Distinct Pools of cdc25C Are Phosphorylated on Specific TP Sites and Differentially Localized in Human Mitotic Cells

BACKGROUND: The dual specificity phosphatase cdc25C was the first human cdc25 family member found to be essential in the activation of cdk1/cyclin B1 that takes place at the entry into mitosis. Human cdc25C is phosphorylated on Proline-dependent SP and TP sites when it becomes active at mitosis and the prevalent model is that this phosphorylation/activation of cdc25C would be part of an amplification loop with cdk1/cyclin B1. METHODOLOGY/PRINCIPAL FINDINGS: Using highly specific antibodies directed against cdc25C phospho-epitopes, pT67 and pT130, we show here that these two phospho-forms of cdc25C represent distinct pools with differential localization during human mitosis. Phosphorylation on T67 occurs from prophase and the cdc25C-pT67 phospho-isoform closely localizes with condensed chromosomes throughout mitosis. The phospho-T130 form of cdc25C arises in late G2 and associates predominantly with centrosomes from prophase to anaphase B where it colocalizes with Plk1. As shown by immunoprecipitation of each isoform, these two phospho-forms are not simultaneously phosphorylated on the other mitotic TP sites or associated with one another. Phospho-T67 cdc25C co-precipitates with MPM2-reactive proteins while pT130-cdc25C is associated with Plk1. Interaction and colocalization of phosphoT130-cdc25C with Plk1 demonstrate in living cells, that the sequence around pT130 acts as a true Polo Box Domain (PBD) binding site as previously identified from in vitro peptide screening studies. Overexpression of non-phosphorylatable alanine mutant forms for each isoform, but not wild type cdc25C, strongly impairs mitotic progression showing the functional requirement for each site-specific phosphorylation of cdc25C at mitosis. CONCLUSIONS/SIGNIFICANCE: These results show for the first time that in human mitosis, distinct phospho-isoforms of cdc25C exist with different localizations and interacting partners, thus implying that the long-standing model of a cdc25C/cdk1 multi-site auto amplification loop is implausible.