Programmable gene insertion in human cells with a laboratory-evolved CRISPR-associated transposase

RESEARCH ARTICLE SUMMARY Programmable gene insertion in human cells with a laboratory-evolved CRISPR-associated transposase

INTRODUCTION: The efficient insertion of gene-sized DNA sequences at user-specified genomic sites is a long-standing goal in genome editing. Although current editing methods can correct most disease-causing mutations, the genetic diversity underlying many disorders will require the design and regulatory approval of many mutation-specific strategies-substantially limiting the number of patients who can benefit from therapeutic genome editing. Programmed genomic integration of a healthy gene copy could offer a mutation-agnostic treatment for loss-of-function genetic diseases. Additionally, targeted gene integration enables other applications, including cancer immunotherapies, transgenic cell and animal models for basic research, and metabolic engineering.

RATIONALE: CRISPR-associated transposases (CASTs) are naturally occurring bacterial systems that exploit nuclease-deficient CRISPR machinery to integrate DNA at genomic locations specified by guide RNAs. CASTs offer many attractive qualities as a genome editing tool, including facile programmability, compatibility with multi-kilobase-scale DNA cargo, and avoidance of genomic double-strand DNA breaks. Despite this promise, wild-type CASTs reported to date support minimal integration in human cells (often less than or equal to zero point one percent of treated cells). We reasoned that this low efficiency may stem from naturally evolved, suboptimal transposition catalysis that mitigates mobilization-induced fitness cost to the host. To enable efficient CAST integration in human cells, we developed a phage-assisted continuous evolution (PACE) system that rapidly evolves CAST variants capable of fast targeted transposition and applied CAST-PACE to a prototypical Type One - F CAST system from Pseudoalteromonas.

RESULTS: We linked on-target DNA integration in Escherichia coli to the propagation of continuously mutating phage genomes encoding evolving CAST components. After hundreds of generations of continuous selection, replication, and mutation in which the resulting phage survived an overall ten to the power of three hundred twenty-two dilution, we generated an evolved variant of the CAST transposase protein TnsB that mediated greater than two hundred fold improved integration activity in human cells. The evolved TnsB contains ten activity-enhancing mutations located throughout the protein, which likely modulate several distinct interactions with other CAST components. Notably, the evolved TnsB mediated efficient integration activity in human cells without requiring codelivery of the bacterial CAST accessory protein, ClpX, which is cytotoxic. We combined this evolved TnsB with other PACE-evolved and rationally engineered CAST components to yield evoCAST, a system optimized for human-cell integration activity. EvoCAST achieved ten to thirty percent integration efficiencies across fourteen genomic targets in human cells, representing a four hundred twenty-fold average improvement over wild-type CAST. EvoCAST supported large DNA cargoes greater than ten kilobases and mediated the integration of several therapeutic payloads at disease-relevant genomic sites, including safe harbor loci, sites for cancer immunotherapy engineering, and genes implicated in loss-of-function genetic diseases. EvoCAST also performed targeted integration in multiple human cell types, including primary human fibroblasts, and exhibited high product purity, with no detected insertions and deletions, predominantly unidirectional cargo insertion, single-base pair precision of integration, and low levels of off-target integration.

CONCLUSION: This work establishes CAST as a powerful platform technology for efficient, RNA-guided gene integration in human cells. The advantages of evoCAST-including its simple programmability, single-step integration mechanism, and avoidance of genomic double-strand breaks-make it well-suited for many applications in the life sciences and therapeutics, including the capability to address genetically diverse patient populations through a single editing agent. The CAST PACE system developed in this work also provides a strategy for improving the properties of other naturally occurring CASTs toward their use for efficient human-cell genome editing.

Programmable gene insertion in human cells with a laboratory-evolved CRISPR-associated transposase

Programmable gene integration in human cells has the potential to enable mutation-agnostic treatments for loss-of-function genetic diseases and facilitate many applications in the life sciences. CRISPR-associated transposases (CASTs) catalyze RNA-guided DNA integration but thus far demonstrate minimal activity in human cells. Using phage-assisted continuous evolution (PACE), we generated CAST variants with greater than two hundred fold average improved integration activity. The evolved CAST system (evoCAST) achieves approximately ten to thirty percent integration efficiencies of kilobase-size DNA cargoes in human cells across fourteen tested genomic target sites, including safe harbor loci, sites used for immunotherapy, and genes implicated in loss-of-function diseases, with undetected insertions and deletions and low levels of off-target integration. Collectively, our findings establish a platform for the laboratory evolution of CASTs and advance a versatile system for programmable gene integration in living systems.

Advances in programmable nucleases, base editors, and prime editors have enabled the disruption, installation, or correction of virtually any specified genomic DNA sequence less than two hundred base pairs in size. These technologies have been effectively deployed in the clinic as one-time treatments for various genetic disorders, with more than sixty clinical trials underway.

Despite this progress, the targeted insertion of gene-sized (greater than or equal to one kilobase) DNA sequences into specified genomic sites in mammalian cells remains a long-standing challenge in genome editing and gene therapy. The mutational heterogeneity underlying many genetic diseases, such as cystic fibrosis, Stargardt disease, and hemophilia B, complicates maximizing the fraction of patients that can benefit from therapeutic genome editing. Individual nuclease, base editing, and prime editing approaches that target pathogenic alleles typically cannot benefit patients with other mutations in the same gene, necessitating the development and regulatory approval of many different genome editing strategies to treat diverse patient cohorts.

