{"label":"Human MDM4 local RNA alternatives during splice-modulator exposure — methods and results","axis":"A13","date":"2026-11-03","summary":"Targeted reanalysis of 15 libraries from one human GM03813 fibroblast line separates 68-nt, 39-nt and full-skip local MDM4 RNA paths. The largest existing model-based inclusion difference is −20.07 percentage points with Branaplam 40 nM (original JC FDR 0.02012). These data do not identify complete sample transcripts, measured proteins or SMA rescue.","sections":[{"heading":"Research question and contribution","text":"This is a targeted reanalysis of previously published RNA-seq data from the SMA-derived fibroblast line GM03813. It asks which local RNA alternatives contribute to a defined MDM4 inclusion measurement, and whether that measurement tracks the gene-abundance response of MDM2, CDKN1A and BBC3 across the available exposure conditions. It is not a new experiment, a neuronal validation or a therapeutic efficacy study.\n\nThe established SMN–MDM–p53 connection is prior published work, including Van Alstyne, Simon et al. 2018. The present contribution is the source-linked identification of the human local event, inspection of existing alignment evidence and reconstruction of distinct annotated coding contexts."},{"heading":"Source data and observation units","text":"The source is Ottesen et al. 2023, study SRP334251: 15 GM03813 libraries after 24 hours, with three deposited libraries in each of five conditions. All four treatment comparisons share the same three DMSO libraries. The deposited library count does not establish independent donors or independently mapped culture units. Alignment and genome-wide event-model outputs were already available; the targeted follow-up did not realign the study."},{"heading":"Exact local event and assay specificity","text":"In GENCODE v33 / GRCh38, the positive-strand 68-nt segment is chr1:204537430–204537497, within MDM4 ENSG00000198625.13. A 39-nt alternative starts at chr1:204537459 and ends at the same downstream boundary. These forms share a downstream junction.\n\nThe primary human primer/probe context comes from Dewaele et al., JCI 2016, supplementary table ST6. Reference reconstruction supports products of 162 nt for the 68-nt context, 133 nt for the 39-nt context and 94 nt for the complete skip. The checked inclusion probe is compatible with both included contexts. Consequently, the SE-model inclusion value below cannot be read as the proportion of complete transcripts carrying only the 68-nt form. No new PCR assay or protein measurement was performed."},{"heading":"Existing event-model results","text":"DMSO: Mean model-based inclusion 71.19%; Difference from DMSO reference; Original genome-wide JC FDR —. Branaplam 2 nM: Mean model-based inclusion 69.54%; Difference from DMSO −1.65 percentage points; Original genome-wide JC FDR 1.000. Branaplam 40 nM: Mean model-based inclusion 51.12%; Difference from DMSO −20.07 percentage points; Original genome-wide JC FDR 0.02012. Risdiplam 50 nM: Mean model-based inclusion 64.75%; Difference from DMSO −6.44 percentage points; Original genome-wide JC FDR 1.000. Risdiplam 1000 nM: Mean model-based inclusion 64.07%; Difference from DMSO −7.12 percentage points; Original genome-wide JC FDR 1.000.\n\nAll three Branaplam 40 nM library values are below all three DMSO values. Single-library omission gives descriptive differences of −25.71 to −17.20 percentage points for that comparison. These ranges are not confidence intervals. FDR values are retained from the existing genome-wide rMATS v4.3.0 outputs; the statistical model was not independently refitted in this follow-up. JC and JCEC contain identical counts and inclusion values here and are not independent confirmations; the JCEC FDR for Branaplam 40 nM is 0.01962. FDR 1 in another condition is not evidence of equality.\n\nThe separate A3SS inspection supports the alternative 39-nt acceptor context. The SE result must therefore remain distinct from a direct 68-versus-39-nt comparison. Event selection was a retrospective, literature-guided follow-up, and the high-Risdiplam row had been seen before extraction of all four comparisons. It is not a blinded validation."},{"heading":"Alignment evidence and distinct coding references","text":"Across the 15 existing BAMs, 1,447 qualifying alignment records support local paths through the 68-nt segment, the 39-nt alternative and the complete skip. Alignment records are not independent molecules; local short-read paths do not determine complete transcript structures or expressed proteins.\n\nThe existing HISAT2 2.2.1 / GENCODE-v33 alignments were queried only in the 0-based half-open interval chr1:204532180,204538320). Primary mapped records were required to have MAPQ ≥20 and an explicit NH=1 tag; secondary, supplementary, QC-failed and duplicate-flagged records were excluded. An exact CIGAR-N intron needed at least eight immediately adjacent aligned bases on both sides. Adjacent insertions/deletions were not counted as matching anchor bases. These are technical readout rules, not universal scientific quality cutoffs.