CellPress Webinar Spatial and Temporal Genomics in Cancer Research

Show video

OK, hello, everyone, very happy to be  here and thank you for joining us for   our Cell press webinar on spatial and temporal  genomics in cancer research to tie Miao-chih Tsai,   senior Editor, cell. And I'll be your moderator  for today's webinar. Technology advancement   allows us to dissect the spatial and temporal  changes in genomics compositions gene expression.   Cellular morphologies and cellular  environments during development and disease   progression of various data collections led to  the emergence of spatial and temporal genomics   as a new Avenue of research. There's a significant  interest in method development and clinical use   of information for cancer diagnosis and  treatment as they may help to elucidate the basis   for heterogeneity of cancer. So in this Webinar,  Miriam Merad from Mount Sinai, Fei Chen from   Broad Institute of MIT and Harvard and Steven Wang  from Yale University will discuss the directions   of its further development. So before we get  started with the talks, we have a few political   points. So each of our speakers will talk about  12 minutes following the talk. We will have  

a Q&A session that covers all three presentations.  If you have a question for any of the speakers,   just click on the question button, and then  type it in. You can enter a question at any   time. No need to wait until the end. We will try  our best to parse out as many questions as we can.  

It is helpful if you can include the name  of the speaker to whom you would like to   the direct your question. So before we begin, I  would like to thank our speakers for making the   time to join us today and also our sponsor  OriGene for supporting the Webinar. So and   now kicking it off is our first speaker, Miriam  Merad. Miriam, thank you for joining us today.   OK, so my great pleasure to be here. Always a  pleasure. Let me just make sure I'm sharing my   screen. Can you guys see my screen? Not yet,  not yet. OK, we do this because even when you.   It's working now. Yes all right, perfect.  

OK, sorry. I think we just lost the Mriam, but  we will have our second speaker to go first. Fei   Chen, who is with us now so Fein? Thank you. Thank  you for the introduction. And I'm very excited to   participate and thank you for organizing. And  really, as Mel said, the session, I think,   is divided into both biology and also really the  recent interest in New tools for our ability to   map, basically bring genomics into the spatial  context of tissues. And that's what I'll talk to  

you about today. Just a quick disclosure slide.  And I think, you know, I think there's no way   to really talk about this without first being into  the discussion of single cells, transport mix. And   I think one single cell transorbital mix is one  recent technological advance that has recently   really revolutionized our ability to study  molecular mechanisms in cells and tissues and   in tumor biology. And that's because you can take  a tissue and break it down into its constituent  

cell types and sample the transcript of each  cell. And this is comprehensive and it gets   at the molecular programs that are driving the  function of cells and the dysfunction of cells   in disease. But I think fundamentally we have to  remember that to collect this type of data, we   need to dissociate the tissue. And this is where  we lose the spatial and temporal context about   where cells come from. And I think that context  is very paramount to studying. Basically, health   issues are organized, how cells communicate with  each other and in particular in cancer. It's very  

important for understanding the disorganization,  heterogeneity of cellular structures, as well as   maybe there's emergent multicellular niches and  hubs which are really important for, I guess,   the immune response or also tissue dysfunction.  And so the other thing I think we got we have   to remember is that cells and tissues are  very also very dynamic. They might migrate.   They differentiate a divide. And almost all of our  tools that we have are at our disposal today are  

kind of snapshots. Right we can freeze the tissue  or we associated with it and we lose the context   of those dynamics. I think those dynamics are  very important to excess. But today I'll really   talk about my lab work on spatial genomics,  although we're very excited about the context   of using molecular recorders like lineage tracing  to study the context of dynamics within tissues.   So so today I'll mainly focus on a technology  for assessing tissue scale, spatial and tissue   spatial genomics, wherein you can retrieve the  transcript terms of cells across tissues. I think   it's also very interesting to think about the  structures of molecules within cells, such as   the genome. And we're very interested in being  able to do that as well. I think that will be  

maybe the topic of the next talk. And so I think  why why are we interested in resolving tissue   organization with spatial genomics? Well, I think  we're motivated by three main classes of problems.   The first is, you know, as we are  becoming so adept at collecting single   cell transcript terms, I think the most na ve  question is like, where are these molecularly   defined cell types located within tissues and  how does that organization relate to function?   Beyond that, I think we're also interested  in how are within these cell types,   what is spatially differential gene expression  patterns? Right and within a cell type like this,   it's location in the tissue explains some  of the variability in its gene expression   and how what percentage or what aspects of its  variability in gene expression can be explained.   And then, of course, the lastly is like, what  are the gene expression changes with respect to   the pathology? Right we've known for a long  time that we can go in and annotate tissue   changes morphologically, such as a case of acne.  But how do we connect those morphological changes   with kind of the molecular mechanisms that we  can discover through sequencing? So these are   kind of the three motivating problems today. I'll  really talk to you about how to address the loss  

of spatial context when performing single cell  sequencing by discussing a tool called slide seek,   which is enables genome wide expression  profiling in tissues at 10 micron resolution.   And a lot of this work is you know, this is a  collaborative effort between my group and then   Costco's my neighbor at the road. And the picture,  big picture of six effectively lets you do,   un-targeted single cell sequencing. And at the  same time, it lets you know where those cells   came from in your tissue. And how do we do this?  Well, inside we deposit individually, barcoded   10 micron polystyrene beads in a mono-layer  on a surface. And and the key is that in each   one of these beads, there's millions of oligos  nucleotides with kind of a similar structure,   a PCR handle or constant sequence, a spatial  barcode. What this is, is a clonal sequence  

for each bead that's different among beads.  And you'll see how we use that in a second.   A unique identifier for counting transcripts and  a TV handle for capturing my name. And what we do   is we deposit these beads into a month later on a  slide. And here's actually a cover slip. Here's a  

