Characterizing coexisting clones within a tumor
One reason tumors can adapt to our treatment is that they are composed of heterogeneous cells. The relative frequencies of subclones within a heterogeneous population changes in response to treatment. The progress in quantifying cellular diversity in the past years has been remarkable, in part thanks to advances in single cell sequencing (sc-Seq). But scDNA-Seq for example is just one perspective on a tumor cell, in turn yielding only one of many perspectives on the tumor’s population composition. Looking at the cell’s RNA instead of the DNA, we will likely see a different population structure. And yet another if we look at imaging data and the cells' microenvironment.
Awards: AACR-Bristol-Myers Squibb Oncology
Clone specific vulnerabilities
Knowing what a clone really is – knowing what constitutes a clone in a tumor – is a powerful metric that informs parameters of tumor evolution that are otherwise difficult to measure directly. Clone characteristics can help inform mathematical models and prediction of therapeutic strategies to steer clonal dynamics into the desired direction. Our lab investigates changes in the clonal composition over time in the face of defined selective pressures. Two projects to that end define the selective pressure as either: DNA-damaging therapies (1) or resource limitations (2, 3).
But what we ultimately want to know is not what subpopulations differ in their DNA or RNA, but we’re interested in that outline of subpopulations that defines differential drug sensitivities. Data integration is likely to bring us closer to this goal than looking at a single angle. Integrating multiple perspectives – we can better answer the question: what really is a clone?
1. A clone's genomic stability as biomarker of its DNA-damage resilience
Awards: Pathway to Independence Award (K99/R00), NCI; Dean’s fellowship, Stanford; Best poster award, Cancer Target Discovery.
A main problem in the treatment of advanced cancers, including gastric cancers and glioma, is the incertitude at which we predict how individual patients will respond to DNA-damaging agents, especially on the long run. Knowing the mechanism behind a patient’s response, or the lack thereof, will help us depart from the oversimplified “more-is-better” and “one-size-fits-all” principles according to which DNA-damaging agents are administered. This will improve clinical outcome by allowing us to pinpoint those who would respond better and longer to lower doses of DNA-damaging agents, than to higher doses. Under the assumption that the success of DNA-damaging therapy increases with the proliferation rate of a relatively homogeneous tumor population, there was little
reason to assume anything other than monotonic dose-response relations. But with the recent paradigm shift that most cancers are in fact DNA mosaic products of ongoing evolution, comes the urgency to reconsider these fundamental principles behind DNA-damaging therapy administration. As the developers of one of the first DNA deconvolution methods and with access to technologies to profile the genomes and transcriptomes of up to 10,000 cells simultaneously, we are equipped to embark on first personalized dose-finding strategies for DNA-damaging therapies. We will test the potential of the very long-term legacy that DNA-damage entails on a cell – genomic instability – as new biomarker of DNA-damage response. Our preliminary studies showed that, for most cancer types, DNA-damaging agents change a clone’s genomic instability and that clones succumb to a limit in the amount of genomic instability they can tolerate. In particular, our results showed that patients with intermediate genomic instability have a very poor outcome and that this relation is only evident among treatment-naïve patients,
but not among patients treated with DNA-damaging agents. Further they show that we can measure genomic instability per clone and that clones with extreme genomic instability typically don’t grow large. Our hypothesis that genomic instability, rather than proliferation rate, determines how sensitive a tumor is to DNA damaging agents on the long-term, is founded on two unexpected findings: (i) Patients with extremely high genomic instability per tumor clone have an exceptionally good outcome. We integrate single cell DNA- and RNA-Seq data to characterize clones and to measure how much genomic instability they can tolerate. (ii) Low genomic instability is associated with reduced benefit from DNA-damaging agents. We quantify DNA damage per clone, relating it to the clones’ ability to tolerate DNA damage and to changes in the genomic instability of therapy-surviving clones. By measuring the proximity of tumor clones to the maximum genomic instability a cell can tolerate, we are developing one of the first models that base clinical decisions on the characterization of clones rather than tumor bulk.
2. The evolution of ploidy in a resource restricted environment
Most gliomas are diagnosed as either lower-grade lesions (grade II) or Glioblastoma (grade IV). Progression of lower-grade gliomas (LGG) to Glioblastoma (GBM) is accompanied by a phenotypic switch to an invasive cell phenotype. Converging evidence from colorectal-, breast-, and lung- cancers, suggests a strong enrichment of high ploidy cells among metastatic lesions as compared to the primary (Brastianos et al., 2015; Angelova et al., 2015).
Even in normal development: trophoblast giant cells are responsible for invading the placenta during embryogenesis and strikingly these cells can have up to 1000 copies of the genome (Hannibal et al., 2014). All this points to the existence of a ubiquitous mechanism that links high DNA content to an invasive phenotype. We formulate a mechanistic Grow-or-go model that postulates higher energy demands of high-ploidy cells as a driver of invasive behavior.
High energy environment
Low energy environment
CNVs are a phenotypically effective form of genomic instability, leading to changes in the expression of a lot of genes simultaneously, even affecting the size of the cell. Our model proposes CNVs as an efficient route for cells to switch back and forth between migration and proliferation. This mechanism may contribute to the quick recurrence of GBMs after surgery (Hatzikirou et al., 2012) and explain differences in the prognostic power of integrin signaling and cell cycle progression between males and females (Yang et al., 2019).
3. Competition and cooperation among co-existing tumor subpopulations
The transcriptome is part of a cell's "vocabulary" that is used to communicate with its environment. As such, not only the fitness of a cell, but also its transcriptome, is context-dependent -- a function of environmental changes. Here we evaluate how environmental fluctuations – changes in nutrient availability and changes in the cellular population structure – affect the fitness of these subpopulations. If coexisting subpopulations interact with one another, then the selective forces acting on these subpopulations will no longer be constant, but will change with changes in the cellular fraction of each population.
Evolutionary game theory (EGT) is a mathematical framework designed to model this scenario of dynamic selection. We integrate single-cell sequencing into an EGT framework to restrict the search space of possible interactions between subpopulations and to characterize the nature (i.e., the strength and direction) of these interactions.