CDK Substrate Phosphorylation and Ordering the Cell Cycle

Abstract

S phase and mitotic onset are brought about by the action of multiple different cyclin-CDK complexes. However, it has been suggested that changes in the total level of CDK kinase activity, rather than substrate specificity, drive the temporal ordering of S phase and mitosis. Here, we present a phosphoproteomics-based systems analysis of CDK substrates in fission yeast and demonstrate that the phosphorylation of different CDK substrates can be temporally ordered during the cell cycle by a single cyclin-CDK. This is achieved by rising CDK activity and the differential sensitivity of substrates to CDK activity over a wide dynamic range. This is combined with rapid phosphorylation turnover to generate clearly resolved substrate-specific activity thresholds, which in turn ensures the appropriate ordering of downstream cell-cycle events. Comparative analysis with wild-type cells expressing multiple cyclin-CDK complexes reveals how cyclin-substrate specificity works alongside activity thresholds to fine-tune the patterns of substrate phosphorylation.

Keywords: CDK; S phase; cell cycle; cyclin-dependent kinase; kinase; mitosis; phosphoproteomics; phosphorylation.

Graphical abstract

Figure 1

CDK Substrate Dephosphorylation after CDK Inactivation (A) The cumulative frequency of the relative phosphorylation of all detected phosphosites at time points after CDK inactivation in mitosis. (B) Plot of CDK substrate sites half-lives after CDK inactivation in mitosis against the half-lives calculated from a biological repeat (three sites with values < 15 s were excluded from this analysis). (C) Heatmap of CDK substrate site phosphorylation (n = 274) and protein (n = 146) levels after CDK inactivation in mitosis. Each row corresponds to a single phosphosite or protein, respectively. Rows are ordered by hierarchical clustering after imputation to replace missing values. Values outside the display range are set to the closest extreme. (D and E) The relative phosphorylation of individual CDK substrate sites after CDK inactivation in (D) mitosis and (E) S phase. Curves are a one-phase exponential decay fit to the data. See also Figure S1. See Figure S1A for experimental design. See STAR Methods for details of phosphorylation half-life calculation.

Figure 2

CDK Substrate Phosphorylation Dynamics during the Cell Cycle (A and B) The relative phosphorylation of individual CDK substrate sites after release from G2 arrest for mitotic (A) and S-phase (B) related proteins. Spline connects points. (C) Heatmap of CDK substrate site phosphorylation (n = 256) and protein (n = 149) levels after release from G2 arrest. Each row corresponds to a single phosphosite or protein, respectively. Rows are ordered according to hierarchical clustering, after imputation to replace missing values. Clusters of early (light gray, n = 16), mid (dark gray, n = 12), and late (black, n = 169) substrate sites are annotated. Values outside the display range are set to the closest extreme. (D) The median relative phosphorylation of early, mid, and late CDK substrate sites after release from G2 arrest. Spline connects points. See also Figure S2. See Figure S2A for experimental design.

Figure 3

In Vivo Substrate Phosphorylation Rates as a Measure for the CDK-to-Phosphatase Activity Ratio (A) The relative phosphorylation of individual early and mid CDK substrate sites after transient CDK inactivation and then restoration in G1, S phase, G2, and mitosis. (B) The median relative phosphorylation of early CDK substrate sites after CDK restoration in G1, S phase, G2, and mitosis. (C) The relative phosphorylation rates (% of mitotic rates) calculated from the slope of the linear regressions in (B). (D) The median relative phosphorylation of early, mid, and late CDK substrate sites after CDK restoration in mitosis. See also Figures S3A–S3D. See Figure S3A for experimental design. See STAR Methods for details of phosphorylation rate calculation.

Figure 4

In Vivo Substrate Sensitivity to CDK Activity (A and B) The relative phosphorylation of individual CDK substrate sites over a range of 1-NmPP1 concentrations for mitotic (A) and S-phase (B) related proteins (normalized so L:H [DMSO] = 1). Curves are a four-parameter sigmoidal fit to the data. (C) The cumulative frequency of 1-NmPP1 IC₅₀ values for early (n = 12), mid (n = 12), and late (n = 116) CDK substrate sites. (D) Plot of CDK substrate site 1-NmPP1 IC₅₀ values from biological repeats. Two values below the axes limit are excluded from this analysis. (E) Plot of 1-NmPP1 IC₅₀ values against total phosphorylation during G2/M (AUC, 50–100 min) for late CDK substrate sites. Two values are below the axis limits. See also Figures S3E–S3J. See Figure S3E for experimental design. See STAR Methods for details of IC₅₀ and AUC calculation.

