Tumor heterogeneity is a fundamental characteristic of cancer and plays a pivotal role in disease progression, metastasis and the emergence of therapy resistance. As cancer develops, tumor cells gradually accumulate genetic and epigenetic changes, resulting in various clonal lineages and subgroups. These linages undergo selective pressure based on growth and fitness advantages, influencing their disease progression or treatment resistance.
Tumor heterogeneity can be investigated with cellular barcoding. This technique uses a short (semi-)random DNA sequence, or barcode, which is stably introduced into the genome of target cells and allows for the identification and tracking of individual cancer cells and subsequent progeny. Due to
its broad range of applications, DNA barcoding has become a powerful tool for
identifying novel therapeutic compounds and drug targets. Additionally, DNA barcoding
can provide valuable insights into tumor heterogeneity, tumor evolution, and clonal
dynamics, with potential implications for personalized cancer therapies in the future.
With this in mind, two barcoding approaches were developed in this thesis to
investigate tumor heterogeneity in solid tumors and serve as tools to study clonal
dynamics and ultimately uncover new treatment options.
In the first approach, the high-complexity CORE barcode library was developed to label
and track tens of thousands of uniquely marked pancreatic cancer cells. The CORE
barcode library was further used to monitor the clonal dynamics of PDAC cells under
FOLFIRINOX treatment. It was observed that the ASPC1 cells have a high intrinsic
resistance to FOLFIRINOX. In contrast, cell expansion after treatment followed a
stochastic pattern. Using the CORE system to track clonal dynamics during metastasis
revealed that ASPC1 cells are characterized by a consistently high inherent metastatic
ability regarding their metastatic seeding. This observation contrasts a patient-derived
cell line that displayed a significantly higher degree of heterogeneity and consisted of
subpopulations with varying metastatic abilities. Lastly, the expertise gained from
developing the CORE library was further applied to conceptualize the sCORE
approach based on mathematical modeling to allow the integration of clonal tracking
into single-cell transcriptome analysis to enable the longitudinal tracking of
transcriptome changes.

Following the comprehensive exploration of cellular barcoding's potential in
understanding tumor heterogeneity and clonal dynamics in PDAC, the second
approach delved into understanding drug resistance complexities in gastrointestinal
stromal tumors. For this, a preclinical GIST barcoding platform was developed that
provided a lens to study genotype-specific drug responses and identify promising drug
candidates. Although the introduction of selective tyrosine kinase inhibitors such as
imatinib has revolutionized GIST therapy, resistance eventually emerges in nearly all
patients. Overcoming resistance and developing new drugs are complicated by highly
diverse resistance mechanisms mediated by a broad range of genetic mutations. The
preclinical GIST barcoding platform was deployed to further study the secondary drug
resistance landscape regarding their overall fitness and genotype-specific drug
response. It allowed to assess the individual fitness of various sublines for the first time
in a competitive setting and revealed a fitness hierarchy; ultimately demonstrating that
drug resistance comes at a fitness cost. A sophisticated Bayesian model was
employed to allow multiplex screening of drugs and to monitor genotype-specific
susceptibility. This joint approach of the preclinical GIST screening platform, combined
with the Bayesian model, was subsequently validated by in vitro and in vivo drug
screens of the standard approved treatment schedules. The expected genotypespecific effects, as seen in preclinical and clinical studies, could be faithfully
recapitulated, thus confirming the effectiveness and accuracy of the platform.
Additionally, the platform was used to screen the novel compound M4205 in vitro and
revealed a strong inhibition in all sublines. These data support M4205 as a promising
drug candidate for further preclinical studies and demonstrate that the preclinical GIST
barcoding platform can also be used to identify novel compounds.
Overall, DNA barcoding technologies are emerging as a powerful tool for studying
tumor heterogeneity as it allows for the tracking of clonal dynamics and the evolution
of cellular states during metastasis and/or under treatment. The barcoding systems
established in this thesis further progress the evolving technology and provide
excellent methodological approaches to better understand the heterogeneity of solid
tumors and to help decipher vulnerabilities for improving patient care.