Novel molecular techniques for diagnostics and cancer biology

Abstract

Molecular biology is reliant on a large set of increasingly complex methods. The development of high-throughput DNA sequencing almost 20 years ago kicked off a revolution in method development due to its incredible versatility. Besides determining the genomic DNA sequence itself, sequencing has been used to profile gene expression, investigate binding of proteins to DNA and RNA, trace cell lineages, screen for genes involved in biological processes, assay 3D organization of chromatin, and much more. Most of these methods have been immensely useful in cancer biology, helping us to understand the mechanisms of this complex disease and find new ways to battle it. But sequencing is not necessary if the mere presence or absence of a nucleic acid is impotant. In order to be able to rapidly diagnose viral diseases, crucial during pandemics such as the recent COVID-19, simpler methods are more useful. Various nucleic acid detection methods have been developed for molecular diagnostics, which can provide an answer within minutes. In this thesis, the fields of high-throughput sequencing, cancer biology, and molecular viral diagnostics are reviewed, since the work presented here consists of three projects dealing with these different topics.

In Paper I, we present a novel method for detecting low frequency variants in DNA. Such variants are important in applications such as genetic heterogeneity or minimal residual disease in cancer. However, their detection is hampered by the errors in sequencing data. To circumvent this, one approach is to attach double-stranded unique molecular identifier sequences (dsUMIs) to the ends of each DNA fragment before sequencing. This allows to compare reads originating from the same original molecule and form consensus sequences, removing most errors in the process. However, protocols that achieve this are challenging to perform. We developed a novel, simplified library preparation approach called one pot double-stranded UMI sequencing (OPUSeq) that adds dsUMIs to DNA in the same reaction as the PCR. We demonstrate that OPUSeq efficiently removes errors in sequencing data and can be used to detect variants down to 0.01% variant allele frequency. Using OPUSeq, we also found a novel type of artifact that arises when fragmentase enzyme mix is used in library preparation.

In Paper II, we investigated the existence of genetic factors that regulate cell state plasticity in cancer. Cancer cells are known to be capable of phenotypic cell state transitions that help them evade treatment. In certain cancer cell line models, such as the chronic myeloid leukemia (CML) K562, the cells are observed to adopt and switch between different states even in the absence of any specific stimuli. As our model system, we used the heterogeneous expression of CD24 protein in K562 as a marker for differential cell states. We designed two orthogonal genome-wide CRISPR-Cas9 knockout screening approaches to look for genes which regulate the spontaneous transitions between CD24-positive and CD24-negative states. We performed both screens and combined the data to produce a list of 49 plasticity regulator candidate genes. We further showed that seven of these genes are differentially expressed between CML patients exhibiting early molecular response to imatinib and those who do not, indicating a connection between plasticity and drug resistance. Finally, we validate one of the plasticity impeding candidates, ALDOB, by generating a single knockout model and demonstrating the increased ability of these cells to undergo state transitions.

In Paper III, we present a protocol for detection of SARS-CoV-2 RNA in unextracted patient samples using reverse transcription loop mediated isothermal amplification (RT-LAMP) with non-commercial enzymes. This protocol provides an alternative diagnostic method for situations where RT-LAMP and RNA extraction reagents are scarce. First, we showed how reverse transcriptases (RT) and strand-displacing polymerases necessary for RT-LAMP can be expressed and purified in-house. We tested different enzymes and LAMP primer sets and optimized the reaction conditions. Benchmarking showed that our in-house mix performs similarly to or even better than commercial alternatives. Finally, we tested our protocol on heat-inactivated, unextracted nasopharyngeal samples from patients and found that it exhibited good specificity as well as good sensitivity in samples with moderate to high viral load.