A cut above: Making drug discovery more efficient using CRISPR

Drug discovery is, notoriously, a difficult and lengthy process. Progressing a drug from bench to bedside can take up to 15 years and cost $1 billion.

Whilst research and treatments have improved cancer survival hugely in the past 50 years, the disease remains the second leading cause of death worldwide, with 166,000 deaths in the UK each year. It’s clear that one key contributor to cancer’s mortality is the low success rate of drug discovery and so there is clear motivation to make this process more efficient. One way this is being done is by establishing a genetic association between drug target and disease, which can significantly boost drug development success. Applying functional genomics, the study of how DNA changes translate to cell functions, to cancer models is crucial to uncover key genetic changes that could be cancer drug targets.

“Incorporating new technologies will enable us to perform CRISPR screens in increasingly complex models that more accurately represent human disease.”

The gene editing tool CRISPR-Cas9 has revolutionised functional genomics. It consists of two components – a Cas9 nuclease and a guide RNA (gRNA) molecule. The Cas9 nuclease induces a double strand break at a target sequence complementary to the directing gRNA molecule. Typically, the DNA damage is repaired by an error-prone DNA repair mechanism – non-homologous end joining. As a result, insertion or deletion (indel) mutations form. In protein-coding regions, indels create frameshift mutations and subsequently gene knockouts. When a gene is knocked out, the consequential effects on cell functions can be observed, helping decipher a gene’s function.

Functional genomics in cancer drug discovery

At the joint AstraZeneca-Cancer Research Horizons Functional Genomics Centre (FGC), we perform pooled CRISPR screens with a genome wide gRNA library. This library contains pools of gRNA molecules targeting thousands of genes across the genome. Amazingly, this enables the phenotype of each gene knockout to be observed in a single experiment and requires no prior knowledge of each gene’s function. This, then, can be seen as an unbiased approach to identifying new cancer vulnerabilities and potential drug targets.

To conduct a CRISPR screen, cancer cells need to express the Cas9 enzyme and gRNA library. At the FGC, we transduce our cancer cell models with lentiviral vectors carrying these CRISPR components. Both the Cas9 and gRNA vector also contain an antibiotic resistance gene, which enables us to select transduced cells by culturing in antibiotics.

Figure1: Pooled CRISPR screening for essentiality and drug-resistance.

CRISPR editing results in a mixture of cancer cell populations, each with a different gene across the genome knocked out. These cancer cells are then subject to different assay conditions. When searching for genes essential for survival – an essentiality screen – cells are expanded for several cell divisions. The same is done when looking for genes that change drug sensitivity – gene-drug interaction screens – but with the additional step of drug treatment.

By the end of a screen, cells with genes knocked out essential for survival or drug resistance will die and thus be lost from the cell population. This change can be detected with next generation sequencing (NGS). For example, in an essentiality screen, you would expect a decreased abundance of gRNAs targeting essential genes in the genomic DNA cancer cell population (Figure 1).

Advancing pooled CRISPR screening

Fluorescence-activated cell sorting (FACS) screens are more complex than essentiality or drug resistance screens, separating cells based on fluorescent properties. Fluorescently labelling a cell surface protein – PD-L1 – means that any cancer cells with gene knockouts that change PD-L1 levels will exhibit a change in fluorescence.

PD-L1 is a checkpoint protein commonly found on cancer cells, which inhibits the immune system when overactivated. This negative checkpoint signal is sent when PD-L1 binds to its cognate PD-1 receptor on T cells. Recently at the FGC, we performed a FACS screen in a pancreatic cancer cell model and identified cancer cell populations with gene knockouts that contribute to tumour-immune evasion.

This screen successfully identified all gene hits altering PDL-1 levels from a previous pancreatic cancer study. Different patient tumours have varying levels of immune attack; therefore, we also added interferon-gamma (IFN-γ), found in immune-rich tumours. This further enabled us to uncover novel genes that regulate PDL-1 levels in the presence and absence of a strong immune response, comparing genes between IFN-γ positive and negative environments (Figure 2).

Figure 2: Cancer genes identified from FACS screen altering tumour immune response via PDL-1. A. Volcano plots with known PDL-1 hit genes in absence (-) and presence (+) of IFN-γ. B. Venn diagram with known (blue) and novel (purple) genes identified in each condition.

Importantly, this work demonstrates a FACS screening capability for the FGC. Separating cells based on fluorescent properties means we can investigate which genes alter the expression of any protein we fluorescently label. Compared to identifying essential or drug interaction genes, this expands our ability to identify genes crucial to virtually any biological process.

Where next?

The future of the FGC rests in the technology development programme, which continuously evaluates new CRISPR technologies. Incorporating new technologies will enable us to perform CRISPR screens in increasingly complex models that more accurately represent human disease.

A good example of this is the culture conditions of cells. Currently, most cancer cells are grown in 2D – flat along a plastic flask. This, of course, poorly recapitulates the 3D tumour structure, which contains a mixture of cancer, fibroblast and immune cells, held by scaffolding proteins like collagen. In contrast, 3D models, such as cancer organoids, derive from stem cells and grow in protein gel structures, which more closely mimics a tumour. Thus, performing CRISPR screening in organoids will capture genes regulated in a 3D environment, where gene expression profiles are closer to patient tumours than 2D models. This will increase chances of identifying genes that are drug targets and the resulting drugs progressing to later drug development stages.

Authors:
Rebecca England is Associate Scientist at Cancer Research Horizons,
Khalid Saeed is a Senior Scientist at Cancer Research Horizons,
David Walter is a Group Leader at Cancer Research Horizons
and Sebastian Lukasiak is Associate Principal Scientist at AstraZeneca