clinic, and in order to be able to go to clinic and do
96:57 - 97:02: everything in the clinic with some of the drugs that we believe that we develop, we need to really
97:02 - 97:09: understand what the drug is doing. We need to understand the mechanism. We need to understand
97:09 - 97:14: what is a stable modification that is inherited, and so on.
97:15 - 97:22: So, what I'm going to try to convince you today is two things. First, which of the many histone
97:22 - 97:29: modifications are epigenetic? And secondly, I'm going to correct something that I have here twice,
97:30 - 97:37: which is that we know quite well, and we have a lot of knowledge, not only from my lab,
97:37 - 97:44: but from other labs, how PRC2 is recruited to specific genes. There are different mechanisms,
97:44 - 97:51: and we also know how, from the recruitment side, which is rich on prime methylation,
97:51 - 97:59: then PRC2 spreads and establishes dimethylation. That's in some genes. There is an involvement
97:59 - 98:06: of RNA that Tom Cech, myself, and others have studied, and also involvement of RNA in
98:08 - 98:15: some inbreeding genes. But, well, despite that, I'm going to now tell you that since
98:15 - 98:20: we discovered, or since the field discovered, that there are many histone modifications,
98:20 - 98:30: the field has started speculating or believing that each of these modifications are epigenetic.
98:31 - 98:38: And the question is, what is epigenetic, and which histone modification is epigenetic?
98:38 - 98:43: So, what I'm going to say is that histone modifications within nucleosomes,
98:43 - 98:49: which is the repeating unit of chromatin, are associated with the formation of different
98:50 - 98:58: chromatin states. Are these chromatin states inherited? Well, there are different chromatin
98:58 - 99:03: states, as I said. One of them is transcriptionally active, or known as euchromatin,
99:04 - 99:13: and this is the active one that generates a transcript or many transcripts. And that is
99:13 - 99:19: all dependent on the ability of an activator or regulator to bind to the promoter sequence.
99:20 - 99:27: When this happens, now the activator recruits histone acetyltransferase coactivators,
99:27 - 99:34: which acetylate residues around the promoter, resulting in the recruitment of the RNA polymerase
99:34 - 99:40: these two transcription complexes. Upon initiation of transcription, very early,
99:41 - 99:47: concomitant with the formation of the first phosphodiester bond, there is the recruitment
99:47 - 99:53: of a specific methyltransferase that methylates lysine 4 of the histone H3,
99:54 - 99:59: and this serves as a docking site for other proteins
100:00 - 100:09: that mediates or facilitates the downstream effects such as elongation, splicing, etc.
100:10 - 100:16: Soon after initiation, the polymerase stops, and now it's going to be converted into a
100:16 - 100:24: elongation complex, and that is mediated in part by the recruitment of a
100:24 - 100:30: different methyltransferase that catalyzes lysine 36 of the histone H3.
100:32 - 100:40: So what I'm trying to say here is that the transcription machinery recruits the histone modifiers.
100:42 - 100:50: Activator binding to its cognate DNA site is necessary to establish and maintain
100:50 - 100:57: these histone modifications. All this process and all these modifications that you hear,
100:57 - 101:07: methylation of lysine 4, lysine 36, acetylation, and so many others, are directed by transcription,
101:07 - 101:15: which is directed by the activator that binds to DNA. It's a DNA mediated process. By definition,
101:15 - 101:22: it is not epigenetic. So none of the modifications that are associated with transcription
101:24 - 101:27: are transmitted epigenetically. They are short-lived.
101:29 - 101:35: One, is there any evidence for histone post-translational modification attaining an
101:35 - 101:42: epigenetic inheritable cellular state? I would say yes. Two modifications, at least two modifications,
101:42 - 101:49: are inherited. Those implicated in the formation of repressive chromatin states.
101:50 - 101:58: One of them is the machinery that forms what is called facultative heterochromatin,
101:58 - 102:03: which is primarily on the arms of the chromosome, and is mediated by the polycomb repressive
102:03 - 102:11: complexes 1 and 2. And in here, I depict that one of the enzymes or one of the components of the
102:11 - 102:19: complex PRC2, it's an enzyme, methylates lysine 27 of the histone H3. The other one is the
102:19 - 102:26: constitutive heterochromatin that is mediated by the SUV3-9 family of enzymes, which are two,
102:26 - 102:35: SUV3-9-1, H1, and H2, and that they methylate lysine 9 of the histone H3.
102:36 - 102:43: Now, what is unique to repressive histone modifications consistent with inheritance?
102:45 - 102:53: Well, PRC2, as well as the canine trimethylase SUV3-9 H1 and H2,
102:55 - 103:03: contains a unique write and read mechanism. The PRC2 core subunit, which I depict here,
103:04 - 103:12: as you have heard, methylates lysine 27 of the histone H3. In this case, it's a trimethylation
103:13 - 103:21: which is established, and the product of this reaction is bound by an essential subunit of PRC2
103:21 - 103:33: EED, and upon binding to the product of the ECH2 or ECH1 enzyme. Now, the entire complex
103:34 - 103:42: goes and has a conformational change, which results in a stimulation. This binding is
103:42 - 103:49: mediated through an aromatic cage that, in collaboration with Steve Kamplin, we discovered
103:49 - 103:56: many, many, many years ago, is shown in here. And binding, as I said, results in a stimulation
103:56 - 104:05: of the complex allosteric activation. Now, to further scrutinize which modifications
104:05 - 104:14: are inherited, we studied parental nucleosome segregation in vivo. We devised a novel approach
104:14 - 104:22: to mark nucleosomes at single gene resolution in vivo, never done before, and follow their
104:22 - 104:34: fates during and after DNA replication. What we did was to prime the endogenous canonical H3.1
104:34 - 104:46: and H3.2 for biotinylation. So, any histone H3.1 or H3.2 contains a pack that can accept
104:46 - 104:54: biotinylation. So, all the chromatin, in principle, can be biotinylated. Now, we defined specificity
104:54 - 105:04: by creating a cell where we can dox-induce the expression of a modified Cas9 containing a
105:04 - 105:12: degradation domain that is fused to the enzyme that biotinylate here, BuA. And this happens
105:12 - 105:22: exclusively at G1. Now, specificity is directed by Cas9, which we direct to specific genes,
105:23 - 105:32: usually spanning 5, 10, or 15 KB. And, in addition, as I said, the specificity is induced
105:32 - 105:41: by nucleosome biotinylation at the selected single locus at G1 because we direct Cas9 to a specific
105:41 - 105:49: locus, yes, one locus. Then we follow nucleosome segregation during and after DNA replication
105:49 - 105:59: by qPCR and ChIP-seq. And what we encounter was actually quite surprising and very telling to the
105:59 - 106:06: process of epigenetics. If we look at ES cells and we look at repressed genes, in this case,
106:06 - 106:13: I'm going to show you two examples, GATA2 and GATA6. I chose these two because they can be
106:13 - 106:19: induced to be activated in ES cells, as you will see in a second. And then we have the cells
106:19 - 106:31: stop at G1. We induce the biotinylation of the GATA2 or GATA6. And this is what you observe.
106:32 - 106:41: Here is biotinylation specific at the promoter of the GATA2 and also around regulatory sequences
106:41 - 106:49: and the promoter of the GATA6. This biotinylation included not a large, large number of nucleosomes
106:49 - 106:56: simply because we're dealing with repressive chromatin. Now, when we release the cells from
106:56 - 107:03: G1 and we allow them to go through replication, and we now follow the fate of the nucleosome that
107:03 - 107:12: were labeled with these genes, we got the surprise that they segregate to their same location.
107:15 - 107:23: Now, when we repeat the experiment, but in this case, we activate the gene by adding retinoic
107:23 - 107:31: acid. We did the same experiment. We stopped the cells at G1. We allowed the cells to be
107:31 - 107:39: biotinylated. And what we observe first is a large number of nucleosomes that get biotinylated,
107:40 - 107:44: most likely a reflection that the chromatin is open. We're dealing with euchromatin.
