CRISPR screening in the third dimension

Listen to this article on the DDW Podcast: Rosie Bryan & Carlos le Sage, Horizon Discovery, explain how CRISPR-Cas9 technology has become a gold standard for functional genomics. While efforts in the cancer genomics field have uncovered the mutational landscapes in tumours covering many types of cancer, the functional contribution of the mutated genes to cancer onset and progression remains under explored. Functional genomics investigates the molecular function of genes and genetic interactions and has been enabled by developments in gene editing and DNA sequencing technologies. Advances made in the discovery and application of RNA interference, and more recently, CRISPR-Cas9 technology, have enabled the systematic analysis of genome-wide perturbations towards any phenotype of interest in a single pooled screening experiment. In cancer biology, CRISPR-Cas9 based functional genomic screening approaches are often applied to the drug discovery process, and while these studies produce high-quality and biologically relevant datasets based on screening in immortal cancer cell lines, in vitro 2D model systems generally do not fully recapitulate the complex nature of cellular interactions occurring in human disease. Various 3D models, including organoids and in vivo models, offer new strategies for CRISPR-Cas9 based screening to identify novel target genes relevant for the advancement of cancer therapy development. Functional genomic screening is generally based on three components, the ability to specifically mutate or modulate target genes; an assay that allows the selection of cellular phenotypes of interest, and the ability to link the phenotypes to the perturbed genes of interest in order to functionally annotate those genes. The CRISPR-Cas9 system, originally discovered in bacteria and archaea, is based on the endonuclease Cas9 which is guided by a short guide RNA (sgRNA) sequence to a specific locus in the genome where it creates a local DNA double-strand break. The resulting DNA damage is often resolved by non-homologous end-joining, an error-prone repair mechanism, that results in a functional knockout of the targeted gene1,2. Since its discovery, the CRISPR knockout system has been quickly harnessed and adapted to enable precise gene editing at defined genomic loci in mammalian cells allowing systematic functional analysis of genes in healthy and disease tissue types. Furthermore, catalytically inactive versions of Cas9 have been engineered to bring transcriptionally interfering (CRISPRi) or activating (CRISPRa) domains to target loci, thereby providing opportunities to study the impact of knocking down or upregulating gene output. The majority of functional genomic screening methodologies use two-dimensional (2D) models, which would normally consist of culturing an immortalised cell line in vitro either attached to a culture surface (adherent) or in a suspension. This form of in vitro cell culture has been utilised since the 1900s and comprises the majority of cancer biological research methodology3. Cancer cell lines are often easy to culture and can grow indefinitely. These cell lines usually tolerate the introduction of reagents through transfection or transduction well, including gene editing tools, and are generally susceptible to gene editing. Crucial for pooled screening, cancer cell lines allow experimentation at scale. Currently, efforts to extend functional genomics approaches to enable discovery in primary cell types, such as immune cells isolated from blood, are well underway (reviewed in 4). In pooled screening the reagents are usually delivered as a pool or library of lentiviral constructs to a population of cells at a low multiplicity of infection to ensure a single unique construct is introduced per cell; enabling direct linkage of the selected phenotype and the reagent. The introduction of the CRISPR reagents into the cells can be done in a single transduction round, ie., the lentiviral particles contain both Cas9 and CRISPR library components, or in two sequential rounds where the CRISPR library is delivered to a cell population previously transduced to express Cas9. The resulting pooled CRISPR edited population is then subjected to an assay with a defined phenotypical readout, and at the end of the assay the quantification of the sgRNAs present in the screen population will aid the interpretation of the impact specific gene perturbations on the measured phenotype. Commonly used pooled screening approaches make use of cell proliferation, cell death, or the expression of a biological marker or reporter as defined phenotypic readouts. Such approaches allow for the swift identification of gene perturbations that affect the phenotype in a population of cells by simply measuring changes in abundance of the sgRNAs between assay conditions by next-generation sequencing. 