In this commentary we discuss our recent work on delivering an anti-GD2 CAR (chimeric antigen receptor) via homology independent targeted insertion (HITI) using the CRISPR/Cas9 technology [1]. HITI relies on Non-Homologous End Joining (NHEJ) that is predominantly exploited by both dividing and non-dividing cells to repair double stranded DNA breaks (DSBs). We explore considerations when using HITI based strategies. Furthermore, we discuss a method for post-HITI CRISPR EnrichMENT (CEMENT) within the context of large-scale clinical manufacturing of non-viral CAR-T cells [2].

NHEJ is the primary mechanism for DNA repair of DSBs throughout the course of the cell cycle program. Unlike homology directed repair (HDR), which is active only during the G2 and S phase of the cell cycle, NHEJ is independent from the cell cycle. Therefore, it presents a unique strategy for therapeutic editing of non-activated T-cells. Many groups have described the use of HDR for performing targeted genomic insertion of a CAR into a variety of loci including TRAC, AAVS1, B2M, or PDCD1, where the activation state of the T-cell is imperative for a successful insertion [3-5]. In contrast, HITI, which utilizes NHEJ for target gene insertion, can be explored in non-activated T-cells, thereby facilitating rapid manufacturing of CAR-T cells. Outside T-cell editing, HITI has been explored for applications of knock-in of large reporter genes in ESCs (Embryonic Stem Cells) and post-mitotic cells [6]. Most importantly, HITI has also been explored in vivo in rat models [7]. Recently, in addition to the NHEJ and HDR pathways, HMEJ (Homology Mediated End Joining) has also been explored for insertion of therapeutic CARs into T-cells. This approach relies on short homology arms (~48 bp), which has been shown to have more efficient integration of larger genetic cargo compared to HDR-based approaches [8].

In the last decade various vectors for delivering transgenes using nuclease-based genome editors have been established (Adeno associated viruses, mRNA, transposons, single stranded and double stranded DNA and plasmid donor DNA vectors) [1]. Unlike viral vectors, manufacturing plasmid DNA from bacteria is relatively easy and low-cost. Therefore, plasmid DNA has recently been explored more for T cell therapy. We employed the Nanoplasmid backbone which is only 450bp. The use of Nanoplasmid DNA technology is desirable for T cell therapy applications as it has been reported to have higher expression levels and reduces the cell transfection related toxic effects in comparison to dsDNA templates [9]. Furthermore, the timeline for production of clinical grade Nanoplasmid vectors is anticipated to be shorter and less cost intense as compared to viral vector based on our experience with non-GMP grade nanoplasmid [8].

For HITI, we and others have demonstrated 1 cut site (cs) to yield in higher KI efficiencies as compared to 0cs and 2cs [2,7]. In turn, for HDR some groups have been able to insert at high efficiencies without employing any cut sites [5,10]. Alternatively, Chavez and colleagues reported simultaneous editing of the insertion site locus and cutting within their HDR template delivered via an integrase deficient lentiviral vector to enable targeted insertion of large transgenes. Here, they compared donor DNA with regular and truncated cut sites and determined that processing and linearization of donor DNA caused by the Cas9 RNP complex results in an enhanced knock-in efficiency [11].

To achieve higher purity and CAR+ cell yield, enrichment methods to increase the target population needs to be considered. Early approaches in non-viral gene editing used feeder cells to aid the growth of the positively edited cell populations [12]. However, these methods are inconsistent and complex. Depending on the manufacturing feasibility and selection strategies, it can include metabolic selection through integration of a methothrexate (MTX) resistant version of the Dihydrofolate Reductase (DHFR-FS) or more complex methods through surface markers including, but not limited to tNGFR (truncated Low-affinity Nerve Growth Factor Receptor) or tEGFR (truncated Epidermal Growth Factor Receptor). We evaluated the DHFR-FS system versus separation column based tEGFR (truncated Epidermal Growth Factor Receptor) and tNGFR (truncated Low-affinity Nerve Growth Factor Receptor) based enrichment to compare the different approaches to selection of edited cells [13,14]. Importantly, we optimized the MTX treatment schedule and shortened the duration of MTX exposure. Our work demonstrates that by using the DHFR-FS system we can enable an efficient MTX-based enrichment of transgene positive cells and generate anti-GD2 CAR-T cells with up to ~80% purity. The DHFR-FS system offers a more controlled, consistent and scalable alternative and allows for an easier implementation into large-scale manufacturing process, whereas the use of cell surface selection markers requires additional cell processing and purification steps, which can significantly decrease the cell yield and increase manufacturing complexity and cost.  Alternatively, transgene knock-in into essential gene loci has also been explored for enrichment of transgene positive populations [15]. Recently, others have explored the use of HDR repair templates containing a splice acceptor and splice donor targeting the intronic sequence of endogenous surface receptor loci, thereby enabling the magnetic depletion of unedited cell populations [16]. However, both technologies are limited to targeted insertion into essential genes and surface receptors respectively.

