# SRSF1 and healthy aging: evidence-audited reanalysis and safety-window hypothesis

**OERSI Research Note 01 · 2 November 2026**

**Evidence status:** secondary computational reanalysis of a published preclinical study; hypothesis-generating. This is not a new experimental discovery, independent biological replication, clinical result, or treatment recommendation.

## Decision

Prioritize SRSF1 as a high-value research lead for healthy-aging biology, but do not pursue indiscriminate upregulation. The useful question is whether a dose-, time- and tissue-specific induction window can preserve reported aging-associated benefits without provoking stress, inflammatory or oncogenic-context signals.

## Why this lead matters

Plesa and colleagues reported that SRSF1 overexpression shifted several cellular aging-associated readouts and, in an inducible aged-mouse experiment, was associated with 44% lower frailty progression and a 14% increase in median lifespan. The mouse groups were small (approximately 9–10 animals per group). These are preclinical results and do not establish human benefit or safety.

Primary paper: https://doi.org/10.1038/s43587-026-01234-8  
Public RNA-seq data: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE325735

## What OERSI reanalysed

We reanalysed the public human raw-count matrix for the 3-day, 2,000 ng/mL doxycycline condition, comparing SRSF1-transgene samples with mTagBFP2/BFP controls.

- 24 selected raw-count columns: 12 SRSF1 and 12 BFP.
- Six independent donor-derived normal human dermal fibroblast lines.
- Two technical replicates per group within each donor. Donor line, not technical library, was treated as the independent unit.
- Sample aliases were mapped directly from the article's Supplementary Table 15, sheet 3a, to raw-count matrix column aliases. The 24 selected aliases were each matched to exactly one GEO sample using donor line, transgene, passage, replicate, dose and treatment duration metadata.
- A targeted 18-gene panel was analysed using log2(CPM + 0.5). Technical replicates were averaged within each donor and group, then paired donor-level contrasts were calculated.
- The authors' own DESeq2 results in Supplementary Table 15, sheet 3e, were extracted as a separate methodological cross-check.

This is a secondary analysis of the originating study's data, not a new cohort or independent biological replication.

## Main result: a mixed expression signature

The authors' DESeq2 table reports the following fold changes for the SRSF1 perturbation. The OERSI column is a separate exploratory CPM-based calculation on the directly mapped 24-sample subset; estimates differ because the methods and data subsets are not identical.

| Marker | Authors' DESeq2 fold change | Adjusted p-value | OERSI CPM fold change | Direction across six donor lines |
|---|---:|---:|---:|---|
| CXCL8 / IL-8 | 4.98× up | 1.02 × 10⁻⁵ | 2.61× up | up in 6/6 |
| CDKN1A / p21 | 1.81× up | 2.92 × 10⁻⁴ | 2.21× up | up in 6/6 |
| CDKN2A / p16 | 0.238× | 9.69 × 10⁻⁷ | 0.486× | down in 6/6 |
| SERPINE1 / PAI-1 | 0.466× | 2.88 × 10⁻⁴ | 0.550× | down in 6/6 |
| IL1R1 | 0.556× | 6.20 × 10⁻⁵ | 0.657× | down in 6/6 |
| MYC | 0.647× | 0.00346 | 0.837× | down in 6/6 |
| COL3A1 | 0.338× | 3.04 × 10⁻⁶ | 0.369× | down in 6/6 |
| COL1A1 | 0.422× | 0.0610 | 0.509× | down in 6/6; author FDR threshold not passed |
| IL1B | 0.939× | 0.845 | 1.008× | no consistent direction |
| IL6 | 0.637× | 0.133 | 0.766× | no consistent direction |

The authors' values are from their Supplementary Table 15, sheet 3e. The OERSI values are a separate exploratory log2(CPM + 0.5) calculation on the directly mapped 24-sample subset, not a replacement for DESeq2.

The profile is not a uniform “younger is better” signature. CXCL8 and CDKN1A rise in this condition, while several senescence-associated and extracellular-matrix markers fall. The CXCL8 and p21 changes are worth investigating as possible dose-/state-specific stress or inflammatory responses. They do not, by themselves, prove harm, senescence, tumour development or a clinical effect.

