The CNNM-binding interface drives PRL-3-dependent tumor progression in de novo cancer models.

Phosphatase of regenerating liver 3 (PRL-3) is a cancer-associated protein linked to tumor progression, metastasis, and poor patient outcomes, but the mechanism underlying its oncogenic activity remains unresolved. PRL-3 functions both as a phosphatase and as a binding partner of CNNM magnesium transporters, raising the question of which activity drives cancer progression. Here, we combine structure-guided separation-of-function mutants with biochemical and structural analyses, de novo zebrafish models of rhabdomyosarcoma and T cell acute lymphoblastic leukemia, and complementary human cancer cell assays to distinguish these mechanisms. PRL-3 wild type and the phosphatase-deficient, CNNM-binding-competent C104D mutant enhanced magnesium retention, stress tolerance, tumor-associated cell behaviors, tumor growth, and leukemia progression, whereas the CNNM-binding-deficient R138E mutant did not. These findings identify the CNNM-binding interface as a central mechanistic requirement for PRL-3-driven malignancy and suggest that therapeutic strategies focused solely on catalytic inhibition may not fully suppress its oncogenic functions.