henhaobio
Sign in

bioRxivCancer BiologyPreprint

An integrated, transcript-resolution atlas of prostate cancer progression unmasks associations with oncogenic pathways

Jichang Zhang, Claudio LorenziiD, Daniela BossiiD, Simone Nicastri, Matteo Corpetti, Nicolò Formaggio, Jacopo SgrignaniiD, Andrea CavalliiD, Marco BolisiD, Jean-Philippe Paul. TheurillatiD

PDFDOIData

In-depth summary

Claude Sonnet 4.6 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (22 Sept 2026) · CC BY 4.0

A transcript-resolution prostate cancer atlas, built by uniformly reprocessing public RNA-seq cohorts, links isoform switches in cancer genes to oncogenic pathway activity and some to CRPC survival.

Existing prostate cancer resources harmonize cohorts only at gene or cell-type level, or reach transcript resolution without cross-study harmonization, so isoform-pathway links could not be mined systematically. This atlas combines transcript resolution, harmonization and full disease-stage coverage to nominate isoform switches.

Key results

  1. ShownUniform FASTQ reprocessing separated disease stages along a pseudotime trajectory without extra batch correction. Single-end libraries caused a layout batch effect and were removed for transcript-level work. New external hormone-sensitive samples projected between primary tumors and early CRPC.

    Gene-level atlas: 1,365 samples from 14 studies. Transcript-level atlas: 1,140 paired-end samples from ten datasets after dropping 225 single-end samples. Transcript-level adjusted Rand index was lower than at gene level. · A reprocessed, harmonized atlas resolves disease…; Extending the atlas to transcript resolution; The atlas accurately contextualizes newly…; Clinical RNA-sequencing cohorts and metadata

  2. ShownSalmon short-read quantification agreed reasonably with matched MAS-Seq long-read data in prostate cancer cell lines, with weaker agreement for lncRNAs.

    Long-read vs short-read Pearson R = 0.8–0.83 for genes and 0.65 for transcripts. For lncRNAs, R = 0.59 at gene level and 0.41 at transcript level. Panel: LNCaP, VCaP, 22Rv1, PC3, H660; two biological replicates each. · Extending the atlas to transcript resolution; Matched long-read and short-read RNA sequencing; Long-read/short-read concordance

  3. ShownCanonical isoforms drove most progression-associated changes among protein-coding transcripts, and AR-V7 was the only non-canonical transcript that progressively dominated among upregulated ones. Pathway-linked isoform switches were catalogued, with oncogenic C6 signatures mostly in metastatic disease.

    A switch was called when two transcripts of one gene had opposite-sign correlations with a target, one above r 0.5 and the other below r −0.5. Manuscript subset: 41 genes (34 prespecified candidates plus ten top delta-correlation genes; three overlapped). Descriptive correlations, no significance test for the subset. · Canonical transcripts dominate…; Isoform switches nominate oncogenes and tumor…; Identification and prioritization of…

  4. ShownThe short RNF43 isoform (RNF43-207) boosted WNT3A signaling and gained little from R-spondin-1 in a reporter assay, whereas canonical RNF43-208 repressed signaling. High short-isoform expression tracked worse overall survival in CRPC.

    TOPFlash, five biological replicates, paired t tests with Šidák correction, adjusted p under WNT3A: control vs 208 0.0022; 208 vs 207 0.000371; R-spondin-1 effect within 208 0.0109 vs within 207 0.167 (ns). CRPC survival: 97 tumors, 96 deaths, nominal log-rank. · RNF43 isoform switching uncouples WNT restraint…; RNF43 TOPFlash reporter assays; Survival analysis; Figure legends

  5. ShownCanonical TRIM11-201 lowered androgen receptor protein through proteasomal degradation requiring the SPRY domain, while the RING-lacking TRIM11-204 raised AR protein and bound AR, nominating AR as a TRIM11 substrate.

    In HEK293T-AR cells, 204 raised and 201 lowered AR protein. The quantification panel carries significance marks of **p < 0.01 and ns by one-way ANOVA, from at least three experiments. Cycloheximide chase showed faster AR turnover with 201, rescued by MG132. SPRY deletion abolished the 201 effect. · TRIM11 isoform switching uncovers a RING-and…; TRIM11 immunoblotting, co-immunoprecipitation…; Figure legends

Methods, data and reproducibility

Raw FASTQs from 14 bulk RNA-seq studies were reprocessed with nf-core/rnaseq (STAR, Salmon, GENCODE v39). Transcript usage was correlated with slingshot pseudotime and ssGSEA pathway scores to nominate isoform switches, followed by selective cell-based and long-read validation.