107:45 - 107:50: But when we follow the fate of these nucleosomes, we couldn't follow them.
107:51 - 107:56: We concluded that these nucleosomes segregate randomly through the genome.
107:57 - 108:04: And our overall conclusion for this part is that nucleosomes in repressed chromatin domains
108:04 - 108:12: in contrast to active chromatin domains segregate to their same domain. Thus, the structural
108:12 - 108:21: feature of repressed chromatin are inherited. Unlike activation, repression is not reinitiated.
108:22 - 108:28: It is maintained as what we learned from the early studies on polycomb in *Drosophila*.
108:28 - 108:31: Repressive histone modifications are inherited.
108:34 - 108:41: And the write and read mechanism of PRC2 fully restores repressive chromatin domains after the
108:41 - 108:47: end of replication. This is emphasizing what I said. We know how PRC2 is recruited at certain
108:47 - 108:53: genes. We know that it's the positive trimethylation. Then it starts expanding,
108:55 - 109:01: moving through the gene, establishing dimethylation. And this is mediated by the
109:01 - 109:09: read and write mechanism of PRC2, and most likely similar to SUV3-9. Now,
109:09 - 109:16: we want to make sure that what I told you is correct. How can we reinforce the epigenetic
109:16 - 109:20: process that we convince ourselves that what we're looking at is correct?
109:21 - 109:27: So, do specific chaperones associated with the replication formulate S phase? That was the first
109:27 - 109:33: question that we asked. Why? Because we know that most of the transactions of the histones
109:34 - 109:42: and nucleosomes from chromatin are mediated by chaperones. So, what we did is represented here.
109:43 - 109:55: You've seen endogenous FLAG-TAG3 histone. We release the cells, sorry, we stop the cells at
109:55 - 110:02: G1, we accumulate at G1, and then we release them for two hours, where we remove an
110:02 - 110:09: aliquot of the cells. These two hours represent an early S phase where most of the active genes
110:09 - 110:21: are being replicated. And then at six hours, it represents late S phase, again, most of the
110:21 - 110:28: repressed genes are being replicated. And then we did a sequence, we broke the cells, we made extra,
110:28 - 110:35: and we did a sequential IP from both early and late S phase, as well as another control where
110:35 - 110:46: we used just FLAG. And what we observed is we immunoprecipitated the proteins associated with
110:46 - 110:55: FLAG-H3.1, then we FLAG eluted the proteins, and then we divided this into two fractions.
110:56 - 111:02: One fraction was further precipitated by antibodies that are specific for
111:02 - 111:10: di- and tri-methylation of the histone H3. And the other half was immunoprecipitated by a
111:10 - 111:19: component of the replication machinery, MCM2. We did the analysis, the proteomic analysis of
111:19 - 111:26: these experiments, and what we concluded is that there are apparently 17 candidate proteins
111:26 - 111:31: that are exclusively associated with late replicating repressed chromatin.
111:32 - 111:43: Most rewarding was that one of these candidates is a chaperone known as NPM1 nucleophosphin B23,
111:43 - 111:50: which was isolated approximately 20 years ago by Kiyoshi Nagata when he was studying
111:50 - 111:58: replication of ribosomal RNA genes that replicates late in S phase. You can see here
111:58 - 112:07: that NPM1 is associated exclusively with late replicating DNA, not early.
112:08 - 112:17: Now, what is NPM1? It's a protein of about 300 amino acids, contains three main domains,
112:17 - 112:25: an oligomerization domain, which is called the core domain, then contains the central domain,
112:25 - 112:32: which is acidic, is where histone chaperone activity resides, and then at the C-terminus,
112:32 - 112:40: which binds to nucleic acid, and also contains a nucleolar localization sequence.
112:41 - 112:53: NPM1 facilitates DNA replication during late S phase, and NPM1, interestingly, is mutated in 35%
112:53 - 113:01: of all adult acute myeloid leukemia (AML). So the first question to understand what we are
113:01 - 113:08: dealing with is, does NPM1 interact with PRC2 and/or with the replication fork?
113:09 - 113:15: The answer to that is yes, with both. Here you have a reciprocal immunoprecipitation
113:15 - 113:21: where we use purified PRC2 and NPM1, or we supplemented this with extract
113:22 - 113:27: derived from late S phase, and we did an immunoprecipitation using antibodies against
113:27 - 113:35: NPM1, and you can see that EED is specifically immunoprecipitated. The same thing was done
113:35 - 113:43: when we used antibodies against EED. You see that NPM1 is specifically immunoprecipitated.
113:43 - 113:52: So the model that we put is as follows. NPM1 associated with the replication fork, and NPM1
113:54 - 114:03: takes the nucleosome that is ahead of the replication fork, and then it moves it behind
114:03 - 114:14: the replication fork, either to the leading or lagging strand. Now, because PRC2 interacts with
114:14 - 114:27: NPM1, and because PRC2 associates with EED through the primed methylation, then PRC2 recognizes
114:28 - 114:38: the newly deposited parental nucleosome, which when in contact with naive nucleosomes can transmit
114:38 - 114:44: the modification to them, and then we establish again the repressive chromatin domain through
114:44 - 114:52: the read and write mechanism. Now, to further analyze this, we used an inducible Dox system
114:52 - 115:02: for NPM1. NPM1 is an essential gene, so we created homozygous NPM1 genes containing an
115:02 - 115:12: AAD tag, and we used an endogenous titer locus. And here you see that induced degradation of NPM1,
115:12 - 115:20: and by four hours, a large amount of the protein is gone, six hours is gone. Now, what happened?
115:21 - 115:29: Well, without NPM1, the cells accumulate in S phase. And G1, sorry, here's the wild type,
115:29 - 115:38: and here are the DNA contact gene cells without NPM1 accumulated in G1. And here is another way of
115:38 - 115:45: representing the same thing that I told you. Now, cells in the absence of auxin containing NPM1
115:46 - 115:56: have approximately 73% of the cells in S phase, whereas in the absence of NPM1 it is about 43%.
115:57 - 116:05: These coincide with a reduction in the normal wild type cell of cells accumulated in G1,
116:06 - 116:12: and here we've got an increase in the cells accumulated in G1 because they are stuck.
116:13 - 116:22: Now, G2 was not affected. So, what happened to PRC2, and what happened with K27 methylation?
116:22 - 116:28: Another methylation mark? So, here is again the same or similar experiment that I described to
116:28 - 116:36: you. We induced degradation of NPM1, six hours, 24 hours is gone. None of the components of PRC2
116:37 - 116:43: are affected. However, when we look at dimethyl and trimethyl methylation of PRC2,
116:44 - 116:54: we observe surprisingly that at 6 and 24 hours, dimethylation and trimethylation of PRC2
116:54 - 117:01: were gone. Here is the H4 marker. Now, we're loading control. Now, we'll repeat the experiment
117:01 - 117:09: again, and we look for other modifications. Again, after six hours of depletion of NPM1,
117:10 - 117:20: there is a loss of H3K27 trimethylation. Surprisingly, this was unique for K27 trimethylation
117:20 - 117:27: because the other repressive mark, trimethylation at lysine 9 that follows the
117:27 - 117:34: same principles of write and read, was not depleted, suggesting that there must be another
117:34 - 117:43: chapter or another mode of removing that nucleosome. And here, we know that K9 exists
117:43 - 117:50: at telomeres and centromeres, and RNA may be involved, and transcription factors may be involved
117:51 - 117:58: in dealing with the nucleosome. Now, another active mark, like dimethylation at K36,
117:58 - 118:05: was not affected. Now, the last experiment, which I'm going to tell you is that we again
118:06 - 118:16: blocked the cells at G1. We now, upon release at auxin, we allowed them to proceed for six hours,
118:16 - 118:23: and then for another six hours. So, we took aliquots at 6 and 12 hours, and we looked for
118:23 - 118:30: expression of genes that we know that are repressed in S phase. For example,
118:30 - 118:38: GATA2 and GATA6, as I showed you early, and the HOX, the HOX family of genes HOXA and HOXB.