3D screening is gaining traction Three-dimensional (3D) screening is the broad term used for the expansion of these screening techniques into more recently developed 3D cell models, where cells are encouraged via alternative techniques to grow into 3D aggregates or spheroid structures in vitro, or by employing the use of animal models to conduct a screen within an in vivo setting. Currently, the main types of 3D in vitro culture models that are being explored in scientific research are multicellular spheroids, organoids, hydrogels, microfluidic devices, microfibre scaffolds and tissue-engineered scaffolds5, all of which have the potential to be exploited for screening capabilities. Alternatively, in vivo 3D methods can be used, which can be classified into two main groups: autochthonous direct methods where tumours are initiated within the animal model within an existing cell de novo, or more indirect methods which involve transplantation procedures into animal models, such as Cell line-Derived or Patient-Derived Xenografts (CDX, PDX)6. While 2D cell systems can be used for many applications in drug development and discovery, they are somewhat limited by their reduced ability to recapitulate the impact of in vivo microenvironment and heterogeneity7. However, key influences on disease progression that occur within a patient, such as hypoxia, or changes in cell-cell interactions and cellular metabolism8 are not directly considered. Thus, combining functional genomic screening with physiologically relevant 3D models would enable the identification of unique cancer dependencies9, while potentially enhancing the translational accuracy of drug development7. Organoids in 3D screening Organoids are defined as stem cells that have been cultured into self-organising, multicellular, 3D structures, via embedding them into an extracellular matrix (ECM)12. Organoids can be grown from healthy human adult stem cells derived from many types of tissue, or from cancer stem cells, which provide patient-derived tumour organoids (PDTOs). Organoids can also be generated from induced pluripotent stem cells, although this can be more challenging. It is possible to grow organoids over long-term in cultures, and they have been shown to be capable of recapitulating many features of the originating tissue10, 12. 3D, complex cultures such as organoids are a useful way to model in vivo cellular responses, while having the ease of use in vitro. The main benefit is the improvement from 2D cultures to researching cell morphology, growth characteristics and proliferation, protein synthesis and cellular response to introducing various stimuli and drug treatments7. Although organoid models lack components such as stroma, blood vessels, and immune cells (reviewed in 10), they more accurately represent the phenotypic heterogeneity observed in vivo versus an immortalised cancer cell line grown as an adherent layer. Organoids can be highly desirable cancer research models, particularly in disease progression and drug development or response, as they are thought to better mimic the physiological environment in patients7,8. PDTOs could additionally provide key insights into personalised treatment strategies. It is also possible to obtain both healthy and tumour organoid models for a specific tissue of origin, which has implications for screening drug specificity for tumour cells as compared to the normal tissue (reviewed in 10). Organoids have been established and biobanked from a wide range of primary and cancer tissue types and accurately represent the tissue of origin10, 11. By using organoid models, new important sensitivity or resistance factors in drug response can be identified that could translate more robustly to the behaviour in the clinic. A key consideration when expanding research to organoid cultures, especially in the context of functional pooled genomic screening approaches at scale, is that they require considerable investment in time, finances, and resources, when compared to 2D culture methods10. For instance, organoids rely on ECMs to support formation of 3D structures. Matrigel (derived from a murine source) is commonly used for this purpose, which although can be used very effectively for these cultures, needs to be kept cold while handling to prevent premature polymerisation, and can be costly when required in large quantities. Additionally, organoids often require culturing in media with the presence of many additives, such as mitogens, signalling activators, vitamins, and other cofactors. Collectively, these additional requirements are important when conducting a pooled CRISPR screen10, 12. Applying functional genomic screening approaches to organoid cultures is a highly exciting and novel concept (see Figure 1). A critical factor for any pooled CRISPR screen is the screening coverage (i.e. the number of cells per sgRNA) required for proper biological interpretations. Higher levels of coverage could help to minimise the risk of stochastic loss of an sgRNA library in complex organoid systems13, however, high coverage can restrict the use of larger screening libraries in such models. CRISPR-Cas9 organoid screens have recently been expanded into whole-genome scale. In one study a pooled CRISPR screen comparing both wild-type and APC mutant human small intestinal