The OERSI directional counts and sign tests are exploratory (six donor lines, targeted panel, no multiple-testing correction) and must not be treated as confirmatory statistics. The authors' DESeq2 adjusted p-values are reported as published; they are not an independent dataset.

## Important interpretation of SRSF1 RNA

Gene-level SRSF1 RNA was approximately 0.071× in the authors' DESeq2 table and 0.072× in our CPM reanalysis. This must **not** be interpreted as proof that total SRSF1 protein was lower or that the transgene failed. SRSF1 is subject to negative autoregulatory splicing; ectopic cDNA expression can suppress endogenous transcript production. RNA abundance, exogenous transgene protein and total functional SRSF1 activity are different measurements.

Relevant work:
- Sun et al., 2010: https://pubmed.ncbi.nlm.nih.gov/20139984/
- Cell Reports, 2022: https://doi.org/10.1016/j.celrep.2022.110985

## Safety context cannot be ignored

A 2023 *Cancer Discovery* study reported that increased SRSF1 can promote pancreatitis and accelerate KRASG12D-driven pancreatic cancer in a specific experimental context, involving IL1R1/MAPK signalling.

Reference: https://pmc.ncbi.nlm.nih.gov/articles/PMC10330071/ · DOI 10.1158/2159-8290.CD-22-1013.

This does not prove that the fibroblast intervention causes cancer or that SRSF1 is generally unsafe. It does establish that tissue, dose, duration and oncogenic context must be explicit safety variables. The CXCL8 signal in this fibroblast condition deserves follow-up rather than dismissal.

## Research decision for Darkoza/OERSI

**Study the SRSF1 safety window, not SRSF1 maximization.**

1. Reanalyse the complete public dose/time series with explicit sample alias mapping, batch-aware count modelling and isoform-level analysis.
2. Pre-register the comparison, primary endpoints, exclusion rules and statistical thresholds before examining further conditions.
3. In independent cell systems, measure SRSF1 protein and isoforms, secreted IL-8/CXCL8, p21/p16 proteins, proliferation, viability and functional senescence readouts.
4. Compare multiple donor lines and tissue contexts; stratify or separately test KRAS/MYC backgrounds with appropriate biosafety and ethics review.
5. Advance only after independent replication and a demonstrated separation between desired aging-associated outcomes and adverse stress/oncogenic signals.

A useful success criterion is not “SRSF1 went up.” It is a reproducible, pre-specified benefit signal with no unacceptable increase in inflammatory, stress, proliferative or oncogenic endpoints across the contexts being considered.

## What this report does not claim

- No new scientific law or original experimental discovery was produced.
- No independent biological replication was performed.
- No human treatment, gene therapy, life-extension protocol or self-experiment is supported.
- RNA expression does not establish protein activity, organism-level safety, cancer incidence or clinical benefit.
- The study's small animal groups and the six-line secondary analysis limit generalization.

## Competing interests

The primary paper's disclosure states that several authors are inventors on patent US20250313599A1 and that Plesa and Church are cofounders/equity holders in RegulaBio. This does not invalidate the findings; it is material context for independent replication and interpretation.

## Reproducibility artifacts

- Public summary JSON
- Authors' DESeq2 cross-check JSON
- GEO mapping QA summary
- Reanalysis source
- DESeq2 cross-check source
- GEO crosswalk validation source
- Public experiment API record: https://darkoza.com/science-api/v1/experiments/srsf1-aging-safety-window.json

The raw-count data remain at GEO; the article's supplementary workbook remains with the publisher. This report does not mirror either input dataset.

## References

1. Plesa AM, Jung S, Wang HH et al. “Transcriptomic reprogramming screen identifies SRSF1 as a regulator of aging.” *Nature Aging* (2026). DOI 10.1038/s43587-026-01234-8.
2. GEO. GSE325735 public RNA-seq series.
3. Wan L et al. “Splicing Factor SRSF1 Promotes Pancreatitis and KRASG12D-Mediated Pancreatic Cancer.” *Cancer Discovery* (2023). PMCID PMC10330071.
4. Sun S et al. SRSF1/SF2-ASF autoregulation. *Nature Structural & Molecular Biology* (2010). PMID 20139984.
5. “Dynamics and functional roles of splicing factor autoregulation.” *Cell Reports* (2022). DOI 10.1016/j.celrep.2022.110985.