A single infusion of CRISPR/Cas9-edited blood-forming stem and progenitor cells was associated with better movement and tissue measures in a mouse model of Friedreich’s ataxia, including the cerebellum, heart and pancreas. Edited donor cells were also detected in the animals’ blood-forming system and several affected tissues, alongside higher levels of hFXN, the human frataxin marker tracked in the study.
The result was incomplete. Treated mice remained below wild-type levels on cerebellar neuronal density and molecular-layer thickness, elevated left-ventricular mass remained, and fasting blood glucose stayed comparable across groups through six months.
How the test was set up
The experiment used humanized YG8JR mice carrying an hFXN transgene with more than 800 GAA repeats on a murine fxn-null background. The analyzed groups were 13 mPPL-001 recipients, 11 mock-HSPC recipients, nine wild-type controls and nine untreated YG8JR controls. Measurements were taken 6.5 to 7 months after transplantation.
mPPL-001 was generated by electroporating HSPCs with a Cas9 ribonucleoprotein containing the UP4 and DN4 guide RNAs; mock HSPCs underwent the same procedure without Cas9. Recipients were lethally irradiated with 7 Gy 24 hours earlier and received 1.5 million to 1.75 million cells per mouse by tail-vein injection. YG8JR mice were randomly categorized as untreated, mock or mPPL-001 recipients, while wild-type littermates served as controls.
Intended GAA-repeat excision in the edited HSPCs ranged from 7.6% to 30.5%, with a mean of 20.3%. The edited cells retained colony-forming capacity comparable to mock HSPCs.
The edited cells reached several tissues
Donor-derived cells were detected in multiple compartments. Chimerism, the share of donor-derived cells in a tissue, averaged 7.6% in bone marrow, 5.9% in peripheral blood, 26.5% in spleen, 19.8% in thymus, 0.52% in cerebellum and 4.14% in skeletal muscle.
On gait measures, mPPL-001 recipients had better results than disease and mock groups. Cerebellar neuronal density in the granule cell layer and molecular-layer thickness were better preserved, but treated values remained below wild-type levels, and existing cerebellar damage was not reversed.
Brain tissue also showed a molecular shift associated with the edited-cell recipients. Bulk RNA sequencing, which measures gene activity across a tissue sample, found a global cerebellar shift. Of 709 significantly altered genes at a P threshold of 0.01, 102 were treatment-associated, including 24 described as rescued: 17 fully restored and seven partially restored. Microglia, the brain’s resident immune cells, shifted in morphology from amoeboid and activated toward ramified and homeostatic patterns.
Signals appeared beyond the brain
Outside the brain, ventricular dimensions in mPPL-001 recipients were nearer wild-type values. Their mitochondria had morphology similar to wild-type controls, mitochondrial hFXN was higher than in mock recipients, and CD68-positive macrophage accumulation was lower. Elevated left-ventricular mass, however, remained above wild-type levels.
In the pancreas, insulin-producing beta-cell abundance in mPPL-001 recipients was comparable to healthy controls, while CD68-positive macrophages were fewer and CD206-positive anti-inflammatory macrophages were more common. Fasting glucose remained comparable with wild-type mice across all groups through six months, so the tissue findings did not demonstrate improved glucose regulation.
What remains unresolved
The study also points to a possible editing threshold, but not a general one. Cerebellar hFXN was higher in mPPL-001 recipients than in untreated and mock cohorts and correlated with bone-marrow engraftment, with a reported P value of 0.04. Cohort comparisons described 7.6% editing as limited benefit, approximately 14% as partial restoration, and more than 14% as greater and more consistent improvement. That threshold was derived from cohort-level comparisons and was not prospectively validated.
The safety picture was less complete. ONE-seq nominated 52,109 potential UP4 sites and 45,998 potential DN4 sites. Follow-up sequencing found fewer than five tier 1 sites and fewer than 71 tier 2 sites above 1% of the on-target editing rate, mostly in intronic or noncoding regions. The findings support a low apparent off-target profile for the two guides in the reported assays, but they do not establish in vivo or human safety.
Important limits remain. The experiment used a murine surrogate product made from mouse Sca1+ HSPCs, not a human product, and the four groups were modest in size; follow-up was limited to the reported 6.5 to 7 months. Cerebellar pathology and cardiac hypertrophy were not completely reversed. The findings therefore remain model-specific biological associations, not evidence of efficacy or safety in human patients.
Funding and data
The study reports California Institute for Regenerative Medicine support for IND-enabling PPL-001 preclinical studies and NIH grants including R01NS108965, R01NS135162 and R01AG086443. Its conflict statement says one author was a Papillon Therapeutics shareholder and board member, while another was a paid consultant; UC San Diego reviewed and approved the arrangement. RNA-seq data were deposited in the Gene Expression Omnibus under GSE326123, with custom analysis scripts and graph data values reported as available.
Paper data and sources
Original title: Edited HSPCs rescue frataxin deficiency and inflammation-driven multisystemic pathology in Friedreich’s ataxia mice
Authors: Sivakumar A, Corl A, Muñoz K et al.
Journal/Repository: The Journal of clinical investigation
Status: Peer-reviewed
First online: 2026-10-19
DOI: Not available
Original paper · Full text