organoids was conducted14. The authors employed this screen, at around 50-fold coverage per sgRNA, to investigate genes driving resistance to TGF-β-mediated growth restriction, which has important implications for colorectal cancer progression. Additionally, a whole genome CRISPR screen in murine gastric epithelial organoids was conducted to gain further insight into the mechanism whereby Wnt signalling regulates stem cell-dependent epithelial renewal in the gastric mucosa15. For this screen, the authors made use of a compact three-guide-per-gene library. These two studies demonstrate that improvements in CRISPR technology have allowed whole-genome pooled CRISPR screens to be run in organoid models and meaningful results can be identified using lower coverage or smaller libraries. In vivo CRISPR-Cas9 screening Similar to conducting CRISPR screening with in vitro cell cultures, in vivo screening starts with careful consideration and selection of the appropriate model, in this case an animal model, and experimental approach to answer a research question (reviewed in 16,17). In CDX models, human or mouse cancer cell lines grown in cell culture are implanted into immunodeficient mouse or rat models, allowing the efficacy of novel therapeutic compounds to block or irradicate the formation of tumours to be tested. Depending on the study requirements, the cancer cell lines can be engrafted subcutaneously (into the animal flank), orthotopically (into tumour-specific tissue), or injected into the tail vein. Implementing CRISPR-Cas9 technology in CDX model studies allows for the identification of genes and pathways that could alter the drug response following gene modulation, or, in a drug-unrelated context, could aid in the discovery of synthetic lethal targets to drive subsequent drug development, or study metastasis drivers18 (see Figure 2). These types of screens are indirect in vivo CRISPR screens, where cancer cell lines are engineered with CRISPR reagents before injecting into the animal. The immediate advantages of this approach are the straightforward and controlled delivery of the CRISPR system to the human cell line and scalability of the experiment. Importantly, and from a technical perspective, cell line take-rates, and the size and coverage of the CRISPR library will have to be factored into the experimental design when considering the use of CDX models for biological discovery and can be defined through barcode experimentation19. PDX models are based on the implantation of tumour material from a patient directly into an immunosuppressed mouse and propagated to allow studying of cancer progression. PDX models can support large-scale drug and CRISPR screens (ex vivo delivery of the gene editing components and implantation of edited cells) but, compared to CDX, the identification of druggable targets in PDX is more likely to provide opportunities for patient-relevant therapy. Similar to the CDX approach, PDX models are often conducted in immunosuppressed animals, and as such any contribution the immune system could have in driving and shaping tumorigenesis cannot be recorded. The development and use of direct in vivo screening models can overcome some of the known limitations of screening in CDX and PDX models. By targeting cells of interest in their natural habitat, which is characterised by the presence of an intact extracellular environment and impacted by signals both from the tissue surrounding target cells as well as from a functional immune system, novel insights on the contribution of the physiological environment to health and disease can be gained. In cancer research, direct in vivo screening involves the de novo tumour generation in, for instance, transgenic cancer mouse models, where the formation of tumours can be induced and their response to new therapeutics can be studied in detail. To date, this approach has been successfully applied to study cancer onset and progression in tissues such as liver, lung, and brain, that are more accessible and responsive to CRISPR editing reagents compared to other tissues16. While in vitro and in vivo pooled CRISPR screens share similar technical and statistical requirements, particularly in relation to the size and diversity of the CRISPR library and associated coverage of each of the sgRNA elements to produce high quality datasets20, there are a number of unique challenges to consider when screening in animal models. Depending on the type of screen and study model, delivery, and expression of the CRISPR reagents, an appropriate balance between achievable cellular coverage and library size must be considered. Conclusions Functional genomics has been pivotal to understanding the functionality of genes both in health and disease. CRISPR-Cas9 technology has become a gold standard for functional genomics, not only because of its efficiency and precision to perturb any target gene of interest, scalability to include whole-genome approaches, and relatively low cost to carry out such screens in a vast number of cell systems, but also based on the establishment of recent expansions to the CRISPR