118:38 - 118:46: And here, you can see that at 6, 12, in each case, we saw an increase in gene expression as expected,
118:46 - 118:54: because there was a loss of H3K27 trimethylation. Now, a control gene, like OPT4,
118:54 - 119:00: was not affected. Now, I'm going to skip these slides, but show you that it's involved in
119:00 - 119:08: AML and disease, and that this is because there's a mutation at the C-terminus of NPM1
119:09 - 119:17: that precludes the accumulation of NPM1, and the nucleus is accumulated in the cytoplasm.
119:18 - 119:24: And this is one of the driver mutations in AML. And I'm going to stop here,
119:25 - 119:32: and acknowledge the people that did the work. Actually, all the work that I presented was done
119:32 - 119:41: by Thelma Escobar, and when she joined the lab, was helped by Ozgur Oksuz, who also trained her
119:41 - 119:49: in the chromatin, as she came from an immunology lab, and also trained her in how to look at genes,
119:49 - 119:58: how the genes of PRC2 are recruited, and how PRC2 moves away from the recruitment sites. After that,
120:00 - 120:06: Thelma continued with her own studies, which I described here, and in here was helped by
120:06 - 120:14: bioinformaticians, Ricardo Saldaña-Meyer and Nicolas Decostes, which are two bioinformaticians,
120:14 - 120:22: now NIC at EMBL in Rome, Ricardo's company in Boston, and all this work was also
120:22 - 120:30: influenced by Roberto Bonasio, which we discussed tremendously, because he likes to
120:30 - 120:35: science and talk about science, and he's a very good bioinformatician too. Thank you very much,
120:35 - 120:44: and I'm happy to answer any questions. And, sorry, a very important collaborator is from the
120:44 - 120:50: Nucleus Lab, which is Nikita Vasilat, which we're using a cross-linking approach that they have
120:50 - 120:58: to map exactly the interactions of the proteins that we talked about, PRC2 and NPM1,
120:58 - 121:07: with the actual replication fork. Thank you. Thank you, Danny. All righty, so we have a lot
121:07 - 121:14: of questions here for you. We'll start with Anne Lawson, who's asking, do you have a mechanistic
121:14 - 121:21: explanation for the very rapid loss of global di- and trimethylation as a result of NPM1
121:21 - 121:27: degradation, and is this a demethylase-induced process, or is it like a histone replacement?
121:28 - 121:34: I wish I had the answer. We do not know. This was a very surprising result for us,
121:34 - 121:41: because the rate of disappearance of K27 is much, much faster than what we have seen with
121:41 - 121:49: any inhibitor of K27. We do not believe that it is due to histone replacement. We do believe,
121:49 - 121:56: perhaps, we have some preliminary results that some of the KDM6s, which are the demethylases,
121:56 - 122:02: okay, are being affected, activated here. But this needs to be further expanded.
122:03 - 122:11: All right, Noel Sandovich wants to know if the H3K9 trimethylation and the H3K27
122:11 - 122:17: trimethylation, to your opinion, are the only epigenetic marks. And I guess he's asking,
122:17 - 122:21: what's up with the histone code, if there is only one or two here?
122:21 - 122:28: Once upon a time, I believe there was a histone code very early when it was proposed. Then,
122:28 - 122:35: after what the entire field has come, okay, there is no histone code, unless it's very hidden under,
122:36 - 122:44: I don't know, maybe the water, okay. Now, I believe that at this moment, we have the
122:44 - 122:53: mechanism by K9 and K27 being transmitted unepigenetically. They are long-lived marks.
122:54 - 123:01: All the other marks that people call epigenetic acetylation, K4 methylation, K36 methylation,
123:01 - 123:10: and so on, they are very short-lived, and they are deposited by the process of transcription.
123:10 - 123:17: You cease, you stop transcription, they disappear. And most importantly, they are mediated by the
123:18 - 123:24: transcription factor that binds to a specific sequence, and therefore is DNA mediated,
123:25 - 123:28: and not epigenetic, therefore.
123:29 - 123:32: So, Ebrahim Natu wants to know,
123:33 - 123:38: how is euchromatin conserved, and what are the activating histone marks that are conserved?
123:39 - 123:45: How is euchromatin conserved, and what are the activating PTMs that are conserved?
123:45 - 123:50: Yeah, well, first, the activator will recruit the transcription machinery, will recruit
123:50 - 123:55: histone acetylases that will acetylate nucleosomes surrounding the transcriptional start site,
123:55 - 124:01: then the polymerase gets recruited. The polymerase has the CTD, which gets phosphorylated at multiple
124:01 - 124:09: sites, but two, serine 5 is recognized by one of the methyl transferases that methylates lysine
124:10 - 124:19: 4 of the histone H3. Moving from the promoter towards the establishment of
124:20 - 124:25: an elongation complex, promoter escape, there is the recruitment of another histone
124:25 - 124:32: methyltransferase, now the CTD that is phosphorylated serine 2, and that establishes
124:32 - 124:41: K36. All of this is accomplished with acetylation and establishment of other modifications that are
124:41 - 124:50: associated with activation. But we need to now start going, leaving ES cells, and start looking
124:50 - 124:57: at specific cell types, and try to understand which of these modifications, active modifications,
124:57 - 125:06: are operating in different cell lineage cell types. Right, so a question from your former
125:06 - 125:12: graduate student, Rima, who wants to know that you have shown the complete loss of trimethylation of
125:12 - 125:20: K27 after six hours, after auxin induction of the NPM1, given that this is a very short time point,
125:21 - 125:25: could this be induction of protection rather than transmission of the mark?
125:26 - 125:33: Protection, protection of what? I guess maybe she's thinking that something binds to H3K27,
125:33 - 125:38: now you cannot see it anymore, a lot like that INHATS story, you remember? Yeah, the INHATS story,
125:38 - 125:43: of course I remember. I think, I think, I don't know, this is what I'm surmising from her question,
125:43 - 125:49: that maybe the mark is protected and your antibody can't get to it. Well, it would be very surprising,
125:49 - 125:55: but I cannot rule it out. I think that the evidence points to perhaps
125:56 - 126:00: something that we do not know, something that's going to be completely novel,
126:00 - 126:09: as well as the induction of some of the KDM6, which are important for demethylation. Yes, give us time. I guess protection,
126:09 - 126:14: if you have a Western blot with the data showing that K27 went away, protection doesn't make sense in
126:14 - 126:20: average. All right, I'm going to ask one more question. There's a lot of questions here for you,
126:20 - 126:24: Danny, I want to open up the session to everyone, so we have, I guess, a joint question session.
126:24 - 126:30: I'm going to read one question from an anonymous attendee, but I encourage you to go
126:30 - 126:35: ahead and answer everybody else's online. It says, one very detailed technical question,
126:36 - 126:43: could you suggest a commercial EED antibody for ChIP? So, let me send an email to the lab,
126:43 - 126:51: and they will be able to tell you, okay. We do have our own EED antibody, which we use sometimes.
126:51 - 126:57: I think that there are other EED antibodies that are better than ours, so I have
126:57 - 127:06: to ask the lab. I do not remember which EED antibody we buy, maybe from Abcam, maybe from
127:06 - 127:10: some other place, I don't know. All right, and Rima writes, I think she means by her protection,
127:10 - 127:16: protection from demethylation, maybe something happens on chromatin that you cannot demethylate.