\n\nThe exact 0-based half-open introns were [204532246,204537429) for entry to the 68-nt segment, [204532246,204537458) for entry to the 39-nt segment, [204537497,204538208) for their shared downstream junction and [204532246,204538208) for the complete skip. Within-read-name groups, overlapping mate support was counted once. Combining two mates required individually qualifying, properly paired reads with distinct mate designations. Conflicting local alternatives were not assigned to a unique path.\n\nThe following table reports local support after read-name grouping, not the total 1,447 alignment records. Each triplet lists the three source libraries in this retained order from [the original SRA study: DMSO, SRR15622470 / SRR15622469 / SRR15622463; Branaplam 2 nM, SRR15622466 / SRR15622465 / SRR15622464; Branaplam 40 nM, SRR15622456 / SRR15622468 / SRR15622467; Risdiplam 50 nM, SRR15622459 / SRR15622458 / SRR15622457; Risdiplam 1000 nM, SRR15622462 / SRR15622461 / SRR15622460. Support is neither length-normalized isoform abundance nor an estimated treatment effect.\n\nDMSO: 68-nt local path 12 / 22 / 19; 39-nt local path 4 / 0 / 5; Direct skip 19 / 17 / 13; Shared downstream only under the fixed rules 18 / 40 / 26. Branaplam 2 nM: 68-nt local path 13 / 24 / 12; 39-nt local path 2 / 0 / 0; Direct skip 10 / 16 / 16; Shared downstream only under the fixed rules 17 / 27 / 21. Branaplam 40 nM: 68-nt local path 13 / 11 / 8; 39-nt local path 2 / 1 / 4; Direct skip 20 / 36 / 18; Shared downstream only under the fixed rules 10 / 19 / 13. Risdiplam 50 nM: 68-nt local path 13 / 21 / 17; 39-nt local path 1 / 1 / 3; Direct skip 15 / 18 / 21; Shared downstream only under the fixed rules 9 / 24 / 19. Risdiplam 1000 nM: 68-nt local path 18 / 15 / 10; 39-nt local path 0 / 2 / 1; Direct skip 14 / 21 / 19; Shared downstream only under the fixed rules 18 / 15 / 16.\n\nA zero denotes no fully classifiable local path under these rules, not absence of the RNA. The downstream-only category is the residual assignment under the anchor rules: an additional short-anchored alternative may still be present in a record. A sensitivity readout using MAPQ ≥30 and ten-base anchors was derived from the same pass. Some path counts change, including loss of single-read long-path support in individual libraries while properly paired support remains. No treatment p-value was calculated from this readout, and SE/A3SS FDRs are not transferred to these counts. Read-name grouping is not UMI-based molecule counting.\n\nThe annotated 39-nt context ENST00000621032.4 / ENSP00000482479.1 reconstructs a 153-aa reference product. Explicit computational removal of the first 29 nt of the 68-nt segment from the 490-aa ENST00000367182.8 reference yields that same 153-aa sequence. Removing the complete 68-nt segment from the same reference instead yields a different 140-aa predicted product. These are coding-reference comparisons, not newly measured sample proteins or experimentally made constructs. ENST00000463049.5 has no CDS in this annotation; no protein was assigned to it.\n\nThe 39-nt / 153-aa reference must not be equated with every product called MDM4-S. Earlier studies use different species, exon conventions or constructs. McEvoy et al. 2012 provides human MDM4-S context, but does not establish function for this exact 29-nt-shortened reference. A bounded literature search did not locate a sequence-bound functional intervention for that exact form; this is not proof of novelty or absence of earlier work."},{"heading":"Interpretation and unresolved questions","text":"The larger SE inclusion difference occurs with Branaplam 40 nM, whereas the stronger combined MDM2/CDKN1A/BBC3 gene-abundance response occurs with Risdiplam 1000 nM in the same source study. This does not support a simple explanation in which this single event alone accounts for the relative marker response. It neither excludes MDM involvement nor establishes an alternative causal pathway. Concentrations are not potency-equivalent, and gene abundance is not p53 activity.\n\nThe exact human MDM2 event-source bridge, complete RNA and protein identities, culture independence and biological function remain unresolved. The data do not demonstrate neuronal rescue, clinical benefit, safety, or a new therapeutic hit. The concrete interpretive question is whether distinguishing the local MDM4 alternatives changes how the human fibroblast inclusion signal should be read before any transfer to a neuronal SMA context."}],"references":[{"label":"Ottesen et al. 2023: original GM03813 RNA-seq study","url":"https://pmc.ncbi.nlm.nih.gov/articles/PMC10325915/","type":"source"},{"label":"SRP334251: deposited library accessions","url":"https://www.ncbi.nlm.nih.gov/Traces/study/?acc=SRP334251","type":"source"},{"label":"Dewaele et al. 2016: human MDM4 primer and probe context","url":"https://www.jci.org/articles/view/82534","type":"source"},{"label":"Van Alstyne, Simon et al. 2018: SMN–MDM–p53 context","url":"https://pmc.ncbi.nlm.nih.gov/articles/PMC6075148/","type":"source"},{"label":"McEvoy et al. 2012: human MDM4-S context","url":"https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0042739","type":"source"}],"id":"ec91ba2d55c4e7977332","originalDate":"2026-11-03","updatedAt":"2026-11-03T11:16:41.442448+00:00"}