toolkit that have enabled controlled modulation of genes, as well as introduce multiplexing of readouts beyond sgRNA abundance. With the development of protocols to establish spheroids, organoids, and in vivo models, 3D systems that in different ways more closely recapitulate tissue complexity and responses occurring within the natural tissue and organ niches in living organisms, new opportunities have arisen to allow a better understanding of human physiology and aetiology. Volume 23, Issue 1 – Winter 2021/22 Figure 1: An example of a workflow for setting up CRISPR-Cas9 screens in an organoid model. Figure 2: Pooled screening workflow where CRISPR engineered cell lines are implanted into CDX animal models for drug-gene interaction (top) or gene essentiality (bottom) screening. About the authors Dr. Rosie Bryan is a Senior Scientist in the Functional Genomic Screening Group at Horizon Discovery. Her work focuses on the assay development and subsequent performance of custom CRISPR pooled screens and was involved in the first pooled CRISPR screening project in a cancer organoid cell line model at Horizon Discovery. Bryan completed a Ph.D. in Cell and Molecular Biology at the University of Essex, UK, where her research focus was based on androgen receptor signalling in endocrine resistant breast cancer. Dr. Carlos le Sage is Manager of the Functional Genomic Screening Platform at Horizon Discovery. His research focusses on discovering the mechanism of action of newly developed compounds and defining new synthetic lethal interactions in defined mutant-driven cancer genomes, through unbiased genetic perturbation screens. His experience in the development and use of functional genomic techniques, initially with RNAi and more recently with CRISPR gene modifying and modulating technologies for genome-wide loss- and gain-of-function screening. Dr le Sage completed postdoctoral research in the DNA damage response field at Cambridge University and holds a Ph.D. in Biology from the Netherlands Cancer Institute through Leiden University. References Jinek, M., et al. A programmable dual RNA-guided DNA endonuclease in adaptive bacterial immunity. Science (2012) doi: 11126/science.1225829 Mali, P., et al. RNA-guided human genome engineering via Cas9. Science (2013) doi: 11126/science.1232033Kapałczyńska, M., et al. 2D and 3D cell cultures – a comparison of different types of cancer cell cultures. Arch. Med. Sci. (2018) doi: 10.5114/aoms.2016.63743 Kapałczyńska, M., et al. 2D and 3D cell cultures – a comparison of different types of cancer cell cultures. Arch. Med. Sci. (2018) doi: 15114/aoms.2016.63743 Brucklacher-Waldert, V. CRISPR-Cas9knockout screening using primary human cells in drug discovery. Int. Bioph. Ind.17 (2021) Jensen, C., and Teng, Y. Is it time to start transitioning from 2D to 3D cell culture? Front. Mol. Biosci. (2020) doi: 13389/fmolb.2020.00033 Gargiulo, G. Next-generation in vivo modelling of human cancers. Front. Oncol. (2018) doi: 13389/fonc.2018.00429 Jensen, C., et al. The new frontier of three-dimensional culture models to scale-up cancer research. Methods Mol. Biol. (2022) doi: 11007/978-1-0716-1558-4_1 Han, K., et al. CRISPR screens in cancer spheroids identify 3D growth-specific vulnerabilities. Nature (2020) doi: 11038/s41586-020-2099-x Miller, T.E., et al. Transcription elongation factors represent in vivo cancer dependencies in glioblastoma. Nature (2017) doi: 11038/nature23000 Drost, J., and Clevers, H. Organoids in cancer research. Nat. Rev. Cancer (2018) doi: 11038/s41568-018-0007-6 Botti, G., et al. Organoid biobanks as a new tool for pre-clinical validation of candidate drug efficacy and safety. Int. J. Phys. Path. Pharm.13, 17-21(2021) Barbáchano, A., et al. Organoids and colorectal cancer. Cancers (Basel) (2021) doi: 13390/cancers13112657 Planas-Paz, L., et al. YAP, but not RSPO-LGR4/5, signaling in biliary epithelial cells promotes a ductular reaction in response to liver injury. Cell Stem Cell (2019) doi: 11016/j.stem.2019.04.005 Murakami, K., et al. A genome-scale CRISPR screen reveals factors regulating Wnt-dependent renewal of mouse gastric epithelial cells. PNAS (2021) doi: 11073/pnas.2016806118 Ringel, T., et al. Genome-scale CRISPR screening in human intestinal organoids identifies drivers of TGF-beta resistance. Cell Stem Cell (2020) doi: 11016/j.stem.2020.02.007 Kuhn, M., et al. Moving from in vitro to in vivo CRISPR screens. Gene and Genome Editing (2021) doi: 11016/j.ggedit.2021.100008 Li, C., and Kasinski, A.L. In vivo cancer-based functional genomics. Trends in cancer (2020) doi: 11016/j.trecan.2020.07.004 Chen, S., et al. Genome-wide CRISPR screen in a mouse model of tumourgrowth and metastasis. Cell (2015) doi: 11016/j.cell.2015.02.038 Nolan-Stevaux, O., et al. Measurement of cancer cell growth heterogeneity through lentiviral barcoding identifies clonal dominance as a characteristic of in vivo tumor engraftment. PLOS One (2013) doi: 11371/journal.pone.0067316 Joung, J., et al. Genome-scale CRISPR-Cas9knockout and transcriptional activation screening. Nat. Prot. (2017) doi: 11038/nprot.2017.016