Individual Final Project

p53 GOF Interactome Cancer Vulnerability Map

María José Bolivar | Ottawa Bioscience Node | HTGAA 2026

The Problem

TP53 is mutated in more than 50% of all human cancers. A subset of these mutations are gain-of-function (GOF), meaning the mutant protein does not simply lose its tumor suppressor activity — it actively hijacks new protein partners and drives tumor growth. Despite this, no tool currently maps which cancer types are most dependent on specific p53 GOF hotspot mutations or ranks their predicted sensitivity to p53 reactivation therapy.

The Insight

Cancers where mutant p53 sits at the hub of a dense oncoprotein network are the most dependent on its GOF program and would respond best to p53 reactivation drugs like APR-246. The key is identifying which cancers are most “committed” to the GOF program — and which specific hotspot mutation drives that commitment.

The Approach

A fully computational pipeline integrating four public genomic databases to score each cancer type’s dependency on specific p53 GOF hotspot mutations and rank their predicted drug sensitivity — with no wet lab required.

Workflow

Step 1 — Mutation Frequency Mapping

Using COSMIC and cBioPortal TCGA PanCancer Atlas (10,953 patients, 32 cancer types), we mapped the frequency of six confirmed GOF hotspots across all cancer types:

• R175H — conformational mutant, most frequent overall (129 samples)
• R248W — contact mutant, prevalent in pancreatic and lung cancer
• R248Q — contact mutant, prevalent in leukemia and colorectal cancer
• R273H — contact mutant, prevalent in colorectal and lung cancer
• R273C — contact mutant, dominant in Brain Lower Grade Glioma (40 samples)
• R249S — structural mutant, nearly exclusive to liver cancer via aflatoxin B1 exposure

Key finding: Colorectal cancer showed the highest GOF diversity across hotspots. Brain Lower Grade Glioma showed the highest concentration of a single hotspot (R273C). Liver cancer confirmed the known R249S-aflatoxin B1 association, validating the pipeline.

Step 2 — Structural Modeling & Interactome Mapping

Each GOF hotspot was modeled using AlphaFold Server to visualize how each mutation structurally reorganizes the p53 DNA-binding domain. All six mutant structures were compared against the wildtype using Mol Viewer*.

The p53 interactome was then mapped using BioGRID (6,106 raw interactions, 2,585 unique interactors, 1,750 publications) and STRING (combined score ≥ 0.7), enhanced with Boltz.bio structural modeling.

Key finding: TP53 shows a betweenness centrality of 0.956 in the combined network — meaning 95.6% of all communication paths between proteins in the network pass through p53. This confirms p53 as the absolute regulatory bottleneck of its interactome and the most impactful therapeutic target in the network.

Step 3 — TCGA Expression Cross-Reference

(In progress) Cross-referencing with TCGA RNA-seq expression data to identify which cancer types show the highest co-expression of p53 GOF interactors, identifying the most “committed” tumors.

Step 4 — Vulnerability Scoring

(In progress) Integrating mutation frequency (Step 1), network centrality (Step 2), and expression data (Step 3) to generate a composite vulnerability score per cancer type. DeepCure AI will be used to predict APR-246 drug sensitivity based on the mutational and network profile of each tumor type.

Step 5 — Validation

(In progress) Validating predicted vulnerability rankings against APR-246 drug sensitivity data from GDSC and CTRP pharmacogenomic databases. Public proteomics data from Waters Corporation mass spectrometry repositories (PRIDE, ProteomicsDB) will be used to cross-reference predicted protein interactions.

Tools & Data Sources

Why This Matters

APR-246 (eprenetapopt) failed Phase III clinical trials for TP53-mutated cancers — not because the drug doesn’t work, but because patients were not stratified by GOF hotspot subtype. This project builds the missing vulnerability map: a ranked list of cancer types by their dependency on specific p53 GOF hotspot mutations, designed to guide more precise patient selection for p53 reactivation therapy.

Current Status