Recent advances in technology that enables us to read, write, and edit genetic information have revolutionized our understanding of biology. For example, the introduction of next-generation sequencing (NGS) beyond Sanger sequencing in the mid-2000s has transformed the field of genomics because we can retrieve vast amounts of cellular genomic, epigenomic, and transcriptomic information at high speeds and low costs (1). In addition, decades of research efforts on synthetic genomes have enabled the complete synthesis of bacterial genomes (2). For genome editing, engineered nucleases, such as zinc finger nucleases (ZFN) and transcription activator-like effector nucleases (TALEN), have laid the foundation for executing site-specific genetic alterations (3). In contrast to ZFNs and TALENs, which require complicated protein engineering for customizing DNA-binding domains, the CRISPR/Cas system relies on short guide RNAs (gRNA) to target any desired location in the genome (4). As such, this system has popularized genome editing for various biological studies, from basic research to therapeutic applications.
Further, the Cas nucleases have been modified to include fused adapter proteins, such as deaminases, transcriptional activators/repressors, and reverse transcriptase (5). With these adaptations, CRISPR/Cas systems have gained popularity in genome-wide screening, wherein individual or epistatic factors involved in disease can quickly be discovered for future therapeutic strategies. In parallel, delivery methods have been optimized for more efficient gene editing, allowing for
In this review, we discuss recent advances and potential strategies for engineering immune cells using CRISPR/Cas9 with a therapeutic emphasis. We review its brief history and functional mechanism and explore how this system has been applied for engineering T and non-T immune cells. Last, we review how CRISPR/Cas9-based screening has been used to find potential targets for immunotherapy and to uncover previously unknown immune pathways.
In 1987, Ishino
The CRISPR/Cas system can be classified into Classes I and II. In the class I system, multiple Cas effector proteins mediate interference with invading genetic material, whereas in the class II system, a single Cas effector fulfills all effector functions (11). Of particular interest, the
Further, the Cas9 protein has two independent nuclease domains: RuvC and HNH. If either one is mutated, Cas9 exhibits only nickase activity (3). This mutant form can further be fused with deaminases for adenine/cytidine base editing and reverse transcriptase for prime editing (5). Similarly, mutation of both nuclease domains generates completely inactive Cas9 (dCas9) that can further be fused to other effector proteins, such as transcriptional activators or repressors, for programmable RNA-guided epigenetic regulation (14).
Next, many CRISPR/Cas9 delivery methods have been developed. Cas9 can be delivered into cells by electroporation or chemical transfection of plasmid, RNA, or protein, or by using viral vectors, such as lentiviruses (LV) or adeno-associated viruses (AAV) (15). The gRNA can be delivered in a similar fashion: plasmid, viral vector, or as a ribonucleoprotein (RNP) complex with Cas9. Various combinations can be used, but for immune-cell engineering, the most widely used method involves delivering RNP using electroporation, because of its instant activity and short half-life, which reduces off-target effects (16). Additionally, introduction of RNP can reduce cellular stress and toxicity better than can plasmid delivery (17) which can activate innate immune pathways within cells (18). For genome-wide screening, the most common format involves the lentiviral delivery of sgRNA libraries into cells that stably express Cas9.
Immunotherapy uses the immune system to fight multiple diseases, such as cancer, autoimmunity, and neurodegenerative diseases (19). For cancer, therapies that use T cells have been the most successful because of their target-specific and robust cytotoxic ability, rapid expansion, and long-term persistence
CRISPR/Cas9-based gene disruption has been used to enhance T cell activity, thereby increasing its efficacy. Among the targeted genes have been those involved in T-cell exhaustion, which is a dysfunctional state induced by repeated and chronic exposure to target antigens, as is often seen in many chronic viral infections and cancer (26). One major phenotypic change in exhausted T cells is the persistent expression of inhibitory checkpoint receptors, such as PD-1, CTLA-4, LAG-3, and TIGIT, which, upon ligation with their cognate ligands expressed on cancer cells or antigen-presenting cells, dampen the activation of T cells (27). Indeed, PD-1 blocking antibody therapy in the clinic has been effective against many cancers (20). PD-1 knockout by CRISPR/Cas9 has therefore been attempted and has shown increased activity of CAR-T cells in preclinical models and is also undergoing clinical evaluation (28-31). Likewise, its disruption is under investigation for TCR T therapy (32).
