Retroelements have a critical role in shaping eukaryotic genomes. For instance, site-specific non-long terminal repeat retrotransposons have spread widely through preferential integration into repetitive genomic sequences, such as microsatellite regions and ribosomal DNA genes1–6. Despite the widespread occurrence of these systems, their targeting constraints remain unclear. Here we use a computational pipeline to discover multiple new site-specific retrotransposon families, profile members both biochemically and in mammalian cells, find previously undescribed insertion preferences and chart potential evolutionary paths for retrotransposon retargeting. We identify R2Tg, an R2 retrotransposon from the zebra finch, Taeniopygia guttata, as an orthologue that can be retargeted by payload engineering for target cleavage, reverse transcription and scarless insertion of heterologous payloads at new genomic sites. We enhance this activity by fusing R2Tg to CRISPR–Cas9 nickases for efficient insertion at new genomic sites. Through further screening of R2 orthologues, we select an orthologue, R2Tocc, with natural reprogrammability and minimal insertion at its natural 28S site, to engineer SpCas9H840A–R2Tocc, a system we name site-specific target-primed insertion through targeted CRISPR homing of retroelements (STITCHR). STITCHR enables the scarless, efficient installation of edits, ranging from a single base to 12.7 kilobases, gene replacement and use of in vitro transcribed or synthetic RNA templates. Inspired by the prevalence of nLTR retrotransposons across eukaryotic genomes, we anticipate that STITCHR will serve as a platform for scarless programmable integration in dividing and non-dividing cells, with both research and therapeutic applications.