  • Harmonized gene-level prostate cancer RNA-seq atlas (normal, primary, hormone-sensitive, CRPC) from public studies · human · n = 1,365 samples from 14 studies
  • Paired-end transcript-level atlas subset used for isoform analyses · human · n = 1,140 samples from ten datasets
  • External cohort of primary and hormone-sensitive de novo metastatic prostate cancer, projected onto the atlas · human
  • Matched Kinnex/MAS-Seq long-read and short-read RNA-seq of prostate cancer cell lines (LNCaP, VCaP, 22Rv1, PC3, H660) · human · n = two biological replicates per cell line · PRJNA1519081
  • LNCaP BRCA1-Δ11q overexpression vs empty-vector RNA-seq · human · n = 3 biological replicates per condition · PRJNA1519081
  • BRCA1-Δ11q BioID vs GFP BioID proteomics in LNCaP cells · human · n = four samples per group
  • Survival subsets: primary tumors (progression-free) and CRPC (overall survival) · human · n = 359 primary tumors (35 events); 97 CRPC tumors (96 deaths)
  • RNF43 TOPFlash reporter assays in HEK293T cells · human · n = five biological replicates
Methods & tools
nf-core/rnaseq v3.6, STAR v2.6.1d, Salmon v1.5.2, tximport, GENCODE v39, slingshot, GSVA ssGSEA, ComBat-seq, DESeq2, AlphaFold 3 and UCSF ChimeraX v1.12, MaxQuant v1.6.7.0, PacBio Kinnex Iso-Seq (SMRT Link v26.1.0.284828)
Reagents & environment
nf-core/rnaseq v3.6 (public cohorts) and v3.15.1 with Singularity (new RNA-seq); STAR v2.6.1d; Salmon v1.5.2; GENCODE v39; SMRT Link v26.1.0.284828; ChimeraX v1.12. Workstation package inventory is not the historical environment of every run.
Assay conditions & compute
BRCA1 perturbation RNA-seq processing used a maximum of 16 CPUs and 100 GB memory.
Reproducibility gaps
Analysis code package is not yet public (release promised by publication). SRA BioProject PRJNA1519081 is private until publication. Raw LC–MS/MS files are available on request. Controlled source cohorts remain under provider access rules.
Limitations (5)
  • Transcript quantification relies on short reads, which cannot always resolve isoforms unambiguously; concordance with long reads was weaker for lncRNAs, likely because of missing poly(A) tails. · stated by the authors
  • Transcript-pathway associations are correlative and come from bulk tissue, so cell-intrinsic isoform switching cannot be separated from shifts in cellular composition. · stated by the authors
  • Functional validation is deliberately limited and punctual (RNF43, BRCA1, TRIM11); broader validation in additional cell lines and patient-derived models is required to establish causal roles. · stated by the authors
  • The BRCA1-Δ11q immunoblot and immunofluorescence were each performed once in a single cell line, and TRIM11 and RNF43 assays used HEK293T overexpression, which limits generalization to endogenous prostate cancer biology. · noted in this summary
  • Survival comparisons use median splits with nominal log-rank tests and no multiplicity correction. The CRPC cohort has 97 tumors, so isoform survival associations are exploratory. · noted in this summary

Read this if you want a harmonized transcript-level prostate cancer RNA-seq resource, or isoform-switch candidates (RNF43, BRCA1, TRIM11) tied to oncogenic pathways and AR regulation.

Citations
0

Abstract

Most genes produce multiple non-coding and coding RNA transcripts with potentially divergent biological functions, yet their contribution to cancer remains largely unexplored. Here, we built an integrated, transcript-resolution RNA sequencing atlas of prostate cancer spanning 1,140 clinical samples from 10 independent studies, covering the full range of disease progression from normal tissue to primary, hormone-sensitive, and castration-resistant disease. Cross-study integration removes technical batch effects while retaining the biological heterogeneity that distinguishes these disease states, a property we validate against long-read sequencing and leverage to place newly incorporated external cohorts within this disease-progression context. We describe transcript patterns of oncogenic and tumor-suppressive long non-coding RNAs along disease progression and identify 41 protein-coding genes with significant isoform switches in cancer-relevant pathways, including WNT2B, RNF43, RELA, BRCA1, ROCK2, and TRIM11. Finally, we provide the Prostate Cancer Atlas (www.prostatecanceratlas.org), an interactive web resource that lets researchers contextualize their own transcriptomic data against this atlas.

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
THISLi '21Cho '26Negrin '20Hasan '20Song '26Lelong '21
62% semantic matchBioinformatics
Ruidong Li, Jianguo Zhu, Weide Zhong+1 more·bioRxiv·Jun 29, 2021· 31 citations