A CDK-mediated phosphorylation switch of disordered protein condensation

Abstract

Cell cycle transitions arise from collective changes in protein phosphorylation states triggered by cyclin-dependent kinases (CDKs), but conceptual and mechanistic explanations for the abrupt cellular reorganisation that occurs upon mitotic entry are lacking. Specific interactions between distinct CDK-cyclin complexes and sequence motifs encoded in substrates might result in highly ordered phosphorylation1, while bistability in the mitotic CDK1 control network can trigger switch-like phosphorylation2. Yet the dynamics of mitotic phosphorylation has not been demonstrated in vivo, and the roles of most cell cycle-regulated phosphorylations are unclear. Here, we show evidence that switch-like phosphorylation of intrinsically disordered proteins (IDPs) by CDKs contributes to mitotic cellular reorganisation by controlling protein-protein interactions and phase separation. We studied protein phosphorylation in single Xenopus embryos throughout synchronous cell cycles, performed parallel assignment of cell cycle phases using egg extracts, and analysed dynamics of mitotic phosphorylation using quantitative targeted phosphoproteomics. This provided a high-resolution map of dynamic phosphosites from the egg to the 16-cell embryo and showed that mitotic phosphorylation occurs on entire protein complexes involved in diverse subcellular processes and is switch-like in vivo. Most cell cycle-regulated phosphosites occurred in CDK consensus motifs and located to intrinsically disordered regions. We found that substrates of CDKs and other cell cycle kinases are significantly more disordered than phosphoproteins in general, a principle conserved from yeast to humans, while around half are components of membraneless organelles (MLOs), whose assembly is thought to involve phase separation. Analytical modelling predicts modulation of homotypic IDP interactions by CDK-mediated phosphorylation, which was confirmed by biophysical and biochemical analysis of a model IDP, Ki-67. These results highlight the dynamic control of intrinsic disorder as a conserved hallmark of the cell cycle and suggest a mechanism for CDK-mediated mitotic cellular reorganisation.

A Mechanistic Model for Cell Cycle Control in Which CDKs Act as Switches of Disordered Protein Phase Separation

“…By varying positions P +1 and P +3 (where P is the phosphorylation site) of a substrate and comparing crystal structures of CDK1 and CDK2, Endicott and colleagues showed that CDK1-cyclin B can readily target non-proline directed sites thanks to higher flexibility of the activation segment, which forms a platform for substrate recognition on both sides of the phosphorylated residue [ 8 ]. Indeed, the lack of absolute requirement for a proline in the +1 position has been largely confirmed by studying defined CDK substrates [ 9 , 10 , 11 , 12 ] or identifying CDK targets on a large scale [ 1 , 2 , 13 ]. For example, of the 19 mapped CDK1 sites on the budding yeast Wee1 homologue Swe1, 11 do not conform to the minimal consensus motif S/T-P [ 11 ].…”

“…Furthermore, in a recent phosphoproteomic analysis of CDK1 targets using chemical genetics in mouse embryonic stem cells, around 30–40% of phosphorylated serines and threonines were not followed by proline [ 1 ]. A new compilation of published human CDK sites found that, similarly to yeast [ 2 ], 10–20% are non-proline directed [ 13 ]. Finally, an in vitro study showed that sites with multiple K/R after the +1 position are more readily phosphorylated by CDK1 than S/T-P sites [ 14 ].…”

“…And are the biochemical and biophysical effects of CDK-mediated phosphorylations on diverse substrates similar? Recent work by Valverde, Dubra and colleagues addressed these exact questions [ 13 ].…”

“…The examples described above demonstrate that CDKs can regulate the organisation and assembly of specific MLOs. However, given the predominance of CDK-mediated sites among cell cycle phosphorylations, the general high intrinsic disorder of CDK substrates [ 13 ], and the switch-like reorganisation of the cell at mitotic entry when most of the biological condensates dissolve, one might hypothesise that CDKs evolved to phosphorylate IDRs and regulate phase separation during the cell cycle ( Figure 1 B). So far, this has not been demonstrated.…”

“…To test the hypothesis that CDK might act as general MLO dissolvase, Valverde, Dubra and colleagues [ 13 ] assembled a human MLO proteome and found that over one third of these proteins are CDK targets. These included major factors of the analysed condensates, such as coilin of Cajal bodies, numerous nucleoporins (NUP53, NUP98, NUP153, ELYS, and others); nucleolin and nucleophosmin of nucleoli; 53BP1, RIF1 and MDC1 of 53BP1 bodies; promyelocytic leukaemia protein of PML bodies; and MED1, which drives phase separation of the transcriptional apparatus [ 76 ].…”