Figure 5

CDK Substrate Dephosphorylation Rates, Cyclin-Docking Motifs, and Substrate Phosphorylation during Reordered Cell-Cycle Experiments (A and B) Dephosphorylation rates. (A) Plot of the CDK substrate site dephosphorylation rates after CDK inactivation in S phase against in mitosis. Three phosphosites with values < 15 s were excluded. (B) Histogram of the relative frequency of dephosphorylation rates after CDK inactivation in mitosis for early and late CDK substrate sites. See also Figures S3K and S3L. See Figure S1A for experimental design. See STAR Methods for details of dephosphorylation rate calculation. (C) The cumulative frequency of the difference between the 1-NmPP1 IC₅₀ values (ΔIC₅₀) for all phosphosite pairs (inter- and intra-protein comparisons, n = 1711) and phosphosite pairs in the same protein (intra-protein comparisons, n = 62). ΔIC₅₀ values were only calculated for pairs of phosphosites from substrates with more than one phosphosite IC₅₀ value. ΔIC₅₀ populations are significantly different (p < 0.001, two tailed Mann-Whitney U test). (D) The cumulative frequency of the protein copies per cell (Marguerat et al., 2012) for early (n = 6), mid (n = 9), and late (n = 100) CDK substrates; all phosphoproteins detected in this study (n = 1209); and all proteins (n = 3301). 178 values are above the axis limit. (E) Western blot analysis of candidate early substrate (Sld3, Drc1, and Mdb1) phosphorylation. Phosphorylation is monitored by mobility shift. Wild-type (WT) and putative cyclin docking motif mutant (RxL:AAA) versions of Drc1, Mdb1, and Sld3 were analyzed after release into DMSO, 2 μM or 20 μM 1-NmPP1 (MS306-311). See Figure S3E for experimental design. (F) Cell length at division (mean + SEM, n = 50) after Drc1, Sld3, or Sld3(RxL:AAA) overexpression in wt or checkpoint-insensitive AF strains. See also Figures S3M–S3O. (G and H) Median relative phosphorylation of early and late CDK substrate sites during two reordered cell cycles. Cell cycle stages and 1-NmPP1 concentrations are denoted above the graphs. See also Figures S4, S5, and S6. See Figure S4A for experimental design.

Figure 6

Comparative Analysis with Wild-Type Cells Expressing Multiple Cyclin-CDK Complexes (A and B) The relative phosphorylation of individual (A) early and (B) mid CDK substrate sites, after release from G2 arrest, in wild-type cells (wt) and cells deleted for G1/S cyclins (ΔCCP and ΔCCP rum1Δ). Spline connects points. Imputation to replace missing values and data smoothing were applied. NB sites in Rad4 and Dfp1 were redefined as mid substrates after visual inspection of the data. (C–E) The median relative phosphorylation of early, mid, and late CDK substrate sites after release from G2 arrest in wt (C), ΔCCP (D), and ΔCCP rum1Δ (E) cells. (F and G) Plot of the G1/S cyclin specificity score for (F) early and (G) mid substrates. (H) The relative phosphorylation of all detected S/T-P phosphosites (n = 1279) (gray), including CDK substrate sites (n = 170) (dark blue), and G1/S cyclin phosphosites (orange) in ΔCCP and cdc13-L-cdc2 Δ2 Δ13 ΔCCP cells during S-phase arrest. Sites for which 0.5 > L:H > 2 in a wt versus wt control were excluded. See also Figure S7. See Figures S7A and S7E for experimental design. See STAR Methods for G1/S cyclin-specificity score calculation.

Figure 7

Summary Model (A) Simplified CDK network: a single CDK activity resolves substrates phosphorylation at distinct activity thresholds, resulting in the temporal ordering of downstream cell-cycle transitions. Multi-cyclin network: G1/S cyclins tune the profile of generic CDK activity and combine this with an enhanced specificity toward mid substrates. (B) Other kinase-regulated processes with multiple cellular outputs resolved in time or space could be organized by the same regulatory principle.

