T-PGI: an engineered STITCHR system for scarless, programmable genome integration

Joo‐Hyung Lee, Jonathan D. Finn, Sandeep Kumar, K. Daneshvar, Yizhao Ni, Dev Paudel, Hyaeyeong Kim, Seppo W. Langer, Sara Ashrafi Kakhki, Omar O. Abudayyeh, Jonathan S. GootenbergbioRxiv 2025 · DOI

TL;DR

Researchers have engineered a new genome editing system called T-PGI that significantly improves the efficiency, specificity, and versatility of scarless DNA insertion. This advanced system builds upon previous work and offers a more robust and adaptable platform for therapeutic genome engineering.

What Problem Does This Paper Address?

Precise and scarless insertion of DNA into the genome is a critical goal for therapeutic genome editing. While CRISPR-Cas9 has revolutionized gene editing, current methods still face challenges like variable efficiency, reliance on cellular repair pathways, and potential for unintended genomic alterations. The STITCHR system, which uses target-primed reverse transcription (TPRT) for scarless insertion, showed promise but suffered from relatively low efficiency, limiting its immediate clinical applicability. This paper addresses these limitations by engineering a significantly improved version of STITCHR, dubbed T-PGI, to overcome these hurdles and make scarless, programmable DNA integration more practical and efficient.

Methods

The T-PGI system is built upon the STITCHR concept and involves several key engineering advancements:

  • Optimized Retrotransposon Enzyme (R2Tocc): The researchers selected and further engineered R2Tocc, a retrotransposon-derived reverse transcriptase, known for its lower background activity compared to other variants. This involved rational point mutations and modular domain insertions to enhance its performance.
  • Paired nCas9 Nicking: Instead of creating double-strand breaks, T-PGI uses two Cas9 nickases (nCas9) guided by sgRNAs to introduce single-strand nicks flanking the target integration site. This approach avoids reliance on error-prone cellular repair pathways.
  • Engineered RNA Donor Templates: The system utilizes RNA donor templates that contain homology arms flanking the desired DNA cargo. These templates are designed with optimized UTRs and homology arm lengths.
  • Enhanced Protein Expression and Localization: To improve the efficiency of the core enzyme, the researchers explored various strategies including bicistronic co-expression with P2A linkers and the addition of nuclear localization signals (NLSs) to the R2Tocc protein.
  • Combinatorial Screening: Extensive screening of guide RNA (sgRNA) pairs and donor templates was performed to identify optimal combinations for efficient and specific targeting. This included developing a method to express sgRNAs from linear dsDNA fragments, streamlining high-throughput screening.
  • Internal Protein Engineering: The researchers delved into internal modifications of the R2Tocc enzyme, identifying a flexible region that could tolerate deletions and insertions, leading to further performance enhancements.

Datasets

Not explicitly discussed in the paper as distinct datasets for training or evaluation in the machine learning sense. The experiments utilize various cell lines (HEK293T) and genomic loci (AAVS1, NOLC1, OTC) for testing and validation of the T-PGI system.

Key Insights and Ideas

  • R2Tocc is the superior ortholog: The R2Tocc variant of the reverse transcriptase showed significantly reduced off-target integration at the native 28S rDNA locus compared to R2Tg, making it a safer choice for therapeutic applications.
  • TPRT handles large payloads: Contrary to initial concerns about reverse transcriptase truncations, T-PGI demonstrated efficient insertion of DNA payloads ranging from 26 bp to over 4 kb, showcasing broad cargo flexibility.
  • Large-scale genomic rewriting is feasible: T-PGI can precisely replace large genomic segments, including deletions up to 5 kb, opening doors for applications like exon replacement and gene repair.
  • Internal engineering unlocks enzyme potential: Modifying internal regions of the R2Tocc enzyme, particularly an intrinsically disordered region, proved more effective than terminal modifications for enhancing activity and stability.
  • Dual-guide strategies are key: Using paired sgRNAs to create nicks on both sides of the target site significantly improved editing efficiency.
  • sgRNA expression from dsDNA is efficient: Expressing sgRNAs from linear dsDNA fragments offers a practical and high-throughput alternative to plasmid-based expression.

Key Results

  • T-PGI achieved robust integration across diverse cargo sizes, from 26 bp to 4.1 kb, with minimal loss in efficiency.
  • The system successfully supported large-scale sequence replacement, including deletions up to 5 kb.
  • Insertion fidelity remained high, with minimal local indels or structural variants detected across various editing outcomes, even for large insertions.
  • Dual-terminal NLS tagging enhanced R2Tocc activity by 4-5 fold.
  • Bicistronic expression via P2A outperformed most direct fusions, yielding approximately two-fold higher insertion efficiencies.
  • Internal deletion of a 50 amino acid segment (Deletion 4) in R2Tocc enhanced insertion efficiency.
  • Insertion of the Mu Gam protein into the R2Tocc deletion site conferred a ~30% increase in insertion efficiency.

Limitations and Future Work

  • Variability in insertion efficiencies: The authors note that insertion efficiencies were occasionally variable across replicates, largely reflecting differences in transfection efficiency and cellular status in HEK293 cells. This highlights the importance of optimizing cellular context and delivery.
  • Decreased fidelity with increasing cargo length: While overall fidelity is high, a modest decrease was observed with increasing cargo length, attributed to the intrinsic error rate of reverse transcription. Future work could focus on further improving the fidelity of the reverse transcriptase.
  • Further optimization of enzyme configuration: While significant improvements were made, ongoing research could explore additional protein engineering strategies for further enhancing enzyme stability, activity, and specificity.
  • Broader application testing: The authors suggest that the principles developed for T-PGI can be applied to other genomic targets, implying future work will involve testing the system across a wider range of loci and cell types for therapeutic applications.

Summary updated

Abstract

Abstract Scarless, programmable insertion of defined DNA remains a central goal for therapeutic genome editing. We introduce T-PGI (Transposon-mediated programmable genomic integration), an engineered implementation of STITCHR that preserves target-primed reverse transcription (TPRT) while substantially improving efficiency, specificity, and modularity. T-PGI uses R2Tocc, a low-background ortholog that we further engineered through defined deletions, rational point mutations, and modular domain insertions to enhance performance. The system employs paired nCas9 nicks flanking homology arms and bicistronic co-expression with dual NLSs, together with optimized RNA donor designs. Using combinatorial guide–template screening, T-PGI achieves robust integration across diverse cargos, including precise cassette insertion following multi-kilobase deletions. Short- and long-read sequencing confirm high-fidelity insertion with minimal local indels or structural variants. Collectively, these advances establish T-PGI as a practical and adaptable platform for accurate, scarless genome integration and provide concise design principles spanning enzyme architecture, donor configuration, and guide pairing for next-generation therapeutic editing.
BibTeX
@misc{lee2025t,
  author = {Joo‐Hyung Lee and Jonathan D. Finn and Sandeep Kumar and K. Daneshvar and Yizhao Ni and Dev Paudel and Hyaeyeong Kim and Seppo W. Langer and Sara Ashrafi Kakhki and Omar O. Abudayyeh and Jonathan S. Gootenberg},
  title = {{T-PGI: an engineered STITCHR system for scarless, programmable genome integration}},
  year = {2025},
  doi = {10.1101/2025.09.28.679074},
  howpublished = {bioRxiv},
}
Figure