Next, transforming growth factor-β (TGF-β) signaling in the tumor microenvironment suppresses the cytotoxic function of CD8+ T cells and promotes the conversion of CD4+ T cells to the regulatory phenotype, negatively affecting anti-cancer immunity (33). Thus, blocking TGF-β signaling might maintain the potency of transferred T cells within a suppressive tumor microenvironment. Indeed, the CRISPR/Cas9-mediated knock-out of the TFG-β receptor 2 (TGFBR2) in CAR-T cells decreased induced regulatory T-cell (Treg) conversion and decreased the state of exhaustion during coincubation with cancer cells in the presence of TGF-β1(34). In addition, TGFBR2 knock-out in CAR-T cells showed improved tumor control in cell-line-derived and patient-derived xenograft mouse models with faster tumor clearance, increased proportions of circulatory memory T-cell subsets, and increased persistence.
Activation-induced cell death (AICD) by death-receptor signaling is essential for central and peripheral tolerance by triggering T-cell apoptosis (35). The major death receptor involved in this process is Fas (CD95), which binds to Fas ligands (Fas-L). Repeated exposure of antigens to CAR-T cells activates Fas signaling, which induces apoptosis, and thereby hampers their anti-tumor activity. To address this, Ren
Diacylglycerol kinases (DGKs) are enzymes that catalyze the conversion of diacylglycerol (DAG) to phosphatidic acid (PA) (37). DAG is a central second messenger that is indispensable for CD3 signaling, by interacting with protein kinase C theta (PKCθ) and Ras guanyl-releasing protein 1 (RasGRP1). Therefore, its conversion leads to decreased T-cell activation. DGKα and DGKζ, two isoforms of DGK, are dominantly expressed in T cells, and their simultaneous deletion led to increased effector function and proliferation of CAR-T cells and increased resistance to immunosuppressive factors, such as TGF-β and prostaglandin E2 (PGE-2) (38).
In addition to enhancing the efficacy of adoptive T-cell therapy, CRISPR/Cas9-mediated gene knock-out can be used to increase its safety. The rapid expansion of CAR-T cells in patients often leads to cytokine-release syndrome (CRS) and neurotoxicity due to the putative increase of inflammatory cytokines, such as IL-1 and IL-6 (39). In particular, IL-6 is mainly secreted by myeloid cells, such as monocytes. Rapidly expanding CAR-T cells secrete more granulocyte-macrophage colony-stimulating factor (GM-CSF), which likely favors the activation of IL-6-secreting myeloid cells (40). Indeed, Sterner
Next, CRISPR/Cas9 can be applied to the manufacturing of fratricide-resistant CAR-T cells, which is particularly important in the treatment of T leukemias and lymphomas. The most frequently targeted surface antigens for the treatment of these T-cell malignancies are the pan-T-cell markers CD3, CD5, and CD7 (42), which are also expressed on CAR-T cells, therefore making their manufacturing a challenge. Gomes-Silva
Additionally, CRISPR/Cas9 can be used to create allogeneic off-the-shelf CAR-T products that have advantages over autologous products, potentially decreasing the manufacturing cost and narrowing the window between patient administration and treatment. The two main hurdles in allogeneic CAR-T therapy are graft-versus-host disease and graft rejection, both of which can be prevented by deletion of the TCR α and β chains and B2M in donor T cells (44). Further, the disruption of endogenous TCR α and β genes can be used to prevent mispairing with exogenous TCR, potentially enhancing the function and safety of TCR-T therapy (45).
In addition to disruption, CRISPR/Cas9 can be used to introduce genes of interest. Currently, the most widely used method in engineering T cells for this purpose involves retroviral vectors such as γ-retrovirus (46) and lentivirus (47). Indeed, the two abovementioned approved CAR-T products were produced using these vectors. These retroviral vectors can deliver transgenes with high efficiency while maintaining stable expression, but their random integration into the host genome comes with several challenges, such as the potential for insertional oncogenesis (48), and clonal expansion (49), and the variation in the integration copy number that results in heterogeneity of the engineered cell products (50), both of which can be overcome using site-specific integration with the CRISPR/Cas9 system.