40 millimeter cover slip. With each one of  these white circles is a 3 millimeter diameter   array with about 100,000 of these 10 micron beads  on it. And because these beads deposited randomly,   we then take a microscope and sequence the clonal  B barcodes for each bead in the array. I won't   go into detail exactly how we go do this, but  basically we're trying to reconstruct the sequence   of A s C s G s and t s that make up each barcode  for each bead. And we do that by taking a series   of images where each color of one of four colors  represents each base. So at the end of the day, we   get kind of the xy locations of each barcode. So I  think one of the main advantages of these special  

capture, spatial transfer approaches that are  very easy to do once you have made the array here,   we can once you make the array, it's very easy to  the experiment. You can take fresh frozen tissue   and resection it directly onto the  array, a 10 micron section. Then   we basically perform molecular biology protocols.  That's very similar to single cell sequencing,   such as reverse transcription and PCR. And we end  up with kind of like a sequencing library. And   the sequencing library is very much like what  the libraries you would like you get in single   cell sequencing for each bead barcode, you have  a digital gene expression vector. So you end up a  

matrix of a cross barcode that counts where the  columns are each gene and accounts of the gene   that we're seeing for each bead. But now, if  you remember, we have the spatial location   of each barcode so you can just match them  up. So you match up kind of the sequenced xy   locations of each bead and now you have a spatial  location for each read in your sequencing library.   So what is the data look like? This is kind of  actually one of the first experiments we ever   did with sensing data has the same structure as  single cell data. And what we can do is we can   apply the same similar computational tools such  as dimensionality reduction and clustering. And  

actually here we ran slightly on the mouse  hippocampus, but in the very left, what we've   done is we've just done unsupervised clustering  and colored beads by the cell type they belong to.   In the middle. We haven't performed any imaging.  We just colored the beads on the array by which   cell type they belong to you from the unsupervised  cluster. And they'll vary. Right you actually have  

an architecture of the tissue and you see that  we reconstruct the tissue architecture. We also   reconstruct the localized. of cell types, so the  takeaway is that there's no microscope. In fact,   you've turned the sequencer into a microscope  instead of just taking one image or take many,   many images across the transcript. And  just some ideas about specifications. So   he has very high spatial resolution, which we  validated directly with a single molecule fish.  

And then the resolution of the approach is about  10 microns the size of the beads. It also in some   recent work, we've demonstrated that we can really  improve the molecular sensitivity of the method   about an order of magnitude to about 1,000 unique  molecules per bead. This is similar to first   generation droplet based single cell approaches,  and I think this can be improved further   just in terms of optimizing the molecular biology.  But the nice thing is, is that because it's a   spatial capture approach, it's actually quite  scalable across many tissue types in organisms.   And so we've applied it to many diverse mouse  organs, a variety of human tumor specimens, as   well as many developing embryo samples. And one of  the things that we're focused on right now is how  

do we perform large scale experiments in terms of  batch effects and correlating these measurements   across many sections in. And so the other thing  that I think is very important with respect to   all full transcript data is that we want to  know how to two, there's two questions. One,   we want to know how we can use single reference.  Cell references that we've collected or birdie   annotated cell types. And the second is that even  though the beads are 10 microns, they're about the   size of a single cell. There can be basically more  than one cell captured per beat, about 1 and half,   like most people have. One in some 30% of the  beads have two cells. And we actually like to  

know if we want to estimate the contributions from  individual cell types to each beat on the array.   And so we developed kind of a computational  algorithm called robust cell type composition,   which was recently published, led by Bob  Dylan and in collaboration. Rafael is our lab.   And this algorithm basically given a single  cell reference in a spatial transcriptomics data   can answer what combination of cell types and  what proportion best fit each bead on the array.   So now you can take a single cell reference and  directly map it into space. And this allows us to   do a couple of things. One, it allows for really  high resolution cell type mapping because single  

cell data sets are often much more diverse and now  we can project very fine subtypes directly into   the spatial data. So for here, for example, we've  projected 27 interferons subtypes into a spatial   data set of the massive campus. And here you can  actually see kind of very fine layered excitatory   neuron subtypes projected into the cortex. And you  can reconstruct kind of the layered architecture.   And actually we've demonstrated to work  across many spatial transcription platforms,   including bizim, although VCM obviously has much  lower resolution than so. You can reconstruct the   same type of spatial architectures and as well as  imaging based transcriptase mix such as mahvish.  