127:18 - 127:24: Give us time, Rima, you know that very well. Yeah, all right, so I'm going to open up the
127:24 - 127:30: session to all the five speakers that we have, and I'm going to open up a session with a question.
127:30 - 127:36: for all of you here, and you can answer it whenever you want, and we can move it up to the
127:36 - 127:41: floor, and we have questions from the colleagues there, and here's a thought for you. So, 70 years
127:41 - 127:48: ago, the role of Polycomb was established as a negative regulator of developmental gene expression,
127:48 - 127:53: and then the Trithorax came as a positive regulator, and one of the earlier mutations that
127:53 - 127:58: was seen in the Trithorax was within the SET domain, which later was thought that this mutation
127:58 - 128:05: within the SET domain of Trithorax renders the enzyme inactive, and this catalytic function
128:05 - 128:09: is gone, and this is what you see. Work from our lab and others have shown that actually that
128:09 - 128:14: mutation in Drosophila renders the enzyme null. When you make that point mutation, the enzyme
128:14 - 128:19: degrades, and you don't have any Trithorax, and through a series of papers that we and others
128:19 - 128:26: have published in past few years is that you can go and make a catalytically dead Trithorax, TRR,
128:26 - 128:33: or SET1, and show that the loss of K4 methylation on enhancers, loss of K4 methylation
128:33 - 128:39: on some promoters, do not have any effect on gene expression, and that Trithorax actually,
128:39 - 128:45: in some sense, in the case of the MLL2 and MLL1, which is component related to Trithorax and
128:45 - 128:51: Drosophila, function as a repressor for the Polycomb machinery and DNA methylation machinery,
128:51 - 128:56: and the mark itself is not functioning in that process. I have not seen that in the Polycomb
128:56 - 129:01: group, because you guys do not have a point mutation that render your enzyme dead,
129:02 - 129:05: but your enzyme, the complex, is still functional.
129:07 - 129:16: We do have a mutation that renders the complex inactive, a mutation within the active site of
129:16 - 129:22: PRC2, TCH2, I don't remember exactly. What is that mutation? Because I haven't seen it,
129:22 - 129:28: Danny. Can you tell me what mutation has been in there? I will send you an email.
129:28 - 129:31: Kirshen and Luciano, do you guys know this mutation?
129:32 - 129:36: We used it as well, so I mean, I think Anna, who was asking the question before,
129:36 - 129:39: she can send you the amino acids we had changed.
129:41 - 129:48: So you have mutations within Polycomb that render the enzyme dead, but the complex does
129:48 - 129:56: not fall apart? Yes. And what happens to that thing? Because I know a mutation that makes
129:56 - 130:01: the enzyme hyperactive, but I did not know of a mutation that rendered the enzyme dead.
130:02 - 130:09: It gets recruited to the same sites, and cannot spread and remains in those sites.
130:10 - 130:19: So recruitment and enzymatic activity are independent processes. You recruit PRC2 to the
130:19 - 130:24: sites. In this case, there is no RNA, because there is no RNA being made, so there are other
130:24 - 130:31: recruitment methods, and it just accumulates there. For time, we have not followed for how long,
130:32 - 130:41: but then I believe disappears. And that does not affect the complex formation, the EZ,
130:41 - 130:47: the SUZ12, all of those are stable within the complex? The core complex. Okay, I do not want
130:47 - 131:02: to talk about the non-core components like PHF or GRV2, because we believe, and actually we know,
131:02 - 131:09: that MEPF2 are still associated with the complex, okay? But I do not know about the others.
131:09 - 131:22: Sorry, and it just ended on Q&A. I see two, H6982A and then EZH731. Actually, I have to go
131:22 - 131:28: read about this thing and look at the detail of it. So now, so when you delete EZH1, EZH2,
131:28 - 131:36: you do not affect stem cell renewal, but you cannot differentiate. Yeah, true. So now when
131:36 - 131:41: you make these mutations, that you abolish K27 trimethylation, what happens to differentiation?
131:43 - 131:48: The answer comes in a moment, probably on Q&A. I'm sorry? The answer probably comes from Q&A
131:48 - 131:55: in a moment. Okay. The cell will die. I don't know. No, when you try to differentiate, the cell dies.
131:56 - 132:03: They cannot differentiate, and then they will die. So you would not imagine that if you don't have
132:03 - 132:09: trimethylation of H3K27, the cell would differentiate. No, the cell will not differentiate,
132:09 - 132:12: the cell will die. We have mutations as such, and I'm not going to talk about what they are,
132:12 - 132:18: that we can abolish trimethylation, and we'll see some differentiation. And that's- H3K27M?
132:19 - 132:26: No, no, no, no. This is H3K27 trimethylation loss. But we are not clear on the dye and monomethylation,
132:26 - 132:32: which could there be function associated with the monomethylation versus a trimethylation,
132:32 - 132:36: but there are conditions of mutants that we can see that there is a loss of trimethylation
132:36 - 132:41: with no or actually very little effect in differentiation, and that we need to look at
132:41 - 132:47: in further detail. I think so. I'm going to open up the session to the floor, people who can ask
132:47 - 132:56: questions from Danny, Luciano, and Kristian, and we had two other speakers. Are they online?
132:57 - 133:05: Hopefully they're online. Yes. All right. Nice. Go ahead. I've got a question for you. I mean,
133:05 - 133:11: the meeting also has clinic in its name, and we know that all of you are involved in drug
133:11 - 133:17: discovery. Can you discuss the potential of drugs against the Polycomb complex, both in the clinic
133:17 - 133:24: and in regulation of mechanism and interesting functions? What are the potentials of these small
133:24 - 133:30: molecules, and are they as good as kinase inhibitors, for example? Can you compare them,
133:30 - 133:36: if you can, through your knowledge? Well, I want to start because I can and I have discussed this
133:36 - 133:45: with some fellows from companies and some friends, and yes, we know that these molecules are going to
133:46 - 133:54: be able to, let's say, cure or help in cure some type of cancer. The problem is how do we keep
133:55 - 134:05: the molecule inside the cell? We do have these complexes that try to throw the molecule out,
134:05 - 134:16: but for me, what has become more crucial at this moment is that I'm not going to accuse everybody,
134:16 - 134:25: myself and everybody. We are all stuck. We're all doing the same kind of experiments
134:25 - 134:33: with the same type of approaches, looking at the same thing. We need new brains in here. We need
134:33 - 134:39: to bring chemists. We need to bring mathematicians. We need to bring engineers,
134:39 - 134:44: etcetera, etcetera, that can help us because we know we deliver the molecule into the cell.
134:45 - 134:53: Sometimes we can use an inhibitor that will prevent the removal of the drug from the cell,
134:53 - 134:59: but something else is happening. We need help to understand what is happening and how we can
135:00 - 135:07: approach, which is different than what we do in these days. Honestly, I've been discussing this
135:08 - 135:16: with physicists, with chemists, with mathematicians, with some engineers, and
135:18 - 135:24: we need to have a meeting like that. Actually, I have been discussing this with, for me, a lot too,
135:25 - 135:30: and quantum biology, quantum chemistry, quantum physics, whatever you want, but we need to have
135:31 - 135:41: a meeting where not the same participants that talk about Polycomb or this or the other or the
135:41 - 135:50: other and cancer come together and start looking at different things. We're stuck. That's what I
135:50 - 135:56: said. I was talking to Daphna Bersagli last week or two weeks ago, and I said, since Nixon
135:57 - 136:05: declared the war on cancer, the amount of billions that have been placed in there
136:07 - 136:14: do not justify the little poor advance that we have done. I think that some of the important
136:14 - 136:23: findings were the oncogenes, were the tumor suppressors, and so on, and some of the mechanism
136:23 - 136:31: of what is happening here and in there, but really, unless you have early detection,
136:32 - 136:40: surgery, or for some cancers, very few specific drugs, we're stuck. We need different approaches.
