Alternative splicing (AS) generates multiple RNA isoforms from single gene loci and is a principal source of proteome diversity in eukaryotic cells. Variability in AS can be tissue- or context-specific and is influenced by factors such as temperature, circadian rhythm and the activity of splicing machinery. Changes in splicing can produce isoforms with altered structure, folding and interaction capacity, thereby rewiring molecular networks and affecting cellular function.
Switches in the Most Dominant Transcript (MDT)—the predominantly expressed isoform for a gene that shifts between normal and cancer samples—have been associated with disrupted protein interaction networks and tumour progression. Such isoform switches can arise from mutations in splicing factor genes, altered expression of splicing regulators, or mutations affecting splice junctions and regulatory regions. In cancer, MDT switching can influence treatment response, for example through BRAF splice variants that mediate resistance to RAF inhibitors in melanoma, or soluble splice variants that affect immune checkpoint therapy outcomes.
This study aimed to comprehensively characterise validated MDT-switching events in metastatic melanoma, ovarian cancer and acute myeloid leukaemia (AML), integrating transcriptomic and proteomic data to nominate isoform-specific biomarkers and to assess functional consequences on protein and drug interactions.
The analysis combined bulk RNA-seq data from public repositories (GTEx and TCGA) and matched multi-omics data from the Tumour Profiler (TuPro) project. The authors report analysing 170 patients from TuPro across the three tumour types: metastatic melanoma, ovarian cancer and AML. For normal-tissue or comparator cohorts they used GTEx samples for ovary and skin (and blood samples as the GTEx surrogate for bone marrow comparisons relevant to AML) and TCGA tumour cohorts (TCGA-OV, TCGA-SKCM and TCGA-LAML) for wider reference.
TuPro provides matched RNA-seq and mass spectrometry-based proteomics, enabling validation of RNA-level isoform switches at the protein level. The text excerpt includes numbers of downloaded samples from GTEx and TCGA: 112 ovary and 273 skin GTEx samples and 465 blood GTEx samples (as bone marrow proxies), and TCGA counts of 381 TCGA-OV, 473 TCGA-SKCM and 129 TCGA-LAML samples. The TuPro cohort composition is reported as including 43 ovarian cancer cases and 83 metastatic melanoma cases; the remainder and some cohort details are truncated in the source excerpt and thus are not fully reported here.
The investigators identified MDT switches by comparing dominant isoform expression between tumour samples and corresponding normal tissues. Through this approach they compiled a catalogue of 6,114 unique cancer-associated MDTs (cMDTs) across the three cancer types. Among these, 1,707 cMDTs encode transmembrane proteins, which may have relevance for signalling, cell–cell communication and potential targeting.
A subset of cMDTs showed particularly high prevalence: 131 cMDTs occurred at a frequency greater than 30% in at least one cancer cohort. These recurrent switching events are highlighted as candidate isoform-specific markers deserving further study.
To confirm that RNA-level isoform switches produce corresponding protein isoforms, the study leveraged mass spectrometry-based proteomics from the TuPro project. Overall, 12.1% of identified cMDTs were validated at the protein level by mass spectrometry evidence. The authors provide an example: the long isoform of ADD3 was repeatedly detected in ovarian cancer samples, providing a concrete example of an MDT switch that reaches the proteome.
Proteomic validation strengthens confidence that specific alternative transcripts are translated and can exert functional effects via altered protein interactions or localisation.
To evaluate functional consequences, the authors developed isoform-specific protein–protein and protein–drug interaction networks. Analysis of these networks showed that most cMDTs lose interactions with protein partners and with drug compounds relative to the canonical isoforms. This loss of interactions implies potential impacts on signalling pathways, complex assembly and on the efficacy of drugs that bind particular protein regions.
The prevalence of interaction loss among cMDTs suggests that isoform switching can alter both biological networks and pharmacological landscapes, highlighting the value of isoform-aware molecular profiling in clinical contexts.
The study found correlations between the occurrence of cMDTs and switch events in spliceosome components, consistent with the idea that splicing dysregulation underlies many isoform switches in cancer. Examples cited in the background (from prior literature) include upregulation of SF3B4 in ovarian cancer and consequences for RAD52 splicing, and DHX34 isoform changes in AML that affect differentiation.
Because isoform switching can modulate drug binding sites and interaction networks, the authors propose that cMDTs could serve as diagnostic biomarkers and as factors to consider in clinical decision-making, including potential implications for drug resistance and sensitivity.
The reported methods include bulk RNA-Seq data acquisition and processing from GTEx and TCGA repositories, with raw data downloaded using standard tools (sra-toolkit and the GDC Data Transfer Tool). The authors integrated RNA-seq and proteomics data from TuPro and constructed isoform-resolved interaction networks to assess functional impacts. Specific computational pipelines and statistical thresholds used to call MDT switches, interaction loss quantification and proteomics search parameters are described in the full article but are not fully available in the provided excerpt. Additionally, parts of the TuPro cohort description and downstream methodological detail are truncated in the source text and therefore not reported here.
This multi-omics study catalogues 6,114 cMDTs across melanoma, ovarian cancer and AML, shows that a subset are translated and detectable by mass spectrometry (12.1% validated), and demonstrates that many cMDTs lose protein and drug interactions when modelled at isoform resolution. These findings nominate isoform-specific candidates for diagnostic biomarkers and underscore the potential clinical relevance of profiling alternative splicing in cancer.
Limitations of the excerpted source text: the downloadable article and abstract report many results and examples, but the provided source body is truncated in several sections (including parts of cohort description and methods). Therefore, precise details on statistical methods, full lists of validated cMDTs, and comprehensive proteomics validation metrics are not available in the supplied text and are not invented here.
Overall, the study supports incorporating isoform-aware analyses into cancer genomics and translational workflows to better predict functional consequences and to refine biomarker and therapeutic target selection.