Using a semi-closed system electroporation protocol, we show that the co-delivery of the genome editing components can be performed in a clinically relevant context. We establish a modular manufacturing process using the Maxcyte GTx electroporation unit that generated 5.5 x 10⁸ – 3.6 x 10⁹ GD-2 CAR-T cells from a starting population of 5 x 10⁸ T-cells in a G-REX 100M cell culture system, across 3 independent donors. The Maxcyte GTx, a clinically relevant electroporation platform, enables semi-closed or closed system flow-through electroporation. It can be aseptically connected to the G-REX (Wilson Wolf) platform, offering a highly efficient method for closed system scale-up. To edit T cells on the Maxcyte GTx, we used the preset Expanded T-Cell 4 protocol, which has been optimized for electroporation of T cells. However, one should consider optimization of parameters such as electroporation voltage, pulse width, cell concentration, DNA vector concentration, electroporation buffer parameters, Cas9:gRNA ratio, time of electroporation, temperature, and pre-electroporation and post-electroporation handling of the T-cells to allow for the most efficient editing process.

As CRISPR/Cas9 therapies are more commonly being applied clinically, it is important to ensure their safe and effective use. CRISPR/Cas9 can cause unintended double-stranded breaks at off-target sites, leading to insertions, deletions, or translocations [17]. Careful selection of gRNA should involve an empirical design process. It's crucial to consider challenges like the presence of a PAM sequence near the target site, the gRNA's ability to form secondary structures, mismatch tolerance, delivery efficiency, and accessibility to chromatin. Therefore, the gRNA design process must incorporate in silico, and in vitro methods to screen gRNAs with efficient on-target and minimal to no off-target activity. The recent FDA approval of exagamglogene autotemcel (exa-cel) provides us with more insights into the standards that need to be established in the process [18]. We outline a roadmap for assessing and mitigating CRISPR/Cas9 genotoxicity events. When designing CRISPR/Cas9 gene editing therapies, it is important to consider early on mitigation strategies of off-target editing by carefully selecting and screening guides. We utilize in silico tools like COSMID (CRISPR Off-target Sites with Mismatches, Insertions, and Deletions) and CCTop (CRISPR/Cas9 Target Online Predictor) to exclude gRNAs with predicted high off-target effects [19,20]. We selected a gRNA targeting TRAC, encoding a mismatch base for optimal on-target performance as confirmed via GUIDE-Seq [21]. Sequencing approaches like GUIDE-seq, CIRCLE-seq, Discover-seq, Digenome-seq, and SITE-seq combine sequencing data and in silico analysis to identify off-target effects and should be considered for off-target assessment [22-26]. It is important to note that these tools rely on the human reference genome, and for a more comprehensive off-target nomination, tools like CRISPRme, which account for human genetic diversity and perform variant-aware off-target assessment, are essential [27]. Post editing, next-generation sequencing (e.g. rhAMP Seq) should be used as it allows for quantification and precise identification of off-target mutations induced by CRISPR/Cas9 by providing high-resolution sequencing data [28]. On-target editing outcomes could include Indels, long deletions/truncations, inversions, insertions, copy-neutral LOH (loss of heterozygosity), and chromothripsis [29]. While Indels are easier to detect, standard sequencing methods do not capture the full range of other outcomes. More complex methods, like long-read sequencing and single primer amplification, are required for comprehensive detection [29-31]. Additionally, on and off-target insertions can be assessed via Targeted Locus Amplification (TLA), which provides an unbiased way to assess insertion sites [32].

Previous reports have provided evidence of low-level chromosome 14 aneuploidy due to the editing of the TRAC locus by CRISPR/Cas9 using a clinically relevant gRNA [33]. Therefore, monitoring chromosomal translocations utilizing ddPCR (droplet digital PCR) over time should be considered for long-term patient follow up assessments. Tsuchida et al. conducted a systematic analysis of Cas9-induced chromosome loss events and recommended protocol adjustments to reduce the occurrence of chromosome loss [34]. These improvements include activating/stimulating T-cells after delivering genome editing components into non-activated T-cells, which helps mitigate aneuploidy linked to elevated TP53 expression. This effect may be reduced utilizing the HITI method and provides further rationale for exploring HITI as a T-cell engineering platform.

To conclude, HITI in non-activated T-cells has its advantages, as it can streamline and accelerate the manufacturing process. Importantly, editing non-activated T-cells using HITI may also enhance the safety profile of the engineered T-cells.