136:40 - 136:45: Danny and Kristian and Luciana, can you guys update us on what is the clinical
136:45 - 136:52: outcome of all of these PRC2 inhibitors that many companies have made on different forms of cancers
136:52 - 136:57: from prostate to brain to liver, and how are they looking, and what's the prospect and future of
136:57 - 137:06: inhibition of Polycomb in disease therapy versus using a degron of SUZ12 over EZH in the process?
137:06 - 137:11: Do you want me to sell shares or what? Go ahead, go ahead, Christian.
137:12 - 137:18: No, I mean, there's one company which has two approved, I mean, indications, right? So,
137:18 - 137:24: they have it approved for the diffuse large B-cell lymphoma, and they have it approved for the
137:25 - 137:31: steroid tumors with two specific mutations. So, that's one compound. I know, and you,
137:31 - 137:35: Danny, of course, knows from as well many of the clinical trials which are ongoing,
137:36 - 137:40: that same company, I know they hope to get it into prostate cancer as well,
137:41 - 137:46: and they're doing clinical trials in prostate cancer, and of course, they are also trying
137:46 - 137:51: other indications, but those are the two cancers where it has been approved, and I think, you know,
137:51 - 137:58: the problem in both cases, I think, is that it's not very big areas. I mean, in one of them is
137:58 - 138:05: third-line therapy, and the question is, how well would they work with monotherapies versus
138:05 - 138:10: all the combination therapies? Sure. I think that's very crucial. So, I think there are a
138:10 - 138:17: lot of clinical trials which will prove that point, and then for the point, so, and that's
138:17 - 138:23: what I mean, then the point is, are the inhibitors that are currently testing good enough, and of
138:23 - 138:28: course, that's your, that's where your degradation tax comes into the picture. I think the one which
138:28 - 138:33: has been approved is relatively weak compared to what has been, what is in clinical development. So,
138:33 - 138:38: I think there are four other companies at least who are actually targeting the PRC2 complex. So,
138:38 - 138:44: I think, you know, reasonable, I would say that it's a big success story from that point of view,
138:44 - 138:49: but it's not perhaps as big as you would like it to be, let's say. It's very similar to the
138:49 - 138:55: first wave of epigenetic drugs, the EZH2 inhibitor and the HDAC inhibitor. There have
138:55 - 139:03: been a second wave and third wave, which the compounds were much more efficient and given to the
139:04 - 139:09: several chemical modifications, and the same will apply here. So, this is the first generation of
139:10 - 139:15: drugs. Inhibition inhibitors were being used in the clinic long before we knew the mechanism,
139:15 - 139:20: right? I mean, those were actually working in the clinic, and the concentration was changed,
139:20 - 139:26: and I think there's sort of a synthetic therapy with them affected, but the Polycomb inhibitors
139:26 - 139:30: are totally opposite of DNA methylase inhibitors. We know the function, and we're trying to find
139:30 - 139:40: the cancer. We need to maintain the, we need to inhibit the ABC transporter. We need to combination
139:40 - 139:48: of drugs, okay? Not only directed to the epigenetic machinery, but to other pathways that
139:48 - 139:57: the cell use to either get rid of the inhibitor or to simply just inactivate, okay? I'm coming
139:57 - 139:59: back to what Kristian was saying. Indeed.
140:00 - 140:16: There are two of them, and the inhibitors are not so good, but now there's going to be a new series of inhibitors that are coming that are, I would say, the best that we have seen, and we will be seeing, okay, for PRC2.
140:17 - 140:31: So I have hope on that. The problem is, it's not only that cancer, and it's not only that. You need to be able to maintain the inhibitor in the cell.
140:31 - 140:56: You need to be able to stop other opposite pathways that may be operating. So that's why I believe that we need to get a group of good scientists that know different fields well, okay, and discuss this openly to come up with new approaches.
140:56 - 141:04: All right. I'm going to direct this question to Luciano since he has done some of the related work for the Trithorax, and then Danny and Christian can fill in.
141:04 - 141:15: So the question comes from Jill Dwocker, and she wants to know, do the mutations on PRC2 catalytic dead mutation and also the drugs, do they affect the three-dimensional structure of chromatin?
141:15 - 141:21: And can you obtain any information from changes in the chromatin structure to gene expression?
141:22 - 141:36: This is a much more complex question. It's not only related to inhibition of PRC2 complex. It's the link between how relevant is the overall chromatin organization and gene regulation.
141:36 - 141:40: Are the two things linked? Is one preceding the other one? This is completely unclear.
141:41 - 141:52: There are many reports that suggest that change in gene expression precedes change in topology, and there are a bunch of other studies which suggest topology precedes gene expression and other things which are completely unrelated.
141:52 - 141:57: Topology is a structural element that does not really impinge on gene regulation.
141:57 - 142:07: So the alteration in gene topology is not something that I would specifically look at as a key parameter for selecting eventually or for developing drugs.
142:07 - 142:13: I think the idea is that what Danny was mentioning, so the cells find a way to expel this compound after a while.
142:13 - 142:22: So this is one thing to overcome, and the second is most likely you need epigenetic drugs being combined with chemotherapy or immunotherapy at some point.
142:22 - 142:29: By itself won't be sufficient, or at least for a short time, but then there will be a reversion. And these are things I’m sure are being considered at the moment. We were much more naïve at the beginning.
142:30 - 142:51: I don't want to leave with a negative impression, because that's what I just want to leave with, leaving a positive impression that we are moving forward.
142:52 - 143:19: And a lot, a lot of improvement has been done for new inhibitors, modifying inhibitors, and we will be getting there, but we need, as you said, and as I said and others said, to start looking at the whole thing much more in total, much more to the cell itself.
143:19 - 143:26: Not just, oh, this inhibitor is getting into the pocket here and here and it's going to inhibit this.
143:26 - 143:31: Yes, but it will get repelled by the ABC pathway, for example.
143:31 - 143:43: And there are inhibitors for that, Dr. Rosen, Klebex, others, and we're going to have to do combination.
143:43 - 143:46: None of these alone are going to solve the problem.
143:47 - 143:56: It's not only EZH2, there are DOT1L inhibitors, there are LSD1 inhibitors, there are arginine methyltransferase inhibitors, which have been tested at the moment.
143:56 - 144:01: So there are many opportunities out there.
144:01 - 144:09: All right, so there's a question for Kristian in here, and I think it's a long question, so I'll start with the question, then I'll read the evidence in here for you.
144:09 - 144:25: It's that, do you think that PRC2 can initiate transcriptional repression of active genes in situations that you've talked about, such as PRC2 is required for maintenance, but not establishment of transcriptional repression in mouse and human stem cells?
144:25 - 144:42: But in tumor cells like MPNST, where you have a homozygous deletion of SUZ12, reference the expression of SUZ12 as directly induced transcriptional repression of some PRC2 target genes and reestablishment of K27 trimethylation of these loci.
144:42 - 144:50: The question is, I guess, can PRC2 initiate transcriptional repression of active genes in these situations?
144:51 - 145:03: Well, I don't think we have any data. I mean, it can if you force it to a promoter and you, together with something, so it binds, you know, tightly to a promoter, it would induce transcriptional repression.
145:03 - 145:09: But otherwise, we have no evidence for that the PRC2 complex initiates transcriptional repression.
145:09 - 145:13: So I don't, I don't think so.
145:13 - 145:16: Danny and Luciano, do you guys have any thoughts on that?
145:16 - 145:24: I agree with what Kristian said, the genetics in Drosophila tell us very clearly.
145:24 - 145:28: These things maintain, they do not initiate.
145:29 - 145:42: Now, there may be a protein that is a sequence-specific DNA binding protein that may bind to a specific set of genes and interact with PRC2 and bring PRC2 there.
145:42 - 145:48: Could be, but in the 20 or 30 years of study, that has not been seen.