Figure S1

CDK Inactivation in Mitosis and S Phase and Defining CDK Substrates, Related to Figure 1 The phosphoproteome was analyzed after CDK inactivation in mitosis and S phase to define CDK substrates. (A) Schematic of experimental design: a heavy labeled culture (MS230) was released from G2 arrest. The culture was split and either treated with 10 μM 1-NmPP1 in mitosis (10 min after release) or S phase (65 min after release). Protein samples were taken at 0, 1, 3, 6, 9, 12 and 24 min after addition of 10 μM 1-NmPP1. Protein samples were mixed with a common light labeled reference (MS230, synchronized in mitosis). For the biological repeat protein samples were taken between 0 and 12 min after addition of 1-NmPP1 (MS122). (B-D) Cell-cycle progression was monitored after release from G2 arrest in untreated cells. (B) Quantification of chromosome and cell division. (C&D) DNA content profiles and quantification. (E&F) Scatterplot of the relative site phosphorylation or protein levels after CDK inactivation in mitosis ((E) 12 min after 1-NmPP1 treatment and (F) 24 min after 1-NmPP1 treatment). By 12 or 24 min after CDK inactivation in mitosis a significant proportion of the phosphoproteome is at least twofold decreased (13.2% and 16.9% respectively), whereas only a small fraction of the proteome is similarly decreased (1.25% and 0.78% respectively). This compares to 0.94% of the phosphoproteome and 0.53% of the proteome behaving similarly before CDK inactivation in mitosis (0 min). Right hand panels show a comparison with the phosphoproteome of cells without a cdc2(as) allele after 1-NmPP1 treatment in mitosis, reported by Koch et al. (2011). CDK substrate sites (blue) and sites defined by Koch et al. (2011) as being dephosphorylated due to non-specific 1-NmPP1 dependent effects (orange) are shown. (G) Workflow used to define CDK substrate sites (see STAR Methods for details). (H) western blot analysis of CDK substrate (Orc2, Sld3 and Dis2) phosphorylation in mitosis. Phosphorylation is monitored by mobility shift except for Dis2 where T316-P is directly detected. Cells (MS212) were synchronized in mitosis (10 min after G2 release) and treated with either 10 μM 1-NmPP1 or DMSO: see Figure S1A for experimental design. Protein samples were taken between 0 and 15 min after 1-NmPP1 or DMSO addition. Candidate CDK substrates are dephosphorylated between 10 and 15 min after DMSO treatment (i.e., without CDK inactivation). (I) Amino acid distribution surrounding all detected phosphosites and CDK substrate sites.

Figure S2

Cell-Cycle-Synchronized Cultures, Related to Figure 2 CDK substrate phosphorylation was analyzed during the cell cycle. (A) Schematic of experimental design: a light labeled culture (MS230) was released from G2 arrest and protein samples were recovered at 20 time points over the first and second cell division cycle. Protein samples were mixed with a common heavy labeled reference (MS230, synchronized in mitosis). (B-D) Cell-cycle progression was monitored after release from G2 arrest. (B) Quantification of chromosome and cell division. (C&D) DNA content profiles and quantification. (E) western blot analysis of Cdc13-L-Cdc2 protein levels after release from G2 arrest. (F-I) Candidate CDK substrate (Orc2, Sld3, Bir1 and Dis2) phosphorylation during the cell cycle was analyzed by western blotting. Phosphorylation was monitored by mobility shift except for Dis2, where T316-P was directly. Asterisks (^∗) mark non-specific bands. Cells (MS132) were synchronized as in Figure S2A. (F&G) Cell-cycle synchrony after release from G2 arrest was monitored by quantification of (F) chromosome & cell division and (G) DNA content. (H) western blot analysis of Cdc13-L-Cdc2, α tubulin and the phosphorylation of candidate CDK substrates after release from G2 arrest. (I) Orc2, Sld3 & Bir1 mobility shifts and Dis2-T316-P signal are CDK and phosphorylation dependent. No CDK activity = 20 min 10 μM 1-NmPP1 treatment, low CDK activity = G2 arrest, and high CDK activity = synchronized in mitosis (MS132). Each sample was treated with: (i) mock, (ii) lambda phosphatase + phosphatase inhibitors or (iii) lambda phosphatase.

