Recurrence of tumors following anti-cancer therapy is often the cause of patient mortality. Patient autopsies reveal that cancer patients have metastatic deposits in distant organs from the primary site. It is unclear what factors, both cell autonomous and in the metastatic niche, facilitate either the maintenance or evasion of tumor dormancy.

Created by Sara Gosline sgosline
I lost the connection when we talked about the therapy induced dormancy part. With the EMT model, I have tried to treat the animals with cyclophosphamide. The surviving GFP+(EMT) tumor cells showed convincing dormancy marker expression (Ki67-, p21+, high p-p38/p-ERK ratio). It will be interesting to look some details with 3D models. I will input these in our shared document. Also, I have RNA-seq data of RFP+(Non-EMT) cells, GFP+(EMT) cells, and GFP+ CTX surviving cells. I will be happy to share for preliminary pathway analysis, which may give us some potential targets.
Requesting access to the doc. Thanks!
Thanks @subhode! You are right, this is better to do some edits and follow up.
@sole It is a great start and I agree with the plan. Within the Synapse folder 'Tumor dormancy and the metastatic niche' (which is within Files > IACM Project Workspace) I have put link to a google doc file 'Tumor dormancy and metastatic niche - working draft' where I have put the above write-up of @sole, so that the team can continue to edit and develop the proposal. PS: In case, someone has difficulty accessing the google doc via the link above, it is here: https://docs.google.com/document/d/193HGSHlB7SAKeQrHgloTYG_89blTFdu6GatiAA7V6J4/edit?usp=sharing
Hi guys! yes I will do my best to meet you at 1 pm (I will be traveling in the west cost that day). If for any reason I can not make it I will let you know. Based on the data, ideas, questions and the people that were on the video call last Friday I wrote down this draft of a project. Maybe we could propose 5 initial aims and then narrow them down to 2 aims because we only have 9 months to develop the project. Feel free to change anything, add comments and propose new aims. Remember these are just ideas that we can adjust to everyone's interest. _Background:_ Clinical and experimental data have shown that during the lifetime of a cancer patient there are different populations of DCCs or seeds for metastasis. Some of these DCCs arrive to secondary organs way before a primary tumor has been formed (early DCCs). However, these DCCs remain dormant for almost all the evolution of the primary tumor. Primary tumors are constantly releasing new pools of more genetically advanced DCCs (late DCCs). However, even these late DCCs undergo dormancy by specific signals at target organs and/or therapy. Thus, how these populations of DCCs contribute to metastasis formation remains unknown. We recently identified at least 2 dormant DCC populations (early and late) in lung and bone marrow in a model of spontaneous breast carcinoma (MMTV-HER2). We have developed in vitro, in vivo models and RNA-seq from these early and late precursors of DCCs (bulk and single cell data). These findings support the heterogeneity of DCCs and bring the question of which of them to target and how. **Aim 1: Contribution of DCCs to relapses** Aim1.1-which population of DCCs (early or late DCCs) gives rise to spontaneous metastatic growth? Based on preliminary data (published and unpublished) we propose 4 scenarios: 1-mets developed by early DCCs; 2-mets developed by late DCCs; 3-ealry DCCs favor the growth of late DCCs; 4-late DCCs provide signals to reactivate early DCCs. Aim1.2- how do these early and late DCCs respond to dormancy and reactivation signals? _Note: the majority of research on dormancy has been done in advanced tumor models thus there is little knowledge about dormancy in these early, less genetically altered DCCs._ **Aim 2: Therapy responses.** which population of DCC remains dormant and resistant to chemotherapy/immunotherapy? **Aim 3: Interaction of DCCs with the stroma.** Do they create specific niches within the same target organ? does the niche change overtime (from single cell to small clusters, micro and macromets)? Do these populations interact with each other? Do early cells prepare the niche for later arriving DCCs? Do late DCCs provide the signals for reactivation of early DCCs? _Proposed work:_ In order to address these questions we propose to build testable mathematical (this may be relevant to @subhode) and computational models (this may be relevant to @lanilonzo) using our available RNA-sequencing data from precursors of early vs. late cancer cells (single cell and bulk data). We could also use vitro models (3D platforms @soloriol) and in vivo (@gaodch, @sole) or ex vivo models and complement them with high-resolution imaging (@edondossola, @jjbravo, @milesm (whole organ clearing)) to visualize single and clusters of DCCs and record their interactions, motility (@jjbravo) and surrounded niche over time (the latter may be relevant to @alexandra.naba). We could also interrogate response of dormant DCCs (early vs. late) and their mets to chemotherapy/immunotherapy (@gaodch) and the cytotoxic drugs developed by @milesm.