And so that is why it's really important to  do this sort of cell type composition? Well,   I think it's because there's been a lot of work  in focus on detecting spatially varying genes   from spatial transcripted data. And in general,  these approaches have not been cell type of where.   But I think what you actually want to know in  your tissue is that you want to know how gene   expression changes as a function of space within  a cell type, as a function of tissue organization.   And so using this sort of approach, we're  beginning to be able to do that by discovering   within cell type spatial variation gradients  within a cell type. For example, these two genes,   which are within 3 of the masticate campus or  cellular neighborhood spatial variation like   what? How old is gene expression change in one  cell type as a function of who its neighbors are.   For example, here we discovered specific genes  which change as a function of their proximity to   excitatory neurons. Just for the last one minute  of my talk, I want to mention that in the sort of   slightly captured platform can really be adopted  to many different types of sequencing modalities,   including DNA and in particular in cancer. We  might be interested in collecting multiple McDade  

data when we integrate DNA sequencing data,  as well as RNA sequencing data to reconstruct   kind of basically the clonal variation in  combination with the transcriptomic variation. So   we've adapted the sights you can raised to capture  DNA through kind of a transposes based method   in collaboration with Jason pronouncers lab and  in actually it's easy to run serial section set   serial sections because these beats are the same  indexing. So now we can collect interior sections,   DNA and RNA. And this allows us to do  Novo identification of tumor, chromosomal,   no aberrations as well as carnality, for  example, assigning subclass by lineage,   by their relatedness in mutations. And then what  we can do is we can kind of quantify intrinsic   and extrinsic factors of which contribute  to gene expression. For example, if we have   compared RNA and DNA data and spatial,  we can examine basically like.  

How does clonal assignment versus basically like  environmental assignment affect gene expression,   so here we can group basically genes by  their various variants in gene expression,   explained by either their clonal identity,  which some clone of the tumor it belongs to,   versus like cellular neighborhood affects  one cellular or neighborhood effect that we   look at with tumor cell density. And so we can  assign genes which are associated with sub clone   identity. And this might be associated with kinase  and genes which are associated with, for example,   tumor or immune density, which are more related to  the cellular environment. So we think that's very   exciting. I'm just out of time. So I just want  to summarize that we've developed kind of very   high throughput experimental approaches for high  resolution spatial transcription profiling. I   think the development of computational  algorithms is extremely important   at this point for looking at cell type projection,  differential expression and interaction.  

And we're really leveraging new tools and genomics  to look at like Malcolm X and something. I didn't   talk about receptor sequencing in the context of  T cells in space. So thank you. And I'm happy to   take any questions at this point or actually,  I guess we're taking questions at the end.  

Yes, thank you for the technology is amazing. I  look forward to more development and new insights   on this, so we'll take questions at the end.  Now, is Miriam coming back? Nadia, so maybe we   moved to the two to ourselves because people want  from you. Stephen, Thanks for joining us today.   Thank you, Michel, and thank you and Southwest for  the invitation. And it's a great opportunity. I'm  

so glad to be able to report our work on studying  space, on architecture from prompting, chasing and   amoeba. So as a conflict of interest disclosure,  I'm actually Venter and a patent applied for   a Harvard University related to mirvish. Our lab  studies nuclear architecture. So I guess the first   question is, what are the architectures? We use  this word to describe the spatial organization of   the tea and other nuclear components. So we know  the genome is spatially organized in the folded  

across multiple scales inside of the nucleus.  First of all, DNA wraps around the histones to   form these nucleosomes structures. And then on the  other end of the spectrum, individual chromosomes   actually occupy distinct nuclear space. And  these are called territories. In recent years,  

Thanks to advancements in sequencing technologies,  we now know that at the large scale enough to   several KB there are these topological satiating  domains are that's whistling past that. There are   often something loops, such as promoting has our  looks and that has our father sort of these and B   compartments and a compartment is enriched with  the active chromatin and B compartment. This was   inactive chromatin. Other sequencing technologies  also revealed that there are specific regions of   the genome that are associated with nuclear  Islamiyah and there was nuclear Osler's and   these are termed the labs and that is  respectively. But these structures have  

been shown to control many essential functions,  not just transcription recognition, but also   DNA replication, mutation rate repair  recombination, and also, they show in triggering   dynamics and changes during development, aging  and many diseases, including cancers. However,   there are still many unknowns regarding the  nuclear market in all these biological processes,   largely due to technical limitations, especially  one key biological question is what is this   treaty? Folding parts of commenting at length  scales above the nucleosomes in single cells.   So, from the tests to the AP compartments to come  some territories, our understanding of the nuclear   architecture is largely biology. So, these are  essentially larger and larger blocks. How do we   get the real speedy boarding pass of the single  cells? And commission or sequencing technologies   do not directly measure this really folding  pass and often rely on population averaging. So,  

they don't really answer this question. How much  imaging techniques like the resolution and the   multiplexing ability to reveal this really folding  pass? To tackle this question, several years ago,   we introduced this image-based charity genomics  technique called the plummeting tracing that   images and the pinpoints the positions of numerous  economic loci along the same chromosome and the   link them to review the folding parts of property.  The challenge here is if you simply label so many   genomic loci along the same chromosome, you don't  know which one should be linked with which one.   And also, due to the different element of light  microscopy, there actually are all connected   into one patch. So, you cannot even resolve the  spatial following that. So many genomic loci  

to tackle these difficulties. We designed  this sequential imaging strategy, so we   sequenced only label image and a pin point that  you now make loci one after another. And you, all   of them are a sequential image and the rebuild and  we can link them based on all their anatomic map.   So, this way we can really reveal and resolve this  really boarding pass of individual chromosomes,   single cells. So, this work really opened up this  field called the image basis, really genomics. And   there are many funding studies from many different  labs, including several more works from our   independent lab at Yale, applying this technique  to different skills and to different model   organisms and biological processes. And these  are all based on actually multiple multiplexed  

