Systems Approaches to Understanding Subpopulation Heterogeneity in Therapeutic Resistance

PROJECT SUMMARY
In recent years, improvements in diagnosis and treatment have extended the lives of many patients with triple
negative breast cancer, but resistance to treatment remains a major clinical and scientific challenge. While
standard-of-care treatment and chemotherapy is effective in many TNBC patients, approximately 40% of
patients display resistance, leading to poor overall survival. TNBC are characterized by significant intratumor
heterogeneity, which further complicates treatment. Mechanisms of chemoresistance in TNBC patients
remain poorly understood, in part due to a lack of available methods and models to measure intratumor
heterogeneity and track changes in heterogeneous tumor compositions over time. Here we propose to use a
new technology to track individual cells and clones as they respond to different chemotherapeutic agents; this
more detailed information about the tumor cell population will be used to build mathematical models better
predict and optimize therapeutic response. We first measure individual cell gene expression changes in
response to treatment and then assemble these measurements into cell subpopulation trajectories, taking
advantage of a barcoding technology developed in our lab to quantify clonally-resolved single cell
transcriptomes. These Aim 1 studies will build a compendium of gene expression, cell growth and survival
data that describes how each of the heterogeneous cells in major experimental models of subtypes of triple
negative breast cancer responds to clinically-relevant therapeutic agents. The new ability to layer clonal
identifier information on single cell gene expression data reveals the detailed trajectories of individual cells
that escape therapy. It also distinguishes subpopulations with pre-existing treatment resistance from those
in which a resistant state is induced. At a higher conceptual level, this proposal seeks to also address a broad
practical challenge: the high-dimensional ‘omics’ data collected in many large-scale efforts points often points
to correlations in disease progression but not been informative for building mechanistic models to aid in the
predictive of tumor response. Often, other types of data are more readily available-- lower dimensional data
with more frequent measurements. We therefore next ask: How can these distinct data types be integrated
into a useful framework to build predictive models of tumor cell response to therapy? This seems a fitting goal
for the systems biology of cancer community. We propose to tackle this challenge with our barcode tracking
technology; relative fractions of sensitive and resistance phenotypes, along with separate longitudinal
measurements of cell number (low dimension data), become the inputs for a mechanistic model to predict
therapeutic response and resistance (Aim 2). In Aim 3, we will perform trajectory-mapping and model testing
using patient-derived triple negative breast cancer cells, towards understanding the potential for translational
utility. By integrating different data types into a cohesive framework, we aim to describe how sensitive and
resistant subpopulations in TNBC grow, die, and transition in response to treatment.

Grant Number: 5U01CA253540-05
NIH Institute/Center: NIH
Principal Investigator: Amy Brock