STRC Research

STRC ASO Phase1 Splice-Switch Design

6 min read

STRC ASO Phase 1 Splice-Switch Design

Candidate splice-switching antisense oligonucleotides (ASOs) against STRC designed to block RBM24-driven skipping of the four rMATS-validated cassette exons (Sun et al. 2026). The prime target is E1 (human aa 1047–1102, GRCh38 chr15:43611148–43611315, 168 nt, in-frame, adjacent to the TMEM145 interface): strongest dPSI (−0.542) and a sharp splice-site signature (acceptor score 6.90, donor 8.70). Two 18-mer and two 20-mer ASOs nominated per exon per splice-site. A single high-conviction lead for E1: 5′-CCTTGCCACTCTCATACC-3′ (donor-blocker, Tm 65.6 °C, GC 56%, zero hairpin). E3 (human aa 712–732, 63 nt, frameshift-on-skip, NMD-triggering) is a secondary target where ASOs forcing inclusion rescue full-length stereocilin. Phase 1 specificity is exon-local only; whole-transcriptome BLAST off-target scan is Phase 2.

Motivation

RBM24 loss (Sun et al. 2026 PNAS, PMID 41973913) strips 40–54% of STRC transcripts at four cassette exons in mouse cochlea — see STRC RBM24 Regulatory Hypothesis and 2026-04-17-sun-rbm24-strc-splicing. For compound heterozygotes like Misha, any boost in correctly spliced STRC increases total stereocilin and may cross the HTC coupling threshold (STRC Stereocilia Bundle Mechanics Model). The RBM24 upregulation strategy carries off-target splicing risk (469 other genes regulated). ASOs that bind directly to STRC pre-mRNA at the RBM24-responsive splice junction get the same result without touching RBM24, avoiding the collateral disruption.

Method

  1. Retrieved canonical STRC transcript ENST00000450892 / protein ENSP00000401513 from Ensembl REST (29 exons, strand −1).
  2. Mapped rMATS event aa-ranges to GRCh38 genomic coordinates via /map/translation/{pid}/{aa} endpoint (corrected from initial cumulative-nt approach which ignored UTR boundaries).
  3. Fetched each target exon ± 80 nt of flanking intron with /sequence/region/human/{chr}:{start}..{end}:-1.
  4. Scored splice acceptor (23mer: intron-last-20 + exon-first-3, PWM around polypyrimidine tract + invariant AG) and splice donor (9mer: exon-last-3 + intron-first-6, PWM on CAG|GTRAGT consensus).
  5. Scanned RBM24 motifs (TGTGTG / GTGTGT / GCTCTTC, ENCODE RBNS ENCSR742AEU) across the full intron-exon-intron region.
  6. Generated ASO candidates at lengths {18, 20, 22} covering three windows per exon: acceptor (intron −25..exon +5), donor (exon −5..intron +20), exonic-RBM24 (±8 nt around strongest exonic hit).
  7. Filtered: GC% in 35–65, hairpin self-complement ≤ 5 nt, nearest-neighbor Tm computed with SantaLucia DNA:DNA parameters (proxy for RNA:DNA; fine for ranking).
  8. Ranked by Tm proximity to 55 °C + GC proximity to 48% + hairpin penalty.

Results

Exon mapping and splice-site scores

EventHuman aaEnsembl exonGRCh38 range (−strand)Exon ntIn-frame?AcceptorDonor
E11047–1102ENSE00003647299 (rank 13)43611148–43611315168✅6.908.70
E21311–1376ENSE00003667797 (rank 20)43604650–43604846197✅6.707.80
E3712–732ENSE00003551147 (rank 5)43614414–4361447663❌ frame-shift → NMD6.609.60
E41311–1376ENSE00003667797 (rank 20)43604650–43604846197❌6.707.80

E2 and E4 collapse to the same Ensembl exon — the mouse rMATS events are alternative 5′/3′ splice sites within that exon, not distinct exons. In human, this exon boundary mapping produces a single 197-nt target; any ASO here affects both events simultaneously. E3’s exon rank 5 at 63 nt closely matches the mouse E3 (70 nt) — confirms ortholog exon assignment.

RBM24 motif hits (TGTGTG / GTGTGT / GCTCTTC, ± 80 nt intronic flank)

  • E1: 0 motif hits inside exon or flanks.
  • E2 / E4: 1 hit in flanking intron, 0 exonic.
  • E3: 0 motif hits in exon or flanks.

Sparse — suggests RBM24 regulation at these four cassette exons acts through either (a) more degenerate secondary motifs not in our set or (b) distal intronic binding outside ±80 nt (Sun et al. 2026 did not publish per-event binding coordinates). Splice-site-blocker ASOs remain viable regardless — they hijack the spliceosome recruitment step directly, independent of which upstream regulator drove the skip.