harmonization of fish. So basically, a library of  that primary folks is simultaneously hybridized   to the targeted genomic loci simultaneously. And  these primary pops all have this overhead region   that are unique to each dynamic locus of interest.  And then one can basically sequentially hybridized   bilabial on a secondary process and imaged  them as we move the signal before the next   round of hybridization. So, in such a way,  all these dynamic loci one after another.   But we also see that field is rapidly  expanding to other varieties of technologies,   technologies also such as the very exciting uses  of sequencing technique that they just introduced   that for me, that's where also, you know, our  work, we one thing that we're focused on is to   combine this a highly multiphasic DNA imaging.  Was Holly multiplexed RNA and protein imaging   all to a single integrated platform called MINA  or multiplexed imaging nucleosome architectures?   So, you mean that we achieved multi still and the  multifaceted all imaging across many scales in   mammalian tissue, in a cell type specific manner  in single cells. So, this work, for example,  

studying mosquito liver tissue we first look  at the commenting folding organization across   four orders of magnitude of genomic lens, all  the way from promoting our interactions to cats   to compartments to problems on territory. And then  we asked the how are the folding differ among the   diverse cell types in the field, however. And to  answer that question, we image that 137 different   species was a technique called RNA mirvish,  which is a multiplexed RNA imaging technique   among these RNA species. 55 of them are cell  type marker genes. And we also label the cell   boundaries. So, it can segregated individual  cells and distinguish the cell types and  

are the cell types of specific questions of  nuclear manufacturers. We also try to answer,   like how are the genome folding associated with  expression changes? And to answer this question,   the rest of the RNA species are actually from  genes, from the choice to chromatin region. So, we   measure their expression as well. And remember,  we were interested in other nuclear components  

as well. And in this where we label the cell  nucleus and cell nuclei so that we can profile   the association of all these genomic regions  with the nuclear laminaria and study those   architectures. And finally, because everything  is done in situ, we naturally preserve the cell   positions in this complex tissue and also  the signaling molecule positions so we can   study cell cell interactions. And the underlying  signaling molecules in cell in this neighborhood  

are showing several example images from  the platform, from the same field of view,   in the same cells I'm showing here, the permitting  process is zooming in on two of them. It   looks like this mouse, chromosome 19, folding  trees. And then here are individual molecules   of RNA and each one of them is color coded with  over 100 subtle colors to show their different RNA   species. Cell phones relabeling cell type Markaz,  an expression cell type identification and also   the volume of the nuclei and then hopefully  olai. Next, I hope to show several example   analyses of whether we can derive what kind of  theological questions we can answer. Was this  

high dimensional multi all mic technique and  to data the first fetal liver based on prior   knowledge. We already know that the cells express  this very important and metabolic gene called   SCV to the other cell types. Don't express it that  much harder. If you look at the genome annotation,   you'll see that near the fetal to promoter.  There are several new cancer clusters.   Which one or one of them interact with the city  to promote her liver? Hepatocyte is and knowing   using the media platform, we can basically group  the permitting choices. And the choices of this  

region separately, and we can show that to this  cancer, interact with the city to promote her. So   basically, we're not still permitting folding and  can't discover cell types specifically promoting   cancer interactions in complex tissues. And  then we ask, what about the larger scale?   How much enfolding, particularly how are the  ab compartmentalization scheme differ in the   different cell types in flavor? So, we know  that the tax department comments are this AAA   and B compartments in a whole chromosome and a  compartmentalizing ritual with active promoting   B compartmentalizing ritualist being active  promoting. But these discoveries were made in   homogenizes cell cultures. Now we have a complex  tissue of so many different cell types. With MINA,   we can profile the ab compartmentalization  scheme and show that this scheme differs   between the different cell types. We can also  prove that in a single copy of the chromosome   that the AAA and B compartments do physically  exist as separated, prompting regions and the   changes also the compartmentalization scheme. They  are associated with C and expression changes. So,  

Mina reveals cell type specific to become  capitalization schemes that are associated   with an expression changes, and that's the way  of branching into the other nuclei architectures.   We ask what is the relationship between a, B  compartmentalization and the nucleus or lamina   association of cremini regions where marrying  the rate of the tax base will be associated   with the nucleus? And then we compare that with  the ab compromise or we see predominantly negative   correlation between these measures. However,  you see there are systematic changes between   the different cells. So, these architectures  they are specific to and the further show that  

this is segregation of the compartments, they do  not depend on the laminar or nuclear association.   So, these results show that prompting  compartmentalization. Our association   of associations do not solely determine each  other, and they may offer no controls to genomic   functions such as transcription regulation  in a cell type specific manner. And finally,   we are asking, what are the cell, how the fetal  liver cell types are arranged in space and we   are projecting the type identities back until  the image from these images that we can analyze   which pattern of cell types are preferentially  being neighbors with each other than expected,   and that we can see the particular cell types in  Paris are indeed preferentially being lavers. Some  

of these discoveries are consistent with  prior knowledge. For example, these red   cells seem to surround macrophage and this is  called a residual plastic island structure.   However, some of the cell-cell interactions are  enduring, for example, hepatocytes seem to be   interacting with the janitors and this is indeed  showing in our image as well. We ask her what her   interactions may underlie this arrangement.  And from our hundreds of RNA species that we   profiled on the same image; we find a pair  of legs. And as we set her leg and the kids  

that are preferentially suppressed in these  neighboring hepatocytes and everything else.   So basically, mimic and potentially reveal  cell type interactions. And the underlying   molecular mechanism as a summary tracing allows to  have spatial tracing that is really folding parts   of quality individual chromosomes, single cells, I  mean, enables multi scale integrative nucleophilic   imaging in complex mammalian tissue and allows the  discoveries of the Ormoc architectures in a cell   type specific manner, studying of the relationship  between architecture and transcription regulation,   also studying of cell cell spatial interactions.  This is because we now have one technique to image  

them all and in the darkness of the microscope  room, find them. And of course, our interest   is not just limited to tissue development in the  liver. For example, we have several works in the   lab applying the technique to different types  of cancers. And I hope to report those results   in the near future. Finally, I'd like to thank  my lab at Yale and my collaborators, professor   Samuel , Sherman Weissman, the representative  here where I naturally led by 2 talented   with students and value. And thank you  all for attending this presentation.  

