Covalent capture of kinase-specific phosphopeptides reveals Cdk1-cyclin B substrates

Abstract

We describe a method for rapid identification of protein kinase substrates. Cdk1 was engineered to accept an ATP analog that allows it to uniquely label its substrates with a bio-orthogonal phosphate analog tag. A highly specific, covalent capture-and-release methodology was developed for rapid purification of tagged peptides derived from labeled substrate proteins. Application of this approach to the discovery of Cdk1-cyclin B substrates yielded identification of >70 substrates and phosphorylation sites. Many of these sites are known to be phosphorylated in vivo, but most of the proteins have not been characterized as Cdk1-cyclin B substrates. This approach has the potential to expand our understanding of kinase-substrate connections in signaling networks.

Fig. 1.

Tagging and purification of kinase-specific phosphorylation sites. (A) A kinase is engineered to accept an ATP analog bearing a substitution at the N6 amine by mutation of the conserved “Gatekeeper” residue (as-kinase, for analog-sensitive). Replacement of a γ-phosphoryl oxygen with sulfur gives the ATP analog N⁶-(benzyl)ATP-γ-S, which bears a transferable thiophosphate group. A kinase reaction performed with an as-kinase and N⁶-(benzyl)ATP-γ-S results in specific tagging of the substrates of this kinase with a thiophosphate group. (B) Thiophosphopeptide purification. A protein mixture containing tagged substrates is digested to peptides, and the products are allowed to react with iodoacetyl-agarose. Thiol-containing groups react to form covalent bonds. Unbound peptides are washed away. Thiophosphopeptides are specifically liberated by oxidation-promoted hydrolysis of the sulfur–phosphorus bond. The thiophosphoryl sulfur atom is replaced with oxygen in this step.

Fig. 2.

Specific purification of a model thiophosphopeptide from simple and complex mixtures. (A) Purification of a thiophosphopeptide from a tryptic digest of labeled Cdk substrate Fin1, analyzed by MALDI-TOF mass spectrometry. Ten picomoles of digest were used. The target species was present at m/z 2,122.00 in the load. After 30 min incubation with iodoacetyl-agarose, analysis of the unbound fraction revealed specific depletion of this species. This spectrum represents approximately half of the unbound material, because thorough washing was required to remove the remainder. After washing, the target peptide was eluted by oxidation-promoted hydrolysis. The eluate spectrum showed recovery of the target species, at m/z 2,106.02 due to replacement of the sulfur atom with oxygen. (B) The experiment in A was repeated with the addition of a tryptic digest of 10 mg of whole-cell extract. The target species comprised 2 ng or less; the background was 500,000-fold higher. Analysis of the load (2 μg, 1:5000) confirmed that the target species was undetectable under these conditions. After purification, analysis of the eluate revealed highly specific recovery of the target species.

Fig. 3.

as-Cdks phosphorylate numerous and varied substrates in cell extracts. (A) Cdk substrates in HeLa extract were radiolabeled by addition of as-Cdk–cyclin complexes and N⁶-(benzyl)ATP-γ-³⁵S, resolved by SDS/PAGE and visualized by autoradiography. (1, no kinase; 2, Cdk2-Cyclin E; 3, Cdk2-Cyclin A; 4, Cdk1-Cyclin B). The dark bands at 48, 52, and 55 kDa represent autophosphorylation of the exogenous cyclins E, A, and B, respectively. (B) Labeled as-Cdk1-cyclin B substrates were separated by 2D electrophoresis (pH 3–10) and visualized by autoradiography. (Inset) Signal intensity has been scaled to lower sensitivity to accommodate several very highly labeled proteins.

Fig. 4.

Cdk1-cyclin B phosphorylation sites identified in our experiments with fractionated HeLa cell extracts, grouped on the basis of protein function. Because many of the proteins are involved in multiple processes, some generalizations were necessary. Sites found to be phosphorylated in vivo in phosphoproteomics studies are shown in red. Sites falling within an optimal Cdk consensus sequence are shown in bold; minimal proline-directed sites are shown in plain text. Sites annotated in the Phospho.ELM database as known Cdk1 sites are indicated with an asterisk.

Fig. 5.

A schematic depiction of the nuclear envelope and embedded nuclear pore complex, indicating the phosphorylation sites recovered in our experiments (not to scale). The indicated positions of the phosphorylation sites (red circles) are approximate.