Both times work for me too.
Either 12 or 1pm works for me.
Would Friday at 12:00 or 1:00 be a good time for the next teleconference?
Thanks for the discussion today, everyone! The meeting notes can be found here at this Google doc (starting on page 9): https://docs.google.com/document/d/1dB7DWkZMsN5csAdUrQjh4HPypAb_t3HkNB9ThawTlsc/edit?usp=sharing Feel free to make edits/additions directly to the Google doc.
Hi Nastaran, I am connected to Cisco. Thanks
Just catching up with th ethread but would definitely be interested to discuss more your ECM signatures @sole. I should be able to join the call tomorrow afternoon at 4pm EST.
@JAguirre-Ghiso @jjbravo, exciting!
Hi All, For those of you available to participate in a WebEx teleconference on this topic of tumor dormancy and metastatic niche on Friday, May 5 at 4-5 pm EDT, I will be hosting the call. Here is the link to join the WebEx: https://cbiit.webex.com/meet/zahirn. Click on the link and select for the WebEx meeting to call you directly at the number you provide. Alternatively you may call in to 1-650-479-3207 and provide the access code 731 432 061. By using WebEx you will have the capability to share your computer monitor during the call to share slides, papers, etc. as needed. I look forward to helping facilitate your conversation on Friday!
I can join 4 on Friday.
I can also do 4pm Friday, and updated the online calendar. Seems like it is the common preference for those who answered so far. In case, it is not a good time for others on this thread, I have created a Doodle poll based on the available slots Thu and Fri here: http://doodle.com/poll/87pxnf4emaczhzdq. Feel free to use it and we could see what is a good time for the majority.
I will be happy to join the discussion on Friday.
I think so @jjbravo - strong ECM, adhesion, migration signature.
I can do Friday at 4PM instead.
Hi Luis, I just realized that there is not time slot for Thursday at 4 pm, the only available time that day is at 11 am or 12 pm. Would it be possible for you to meet at 4 pm on Friday? We could also find another day next week. Best, -sole
I think that getting a call together is a fantastic idea. Thursday at 4pm Eastern Time works best for my schedule.
Hi @soloriol, that is really interesting. We could actually study the effect of ECM on dormancy using your 3D platforms. I agree with Sara that we need to start creating a project soon. Based on these discussions I would love to have a teleconference with @subhode, @lanilonzo, @aedinc, @scarc, @jjbravo, @soloriol, @snyderjc1, @Gaodch, and @alexandra.naba and of course anybody else out there. I think we can combine our expertise (dormancy, intravital imaging, computational analyses, mathematical models, ECM interactions, drug resistance) and create a project around characterization of DCC population. The characterization may include: 1- interactions of different DCC population (i.e. early vs late DCCs) with ECM at single cell levels all the way to small clusters and macromets, and how the ECM changes in all these steps, 2-differential motility within niche and in each population of DCCs, 3- comparison of time to reactivation from dormancy in these population of DCCs, 4- drug resistance of these populations of DCCs, 5-generation of mathematical models to predict behavior (all the previous points) of these populations of DCCs once at target organ (we have single cell and bulk RNA-seq from the precursors of these population of DCCs). I would like to propose a teleconference this week on Thursday at 4 pm or Friday at 4 pm. Let me know if you are interested in discussing these ideas or any other ideas you think more relevant. Thanks Sole
Hi Everyone. I'm a bit shy when it comes to commenting on message boards, but I think that this is a fantastic topic. I was wondering if there would be any use for a system that can systematically evaluate the effect of the ECM composition on dormancy. I have a 3D printing platform that allows us to create a polymer skeleton that only takes up about 3% of the free volume, and we can completely embed the scaffolding in different ECM proteins. Thus far we have been able to embed the polymer scaffolding in a native-like fibrillar fibronectin network (similar to the structure observed in tissues), fibrillar laminin, a thin smooth laminin film, collagen type I, as well as mixtures of ECM proteins that include tenascin-C.