Thank you, Stephen, for bringing this  data and technology from talking to us.   Very interesting. I'm a big fan of mirvish, so   I encourage people to put the questions in the  chip box as well. So, we will have a discussion   later on. So, I think we have Miriam back with us.  Very nice to have you back. It's all yours now.   I think you unmute. Yes, I did.  I'm very sorry for what happened,   so what can you tell me? When did when  did you guys lose me? At the beginning.  

The very beginning. So you can stuff. All  right. OK, well, you have to learn to do that.   Will try to be fast because I  understand that we have some time to.   We are on some timeline here. Can you see my  screen? Yes, I am anxious now to do something   wrong. I'm taking your time. OK all right. OK,  so I just. Can you still hear me? You'll hear me.   Right let me try to put these things OK. I'm just  trying to see whether I'm going to do something  

wrong. OK, so I was talking about my little  compartment in criminal. I think you may have   seen this. And specifically, I was emphasizing  that macrophage is the largest human compartment   in incompletions, but there is very little  known about this compartment because most of the   biology of has been obtained in Witold and there's  very little knowledge of it, especially in human   leashes. I don't know whether I've been through  this, but this is an important slide that you guys  

have to remember, because this is touching  to be part of the textbooks. But we now know   that there is a population of macrophage.  We think they do things very differently,   at least in some situations. There's this  big tissue isn't macrophage compartment,   that self-cleaning tissue that  is present higher than any type   of development that they are having printed  by tissue who they killed in tissue repair,   and that there is another population that are  located mostly in response to inflammatory host.   You guys, I'm still with you guys, right? Yes, you  can hear me ok? I just want to make sure you know   that, yeah, perfect. Then me. You hear me? Well,  and the slides are OK. OK, I'll let you know.  

All right, I then the other compartment that is  located on in response to inflammatory. Q so two   Melanesians, for example, a lot of inflammatory  money could in fact very, very early progression.   And we have observed the organization of my  life compatriots awaiting the bone marrow that   promote the good of mine. We put unitards into  treatment. We put into Indonesian s and they also  

are really part of the tumor. So, the first things  we did to learn about the macrophage compartment   is we go from this envoi used to profiling using  sites, which allows us to look at the antibody   profile of cells and a polygon using the next  platform. And we first decided that we were going   to spend a lot of time really profiling during my  confinement of treatment. Na ve criminals. We are  

focusing on criminals, including blankie and this  is what I'm going to focus on today, but also HCC   and where we are going to build this knowledge  base and then build on top of it, then we do.   We are doing some exercise to keep the tumor  fitted with monotherapy, combination therapy,   et cetera but today I'm going to talk about our  efforts to provide the equipment, na f lesions and   here. So, we use that knowledge to look for cells,  express the same antibody and the same kocon. And   we used to call them in cells, ILC T cells or  macrophage, and we identify several macrophage   Qualcomm. Then we look where they are distributed  because here, we are a pathologist. We always   isolate the tumor and the adjacent tissue and then  we purify this compartment from both the tumor and   adjacent that we are dealing with a different  part of the tumor to look at heterogeneity of   also of the regions. And then we do single  cell suspension and decisive analysis. So,  

the first analysis we do is look at distribution  of all these, a molecular and viscously identify   molecular and antibody profile of this minute  cells in the adjacent tissue. And we spend a lot   of time doing these then because we want to know  the medical condition, of course, and the spatial   distribution and the function of this compartment.  We have to have access to some experimental model.   So, we have a model that we imagine that we like  a lot in the lab that was developed by tilo jaxa,   which is this KP line which has extubated mutation  and P53 deletion. And these lines, when injected,   Kelly form these nice adenocarcinoma regions in  the mouse and in the longer of the mouse. And the  

first things we do, this profile is exactly the  same findings. So, committed to look at my view   of these solutions in mice. And then we use these  species, and that is to look for PROC that are   identical in mice and we identify for Pagan that  are similar or we coalesce differently still that   I showed you before. And we are going to focus  on these four different people because we were   able to do some fine tuning, especially using  Google analysis. I don't have the time to go in   deepening the analysis we performed, but we think  that these are programs that are somehow engaged   by 2 molecules, quite similarly in mice. And so  not when we had this program. The first question  

we ask is, what is the origin of these? My glove  compartments? Are these macrophages? Are they   derive from that blood circulating progenitors?  And to do that, we use a faith mapping model   that was developed by this crisis and that we  had, in fact, a profile in these models. What we   use is we use this one to calculate seventeen,  which is early adult film precursors. We develop   this mouse model that allows us to genetically  label both the map Th17 cells conditionally   upon tamoxifen injection. We conditionally label  17 and all the 17 progenies. And the genetic cases   remain that they are completely untouched for the  life of the animal. So, we feed them up the mice.   And then when the mice are six months old,  because it takes a lot of time to really fake my   faithfully the hematopoietic, the right cells, we  inject the tumor and then into tumor. We find the  