145:48 - 145:57: I know that I have seen many publications, not many, a few publications saying PRC2 interacts with this repressor or PRC2 interacts with this, okay?
145:57 - 146:00: None of those studies has been done.
146:00 - 146:06: Danny, I agree with you, but it's also true that Kristian published some data.
146:06 - 146:11: If you target PRC2 to a specific locus, you can initiate transcriptional repression.
146:11 - 146:19: For future generations, maybe it's not perpetuated forever, but that it has the capability of doing it, whether it does it or not is unclear.
146:19 - 146:23: So it's true that in 20 years, we never, we don't have anything in this direction.
146:23 - 146:30: But it's also true that we believed the CBX protein were the only reader for K27 trimethylation.
146:30 - 146:35: And recently, a new genetic paper indicated it's another family protein which can read K27.
146:35 - 146:37: So I might be open for surprise.
146:37 - 146:40: So as today, you're completely right.
146:40 - 146:41: There is no evidence in this direction.
146:41 - 146:46: We have some data that points to this as well, where we come from the Trithorax end.
146:46 - 146:59: And the question that we had is that if you make a catalytically dead Trithorax versus deletion of Trithorax, identifying the target gene for Trithorax function, we put a reporter in front of those genes.
146:59 - 147:04: And we asked the question, when you get rid of Trithorax, this is MLL2 in this case, this is a bivalent cluster.
147:04 - 147:07: The reporter expression is off.
147:07 - 147:17: And we asked in the absence of MLL2 where there is no K4 trimethylation, what gene can we delete that you see turned in gene basically on?
147:17 - 147:27: And deletion of Polycomb and DNA methylation directly resulted in activation of about 2000 genes that follow this rule.
147:27 - 147:33: And our conclusion was that the function of Trithorax is to repel Polycomb and DNA methylation machinery.
147:33 - 147:40: So if the Polycomb and DNA methylation maintain and cannot do it, we would not be seeing what we are able to see.
147:40 - 147:50: Because when we get rid of Polycomb by getting rid of SUZ12 or inhibitor of Polycomb, these genes express now in the absence of MLL or K4 trimethylation.
147:50 - 147:55: So everything else is normal in the cells except loss of Polycomb and the loss of
147:55 - 148:03: So basically, this seems to me that what Kristian is saying is that recruitment of Polycomb to those sites does not initiate silencing and can maintain it.
148:03 - 148:09: That's not what I said. I said the opposite. I said it does not initiate silencing.
148:09 - 148:14: But the data I'm telling you is that if you get rid of Polycomb, you get rid of silencing.
148:14 - 148:23: It doesn't say the same. It basically says that there are transcription factors around which can bind there if you don't have the Polycomb and they're sufficient then to induce transcription.
148:24 - 148:28: So you're saying that these transcription factors require Polycomb for binding to silence?
148:28 - 148:34: No, I think there are thresholds for transcriptional activation, which is activated by cell signaling and so on.
148:34 - 148:38: And what the Polycomb do is to put in thresholds for transcriptional activation.
148:38 - 148:43: So clearly, we know that you can lose Polycomb binding if you induce differentiation.
148:43 - 148:46: For instance, you see a lot of Polycomb in those persons.
148:46 - 148:54: And what happens there is that you basically are losing them, most likely because you see changes in signaling to chromatin.
148:54 - 149:01: And one way to think about it, I think, is that it's not a very stable interaction.
149:01 - 149:08: So you see a very mobile interaction and you get something which is actually binding with higher affinity to the same sites, and then you have transcriptional activation.
149:09 - 149:17: I'll just give you an example of 2000 genes that are silenced in the absence of Trithorax.
149:17 - 149:23: And this silencing depends on Polycomb function, because if you inhibit Polycomb, they are re-expressed.
149:23 - 149:26: That indicates to me that Polycomb is directly silencing them.
149:28 - 149:34: It maintains the silencing, which is different than actually setting up transcriptional inhibition.
149:35 - 149:39: I agree with Kristian here.
149:42 - 149:45: Food for thought. I've got to think about this.
149:45 - 149:49: Go ahead, but you are a slow thinker, so we don't have too much time.
149:49 - 149:51: I'm a fast talker, though.
149:53 - 149:56: All right, I'm going to read some anonymous questions.
149:57 - 149:59: This is for Kristian.
149:59 - 150:04: You mentioned in your talk that EZH2 inhibition effect was rescued.
150:04 - 150:11: So while cancer treatment, top PRC2 inhibition will be more potent than chemo drugs.
150:15 - 150:17: Oh, I don't think I mentioned that.
150:17 - 150:23: I think chemo drugs, they beat everything, but the problem is the side effects of traditional chemotherapy.
150:24 - 150:31: Of course, if you have an EZH2 inhibitor, you can hope to be more specific, and therefore not having the same side effects.
150:32 - 150:36: And if you have a PRC2 inhibitor, you can be even more specific.
150:36 - 150:39: You may affect one part of the PRC2 complex, not all of them.
150:39 - 150:41: You can be even more specific.
150:42 - 150:48: All right, so here's sort of a general question on the function of Polycomb.
150:49 - 150:54: How do you guys think that the Polycomb system maintains gene expression?
150:54 - 150:56: Gene repression, sorry.
150:58 - 151:00: So how does this maintenance work?
151:02 - 151:04: Can you apply genetics?
151:04 - 151:06: Can we apply genetics?
151:06 - 151:08: No, can you fight genetics?
151:08 - 151:16: Isn't that what we learned from the early studies in Drosophila development, early development?
151:16 - 151:23: Well, we've been saying for the past 15 years that epigenetics is a major regulator in many of these processes, right?
151:23 - 151:31: There are repressors that they bring in, maybe PRC2, and then those repressors disappear.
151:31 - 151:33: I think that's the question.
151:33 - 151:35: Mechanistically, how do you describe that?
151:35 - 151:37: I mean, you're the mechanism guy, right?
151:37 - 151:42: So mechanistically, how would you define Polycomb system maintains gene repression?
151:42 - 151:52: By compacting, by preventing engagement of polymerase, while SNF, maybe even condensate phase separation, higher-order chromatin structure.
151:52 - 151:54: All these are possible mechanisms.
151:54 - 151:58: I think what he asked is really, you know, what maintains it?
151:58 - 152:01: So meaning, okay, all that happened.
152:01 - 152:03: So now your cycle, da-da-da-da-da.
152:03 - 152:06: I mean, what we tried to address over the years, right?
152:06 - 152:10: So how do you actually maintain the transcription repression?
152:10 - 152:14: So, you know, K27, how do you maintain K27?
152:14 - 152:21: I mean, while Danny's lab has tried to address that, I mean, basically that's what Danny addresses in his talk with the MPN, right?
152:21 - 152:29: That's basically a mechanism for maintaining transcription repression for K27, or maintaining K27 trimethylation.
152:29 - 152:34: I mean, we don't have the answer, but this is the way we're thinking about it, right?
152:34 - 152:43: And I think that the maintenance is also mediated not only for K27, but for K9 to the right and right mechanism.
152:46 - 152:48: All right, so there are no more questions in here.
152:48 - 152:52: I'm going to just ask sort of a general question for the panel to discuss.
152:52 - 152:55: And some of us are old enough to remember all of these things.
152:55 - 153:06: So the excitement in the field, you know, back in the mid-90s was that there are enzymes that modify chromatin and some of these modifications sort of associated with transcription.
153:06 - 153:10: And there was a race of identifying all of these modifiers.
153:10 - 153:13: And then the next battle came out, who are the demodifiers?
153:13 - 153:17: And we all spent years identifying demodifiers, right?
153:17 - 153:24: And then sort of battle came up, what happens if you inhibit these modifiers and demodifiers, and can you cure all cancers?
153:24 - 153:28: And so what do we spend the past, you know, five,t six years as a community doing that?
