Molecular characterization of disseminated cancer cells isolated from patients with luminal B type breast cancer

Breast cancer (BC) accounts for almost a quarter of reported cancer incidences in women worldwide. It comprises several molecular subtypes, out of which the luminal A (LumA) and luminal B (LumB) types are the most common. Despite being quite similar from a histopathological point of view, the LumB type has a far worse prognosis due to a higher propensity to metastasize and develop therapy resistance. Despite decades of effort, metastasis is still responsible for ~90 % of cancer-related deaths. Metastatic relapse is caused by disseminated cancer cells (DCC) that can lie dormant in distant organs for several years before growing into macrometastases. The epithelial cell adhesion molecule (EpCAM) can be used as a marker to detect DCCs in patient bone marrow (BM). However, there are non-cancer cells (NCC) belonging to the erythroid progenitor lineage in the BM, which also express this marker, thereby representing a confounding factor in our EpCAM+ single cell collective. Regarding treatment resistance, recent studies have implicated several miRNAs in resistance of BC to the most common systemic treatments. The aim of this dissertation was (1) to identify a way to distinguish true DCCs from NCCs, (2) to identify genomic or transcriptomic differences between LumA and LumB DCCs accounting for LumB’s increased malignancy, and (3) to develop a novel extended whole transcriptome amplification (eWTA) to isolate miRNAs along with mRNA and gDNA from single cells to further elucidate the underlying causes for LumB BC’s higher aggressiveness. The data revealed that separation of NCCs from true DCCs was possible using an M0 DCC qPCR signature consisting of four genes, but the distinction was most reliably done using copy number alteration (CNA) profiling. However, the data suggested that the qPCR signature might actually be more precise that the CNA profiling. A detailed analysis of the CNA profiles of true DCCs revealed no differences between LumA and LumB DCCs. Additionally, targeted proliferation marker analyses by qPCR did not reveal differences between LumA and LumB DCCs, neither regarding expression levels nor regarding the percentage of proliferating cells. In contrast to the comparison of LumA and LumB DCCs, there was a pronounced divergence of CNAs in DCCs derived from non-metastatic (M0) and metastatic (M1) BC patients with the latter carrying more genomic aberrations compared to the former. In line with the CNA profiles, targeted proliferation marker analyses showed a difference between M0 and M1 DCCs with the M1 DCCs displaying a lower percentage of proliferating cells. Interestingly, qPCR revealed that half of M0 cells were proliferating. However, there was no correlation between the KI67 status of the primary tumor (PT) and the expression of MKI67 in matched DCCs. Subsequent global transcriptomic profiling by RNA-Sequencing confirmed that all of the DCCs, which were classified as proliferating by qPCR, were expressing cell cycle-associated genes significantly more than the non-proliferating DCCs. Unsupervised hierarchical clustering identified two subgroups among the proliferating DCCs with differing expression of analyzed genes. These groups comprised either a mix of DCCs from all subtypes, or almost exclusively LumB DCCs, suggesting a higher proliferation of LumB DCCs. The RNA-Seq analysis also uncovered a correlation of the overall expression signature of DCCs with the KI67 status of the PT, which indirectly translated to differences between LumA and LumB subtypes in their overall gene expression. Gene ontology (GO) analysis identified several biological processes that were enriched among the up- and down-regulated genes in LumB compared to LumA. Genes related to membranes and transmembrane transport were associated with down-regulated genes, while splicing, ribosomes, and translation were overrepresented among the up-regulated genes. Cell adhesion and extracellular matrix (ECM) pathways were present in both gene lists, indicating a complex differential regulation of these processes in LumB compared to LumA. In parallel to the previous experiments, a preliminary protocol for the novel eWTA was established. This protocol included a polyadenylation step to enable capture of single stranded RNAs. The polyadenylation required introduction of an additional dilution of the cell lysate, in order to prevent denaturation of the Poly(A) polymerase by the lysis buffer. Experiments on artificial long RNA as well as expression changes of ribosomal RNAs demonstrated that the employed polyadenylation strategy worked in principle. However, a detection of short RNAs in the range of miRNA could not be done, as this would have required development of a separate method compatible with our WTA adapters. Nevertheless, the first step towards an eWTA for isolation of the miRNAome alongside the genome and transcriptome of a single cell has been taken. In conclusion, the data suggest that the main factors driving the increased malignancy of LumB cancer is likely a higher proliferation, because it enables faster accumulation of somatic mutations. Two hypothetical scenarios explaining the underlying mechanisms are the following: (1) LumB DCCs may interact with the microenvironment (ME) at metastatic target sites in a different manner than LumA DCCs, which leads to changes in mRNA splicing, which in turn increases proliferation. Or, (2) LumB DCCs may initially display differential mRNA splicing before arriving at the target site, causing altered interaction with the ME at the target site and in response proliferation is up-regulated in the DCCs. More research will be required to provide functional proof of the higher proliferation of LumB DCCs and to determine the exact mechanisms underlying this alleged higher proliferation profile of LumB DCCs. It is also possible that miRNAs are somehow involved in this. Therefore, it will be important to further develop the eWTA protocol, in order to aid in advancing our knowledge of the intricate crosstalk of miRNAs with their target mRNAs and concomitant genomic changes