blood cells from the blood cells and we look at  the particular distribution of this mind. We put   on label, on label. And what we found is there  was one specific cluster that was never reached,   the label cells, which we call cluster 1. And we  these enable those to be densify these cluster   of macrophages as the tissue adjacent macrophage.  These are the tissue regions that are present in  

the tumor. So, the. We do this identified Moscow  that allows us to identify them using flow   cytometry so that we can plug their function or  also look at the official distribution. And what   we found is that this caused a very high level of  severe 169 high level of three to six. And this   enables us to look at the spatial distribution  of tissue into micro fit versus these adult bone   marrow derived macrophages. And what we found  is that the tissue Malkovich, surprisingly,  

are always present outside the criminal. And this  was also the same in human lesions. And then here   we use these molecular analyses, we use the  species. And then he says to see, OK, these   macrophages resemble the tissue of macrophage  that we analyzed using these faith mapping models   in mice. So we think and they also were reaching  to six. But also, when getting to 10, we use these   two molecules to profile the tissue macrophage  and we found are abundant outside the Melanesians.  

That was surprising to us because the  macrophage are part of the lung tissue   and just want to make sure that you guys are  here, OK, that you are here and and then so we   then started to say, well, OK, so let's really  look at the temporal distribution of macrophage.   And we started now to the image, in fact,  the composition, a different time location.   And what we found is that these are the first,  in fact, to interact with tumor cells. This is   what I'm showing you here in red, that these  are the tissue organisms macrophage. And Ingrid,  

these are a KP cell leukemia expressing feel  this is putting so that we can track them.   So, these are the first interactions with  micro fit. So, the experiments we do this   weather is to probe with this a little intimate  position. And here we use the reductionist model   in which we call ourselves with tissue macrophage  or multisided like micro fit, which we find are   abundant in advance to Melanesians. And we ask  whether they play different cording in promoting   Tamil invasiveness. And strikingly, what we  found is that the tissue macrophage where we   put up inducing this common position  where they could use the expression of   an induced expression of some kind, would  pull commonly used the formation of these   branching carinae. So, they promoted really the  capacity of macrophage to invade inside the jail.  

I'm going quickly here because this paper has  been published. I just want to oppose, in fact,   the ability of this issue to induce, in  fact, MTD antiproton. But what we also   found is that this isn't macrophage. We are also  interacting, especially only a joint application   with regulatory cells. And they were deregulatory  said that I'm showing here in red and these early   in fact, that action was very intriguing to us  and suggesting that maybe they were also inducing   a regulatory sale. And indeed, what we see is that  there is a first wave of T cells in Tamil that   we've looked at. And then the cells are dampened.  And I think this is something we think that this  

injection of t like. But here is contributing  to inhibiting this first wave of t relation   to the next experiment we did is we depleted  this macrophage without affecting the macrophage   compartment. And we ask whether this will be  used. In fact, to invasiveness, antiregulatory   cell injection. And this is exactly what we saw,  is that in the absence of tissue macrophage,   Tamils have difficulty to invade despite the fact  that the sieging of Tamil cells was not affected.   And we also saw that there was a reduction of t-  cell, but also the reduction in that direction   of the functionality of T regulatory cells and an  accumulation of THT cells. But this year, really,  

if we now depleted ITSM later, there was no effect  at all until closer. And this is because the serum   at present, as I've excluded then from the  tumor.So in order for the tumor to continue   to grow in order, I mean, something happened and  we are looking into exactly what's happening.   We believe that the chemo then excluded because  they are going to form somehow or build a lethal   hormonal structure that's going to be harnessed  by the tumor to into and are going to look to the   side macrophage, which then become abundant  in advanced stimulations. So, we started then   to look at this one macrophage and indeed the we  saw that in advanced nations monocytes macrophage   dominate the tumor microenvironment and the  outside the tumor microenvironment. However,   this one says the right macrophages that we  identified with our feet mapping methods are also   quite heterogeneous. But there was one program  that was specifically abundant in the tumor. And  

these people can cause very high level of this  walkable and now these two positive macrophages,   we have heard about them mostly SRE. So, this is  what I'm trying to do here is human lung cancer   admissions. There is abundant of two positive  macrophages. And this is a mouse, a lesion. There   is also abundance of temporal macrophage. OK, so  we had two positive macrophages. Sorry, before I  

got here, I'm just showing you something that is  positive macrophage at this particular cluster   that we identified in human was also present in  mice and in fact the outcome was quite identical.   were many that were identical between mice  and human. We also looked at the temporal   accumulation of these two positive macrophages  and what we saw that in contrast to the. That had  

to be quite early due to an application that  came to positive microphages recruited later   during tumor growth. So now we had this to talk  about, shared between mice and human, and they are   abundant in advanced lesions. And we are going to  look at how these two we have no tools to look at   all of these come to fruition. But this  is what they've been trying to tell you,   is that we have four of these symptoms, mostly  in the context of Alzheimer's disease. We've come   to vain have been shown to be associated with  increased receptivity to Alzheimer's disease.  