153:28 - 153:29: What is next?
153:29 - 153:31: What is next for Polycomb?
153:31 - 153:32: What is the burning question?
153:32 - 153:34: Where are we going with Polycomb?
153:36 - 153:38: I mean, those are the ones we just addressed, right?
153:46 - 153:47: No more burning questions?
153:49 - 153:50: The field is over?
153:50 - 153:51: No.
153:51 - 153:53: Right, so we identified the enzyme.
153:53 - 153:55: We identified the demodifying enzyme.
153:55 - 153:57: We got some inhibitors that function in there.
153:57 - 153:58: We know all the complexes.
153:58 - 153:59: What's next?
153:59 - 154:02: I just told you there's something that we don't understand, or I do not understand.
154:02 - 154:11: For example, we know that if you produce a drug, okay, it will take multiple generations for K27 to disappear.
154:11 - 154:18: Now you remove MEP1, and within 12 hours, let's say, okay, or six hours, it's gone.
154:18 - 154:21: So there must be a mechanism that we do not understand.
154:23 - 154:29: So there must be other things that we are going to be discovering as we continue doing science.
154:30 - 154:36: In addition, I think the single-cell ChIP analysis will help us clarify.
154:36 - 154:46: When we do ChIP-Seq, we see PRC2, we drive it, sitting on a promoter, together with PCR3, together with PCR2, all of them on the same promoter, which is impossible.
154:46 - 154:47: Physically impossible.
154:47 - 154:51: So what provides the difference is cell variation.
154:51 - 154:53: It's a kind of continuous turnover.
154:53 - 154:54: Why is that?
154:54 - 154:58: Why do you need so many variations of PRC2 complex, all eating on the same place?
154:58 - 154:59: That also is unclear.
154:59 - 155:02: But ChIP-Seq, you align the track, you see all of them there.
155:02 - 155:07: All the accessory factors are all together there, which is difficult to believe.
155:08 - 155:13: Have you been able to successfully do single-cell ChIP-Seq?
155:14 - 155:15: No, not yet.
155:15 - 155:16: I think that will break.
155:16 - 155:19: That will help us a lot once the technology is there.
155:19 - 155:22: It will help us a lot in dissecting this question.
155:22 - 155:25: The specificity, how this is achieved.
155:25 - 155:29: I actually think what is going to happen is that we have to put it all together. You’re synching Polycomb now, before you were synching Trithorax,
155:30 -155:44: we’re thinking Trithorax now, we were Polycomb before. Now we have to think about all these modifications, put them together
155:44 - 155:50: and, you know, how is transcription really regulated through all these modifications, and can we get a more complete picture

155:50 – 155:59: A big picture, yeah, sure. That’s fair. I think, you know, the same battle was going on when I was a student when there was a race identifying all these basal factors,

155:59 - 156:06: could we identify them, clone them… and basal factor biology sort of died, right? Not much really paid attention on basal.

156:06 - 156:16:But maybe we should go back to see what is the role of these basal factors on initiation of transcription, how they function on this Chromatin mark and what does mean from cell to cell?

156:16 - 156:29: I disagree with you on the basal factors. Why? You haven’t been studying basal factors for 15 years. I know but you know from the basal factors, how do we bring the polymers into it.

156:29 - 156:41: We have the structures. We know a lot, you know, we learned that. How the polymers move we also know it.

156:41 - 156:51:There must be other things we do not know there, or simply there is no more to learn how you bring the polymers to the promoters.

156:51 - 157:12: The question of elongation, what is stopping elongation. Promoter clearance, which has become very confused as you know, with one nucleosome, which has induced whatever you want to call it, has become very confused.
157:12 -n 157:22: Sure, but I think I like what Kristian says. I think bringing it back all together. And he mentioned, you know, we have been paying attention on elongation at Trithorax for almost 30 years now.

157:22 - 157:27 And now we started looking at Polycomb and other labs are looking at Trithorax and elongation
157:27 - 157:30: and I think bringing all of that together and having a bigger picture
157:30 - 157:35: of how these function in cells definitely is an important endeavor.
157:35 - 157:37: I mean, I ask this question daily of myself.
157:37 - 157:39: Where do I want to be 10 years from today?
157:39 - 157:41: What are the areas we don't know?
157:41 - 157:42: Where are we moving?
157:42 - 157:47: And I think for our audience of dwindling from 400 to 176 now,
157:47 - 157:50: they might be interested to know what you guys are thinking,
157:50 - 157:54: and what are your worries and plans for the future of chromatin biology?
157:54 - 157:56: Where are you taking your research?
157:56 - 157:59: The brain.
157:59 - 158:01: He takes it to the brain.
158:01 - 158:02: He thinks about it.
158:02 - 158:04: Well, no, I agree with Danny.
158:04 - 158:07: I think, you know, the memory and the neuroscience,
158:07 - 158:12: it is definitely the final frontier for, or at least the next frontier.
158:12 - 158:15: And Polycombination is very important in there, too.
158:15 - 158:16: You know, I was amazed.
158:16 - 158:24: I went to a meeting, the last meeting we all attended in the Bahamas, I think it was, wasn't it?
158:24 - 158:26: And I went to the very last session.
158:26 - 158:31: And I was amazed how many talks were there on mutations on Polycomb,
158:31 - 158:36: Trithorax, and transcription factors in neurodegenerative diseases.
158:36 - 158:37: I mean, I was shocked, right?
158:37 - 158:40: There's a whole session was on all of these things.
158:40 - 158:42: And, you know, as someone I'm involved with,
158:42 - 158:44: some of my own nitty-gritty details of elongation,
158:44 - 158:46: you don't pay much attention to a lot of those things.
158:46 - 158:50: But it was eye-opening to find out that how much is being done,
158:50 - 158:52: as what Kristian is saying in a bigger picture,
158:52 - 158:56: that, you know, how do these machineries function to develop an organism,
158:56 - 158:59: and how their mutations are associated with some of these diseases.
158:59 - 159:04: Yeah, I think that neuroscience is definitely the next frontier for many.
159:04 - 159:08: Yeah, well, I don't think we're going to have a session on Polycombs next, you know,
159:08 - 159:11: if Abcam organizes this in five years from now.
159:11 - 159:16: What I think we will have is a session on, you know, gene expression.
159:16 - 159:19: You know, in general, like trying to put out the big picture.
159:19 - 159:22: And, of course, Polycombs, but then it would be neuroscience
159:22 - 159:25: and how those chromatin structures, you know,
159:25 - 159:29: involved in the gene expression in neurobiology.
159:29 - 159:31: We're starting to see that, right?
159:31 - 159:36: So now the Polycombs or the so-called Polycomb meetings that we have,
159:36 - 159:41: now we bring in neuroscientists, thank God, and we're learning a lot.
159:41 - 159:45: I realized that I want to go to the brain to study Polycombs, okay,
159:45 - 159:47: and that's the direction I'm going.
159:48 - 159:55: But I insist that we need to bring other people that look at the problems
159:55 - 160:00: that we're facing with a different eye, okay, with different technology.
160:00 - 160:06: What, something different, okay, we're missing something.
160:06 - 160:10: We have all the information from us on front of us, okay.
160:10 - 160:14: We know that there are these transporters
160:14 - 160:17: that move the drugs out here, others.
160:17 - 160:19: We have all this data there.
160:19 - 160:26: We just need people, I don't know, physicists, chemists,
160:26 - 160:29: mathematicians, all of them together, okay,
160:29 - 160:32: and we're biologists, we start looking at that.
160:32 - 160:34: So with that thought, we have a question coming in here
160:34 - 160:38: from an anonymous attendee, it sounds like Rick Young to me,
160:38 - 160:40: and this anonymous attendee is surprised
160:40 - 160:44: that none of you guys have put your data into condensates
160:44 - 160:47: and the function of condensate in the process.
