henhao
Sign in
NeuroscienceNucleic Acids Research
Peer-reviewed journal article

Divalent siRNA for prion disease

Juliana E GentileiD, Taylor L CorridoniD, Fiona E. SerackiD, Dimas EcheverriaiD, Zachary KennedyiD, Corrie L. Gallant‐BehmiD, Matthew R. HassleriD, Garth A. Kinberger, Margaret N Kelemen, Nikita G Kamath, Yuan LianiD, Katherine Y. GrossiD, Rachael MilleriD, Kendrick DeSouza-Lenz, Michael Howard, Kenia GuzmaniD, Nathan Chan, Vanessa LaversenneiD, Daniel T Curtis, Kevin J. Fettes, Marc M. Lemaitre, Aimee L. JacksoniD, Ken YamadaiD, Julia F. AltermaniD, Alissa A. CoffeyiD, , Sonia M. VallabhiD

Abstract

Prion protein (PrP) lowering is effective in animal models of prion disease and is being tested clinically in prion disease patients, but there remains a need for more potent PrP-lowering drug candidates. Inspired by the reported potency and duration of action of divalent short interfering RNA (siRNA), a new oligonucleotide drug modality for the central nervous system, we sought to discover and develop a new PrP-lowering drug candidate. Herein we identify a mouse Prnp-targeting divalent siRNA molecule, 1682-s4, that lowers PrP to 49% residual brain expression in wild-type mice, and, in the context of intracerebral infection with Rocky Mountain Laboratory (RML) prions, achieves a 2.7-fold increase in survival time with pre-symptomatic chronic treatment and 64% increase in survival time with a single dose after symptom onset. We describe the generation of two transgenic mouse lines, Tg25109 and Tg26372, expressing the full human PRNP gene and its non-coding sequence, and demonstrate their utility for in vivo discovery of potent human PRNP-targeting oligonucleotides. We discover siRNA sequence 2439 against human PRNP and compare its potency in different divalent siRNA chemical scaffolds. We determine that both the fixed UU tail and extended nucleic acid linkages of scaffold s4 contribute to superior potency compared to other scaffolds tested, offering 9.4 and 15.9 percentage points respectively of additional PrP knockdown. A single dose of 348 {micro}g of 2439-s4 lowered whole brain hemisphere human PrP in transgenic mice to 17% residual after 30 days, while 52 {micro}g lowered PrP to 49% residual. 1-2% of the dose of 2439-s4 delivered into cerebrospinal fluid is retained in the brain, and the median effective tissue concentration is estimated at 1.2 micrograms per gram of tissue. Good Laboratory Practices toxicology studies identified no significant liabilities, and the U.S. FDA has cleared an Investigational New Drug application to bring 2439-s4 into clinical trials.

PDFDOICodeData
Full-text digest ↓

02 /Specification & Classification

Click any tag to browse matching papers
Citations
28
FWCI
45.75
Field percentile
top 1%
OpenAlex Topics

In-depth summary

✦ Multi-agent verified full-text synthesis

Claude Sonnet 4.6 · xhigh effort · 8 Nov 2026 · from the full text of version 6 (5 Mar 2026) · CC BY 4.0

Divalent siRNA 2439-s4 lowers human PrP to 17% residual in transgenic mouse brain after one dose, and a separate mouse Prnp-targeting siRNA extends survival in prion-infected mice.

Lowering PrP by about half delays prion disease in mice but never prevents death, so deeper knockdown is needed, and only one ASO has reached clinical testing. This work develops a divalent siRNA candidate that reaches deeper, longer-lasting lowering, and the FDA has cleared an IND for it.

Key results

  1. ShownA mouse Prnp-targeting divalent siRNA (1682-s4) lowered brain PrP by about half and extended survival in RML prion-infected wild-type mice, whether dosed pre-symptomatically or after symptom onset.

    1682-s4 gave 49.4% residual PrP. Chronic dosing from 75 dpi: median survival 442 vs 165 dpi (2.7-fold, P = 0.0002, log-rank). Single dose at 126 dpi: median 270 vs 164 dpi (P = 0.0002). N = 6 (early) and 8 (late) for 1682-s4. · Proof of concept in a prion disease model; Figure legends

  2. ShownTwo BAC transgenic lines carrying the full human PRNP gene were generated; Tg26372, which expresses high human PrP, was used for most in vivo screening, where sequence 2439 was the most potent among those tested.

    Tg26372 has 20 transgene copies and 5.4-fold wild-type PrP. In cells, 34 (40%) of 84 compounds gave <10% residual PRNP at 2.0 µM. In vivo screen: 17 groups, 120 Tg26372 homozygotes, 348 µg. In one experiment, regional PRNP RNA residual ranged from 9.6% (thalamus) to 19.9% (cerebellum) versus 28.3% for whole-hemisphere PrP. · Generation of human PRNP transgenic mice; Identification of compounds targeting human PRNP

  3. ShownBoth the fixed 3' UU antisense tail and the terminal exNA/low-phosphorothioate chemistry independently improved potency of the 2439 sequence, making 2439-s4 the lead scaffold.