We have also identified abundant to positive  macrophage in colon cancer. In fact, in the paper   that we published in Cell Malkovich had been found  in liver. That happened also in adipose tissue   with these patients. And there have been also  shown to be abundant in experimental sarcomere   cancellations by markkula and examined recently,  in fact, in back-to-back papers published in Cell.   And so, so this clearly it seems that these  two positive things seem to be accumulating   different inflammatory cues. Something that is  common with all these people is the abundance   of either fat tissue antigen gargle and indeed  simple, easy scavenging receptors. So, the next  

question we ask is whether it is applicable in  the uptake of optimal cell degrees and to ask this   question with really high macrophage. And by now  we just take a mild marker in this case live in B   and that are high because high level  of GFP or low level of GFP. And then we   look at their molecular whole. And what we found  is that the positive and the they could come that   I described to you was fairly strongly enriched  in this GFP high macrophages that we isolated   from a mouse, Tamil. Then we did now another  experiment to further prove that this uptake   of the cargo was inducing these Cocom. We divide,  but not with the right macrophages, if you will.  

And then we Fed them with expressing GFP. And what  we saw is that these feeding process will induce   this program to include the expression  of time to if you want an MBA to eat. I   have been up to all these people that seems  always to be expressed together, but in mice   and human associated macrophage. And now we  see in the picture in macrophages that I have  

cleared or captured apoptotic tumor cells.  OK, so the next experiment we ask is whether   what happened if we now try to delete them  too, will we affect that had shown and that   the deletion of time to reduce the progression of  sarcomere consolidation's? So here we did the same   experiment where we needed to fund the whole stuff  from the microenvironment and injected the visa. I   didn't look at my life and we saw that  in the absence of a positive macrophage,   that was a very strong prediction of tumor.  And this was associated with the organization   of the mildly puddicombe. We saw injection of  this womb pushed up an expansion of one site.   An extension of tissue has been to macrophages in  blue, but also an extension of a specific subset   of this one. And this seems to be very specific  to this compartment, because this is two which   are usually a month and was not affected by the  deficiency. So, something was happening in the  

myeloid compartment of deficient animal. Now,  the problem is when we saw that there was such   as tongji, a reduction at supercool, but the  injection of them could also contribute to all   this mildly changes that we observed. So, we did  a cool experiment. I think when we look at it in   mice with bone marrow cells that are mixed  blood cells with a mixture of so proficient   and efficient bone marrow cogenitor, which will  give rise now to control proficient and deficient   macrophage in the same compartment. Now there  is no effect on millhauser. So we are going to   purify come to a positive and negative macrophage  because we label them differently genetically   and we are going to compare their proconsul  here that is identical. We are just looking   at cell and testing all of time too. And  what we saw was very exciting. The first   we saw that positive and negative macrophage  uptake, the same amount of tumor antigens. So the  

cancer was not with insulating uptake, but we saw  that to regulate regulated very strongly that he   pulled. It's very strongly then the inflammatory  production and expression, of course, stimulate   the production of a lot of inflammatory monikered,  including the production of five, 15 and 18,   which are very important, could to activate and  expand and says, I'm going to show you. They also   come to whitecap was important for the plot of a  scavenger scepter and also a lot of checkpoints,   including this little bit for which we are very  excited about. OK, so that's quite interesting.   So now we go back and ask what happened in the  environment of these two deficient sentiments   and will come to deficient anyone? And what we  saw is there was a very strong expansion that. So,  

I showed you the reorganization of the microwave  compartment. And what we also saw there was a   very strong expansion of NK cells. And  we are very excited about this because   in cases are always depleted from the humiliation  that we have seen that in human lung cancer. Also,   there was a very strong depletion of cells. But  also, we saw in that accumulation of activated   cells, we also saw an accumulation of 80 cell,  but also an accumulation of cells that in   one are excess, also can be suggesting that  these were affected the effect of the cells.   So, the next experiment we do that is depleted  to or 10 cases and ask whether we abrogated   this beneficial effect. And this was indeed  the case when we and not so much. So,  

we are quite interested in this. It also seems  that the beneficial effect was mostly due to the   recruitment of a castellan, their occupation, and  we also obligated these to see one accumulation   here. So, this is quite interesting because first  time we are able to expand in the compartment   and expenditure one, and because of this  result, we are starting our first in human   study in advanced cancellations in patients  at Saini. And this will be done with a small  

biotech that have been acquired by Gilead biotech  that was, in fact, from the coma. So, we are very   excited about this. In conclusion, I'm just going  to summarize my what I showed you. showed you   the tissue relating to Michael promote jam rock  invasive next to the injection of an individual   kinmont position on an injection of only two  legs that then contribute immunity. This is very  

relevant for cancer institutions. We have shown  similar results in so many years ago. There was   a recent study showing that tissue macrophage  also promoting this early invasiveness in   the study seems to be leaving the property of this  intellectual fate. This is very interesting to us   because we have a very strong medicine that we  can detect this early in site link institutions.  

