Avtar Singh is a Senior Scientist in the Cell and Tissue Genomics Department at Genentech, where he develops new approaches for image-based screening and high-throughput biological discovery. DDW’s Megan Thomas spoke with Singh about SLAS2023, where he presented Image-based screening of pooled genetic libraries.
MT: Why is SLAS2023 a prime space for you to share your work?
AS: SLAS2023 is an excellent forum for me to share my research, because it brings together experts in technology development, high-content screening, automation, and functional genomics. After my presentation at SLAS2020, I connected with leaders from >10 biotech companies and had extensive follow-up conversations about tech translation, new directions, and potential collaborations. It also introduced me to new career opportunities and played a key role in my decision to join Genentech. As someone who specialised in physics for my undergraduate and graduate studies, I never planned on a career in the pharmaceutical industry. SLAS2020 showed me how technology development fits into the landscape of drug discovery and stimulated me to further explore those possibilities.
MT: What impact have pooled CRISPR screens had on our ability to perform analyses at genome scale?
AS: In arrayed screens, perturbations are introduced into individual wells of a multi-well plate and a cell-based assay is used to assess their functional impact. This is a challenging way to conduct large genetic screens because it requires careful handling of thousands of samples, often using specialised automation, and introduces batch effects across samples.
Meanwhile, pooled screens have revolutionised our ability to perform genetic analyses at scale. In traditional pooled screens, perturbations are introduced into a single sample using low multiplicity of infection (low MOI) lentiviral transduction, yielding a pooled cell library where individual cells possess different genetic disruptions. Cells are then enriched for phenotypes of interest based on cellular fitness or marker expression. Next-generation sequencing is used to measure the abundance of perturbations before and after enrichment to identify genetic changes that impact cell function.
Optical pooled screens build upon the benefits of traditional pooled screening by expanding the set of phenotypes that can be assayed. A high-content imaging assay can be tailored to the biological questions under investigation, enabling in-depth characterisation of complex cellular phenotypes such as protein/RNA localisation, metabolic state, cell morphology, cell-cell interactions, and live-cell dynamics. In order to match single-cell phenotypes to the underlying genetic changes in a pooled cell library, targeted in situ sequencing is used to read out the identity of genetic perturbations in each cell. This approach has made large image-based genetic screens a much more tractable undertaking. In our first application of this technology, we conducted a ~1000-gene screen for regulators of p65 translocation in order to study NF-kB signalling. In another project led by Luke Funk, we used multichannel image-based profiling to study the functional impact of ~5000 essential gene knockouts.
Looking ahead, we expect that genome-scale imaging screens will become a core component of the functional genomics toolkit. Compared with arrayed approaches and enrichment-based methods that isolate populations of interest for downstream sequencing, optical pooled screens are able to map the phenotype and genotype of each cell in a pooled library, greatly reducing the cost and effort required to conduct comprehensive genetic investigations.
MT: What are the opportunities of this type of image-based screening in years to come?
AS: Optical pooled screens (OPS) have already had a significant impact on our ability to identify the genetic regulators of complex cellular processes, such as immune signalling and mitosis, and will be a key tool for target discovery and validation in the pharmaceutical industry. In the past, many projects have relied on preliminary screens using indirect or proxy phenotypes that could be easily measured at scale (e.g. based on cell viability or marker levels) to identify hits for further investigation. OPS makes it possible to screen directly with precise phenotypes-of-interest that can be tailored to each application. Furthermore, OPS combined with image-based profiling makes it possible to cluster perturbations at scale for unbiased discovery based on cell morphology. These approaches will continue to evolve and improve as OPS is: 1) integrated with methods for highly multiplexed phenotyping based on multi-round immunofluorescence/FISH, 2) extended to additional disease-relevant model systems, and 3) combined with more advanced computational strategies for deep learning of phenotypes and single-cell analysis.
MT: How has this technology and methodology evolved since you first started your research?
AS: Prior to our 2019 paper, most large imaging screens were conducted in an arrayed fashion. Our first project established the optical pooled screening platform by showing that targeted in situ sequencing could be used to identify the genetic perturbations from millions of cells in a pooled cell library. In a second project led by Luke Funk, we showed that multichannel phenotyping and high-dimensional image-based profiling can be used to cluster perturbations and infer functional roles using a “guilt-by-association” approach. We’ve also made improvements to the imaging hardware to increase the throughput of each screening experiment and integrated OPS with several phenotypic approaches, including live-cell reporters, immunofluorescence, and fluorescence in situ hybridisation (FISH).
MT: What challenges need to be overcome in order for image-based screening of pooled genetic libraries to reach its full potential?
AS: Although OPS is already a transformative technology, we’re really excited to continue building upon the platform so that it can realise its full potential. Currently, most OPS projects have relied on cancer cell lines because perturbations can be detected efficiently and reliably at scale. Extending the technology to more challenging cell types, organoids and tissue will have an enormous impact on our ability to conduct genome-scale imaging screens in the most relevant model systems across therapeutic areas. Integration of OPS with multi-round immunofluorescence and FISH will substantially increase the information content of each experiment and enable precision analyses of relevant cell states. More advanced computational strategies will play an essential role in maximising the utility of these rich datasets. There are several opportunities in this space, including 1) deep learning of phenotypes without predefined features, 2) single-cell analyses of subpopulations and trajectories, 3) spatial analyses between neighboring cells, and 4) methods to predict the effects of combinatorial perturbations.
Finally, the number of labs employing OPS is still fairly low. In order for the technology to achieve its full impact, it will be imperative to lower barriers to entry by simplifying and standardising protocols, improving robustness, automating in situ sequencing, and reducing cost to enable larger experiments.
Avtar Singh is a Senior Scientist in the Cell and Tissue Genomics Department at Genentech, where he develops new approaches for image-based screening and high-throughput biological discovery. During his postdoc with Paul Blainey at the Broad Institute, he established optical pooled screens as a method for high-content screening of genome-scale CRISPR libraries. Prior to that, he specialised in super-resolution microscopy and single-molecule imaging during his PhD in Warren Zipfel’s lab at Cornell University.