Figure S3

CDK Substrate Phosphorylation Rates, Dephosphorylation Rates, and Sensitivity to CDK Activity, Related to Figures 3, 4, and 5A–5F (A–D) Related to Figure 3. CDK was transiently inactivated at different cell-cycle stages and CDK substrate re-phosphorylation rates were quantified as a proxy for the CDK-to-phosphatase activity ratio. (A) Schematic of experimental design: a light labeled culture (MS213) was synchronized as in Figure S1A. In mitosis (10 min after release), G1 (35 min after release), S phase (65 min after release) or late G2 (120 min after release) a parallel culture was treated with 10μM 1-NmPP1 to inactivate CDK (15 min). 1-NmPP1 was then washed out (3x wash, 40 s per wash) to restore CDK activity. Protein samples were taken before (0 min) and 0.5, 1, 2 and 10 min after completion of the final wash. Protein samples were mixed with a common heavy labeled reference (MS230, synchronized in mitosis). (B-D) Cell-cycle progression was monitored after release from G2 arrest in untreated cells. (B) Quantification of chromosome and cell division. (C&D) DNA content profiles and quantification. (E–J) Related to Figure 4. Cells were arrested and released into a range of 1-NmPP1 concentrations to quantify CDK substrate phosphorylation across a range of CDK activity levels. (E) Schematic of experimental design: a light labeled cultures were arrested for 1.5 generations (1 generation in 1μM 1-NmPP1 followed by 0.5 generations in 2μM 1-NmPP1) and CDK was then inactivated by the addition of 10 μM 1-NmPP1. Cultures were then washed in and released into media containing DMSO (0.2%) or 1-NmPP1 (5 nM, 15 nM, 50 nM, 150 nM, 300 nM, 1 μM, 2.5 μM, 5 μM, 7.5 μM, 10 μM or 20 μM). Protein samples were taken 10 min after release and were mixed with a common heavy labeled reference (MS230, synchronized in mitosis). An AF strain (T14A, Y15F mutations in the Cdc2 moiety) (MS86) was used to bypass feedback on CDK activity. Protein samples were also recovered from a strain with a wild-type (wt) Cdc2 moiety (MS87) in DMSO and 20 μM 1-NmPP1. rad3Δ was introduced into the wt control to ensure that DNA replication/damage checkpoint signaling did not cause T14/Y15 phosphorylation dependent CDK inhibition. (F) The cumulative frequency (% of CDK substrate sites) of the relative phosphorylation values in AF and wt after release into DMSO or 20 μM 1-NmPP1. There is a major global increase in CDK substrate site phosphorylation between the extremes of the titration series for both AF and wt. AF (DMSO) n = 236, AF (20 μM 1-NmPP1) n = 218, wt (DMSO) n = 242, wt (20 μM 1-NmPP1) n = 220. (G) The cumulative frequency of the 1-NmPP1 Hill slope values for early, mid and late CDK substrate sites. Median Hill slope values are −3.579, −1.325 and, −1.406 for early (n = 12), mid (n = 12) and late (n = 117) substrate sites respectively. See STAR Methods for details of Hill slope value calculation. (H and I) The cumulative frequency of 1-NmPP1 IC₅₀ values for CDK substrate sites at (H) the minimal or full CDK consensus sequence and (I) serine or threonine phosphosites. See STAR Methods for details of IC₅₀ value calculation. (J) The relative phosphorylation of individual CDK substrate sites in Bir1, Pic1 and Plo1 during G2/M (50-100 min after release from G2 arrest, only Spline plotted for presentation) and over a range of 1-NmPP1 concentrations (only sigmoidal fit plotted for presentation). (K and L) Related to Figures 5A and 5B. (K) Plot of phosphorylation half-lives after CDK inactivation in mitosis against 1-NmPP1 IC₅₀ values for CDK substrate sites. Three values below the axis limits are not shown. See STAR Methods for details of IC₅₀ and phosphorylation half-life calculation. (L) The cumulative frequency of phosphorylation half-lives after CDK inactivation in mitosis for serine and threonine CDK substrate sites. See STAR Methods for details of phosphorylation half-life calculation. (M–O) Related to Figure 5F: Drc1-v5, Sld3-v5 and Sld3(RxL:AAA)-v5 were expressed from the full strength nmt1 promoter in cells expressing a wild-type (wt) or checkpoint insensitive (AF) Cdc2 moiety (MS312-319). Cells were initially cultured in EMM4S + thiamine. Thiamine washout was used to induce expression from the nmt1 promoter. (M) Representative photos of calcofluor stained cells 48 hr after thiamine washout. (N) Cell length measurements (mean + S.E.M., n = 50) of dividing cells before (top panel) and 48 hr after (bottom panel) thiamine washout. NB bottom panel is reproduced in Figure 5F. (O) western blot analysis of Drc1-v5, Sld3-v5 and Sld3(RxL:AAA)-v5 expression before and 48 hr after thiamine washout.