Hi @subhode, @jjbravo, @sole, @lanilonzo, and @aedinc, This is a great discussion! To move it along further I suggest signing up for a teleconference this week. The signup sheet can be found [online](https://docs.google.com/spreadsheets/d/1hY53jRaqoBMnb9HhuE8dN3k4ejX37gG-4C4pattxkuM/edit#gid=1701101959) and as many of you can sign up for as many topics as you'd like. Just put the 'call topic' in column F. The workshop organizing committee is ready to host and facilitate these calls as need be, all you need to to do is call in the number in the column A. You can use this forum or communicate offline to find a time when all of you can call in to discuss further. The sooner you circle around a project the more time you will have to prepare for a successful project in June. Please feel free to ask me if you have any questions about the process. -sara
1.1, 1.2 - Can you infer if a metastasis came from early DTCs? Hosseini showed elegantly that aberrations shared between primary tumours and metastases were acquired earlier (Fig 4h), indicating early genetic divergence (and early DCC origin). If we can get copy number data from the primary and mets, we can further deconvolve the bulk copy number signal and infer subclonal architecture within the primary and met(s). _**You should contact Christoph Klein to see if he could share the data for analysis**_ This can help address whether the primary and mets follow similar evolutionary trajectories. Moreover, it can help identify whether there are shared subclones, suggestive of potential cross seeding between primary and met(s) and/or among the mets. This might complement secretome-based analysis and address #1.2. _**__Maybe if everything that cross seeds grows and you do not have dormant DTCs. I guess the data analysis might reveal it. Regarding the secretome I feels you would be missing data to infer those interactions. **_ 1.3 - Do late tumors tap into early dissemination mechanisms to spread even if they are genetically more "evolved"? Do the metastases derived from early and late DTCs differ in their aggressiveness? This is partly related to the original questions presented by soledad sosa (sole) : Do we need to target both populations to extend remission phase? Hosseini et al. showed that animals transplanted with primary tumourspheres had a higher number of metastases compared to those transplanted with primary tumors (Fig 3f). _**__Harper and Sosa showed the same**_ If the mets derived from the primary and tumorospheres are transplanted into WT mice, do they grow similarly? Do they have comparable number of secondary metastases? _**Experiment was not done **_ The observation, when combined with expression profiles, could be compared with computational models guided by different hypotheses. We could use branching process model with data-driven estimates of model parameters (e.g. growth rate, tumor density). From the Fig 1A of Hosseini et al. it appeared that the primary and metastases are considerably more similar to one another, than they are to the early lesions. Do you think this is a microenvironment/density-dependent switch that the cells are more mesenchymal-like in environment with low cell-density, and transition to Her2 activated proliferating state when cell density increases? **_Possibly - metastases and tumors are masses that have switched on the grow programs downstream of HER2. The early lesions are more committed to a movement branching morphogenesis program with components of stem cell programs_** In other words, in this model system, would expression profiles and seeding potentials of nascent and grown mets be similar to early lesions and primary, respectively, irrespective of their early/late DTC origin? _**Hard to tell, but seeding potential is different in early and late lesions... there may be cells with less growth capacity with more seeding potential in some tumors as they grow.**_ _**Good discussion. I suggest you pick the battle that can be fought based on available data that is interpretable.**_
@JAguirre-Ghiso, thanks for putting the ideas cohesively. I very much enjoyed reading Hosseini et al. and I revisit some of the findings in the light of the current discussion.   1.1, 1.2 - Can you infer if a metastasis came from early DTCs? - Hosseini showed elegantly that aberrations shared between primary tumours and metastases were acquired earlier (Fig 4h), indicating early genetic divergence (and early DCC origin). If we can get copy number data from the primary and mets, we can further deconvolve the bulk copy number signal and infer subclonal architecture within the primary and met(s). This can help address whether the primary and mets follow similar evolutionary trajectories. Moreover, it can help identify whether there are shared subclones, suggestive of potential cross seeding between primary and met(s) and/or among the mets. This might complement secretome-based analysis and address #1.2.   1.3 - Do late tumors tap into early dissemination mechanisms to spread even if they are genetically more "evolved"? Do the metastases derived from early and late DTCs differ in their aggressiveness? This is partly related to the original questions presented by @sole: : Do we need to target both populations to extend remission phase? - Hosseini et al. showed that animals transplanted with primary tumourspheres had a higher number of metastases compared to those transplanted with primary tumors (Fig 3f). If the mets derived from the primary and tumorospheres are transplanted into WT mice, do they grow similarly? Do they have comparable number of secondary metastases? The observation, when combined with expression profiles, could be compared with computational models guided by different hypotheses. We could use branching process model with data-driven estimates of model parameters (e.g. growth rate, tumor density). - From the Fig 1A of Hosseini et al. it appeared that the primary and metastases are considerably more similar to one another, than they are to the early lesions. Do you think this is a microenvironment/density-dependent switch that the cells are more mesenchymal-like in environment with low cell-density, and transition to Her2 activated proliferating state when cell density increases? In other words, in this model system, would expression profiles and seeding potentials of nascent and grown mets be similar to early lesions and primary, respectively, irrespective of their early/late DTC origin?