Although the delivery of donor DNA using non-integrating viral vectors such as AAV ensures high efficiency of knock-in, their production is a lengthy and complicated process that reduces their throughput (52). Additionally, the use of viral vectors can increase the manufacturing cost of engineered cell products and might have safety issues. To overcome this, non-viral donor DNA delivery may be preferred. Indeed, one study successfully used RNP and an electroporated DNA donor to insert a new TCR α/β chain (NY-ESO-1) into the TRAC locus (53). They optimized various parameters, including the stimulation conditions, the concentration of RNP and donor DNA template, and the electroporation parameters, thereby increasing the knock-in efficiency and reducing the toxicity in T cells. The knock-in efficiency of this method can be further improved by modifying the DNA donor template and stabilizing the RNP complex by adding anionic polymers (54). Next, the same group used a DNA-based donor strategy to compare the effects of 36 different genes (switch receptors, heterologous transcription factors, or metabolic regulators and receptors) inserted within the TRAC locus on the context-dependent fitness of human T cells (55). As a result, they identified that the insertion of the TGFBR2-4-1BB chimeric receptor improved solid- tumor clearance. Further optimization of DNA donor-based delivery methods will facilitate the basic research and clinical applications of CRISPR/Cas9-based knock-in technology in engineered T-cell therapy.
CRISPR/Cas9 editing is not limited to the engineering of T cells. Indeed, many groups have used this system for disease modelling and therapeutic genome editing in non-T immune cells, such as hematopoietic stem and progenitor cells (HSPC) and B cells (Fig. 2).
Clonal hematopoiesis of intermediate potential (CHIP) is a pre-malignant state that involves the temporal acquisition of mutations in HSC that can lead to clonal expansion and an inc-reased risk for hematological malignancies and cardiovascular diseases. To model CHIP, Tothova
In another study, Wagenblast
Genome editing has been used in HSPCs to correct β-hemoglobin disorders, which includes sickle-cell disease (SCD) and β-thalassemia, the former of which results from a missense mutation in the β-hemoglobin gene, and the latter of which results from a nonsense or frameshift mutation (58). Using the CRISPR/Cas9 system, three different approaches have been used to treat these diseases: 1) the direct correction of mutant β-hemoglobin with HDR (59), 2) the disruption of the erythroid-specific enhancer of BCL11A, a transcriptional repressor involved in the switching of fetal γ-hemoglobin to adult β-hemoglobin expression (60, 61), and 3) the disruption of the BCL11A-repressor binding site in the γ-globin promoter (62, 63). In particular, approach 2) is being evaluated in clinical trials (NCT03745287, NCT03655678).
Next, severe combined immunodeficiency X1 (SCID-X1) is a disease caused by the loss-of-function mutation of the IL-2 receptor γ chain (IL2RG), thereby impairing the development of B, T, and NK lymphocytes. Lentiviral delivery of wild-type IL2RG into HSCs and their subsequent adoptive transfer has been successful in recovering immune function in these patients. However, in previous clinical trials, the engineering of HSC using γ-retroviral vectors in a similar approach has led to T-cell leukemia in some patients due to insertional oncogenesis (64, 65). Although this has yet to be seen with the lentiviral system, Pavel-Dinu
For infectious diseases, genome-editing approaches have been used to mitigate viral infections. For example, C-C chemokine receptor 5 (CCR5), an important co-receptor involved in human immunodeficiency virus (HIV) internalization into T cells and macrophages, is a promising target for therapy (67, 68). One study used CRISPR/Cas9 to delete CCR5 by using a dual sgRNA strategy in HSPCs and transplanted them into immune-deficient NPG mice (69). Following HIV-1 challenge, the group observed a significant decrease in the RNA of a CCR5-tropic HIV-1 strain in the blood plasma. Further, this group evaluated the clinical application of similarly prepared cells in a patient with HIV and acute lymphocytic leukemia and observed long-term albeit marginally efficient engraftment (NCT03164135) (70).