Traditional gene addition therapies use viruses to provide healthy gene copies that rescue loss-of-function mutations, enabling a single-treatment strategy for many mutations in the same gene. Although effectively used in clinical applications, viral gene therapies face limitations, including risks of oncogenic DNA integration

(which can be attenuated by vector design), potential need for redosing, and immune responses to viral vectors. Moreover, genes expressed exogenously or from ectopic genomic loci lack their native regulatory contexts, which can lead to underdosing, overdosing, silencing, or dysregulated function.

Programmable insertion of large DNA sequences at endogenous genomic sites could enable one-time, permanent, mutation-agnostic therapies for loss-of-function diseases through the installation of a healthy gene copy at the native locus or a safe-harbor locus. Additionally, programmable DNA insertion could facilitate many other therapeutic and life sciences applications, including the streamlined production of cancer immunotherapies requiring transgenes (e.g., chimeric antigen receptor T cell therapy) and the simplified generation of transgenic cell lines and model organisms requiring large payloads.

Nucleases such as CRISPR-Cas nine generate targeted DNA double-strand breaks that can stimulate incorporation of exogenous donor DNA through homology-directed repair or end-joining pathways, for example, homology-independent targeted integration. However, homology-directed repair requires cellular machinery typically only expressed in dividing cells, preventing its efficient application in most therapeutically relevant cell types. Although homology-independent targeted integration can occur in nondividing cells, integration events lack both orientation and copy number control. Additionally, DNA double-strand breaks lead to uncontrolled formation of insertions and deletions at rates comparable to or higher than that of desired DNA integration and are associated with undesired cellular consequences, including chromosomal translocations, large deletions, and P fifty-three activation.

Engineered fusions of transposase and recombinase domains to Cas nine can support DNA integration without requiring DNA double-strand break formation, but thus far they have shown low efficiency at genomic loci in human cells and frequent off-target integration. The combination of prime editing and site-specific recombinases can mediate the efficient targeted installation of recombinase attachment sites followed by recombinase-mediated cargo gene insertion. This approach, however, requires coordinated prime editing and recombinase systems to catalyze multiple successive enzymatic steps, which can generate undesired by-products such as insertions and deletions and attachment sites lacking cargo gene insertion. Developing programmable DNA insertion strategies that avoid genomic DNA double-strand break formation, offer high product purity, and proceed in a single enzymatic step would complement existing approaches and potentially enable new research and therapeutic applications.

CRISPR-associated transposases are recently discovered bacterial systems that use RNA-guided, nuclease-deficient CRISPR-Cas systems to direct kilobase-scale transposon insertion by Tn seven-like transposases. Tn seven-like transposons have exapted multiple distinct CRISPR-Cas subtypes, with type one-F and type five-K CRISPR-associated transposases comprising the most extensively characterized systems to date. Type one-F CRISPR-associated transposases are especially promising for genome editing applications, exhibiting high insertion efficiency, high on-target specificity, high directionality bias, high product purity, and low incidence of tandem-insertion by-products in Escherichia coli.

Despite robust efficiency in bacteria, type one-F CRISPR-associated transposases reported to date are minimally active in human cells. Assessment of diverse type one-F CRISPR-associated transposases in human embryonic kidney two nine three T cells identified a Pseudoalteromonas sp. S nine eighty-three system with less than approximately zero point one percent genomic DNA insertion efficiency, which improved to approximately one percent efficiency when supplemented with the bacterial unfoldase ClpX, albeit with increased cytotoxicity. Although the low activity of Pseudoalteromonas sp. S nine eighty-three in human cells could arise from many potential explanations, we reasoned that insertion efficiency might be limited by transposition catalysis or DNA binding, which may have naturally evolved suboptimally to mitigate host fitness costs from excessive transposition.

In this study, we report the application of phage-assisted continuous evolution to evolve CRISPR-associated transposase systems that function efficiently in human cells. We evolved the Pseudoalteromonas sp. S nine eighty-three transposase module toward increased catalysis through hundreds of generations of mutation, selection, and replication, yielding an evolved transposase variant with greater than two hundred-fold average improved integration activity in human cells compared with that of wild-type Pseudoalteromonas sp. S nine eighty-three, without requiring ClpX. Structure-guided engineering of the DNA-targeting module of Pseudoalteromonas sp. S nine eighty-three further improved efficiency, synergizing with the evolved transposase to yield an optimized, evolved CRISPR-associated transposase system. The evolved CRISPR-associated transposase system supported ten to thirty percent insertion efficiencies of kilobase-sized DNA cargoes at fourteen tested genomic loci in human embryonic kidney two nine three T cells, with similar efficiencies in primary human fibroblasts. The evolved CRISPR-associated transposase system retained favorable aspects of wild-type Pseudoalteromonas sp. S nine eighty-three integration, including high regiospecificity, near unidirectional insertion, and undetected genomic insertion and deletion formation. Collectively, these results establish a platform for the evolution of CRISPR-associated transposase systems toward increased activity in mammalian cells and represent a milestone in the development of CRISPR-associated transposases for targeted, DNA double-strand break-free DNA insertion with therapeutically relevant efficiencies.