@JAguirre-Ghiso you are right ,Bova lab showed evidences of clonal cross-seeding ( this is the paper: http://www.nature.com/nature/journal/v520/n7547/full/nature14347.html ) and also Hong MKH ( https://www.nature.com/articles/ncomms7605) . It will be interested to know if that bridgehead metastatic sites provide the tumor cells with a special feature ( migratory, invasive, ) that allow them to land in another niche. As we have been discussing in the other threads, if for example the DTC seeded in the lymph node are altered by the mechanical properties of the microenvironment and help them to travel easily to the BM for example, this changes may be more transient and will require analysis of activation of signaling pathways such as GTPases (using single cell biosensors.) for example.
These are very interesting discussions. Maybe you want to identify 2-3 different questions to address and this may guide to the databases and type of modeling to employ? for example: 1- early vs late DTC biology and metastasis 1.1- can you infer if a metastasis came from early DTCs? 1.2- Do early and late DTCs cooperate to form metastasis? (may need more cell biology modeling first) 1.3- Do late tumors tap into early dissemination mechanisms to spread even if they are genetically more "evolved"? - early lesion vs. primary tumor vs. metastasis signatures are available in the **Hosseini et al., paper in Nature 2016 **as well as genomic profiles of the metastasis and primary tumors. (may help with Q1.1) - Soledad has access to early vs late DTC precursor signatures. (may help with Q1.1 and Q1.3) - existing databases of gene expression profiles may be interrogated for the presence of the early DTC signatures (access to single cell RNA seq may help as in bulk sequencing you may dilute subpopulations. (may help with Q1.1 and Q1.3) 2- inter-metastasis exchange. (@subhode - For instance, do you know how would the inter-met interactions via secretome modulate their pathway-level signatures? Also is there any possibility of cross seeding?) - sequencing studies from the Bova lab have provided evidence for clonal cross-seeding I believe. Cross talk via secretome could be modeled using experimental metastasis in different organs in different mice with some form of tagging or barcoding system to identify the molecules from one animal in the other connected via parabiosis?
@subhode we do not have data sources specifically available for this workshop. If there are specific data types you are looking for, let us know and we can try to locate possible sources. The data related to the PS-ON paper you referenced are available here: https://www.synapse.org/#!Synapse:syn7248578/wiki/405995.
Hi Aedin @aedinc, this is great! let me know about the Kim data! And yes you are welcome to use the single cells RNA-seq data as well. @subhode, I will check the Cancer Research paper. Thanks for sharing. Hi Elana @lanilonzo. Thanks for sharing your knowledge. I think all collaborations are possible. I am really excited to see how these projects develop over time! Happy Friday!
@subhode I wonder if it would be helpful to have more of a computational thread to discuss potential synergy between algorithms as well as the scientific basis of projects. What would folks think of that?
@lanilonzo , of course, I'm very much interested to collaborate with all of you on different aspects of the project. I think as the discussion goes further, we will learn more about the model systems, data, and the scope of project, and how we can contribute. I'm looking forward to it. @nzahir, would there be any biobank, clinical, or epidemiological data sources available for this initiative that could complement resources from the investigators' laboratories or publicly available repositories (similar to that in https://www.ncbi.nlm.nih.gov/pubmed/23618955)?
Also @subhode I would be really interested in seeing whether there is room for collaboration to integrate the mathematical modeling approaches with the sc/bulk RNA-seq data to better characterize the dynamics from the genomics data.
It would be great to be in touch about potential collaborations for scRNA-seq data in this context. It seems like it's a central interest and may benefit from collaboration in this context.
Lots of really cool ideas, it is really exciting! @sole yes, our recent paper in Cancer Research on mathematical approaches to model population dynamics of metastatic clonal populations can be found here: https://www.ncbi.nlm.nih.gov/pubmed/28381541
I have been working working on new methods for analysis of the Kim et al., data. I will know a little later this week, next week whether these outperform existing approaches. But developing approach for scRNA is def an interest for us Aedin
@sole Thanks! That's sounds really interesting. It will also be interesting to analyze how early vs late DCC sense the different types of ECM (FN, collagen) and how they degrade it. Maybe that information ( ECM, actin related genes) can be extracted from your signature and can be a mark to determine niche seeding and fate. Could this information be used to predict invasive properties of eDCC vs late DCC? Do these different populations have changes in the actin cytoskeleton and matrix degradation? Thanks, Javier
Hi @jjbravo, yes, they do indeed. ECM is one of the top hit. Hi @subhode, Thanks for these papers. I'll check them out. Do you guys generate mathematical approaches to predictive cellular behavior right? Thanks.