Further, like the introduction of exogenous TCRs or CARs in T-cell engineering, B cells can acquire exogenous antibody sequences with known antigen specificity. One group used CRISPR/Cas9 to introduce the sequence of a HIV-1 broadly neutralizing antibody that cannot be easily generated with conventional immunization (71). They showed that the adoptive transfer of these engineered B cells provided long-term protection in mice challenged with HIV-1. Similarly, another group knocked-in engineered neutralizing antibodies consisting of a single-chain Fab against respiratory syncytial virus (RSV), HIV, influenza virus, and Epstein-Barr virus in primary human and mouse B cells (72). The RSV-specific mouse B cells were tested in a mouse-infection model and showed neutralizing activity, providing further evidence that engineered B cells might be an alternative against viruses that do not yet have clinically available vaccines despite decades of efforts.
An important bottleneck in the research on immune mechanisms has been the low throughput of conventional genetic knock-out or knock-down strategies. The discovery and development of RNA interference (RNAi), particularly with shRNA technology that exploits endogenous RNAi machinery (73), has allowed for much higher throughput workflows, but these are still limited by the incomplete knock-down of targets or off-target interactions, which can make proper interpretation difficult (74). Furthermore, incomplete knock-down may make it difficult to properly identify lethal genes based on the viable sufficiency of incompletely inactivated gene expression (75). With the progressive development of CRISPR/Cas9 technology, many groups have developed platforms for genome-wide screening in non-immune and immune cells for robust and massively high-throughput experiments that allow for a rich understanding of cellular mechanisms (76).
Many different variations of the CRISPR/Cas9 screening system have been developed: classical gene knock-out using enzymatically active Cas9 (77), transcriptional activation (CRISPRa) with catalytically inactive Cas9 (dCas9) fused to transcriptional activation domains (VP64, p65) (78, 79) or with dCas9 bound to modified gRNA scaffolds (80, 81), and transcriptional repression with dCas9 fused to the KRAB transcriptional repression domain (CRISPRi) (82). With these, three different branches of CRISPR genome-wide screening methods have been developed: arrayed, phenotypic pooled, and single-cell screens. In arrayed screens, the experimental design revolves around multi-well plate experiments, with wells being exposed to a single sgRNA at a time, a costly and time-consuming process with limited throughput. As such, most recent genome-wide screens have relied on the pooled phenotypic approach, with single-cell CRISPR/Cas9 screens becoming increasingly popular.
In pooled screens, a pool of cells is exposed to a library of sgRNAs, after which phenotypic analysis (selection) is done. Importantly, this sgRNA library is both diverse and redundant. Multiple target sequences (exons or regulatory elements of genes) are selected, and multiple guides for each target are designed to mitigate false-negative results. As an advantage, many pre-defined libraries that have been optimized for various applications are available (83). In very general terms, a pooled genome-wide CRISPR/Cas9 screen involves four key steps (Fig. 3): 1) the generation and packaging of an sgRNA library that targets genes of interest, usually using lentivirus vectors, 2) transduction of the sgRNA library into cell lines of interest that express or contain a variation of Cas9, 3) selection of transduced cells following various challenges (such as infection or drug inhibition), and 4) retrieval of genetic information from enriched and control cells using NGS (84).
Pooled screens can rapidly identify therapeutic targets for a multitude of diseases, both genetic and infectious. For HIV, a pooled screen was used in CD4+ cells to identify novel factors involved in internalization. The study identified TPST2 and SLC35B2, two members of the CCR5 sulfation pathway that was previously reported to be crucial for the interactions with the gp120 of HIV. Indeed, the independent knock-out of these proteins significantly decreased the rate of infection of the CCR5-tropic JR-CSF HIV strain against a human CD4+ T-cell line and primary T cells (85). Further, since combination antiretroviral therapy alone cannot cure HIV, many groups have tried to find mechanisms involved in HIV latency that can be targeted using latency-reversal agents with the aim of depleting latently-infected cells (86). One study used a pooled screen to identify a set of deubiquitinases involved in the latency of HIV. Among these were UCH37, USP14, OTULIN, and USP5, with USP14 being confirmed pharmacologically (87). With these results, the authors proposed that these enzymes could have a direct role on HIV-1 latency reversal by regulating NFkB signalling, which has been shown with other deubiquitinses (88).
