CRISPR and Beyond: A Guide to the Next Generation of Gene Editing Technologies

If you are planning a gene editing experiment today, the choice is no longer just CRISPR-Cas9 versus nothing. Base editors, prime editors, and CRISPR-associated transposases (CASTs) each solve specific problems that the original system left unsolved: off-target cuts, double-strand breaks, and the need for donor templates. This guide lays out a practical decision framework for teams moving beyond first-generation CRISPR.

Who needs next-generation gene editing and what goes wrong without it

Research groups working with primary cells, stem cells, or organisms where DNA repair pathways are inefficient often find that standard Cas9 creates more problems than it solves. The double-strand break (DSB) induced by Cas9 triggers the cell's repair machinery—non-homologous end joining (NHEJ) or homology-directed repair (HDR). NHEJ is efficient but introduces small insertions or deletions (indels) that are useful for gene knockout but not for precise corrections. HDR, the pathway needed for precise edits, is inefficient in most cell types, especially post-mitotic cells like neurons or hepatocytes.

Without moving beyond Cas9, projects that require single-nucleotide changes, exon corrections, or targeted insertions without DSBs often stall. A typical failure mode is spending months designing HDR templates and screening clones, only to achieve editing efficiencies below 1%. Another scenario is off-target edits that invalidate phenotypic data. Teams that have tried and abandoned CRISPR for therapeutic applications often cite these bottlenecks.

Next-generation tools address these issues directly. Base editors (cytidine and adenine deaminases fused to a catalytically dead Cas9) convert one base pair to another without cutting both DNA strands. Prime editors use a Cas9 nickase fused to a reverse transcriptase and a prime editing guide RNA (pegRNA) to write new genetic information directly into the genome—no donor template required. CASTs allow RNA-guided integration of large DNA payloads without DSBs. Each tool has trade-offs, but all remove the reliance on HDR.

The reader who needs this guide is the one who has tried Cas9 for a precise edit and found the efficiency too low, the off-target rate too high, or the clone screening too laborious. Alternatively, the reader may be evaluating platforms for a new project and wants to avoid the pitfalls of first-generation systems from the start.

Prerequisites and context to settle before starting

Before selecting a tool, the team must clarify three things: the desired edit type, the target cell type, and the tolerable level of byproducts. These constraints will narrow the options dramatically.

Edit type and precision requirements

Single-nucleotide substitutions are best served by base editors. Small insertions, deletions, or substitutions (up to about 40 base pairs) are the sweet spot for prime editors. Larger integrations—kilobase-scale—require CASTs or, if DSBs are acceptable, HDR with donor templates. If the goal is a simple knockout, standard Cas9 remains the most straightforward and cheapest option; there is no need to use a newer tool just because it exists.

Cell type and delivery constraints

Primary cells, particularly T cells and hematopoietic stem cells, are notoriously difficult to transfect and edit. Base editors and prime editors, because they nick rather than cut, tend to cause less p53 activation and DNA damage response, leading to higher viability in sensitive cells. For in vivo delivery, the size of the editing machinery matters: Cas9 is about 4.2 kb, base editors are slightly larger, and prime editors with the reverse transcriptase can exceed 6 kb. CAST systems are even larger. If the delivery vehicle is an adeno-associated virus (AAV) with a ~4.7 kb cargo limit, split systems or smaller Cas9 orthologs (e.g., SaCas9) may be necessary.

Byproduct tolerance

Base editors can produce bystander edits—unwanted changes to bases adjacent to the target. Prime editors can produce indel byproducts from the nicking step or from pegRNA mispriming. CASTs have the lowest byproduct rate among current tools but are less well-characterized in human cells. The team must decide what frequency of byproducts is acceptable for the application. For a therapeutic product, regulatory expectations are extremely low; for a basic science screen, higher rates might be tolerable.

We also recommend checking the available bioinformatics resources for each tool. For base editors, there are web tools (BE-Designer, CRISPResso2) that predict bystander edits. For prime editors, pegRNA design remains more complex, with tools like PrimeDesign and pegFinder. Teams should ensure they have the computational capacity to design and validate pegRNAs before committing to prime editing.