Hi @sole @subhode this is a great threat,! I enjoy reading you guys! @sole I have a question: do the early DCC signature identify/point towards a microenvironment feature that may feed this early cells (ECM, immune cells...)? Thanks, Javier
These papers might be relevant for you Single-cell RNA-seq supports a developmental hierarchy in human oligodendroglioma http://www.nature.com/nature/journal/v539/n7628/full/nature20123.html Application of single-cell RNA sequencing in optimizing a combinatorial therapeutic strategy in metastatic renal cell carcinoma https://genomebiology.biomedcentral.com/articles/10.1186/s13059-016-0945-9 Visualization and cellular hierarchy inference of single-cell data using SPADE http://www.nature.com/nprot/journal/v11/n7/full/nprot.2016.066.html Subho
Hi @subhode. Thanks for your comment. These are all important questions that we can test in animal models and ex vivo pulmonary assays. I also want to comment that in one of the papers we published last year (Hosseini et. al. Nature 2016) we actually showed that tumor masses released less tumor cells into circulation over time. So, it could be possible that still the burden of late DCCs is higher than early DCCs but not excessively larger as we may think. You are completely right that the growth rates of metastases may differ depending on which cells gave origin to the metastasis and maybe the presence of early and late DCCs in the same niche can actually affect the growth rate. We propose to use the RNA-seq data (from early and late cancer cells) to generate a signature that could predict behavior of DCCs once in secondary organs. But as you mentioned, the interactions between early and late DCCs and even with the stroma can alter their signatures. Thus, after establishing potential mathematical predictions about their behaviors we will experimentally test them by using fluorescence tagged early (CFP) and late DCCs (YFP) and analyzing the fluorescence of resulting metastases in real time. These experiments will reveal the possible cross seeding effect, the frequency of metastasis from each origin, the frequency of single DCCs that remain quiescent, etc. And we can also generate new RNA-seq data from these experiments that will reveal if a CFP+ metastasis derived from early DCC retain a similar mesenchymal-like signature. Do you have a predictive algorithm to implement using the RNA-seq data? Thanks.
Hi Sara, great topic. Hi Sole, as you said, it is likely that the early vs late DCCs have different potential. It is also likely that the proportional burden of early vs late DCCs differ over time i.e. while the burden of early DCC remains the same, the burden of late DCCs increases with time. The frequency of observed mets would probably depend on both the size of the seeding population and their metastatic potential (and then the growth rate of the metastatic clones). Do you think it would be possible to infer the likely origin of a met (late vs early DCC) from molecular signatures? For instance, would mets derived from early DCC retain their mesenchymal like signature? As you suggested, if early/late DCCs (or mets) interact and influence their cellular processes, the pathway-level signatures can be complex. For instance, do you know how would the inter-met interactions via secretome modulate their pathway-level signatures? Also is there any possibility of cross seeding? We are very interested in this topic and have just started to look into bulk and single cell RNAseq data for GI cancer DCCs.
Hi Sara. This is a great topic that needs further investigation. It is also important to consider the cellular identity of those "seeds" that will give rise to new metastasis. If different types of disseminated cancer cells (DCC) arrive to secondary organs over the lifespan of a patient then their interactions with the new stroma could influence it in different ways. Moreover, even when a metastasis is formed there are still dormant single DCCs few inches away from that metastasis. This is telling us that the heterogeneity of DCCs may dictate their own fate (by creating different niches) even in the same target organ. Using mouse models we have identified at least two types of DCCs based on different time points during tumor progression. Very early in tumor progression, even in the absence of a tumor mass (pre-malignant stage), we detected DCCs at secondary organs and we called them _early DCCs_. Then, while a tumor is forming, it constantly releases cells into circulation and these constitute what we call _late DCCs_. By immunofluorescence analysis, we showed that both populations of single DCCs were dormant (during a time frame) and that early DCCs were more mesenchymal than late DCCs. These findings invited us to ask the following questions: **which one of them will escape dormancy first to form metastasis? which one will remain quiescence and resist therapy? do they interact with each other? do we need to target both populations to extend remission phase?** We have generated RNA-seq from these two populations (bulk and single cell analysis) and we think this novel data set will help to create mathematical models that might predict the behavior of these two populations of DCCs and their contribution to metastasis formation. We also expect to use high-resolution imaging technology and secretome screening to define interactions between early and late DCCs. If anybody out there has special algorithms/tools/expertise that could be applied to this type of RNA-seq data please let me know and I'll be happy to share more information.

Topic suggestion: Tumor dormancy and the metastatic niche page is loading…