Pooled screens have also been applied to key regulatory genes in immune cells involved in cancer immunity. Pan
Another study attempted to elucidate factors involved in T-cell exhaustion, one of the major mechanisms in the failure of CD-19-specific CAR-T-cell therapy (CAR-T19). They used the Brunello lentiviral sgRNA library targeting nearly 20,000 genes to probe factors that conferred resistance of a CD19+ acute lymphoblastic leukemia (ALL) to CAR-T19 therapy. They found that FADDKO and BIDKO Nalm-6 cells, a human CD19+ leukemia cell line, were more resistant to CAR-T19 treatment both
Last, another study attempted to identify novel regulators of Foxp3, a master transcriptional regulator in Treg cells. By targeting 489 nuclear factors, they identified Usp22, which is part of the deubiquitination module of the SAGA chromatin-modifying complex, as being a positive regulator of Foxp3 (92). Indeed, Usp22Treg-KO B6 mice displayed a delayed increase in EG7 lymphoma volume, which was associated with an increase in CD4+ and CD8+ tumor-infiltrating lymphocytes, likely because there were fewer tumor-infiltrating Treg cells. These studies and others support the use of genome-wide pooled phenotypic screens in the elucidation of disease-related mechanisms that might lead to the discovery of novel therapeutics.
Single-cell CRISPR/Cas9 screens are used to study the correlation between small genomic perturbations and global transcriptome changes at a single-cell level for a more comprehensive dataset than can be provided by pooled phenotypic screens (93). Several platforms have recently been developed, each of which relies on massively parallel RNA single-cell sequencing (MARS-seq) or droplet-based barcoding (DROP-Seq) (94, 95). For example, DROP-Seq uses microfluidic channels to pair a single cell with a hydrogel-encased bead attached to oligonucleotides containing unique cell barcodes (CBC) and unique molecular identifiers (UMI). The CBC pairs cDNA to a single cell, whereas the UMI is used for quantification. This single-cell-bead pair is then incorporated into a nanolitre-sized droplet containing cell lysis buffer and a RT-PCR enzyme mix for cDNA synthesis and subsequent sequencing. Several single-cell CRISPR/Cas9 platforms using these technologies have been independently developed in parallel and are named Perturb-Seq (96), Mosaic-Seq (97), CRISP-Seq (98), and CROP-Seq (99). Generally, these platforms use a standard sgRNA lentiviral library with a key modification, such as a polyadenylated unique guide index (UGI), to find the sgRNA identity within the transcriptome. The applications of single-cell CRISPR screening has been used extensively in non-immune cells but will not be covered here (94, 100, 101).
In its development, CRISP-Seq was used to probe immune genetic circuits in myeloid cells and identified key transcription factors involved in lineage commitment and response to pathogens. To do this, the authors transduced Cas9+ bone-marrow cells with a panel of sgRNAs targeting transcription factors involved in immune-cell development and responses, and 
Developments in CRISPR/Cas9 technology have revolutionized our understanding of disease-related mechanisms and are allowing research for potentially ground-breaking therapies. Despite its advantages, this system should be further optimized to mitigate safety concerns. For example, the generation of DSBs activates p53 (103) and induces megabase-scale chromosomal truncations (104). This can be overcome with precision editing methods, such as base editing and prime editing, that do not rely on DSB. Indeed, it has recently been shown that multiplexed gene knock-out can be accomplished by using base editing (105). Next, although the AAV-based donor delivery is highly efficient in HDR-based knock-in, the time and costs for its manufacturing limit its broad applicability, and as such, non-viral DNA delivery is being optimized. As an interesting future direction, since the efficiency of HDR mediated knock-in can be limited by endogenous DNA-repair machinery, the use of self-sufficient integration machinery, such as transposases, could be an alternative method, and has been used in
The authors have no conflicting interests.