And we are going to now start a clinical trial to  try to activate or deplete the tissue macrophage   without operating on these lesions and see whether  we can somehow use tumor. Also, I've showed you   that the children are depleted from advanced  lesions, which are dominated by macrophage, so   that provide us with a way of targeting the right  macrophage while sparing tissue isn't macrophage.   We also have a lot of homeostatic property,  including in the lung. They are the one that are  

clearing the surfactant. They could be lung tissue  integrity. So, if we don't have to target them,   then let's don't do that. I showed you that is a  very good target of these macrophage. I've also   showed you the sensing of the bleed and use it.  And these, for example, can be to the depletion  

of encases in d.c., one from the Michelangelo  momentum. And and that came to look to deficiency.   But in collaboration with Michael Goodwin, we are  also using the trip to the blockade, which led   to very similar results where we can put on the  middle landscape. I didn't show you the body of   time to do that. The single celled sequencing of  kintu deficient macrophage, you've seen them. So,  

they start to be inflammatory. They start to  produce inflammatory molecules, cuticle could and   castell in documentation suggesting that we  can reduce their ability to promote immunity.   And we think that this is really a very strong  Avenue to put in hands of responsible and cancer   patients. These are my people. I know we are  out of time. I'm not going to go through all   the names here. Unfortunately, this is our  group wearing our vaccine campaign t shirts,  

and I'd be happy to address any question  at the end of the talk. I hope I didn't go.   I hope it won't be too long, and then with my  presentation, I hope that you referred to me well.   Yes, beautiful talk, Miriam,  quite an exciting discovery.   I'm glad that I was able to finish it. And did I  go over the 20 minutes to much or not? I didn't  

have a timer on it, but I think we are running out  of time. We have a lot of questions in the list.   So, I know that they need to go. Do you  have time to take one question? Yeah, yeah,   of course. I can think a couple questions right  now. So, first question for you is, how do you   quantify the density of tumor cells and stroma  components in the tumor microenvironment from the   data right in there? What we did was we used  the transcript to make signatures to map the   cell types into the space. And it's a spatial  density based on the transcription signatures.   You should stop sharing my slightly. OK, I  should stop shooting for four hours, because   if you have questions for each other, feel  free to interrupt me to ask and share here.  

OK well, I was going to ask my God that I  would love to, I would love to. Now, look at my   she said corruption in the government. And, yeah,  we started to look at it. So I didn't. Yeah to   purify my lord, my Lord. Then he said comportment.  And what's clearly the kind of government is to  

have an impeachment of Michael Vick. He said  intellection that dominate the action. I   said so can we. So to somehow forget  the sequencing is the best way to go.   Can we do some target sick where we just at least  locate my said fact locally and then gloopy? Now,   possible in today's world, there's  many ways it depends on if you want   to ask questions about so it depends on  what your questions are like. Are you   interested in the differential expression  of genes as a function of the interactions?   And you probably already have a list of targets  from single cell, in which case it sounds like   it's a good experiment for targeting imaging  based methods as well, like mahvish or situ   sequencing, which some of the technologies that  we didn't I mean, that Steven talked about,   I didn't get a chance to talk about.  But that's great for targeted methods.   If if you're interested in doing discovery of what  the transcript like, what transcript terms are   changing as a function of interaction, then you  can use like untaken methods, like slides or   speech, which makes it easier to inflation rate.  So it's important to start having a sense of   a single solution just because I'm sure  to rate the genes are present everywhere.  

So so give me a little bit of the resolution that  you guys have now if let's say we want to go.   I mean, I think I mean, Stephen can  talk about more fish more than anyone,   probably, but I think the resolution is  very high, if you know what the targets are.   Yeah, I started the  collaboration with muffie Steven,   as you know, so I'm looking forward to a very  exciting and don't I like we have a molecule   that I'm very excited about and they're  all encompass all these monkeyface   comportments that I described here today.  So, I'm looking forward to this result.  

You're the resolution. Actually, I have a question  for you as well. So, you see the garfish have   this a single molecule resolution. But I think for  a lot of people, actually, they don't need to know   the subsequent our distribution of RNA and wonder  at a wholesale level spatial resolution is enough   and technique, definitely. If you use the smaller  spatial resolution, would it be better and better?  

But I guess the texture of RNA will get lower  and lower. So what's your estimate of how small   the beat is optimal? Like a balance between  a do or just your current size is already   the optimal? That's a good question. I think  probably optimal is, as you said, basically   the capture rate scales with area. Right? like  there's just less RNA. Ernest goes with area.  

And so if you have the if you have the size of the  beads, you get a four times reduction, but then   you can dynamically aggregate. We aggregate the  beads. If you have good computational algorithms,   that's like a little bit like a different version  of the segmentation problem. I actually think that   probably the optimum lies at like  5 to 10 micron 5 microns maybe   you already achieved. Essentially, it's close  to where we are. I mean, we chose, what, 10-4   four? Good balance between those things.  Yeah, but that is going to be impacted by   how crowded places today are still there. Hey, did  you see in the female microenvironments, will we   find that, for example, where there is lymphoid  aggregate resolution that you would require would   be much higher than regions that are sparser?  So, it's important to continue to think about   that. The distribution is not equal and that  solution is going to be affected by that. Mm-hmm  

Yeah, OK, OK, I think we probably need to  end this session, but it's great to see you   guys already talking about this and then  more and more questions are coming up. So,   it's great to have biology and technology all  coming together. And then we can bring cancer   research on many other biological researches to  the next level. And we also have many questions   in the oral question, but I'll send those  questions to our speakers. So that they can   go through them and probably get back to you. So,  with that, I think we have come to the end of the  

session. We would like to say Thanks again to  all of our speakers, Miriam Fey and Steven, for   their engaging presentation. We are very grateful  to our sponsor origin for the contributions that   made these women all possible. We would also like  to thank you, the audience, for tuning in today.   If you missed anything during this, lamina,  or you would like to listen to it again   of the record. The recording will be  available shortly, though. Still to come,   if you have any comments about the women, all  suggestions for future topics would love to hear   from you by email at CP women women. Say thank  you. Bye bye. Thank you. Thank you. Bye bye bye.

2021-08-12

Show video