Top ASO candidates (lead per exon and window)

EventWindowASO 5′→3′LengthTm (°C)GC %Hairpin
E1 (high)donor-blockerCCTTGCCACTCTCATACC1865.655.60
E1donor-blockerTTGCCACTCTCATACCCGAA2069.450.00
E1acceptor-blockerGCTAAAGAGCAAGAGAGA1861.744.40
E3 (high)donor-blockerATGTGCCCTCCACCTGTTAC2070.755.00
E3donor-blockerTGCCCTCCACCTGTTACC1869.761.10
E3acceptor-blockerCTGGTGGGAAGAAAGACAAT2066.245.00
E2 / E4acceptor-blockerGAGAAGAGCATCAGAACT1861.744.40
E2 / E4donor-blockerCGAAGGTCCATACCCAAG1865.655.60

Full 16-candidate CSV at aso_phase1_candidates.csv; all 56 candidates including 22-mers and exonic-RBM24-window options in JSON.

Lead rationale

  • E1 donor-blocker CCTTGCCACTCTCATACC: sits across the exon-13 / intron-13 boundary, covering the GT donor dinucleotide. Blocks U1 snRNP recruitment. 18-mer with 56% GC and zero hairpin is a standard PMO length.
  • E3 donor-blocker ATGTGCCCTCCACCTGTTAC: exon 5 donor splice site. E3 is a frameshift exon — forcing inclusion prevents a PTC at codon 773 that would otherwise trigger NMD (STRC E1 Conservation and NMD Motifs).

Interpretation for Misha

Phase 1 establishes the ASO design is buildable, not that it will work in DFNB16. Key caveats:

  • Genotype fit: Misha is compound het (98 kb deletion + E1659A). E1659A is on exon 29 (aa 1659 maps to chr15:43600550–43600552, far from E1/E2/E3). ASOs here do not touch E1659A — they act on the wild-type portion of the maternal allele. The deleted paternal allele is unreachable.
  • Expected effect: if RBM24 activity in Misha’s cochlea causes even 20% skipping at E1, blocking it recovers ~20% of full-length maternal STRC — potentially meaningful given the bundle-mechanics steep threshold (STRC Stereocilia Bundle Mechanics Model). If baseline skipping is zero, the ASO does nothing useful.
  • Compatibility with gene therapy: ASOs are repeatable, short-halflife, and orthogonal to AAV. Combining ASO (boost endogenous) + mini-STRC AAV (ectopic expression) is mechanism-additive.

Limitations

  • Phase 1 specificity check is local only; off-target against human transcriptome (Phase 2 BLAST) is mandatory before wet lab. A 20-mer has ~10⁻¹² random match probability but the transcriptome is 10⁸ bases — expect real off-targets.
  • Mouse → human exon orthology inferred from aa range intersection, not direct liftOver. E2 / E4 ambiguity (cassette exon with alternative splice sites) is not resolved at the mouse coordinate level.
  • Splice-site scoring is PWM-lite (not MaxEntScan). Use for relative ranking, not absolute strength claims.
  • Tm calculated with DNA:DNA nearest-neighbor parameters as an RNA:DNA proxy; real RNA:DNA Tm is 3–5 °C higher. All values should be read as relative, not absolute.
  • RBM24-binding architecture on STRC pre-mRNA is not published at nucleotide resolution. Our ± 80 nt scan is a first pass; Sun et al. eCLIP data (if released) would replace it.

Next steps

  1. Phase 2 — off-target BLAST scan of each candidate against RefSeq human transcriptome. Drop any ASO with ≥ 15-nt contiguous match to an unintended transcript.
  2. Secondary structure around splice sites via RNAfold / ViennaRNA: if an ASO target region is deep inside a stable hairpin, accessibility is poor → Tm alone overestimates binding.
  3. Gapmer vs PMO chemistry decision depending on cochlear delivery route (PMO favored for intratympanic or round-window administration; gapmer needed if RNase H cleavage is desired for an exonic site).
  4. In-vitro splicing reporter assay at Jeffrey Holt lab: HEK minigene carrying STRC exons 12–14 (covering E1), ASO titration, RT-PCR readout. ~$5–8 k, 4–6 weeks.

Replication

cd ~/STRC/models
/opt/miniconda3/bin/python3 aso_phase1_design.py
# outputs: aso_phase1_design.json + aso_phase1_candidates.csv

Files / Models

  • ~/STRC/models/aso_phase1_design.py — fetch, splice-site score, RBM24 motif scan, candidate generation + ranking
  • ~/STRC/models/aso_phase1_design.json — full candidate list with per-event metrics
  • ~/STRC/models/aso_phase1_candidates.csv — 16-row shortlist for lab order

Connections

Graph View

Backlinks