Core workflow: selecting and applying the right tool

The workflow we describe here is conceptual—it applies across platforms but must be adapted to the specific tool's protocols.

Step 1: Define the edit and choose the tool family

Start with the target sequence. If the desired change is a single base transition (C→T, G→A, A→G, T→C), a base editor is the first candidate. For transversions or more complex edits, prime editing is the primary option. For large insertions, consider CASTs or HDR.

Step 2: Design guide RNAs and repair templates

For base editors, design a single guide RNA (sgRNA) that positions the target base within the editing window (typically positions 4–8 of the protospacer). Use a tool that predicts bystander edits. If bystanders are unavoidable, consider using a different base editor variant with a narrower window (e.g., YE1-BE4max) or switch to prime editing.

For prime editing, design a pegRNA with the desired edit in the 3' extension. The extension must include a primer binding site (PBS) and the reverse transcriptase template. Design tools automate this, but manual inspection is still needed to avoid secondary structures. Also design a nicking sgRNA for the opposite strand to boost efficiency.

For CASTs, the guide RNA and donor DNA must be co-delivered. The system integrates the payload at a specific genomic site defined by the guide. Design the guide to target a safe harbor locus (e.g., AAVS1, safe harbor sites in the genome) if integration position matters.

Step 3: Deliver the components

Delivery can be plasmid transfection, ribonucleoprotein (RNP) delivery, or viral transduction. RNP delivery is preferred for primary cells because it avoids DNA integration and reduces off-target effects. For base and prime editors, RNP delivery is challenging because of the large protein size; mRNA or plasmid transfection is more common. For CASTs, plasmid or mRNA delivery is typical.

Step 4: Screen and validate edits

After editing, cells are harvested for genomic DNA extraction. Targeted deep sequencing of the on-target site and predicted off-target sites is the gold standard. For base editors, also sequence the bystander window. For prime editors, check for indels at the nicking site. For CASTs, confirm full-length integration and absence of vector backbone.

We recommend screening at least 48 hours after editing to allow time for complete editing and repair. Clonal isolation is often needed for therapeutic applications, but for pooled screens, bulk sequencing with a high read depth (>10,000 reads per sample) is sufficient.

Tools, setup, and environment realities

Choosing the right reagents and protocols is as important as choosing the tool. Below we compare the main platforms across practical dimensions.

Reagent sources and quality control

Many labs order custom sgRNAs and pegRNAs from commercial vendors. For base editors, the protein can be purified in-house or purchased as mRNA. Prime editing requires a reverse transcriptase that is not yet widely available as recombinant protein; most labs use plasmids encoding the PE2 or PE3 system. CAST systems are available from a few academic labs and commercial sources, but the field is still early.

We strongly recommend testing each new batch of guide RNA or editing protein in a control cell line (e.g., HEK293T) before moving to precious primary cells. This step catches synthesis errors, poor design, or degraded reagents.

Computational environment

Design tools are web-based, but local installation of tools like PrimeDesign may be needed for bulk design. Teams should have access to a Linux server or cloud instance for running CRISPResso2 on sequencing data. We advise setting up a standardized analysis pipeline early, as manual analysis of deep sequencing data is error-prone.

Safety and containment

All gene editing work with human cells should be done in BSL-2 facilities with appropriate institutional approvals. For projects involving germline editing or heritable modifications, additional ethical and regulatory review is required. This guide is for general informational purposes and does not constitute regulatory or legal advice. Teams should consult their institution's biosafety committee and applicable national guidelines before beginning experiments.

Variations for different constraints

The optimal tool and protocol depend on the specific constraints of the project. Here we describe three common scenarios.

Scenario A: Correcting a point mutation in patient-derived iPSCs

Induced pluripotent stem cells (iPSCs) are sensitive to DSBs and have low HDR efficiency. A base editor is the best choice if the mutation is a transition. For a transversion, prime editing is the only option without DSBs. Delivery via nucleofection of mRNA encoding the editor, combined with chemically modified sgRNA, yields the highest viability. Expect editing efficiencies of 10–30% in iPSCs. Clonal expansion is necessary to isolate corrected lines. A key pitfall is epigenetic silencing of the editor mRNA; using a modified cap and polyA tail can improve expression.