160:47 - 160:49: So what are your thoughts on condensate,
160:49 - 160:53: chromatin structure, and Polycomb?
160:53 - 160:56: Oh, come on, please.
160:56 - 160:59: Condensates were discovered 20 years ago.
160:59 - 161:01: We know what they are.
161:01 - 161:06: You just need to reach a crucial concentration of factors
161:06 - 161:08: that will do a reaction,
161:08 - 161:10: will perform a reaction more efficiently,
161:10 - 161:12: and then that condensate disappears
161:12 - 161:14: and is replaced by others.
161:14 - 161:18: This is all physical biochemistry
161:18 - 161:22: that was there 20 years ago or 15 years ago.
161:22 - 161:25: Tell me, what have we learned from the five years
161:26 - 161:30: that people have been studying condensates,
161:30 - 161:31: other than they are aggregates
161:31 - 161:35: or things that come down in the test tube, you take them.
161:35 - 161:39: Have I seen a reaction that is more efficient
161:39 - 161:40: by a condensate?
161:40 - 161:44: Do I see PRC2 acting much more efficiently
161:44 - 161:46: because it's within a condensate
161:46 - 161:49: or Trithorax or transcription?
161:49 - 161:52: Yes, I know I saw one or two papers saying,
161:52 - 161:55: here's transcription within the condensate, okay?
161:55 - 161:56: Give me a break.
161:56 - 161:58: I look at the paper and I'm not going to go into the details
161:58 - 162:01: because I don't want to start an argument
162:01 - 162:02: with one and one,
162:02 - 162:04: but yes, we need to be more critical.
162:04 - 162:06: What have we learned?
162:06 - 162:07: Tell me.
162:07 - 162:10: I think, you know, the condensate is definitely,
162:10 - 162:12: there's been a lot of great stories on it
162:12 - 162:13: and some genetics being done,
162:13 - 162:15: but the functional significance
162:15 - 162:18: and how do you actually mechanistically address
162:18 - 162:19: their role in gene regulation,
162:19 - 162:21: that's a black box to me.
162:21 - 162:22: And I think that's something
162:22 - 162:24: that probably could be addressed
162:24 - 162:26: once we develop more tools
162:26 - 162:28: besides this hexane dial that people have used
162:28 - 162:30: to address some of these questions.
162:33 - 162:35: The same is true for the Polycomb bodies
162:35 - 162:37: or the stress granules.
162:37 - 162:39: There are bodies in the cells,
162:39 - 162:41: which we know since many years,
162:41 - 162:44: and we still have not really clarified
162:44 - 162:45: what their function is,
162:45 - 162:47: how they are constructed,
162:47 - 162:50: if they split and reform again.
162:50 - 162:52: We know very little about these things.
162:52 - 162:55: And this is another example of condensate.
162:55 - 162:57: Membrane-less organelles.
162:57 - 162:58: P bodies,
162:58 - 163:03: aren't they just a storage of Polycomb components?
163:04 - 163:07: I mean, there's data suggested the Nascent RNA is also there.
163:07 - 163:08: So the Nascent RNA is also there,
163:08 - 163:09: they may come together,
163:09 - 163:11: those can be points of, you know,
163:11 - 163:13: 3D chromatin interaction
163:13 - 163:16: between different sets of genes correlated.
163:16 - 163:18: So there are many possible hypotheses there.
163:18 - 163:20: And none of those have been really nailed down
163:20 - 163:21: to clarify.
163:23 - 163:28: All right, so we have to bring the last questions in here
163:28 - 163:29: and then we can close down
163:29 - 163:31: with maybe some closing statements
163:31 - 163:33: from each one of you guys.
163:33 - 163:36: And this last person wants to know,
163:36 - 163:38: I would like to know what is your opinion
163:38 - 163:39: about the role of PRC-1
163:39 - 163:43: regulating transcription frequency and bursts
163:43 - 163:45: based on some of the recent papers
163:45 - 163:47: that are published in the area?
163:52 - 163:53: What should we say?
163:53 - 163:55: I think transcription factors
163:55 - 164:00: and complexes regulate gene expression,
164:00 - 164:02: they all affect the initiation of transcription
164:02 - 164:04: and bursts of transcription.
164:04 - 164:07: And Polycomb may be in the same bag,
164:07 - 164:08: positively or negatively.
164:10 - 164:11: All right.
164:11 - 164:14: And PRC-1 particularly now
164:14 - 164:15: has been linked to gene activation
164:15 - 164:17: or enhancer activation.
164:17 - 164:20: So there are cases also in the Ring 1B in that case.
164:20 - 164:21: Polycomb.
164:21 - 164:23: That's what I was saying.
164:23 - 164:24: So you're asking what was involved in that?
164:24 - 164:26: No, Ring 1.
164:27 - 164:28: Ring 1B.
164:32 - 164:35: Well, I would think that what happened
164:35 - 164:38: is that by having chromatin structures,
164:38 - 164:40: what you are regulating is basically
164:43 - 164:45: the probability of getting transcribed.
164:47 - 164:50: And so the system does not completely
164:50 - 164:52: allow you to lock the door that the door is closed,
164:52 - 164:55: but you are actually regulating the chromatin structures.
164:55 - 164:57: So you make it much less likely
164:58 - 165:00: that you have activation.
165:00 - 165:02: And so I think it's consistent with the idea
165:02 - 165:04: that you may have a background expression.
165:04 - 165:07: And I think that's what has been seen on Polycomb genes
165:07 - 165:09: is that you do have a background level of expression,
165:11 - 165:15: but that may not be deleterious for the cell at all.
165:15 - 165:15: It doesn't really matter.
165:15 - 165:17: Perhaps it costs too much energy for the cell
165:17 - 165:20: to actually completely close down transcription.
165:21 - 165:22: And so when you're looking at bursting,
165:22 - 165:23: I would agree to the point of view
165:23 - 165:25: that it would make sense
165:25 - 165:27: that what the Polycombs are doing
165:27 - 165:29: is basically making this bursting less frequent.
165:29 - 165:32: I mean, it makes sense.
165:33 - 165:34: So, yeah.
165:36 - 165:38: All right, well, Tony,
165:38 - 165:41: thank you for bringing us all together in here.
165:41 - 165:44: I certainly much prefer it in person.
165:44 - 165:48: So this is probably my last Zoom meeting for anybody.
165:48 - 165:51: I hope we move on to the in-person meeting soon
165:51 - 165:55: because I really liked that discussion afterward
165:55 - 165:56: where we can get to the bar
165:56 - 165:59: and the lunches and dinners to have a lot more interaction
165:59 - 166:02: and possibility of setting up collaborations.
166:02 - 166:04: But this was as good as Zoom can get.
166:04 - 166:06: And I'm very grateful to all of you guys,
166:06 - 166:09: Abcam, Tony Kouzeridi, and all the speakers
166:09 - 166:11: for having a fantastic session.
166:11 - 166:13: And let's thank the big pharma
166:13 - 166:17: that it appears to be a good antibody
166:17 - 166:19: that was announced today.
166:19 - 166:20: Pfizer, yes, 90%.
166:20 - 166:21: Yes.
166:21 - 166:24: And like Pfizer, I'm sure others are gonna come.
166:24 - 166:25: So-
166:26 - 166:27: Antigen.
166:27 - 166:31: Regardless of whatever we used to hear,
166:31 - 166:36: science drives how we cure diseases, okay?
166:37 - 166:38: And now-
166:38 - 166:39: I know, I think you're preaching to the choir
166:39 - 166:40: in that regard, Danny.
166:43 - 166:45: We all are great believers in that.
166:45 - 166:48: Thank our presenters, Ineretz, Paweł, Luciano,
166:48 - 166:51: Kristian, and Danny, and of course, our moderator, Ali.