    Linear model on regional PRNP RNA: s1 fixed vs matched tail 9.4 percentage points deeper (P = 0.0017); s4 another 15.9 beyond s1 fixed tail (P = 5.0e-11). 2439-s4 gave 29.5% residual PrP at 174 µg; N = 8 per group, one sequence only. · Comparison of chemical scaffolds and antisense…; Abstract

  4. Shown2439-s4 shows dose-dependent, durable target engagement with a tissue IC50 of 1.2 µg/g, and deep PrP lowering after a single dose in transgenic mice.

    PD/PK model tissue IC50 1.2 µg/g; 1-2% of dose retained in brain at 30 days. A 348 µg dose gave 17% residual PrP and 52 µg gave 49%. Single 348 or 174 µg doses were active to at least 4 and 6 months. A day 0 and 7 loading regimen of 52 µg gave 5/6 regions below 25% residual RNA. · Durability and dosing regimens for drug…; Abstract; Discussion

  5. ShownSingle-dose intrathecal GLP toxicology in dogs and rats and in vitro safety assays showed no significant liabilities, supporting an FDA-cleared IND.

    Dogs received 0, 20, 60 or 200 mg and rats 0, 0.3, 1 or 3 mg, with necropsy at 1 and 28 days and no adverse findings. No Cyp interaction with IC50 <400 nM; maximum 27.2% OAT3 inhibition; about 1% of dose retained in brain; deep dog regions below the 1.2 µg/g IC50. · Investigational New Drug (IND)-enabling studies; Discussion

Methods, data and reproducibility

Cholesterol-conjugated siRNAs were screened in cell lines, then divalent siRNAs were dosed by ICV injection into wild-type and human PRNP BAC-transgenic mice, with PrP ELISA, RT-qPCR, PK/PD modeling, prion survival, and GLP toxicology in rats and dogs.

  • Cell screens of siRNAs: mouse N2a (20 Prnp sequences), human A549 (24 compounds), human U251-MG glioblastoma (84 compounds) · mouse; human · n = 20, 24 and 84 sequences; triplicate wells
  • Wild-type C57BL/6N RML prion survival studies with 1682-s4, PBS and non-targeting controls · mouse · n = early: 6 to 8 per treated group; late: 8 to 9 per group
  • Human PRNP BAC transgenic lines Tg26372 and Tg25109 on ZH3 PrP knockout background; PrP expression and copy number · mouse (human PRNP transgene) · n = n = 3-6 per group · RRID:MMRRC_075939-UCD (Tg26372); RRID:MMRRC_075940-UCD (Tg25109)
  • In vivo sequence and scaffold screen in Tg26372 homozygotes at 348 µg · mouse (human PRNP transgene) · n = 120 animals, 17 groups
  • Scaffold and 3' tail in vivo dose-response (7, 35, 174 µg) with whole-hemisphere ELISA and regional RT-qPCR · mouse · n = N = 8 per group
  • GLP intrathecal toxicology with biodistribution · dog (beagle); rat (Sprague-Dawley) · n = dogs 2M/2F per group at day 28; rats 6M/6F per non-zero group
Methods & tools
R 4.2.0 custom scripts, drc package (4-parameter Hill dose-response), Linear model (residual ~ scaffold + log(dose) + region), Log-rank test, Dunnett's test, QuantiGene branched DNA assay, TaqMan RT-qPCR (ddCt), In-house PrP sandwich ELISA, PNA-HPLC drug quantification, Targeted locus amplification, NCBI BLAST megablast (refseq_select), Python script written with ChatGPT
Code
https://github.com/ericminikel/divalent
Data & accessions
https://github.com/ericminikel/divalent
Software & environment
R 4.2.0 custom scripts with the drc package; no container or lockfile described.
Reproducibility gaps
Repository holds animal-level data, analysis code and the IND text; the transgene plasmid sequence and characterization reports are also said to be there. Mouse lines are at MMRRC. No compute details given.
Limitations (6)
  • The scaffold comparison used a single sequence (2439) and may not generalize to other sequences. · stated by the authors
  • Hypochromicity of duplexed siRNA means administered doses may differ between scaffolds (s2 possibly 14% lower), so relative potency should be interpreted cautiously; one correction factor was applied to all compounds. · stated by the authors
  • Survival benefit of the deeper PrP lowering from 2439-s4 was not measured: it does not target mouse Prnp, and no human prion challenge in humanized mice was done. · stated by the authors
  • 2439-s4 is not sequence-matched to rat or dog PRNP, so biodistribution and target engagement were not assessed in a pharmacodynamically relevant toxicology species, and uniform brain exposure remains a challenge for intrathecal oligonucleotides. · stated by the authors
  • The ELISA overestimates residual PrP: a sample mimicking 10% residual read about 17%, so reported lowering values are conservative and floor-limited. · stated by the authors
  • Survival cohorts were small (6 to 9 per treated arm) and used one tool sequence, and the Tg26372 model overexpresses PrP 5.4-fold, which may limit translation of potency to human brain. · noted in this summary

Who should read:Read this if you develop oligonucleotide therapeutics for CNS disease and want in vivo scaffold comparisons (exNA, fixed UU tail) plus the IND-enabling toxicology data for a PrP-lowering siRNA.

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISGentile '26Minikel '20Corridon '26Belgrad '23Liu '26Chou '25
79% semantic matchNeuroscience
Juliana E Gentile, Jonathan E. Farley, Margaret N. Kelemen+29 more·bioRxiv·Oct 24, 2026