Scenario B: Knocking in a large reporter gene in cell lines

For integrating a GFP or luciferase cassette into a safe harbor locus, CAST systems offer a DSB-free alternative to HDR. The integration efficiency is typically 10–30% in HEK293T cells, and the cargo size limit is about 10 kb with current systems. If a CAST system is not available, HDR with a donor template and Cas9 is still viable, but the efficiency is lower and requires drug selection. A common mistake is using a donor template with insufficient homology arms; at least 800 bp per arm is recommended for HDR.

Scenario C: Multiplex editing in T cells for cancer immunotherapy

Editing multiple genes simultaneously in primary T cells is challenging because of cytotoxicity from multiple DSBs. Prime editing can achieve multiplex edits with minimal toxicity, but the efficiency per edit drops with each additional pegRNA. Base editors can also be used for multiplexing if all edits are transitions. Delivery by electroporation of mRNA encoding the editors and chemically modified guide RNAs is the current best practice. Off-target analysis must be performed for each guide. A trade-off to consider is that multiplex prime editing requires careful pegRNA design to avoid interference between guides.

In all scenarios, the team should run a pilot experiment with a single edit to validate the protocol before scaling to multiple targets or precious samples.

Pitfalls, debugging, and what to check when it fails

Even with careful planning, experiments fail. The most common issues are low editing efficiency, high byproduct rates, and cell death.

Low efficiency

If editing efficiency is below 1%, first check delivery efficiency. Use a fluorescent reporter or a control guide that targets a well-characterized locus (e.g., EMX1) to confirm that the editing components are entering cells. If delivery is adequate, the guide RNA may be poorly designed. For base editors, check that the target base falls within the editing window. For prime editors, the pegRNA structure is critical: a PBS melting temperature between 30–35°C and a template length of 10–15 nt is a good starting point. For CASTs, verify that the guide RNA and donor DNA are co-delivered in the same cell; using a fluorescent marker on the donor can help.

Another cause of low efficiency is poor cell health. If cells are stressed, they may not express the editing machinery or may undergo apoptosis. Optimize transfection or electroporation conditions to maximize viability above 70%.

High byproduct rates

For base editors, bystander edits are the main concern. If bystanders are unacceptable, switch to a variant with a narrower editing window (e.g., YE1-BE4max) or redesign the guide to shift the target base relative to the window. For prime editors, indels at the nicking site can occur if the nicking sgRNA cuts in the wrong position. Use the PE3b strategy, which nicks only the edited strand, to reduce indels. For CASTs, byproducts are rare, but if integration is not full-length, check the donor DNA for truncation.

Cell death or poor growth

If cells die after editing, the most likely cause is toxicity from the editing protein or from multiple DSBs. Reduce the amount of plasmid or mRNA delivered. For Cas9-based systems, using a high-fidelity variant (e.g., eSpCas9, SpCas9-HF1) can reduce off-target cutting and associated toxicity. For base and prime editors, the nickase activity causes less DNA damage, but high expression can still be toxic. Titrate the dose.

Finally, if nothing works, go back to the basics: sequence the target locus in the cell line to confirm there are no polymorphisms that prevent guide binding. Check that the editing components are full-length by Sanger sequencing of the plasmids. And consider that some cell types are simply recalcitrant to editing; in that case, switching to a different cell line or using a viral delivery method may be necessary.

As next steps, we recommend that teams new to these tools start with a simple base editing experiment in HEK293T cells to build confidence, then move to prime editing for a more challenging edit, and finally explore CASTs if large insertions are needed. Document every parameter—cell density, transfection condition, guide sequence, and analysis pipeline—so that troubleshooting is systematic. The field is moving fast, but the fundamentals of careful design, rigorous validation, and honest reporting of failures will serve any project well.