Figure S4

Reordering S Phase and Mitosis, Related to Figures 5G and 5H (A–D) S phase and mitosis were reordered using 1-NmPP1 treatment regimes as previously described (Coudreuse and Nurse, 2010). (A) Schematic of experimental design: light labeled cultures were reset in G1 (reordered cycle a&b (MS230)), arrested in G2 (reordered cycle c&d (MS230)) or arrested in G1 (reordered cycle e&f (MS108)). Cultures were then washed and released into DMSO (reordered cycle a, c and e) or 1 μM 1-NmPP1 (reordered cycle b, d and f). Protein samples were taken during the arrest or reset and 10 & 15 min after release. Protein samples were mixed with a common heavy labeled reference (MS230, synchronized in mitosis). (B-D) Cell-cycle progression was monitored during reordered cycle a-f. (B) Quantification of chromosome and cell division. (C) Representative photos of DAPI (DNA) and calcofluor (cell septum) stained cells. Aberrant nuclei and cut cells reflect mitotic progression concurrent with ongoing DNA synthesis (Coudreuse and Nurse, 2010). (D) DNA content profiles.

Figure S5

CDK Substrate Phosphorylation during Six Reordered Cell-Cycle Experiments, Related to Figures 5G and 5H (A–F) CDK substrate phosphorylation during reordered cell cycle a-f, respectively. Reordered cycles a-f are outlined in Figure S4 and cell-cycle stages are annotated above the graphs. Left hand panel shows the median relative phosphorylation of early mid and late CDK substrate sites. Remaining four panels shows the relative phosphorylation of individual CDK substrate sites with examples of protein with S-phase-related and mitotic functions presented in separate panels. See Figure S4 for experimental design. NB data in Figures S5B and S5E are reproduced in Figures 5G and 5H.

Figure S6

Candidate CDK Substrate Phosphorylation during Six Reordered Cell-Cycle Experiments, Related to Figures 5G and 5H (A–C) western blot of Cdc13-L-Cdc2, α tubulin and the phosphorylation state of four CDK substrates (Orc2, Sld3, Bir1 and Dis2) during reordered cycle a-f respectively. See Figure S4a for experimental design. Phosphorylation is monitored by mobility shift, except for Dis2, where T316-P is directly detected. Asterisk (^∗) mark non-specific bands. (D&E) Cell-cycle progression was monitored during reordered cycle a-f. (D) Quantification of chromosome and cell division. (E) DNA content. NB for reordered cycle a-d, cells were cultured in YE4S (MS212) due to orc2-v5 dependent retardation of re-replication in EMM4S. For reordered cycles e&f, cells were cultured in EMM4S (MS132).

Figure S7

Comparative Analysis with Wild-Type Cells Expressing Multiple Cyclin-CDK Complexes, Related to Figure 6 (A–D) Related to Figures 6A–6G: CDK substrate phosphorylation was analyzed during the cell cycle in the presence and absence of G1/S cyclins. (A) Schematic of experimental design: light-labeled wild-type (wt) (MS282), cig1Δ cig2Δ puc1Δ (ΔCCP) (MS278) and cig1Δ cig2Δ puc1Δ rum1Δ (ΔCCP rum1Δ) (MS69) cultures were release from G2 arrest. To avoid complications regarding SILAC media differentially influencing the cell-cycle distribution among the above strains, light labeled samples for all three strains were grown in EMM4S. Protein samples were recovered between 30 min (first G1) and 160 min (second G1) and were mixed with a common heavy labeled reference (MS131, synchronized in mitosis). (B&C) Cell-cycle progression was monitored after release from G2 arrest. (B) Quantification of chromosome and cell division. (C) DNA content profiles. (D) Schematics illustrating how CDK activity orders substrate phosphorylation in the minimal CDK network (left panel) and two hypothetical models for a multi-cyclin system (center and right panel). An exclusively qualitative model involves phosphorylation at G1/S occurring due to G1/S cyclin-substrate specify and high generic CDK activity initiating mitosis (middle panel). In contrast, an exclusively quantitative model involves G1/S cyclins only modifying the profile in the rise of total generic CDK activity to promote phosphorylation at G1/S. (E–F) Related to Figure 6H. The phosphoproteome was analyzed during S-phase arrest in the presence and absence of G1/S cyclins. (E) Schematic of experimental design: light labeled wt (MS131), ΔCCP (MS200) and cdc13-L-cdc2 Δ13 Δ2 ΔCCP (MS213) cell were released from G2 arrest and treated with 12mM Hydroxyurea (HU) 10 min after release. Protein samples were recovered during the subsequent S-phase arrest (60 min after release) and mixed with a heavy labeled wt (MS131) protein sample also taken during S-phase arrest. (F) DNA content profiles during S-phase arrest.