A cyclin/cdk is required for degradation of SLBP at S/G2 transition.
SLBP expression is limited to S phase (34
), and at the end of S phase, SLBP is rapidly degraded (37
) (Fig. ). Previously we have shown that SLBP degradation requires phosphorylation of two threonine residues (Thr60 and Thr61) by two different kinases (37
). Further, we have shown that along with Thr60 and Thr61, Pro62 and a downstream KRKL sequence are required for SLBP degradation at the end of S phase (37
). The observation that Thr61 is a possible cdk phosphorylation site and KRKL is a putative cyclin binding site suggested that a cyclin/cdk was involved in SLBP degradation by phosphorylating Thr61. To test this possibility, we examined the affects of roscovitine, a known Cdk1/Cdk2 inhibitor, on SLBP degradation at S/G2
in HeLa cells. At the end of S phase, the level of SLBP remained higher in cells treated with roscovitine than in dimethylsulfoxide (DMSO)-treated control cells (Fig. ). Fluorescence-activated cell sorting (FACS) analysis confirmed that both cell populations successfully completed S phase and accumulated at G2
without a significant difference in their cell cycle profile at the time of collection (Fig. ). Since roscovitine is specific to Cdk1/Cdk2 (7
), we hypothesized that one or both of these cdk's was involved in SLBP degradation, possibly by phosphorylating Thr61. Several other kinase inhibitors were tested by adding them to cells in mid-S phase. However, none of these inhibitors, including other inhibitors of proline-dependent kinases, inhibited the degradation of SLBP at the end of S phase (Table ).
FIG. 1. S/G2 degradation of SLBP is roscovitine sensitive. (A) Western blot analysis of SLBP levels in HeLa cells synchronized by a double-thymidime block and collected at the indicated time points after the release. (B) FACS analysis of the cells collected at (more ...)
Effects of kinase inhibitors on SLBP degradation
In order to determine whether phosphorylation of these two residues and the presence of the KRKL sequence were sufficient to trigger SLBP degradation, we designed a Myc-tagged GST protein fused to amino acids 51 to 108 of an SLBP fragment (mGST-SLBPF
) in which all of the serines and threonines except Thr60 and Thr61 were mutated to alanine. We constructed similar genes encoding mutation of Thr60, Thr61, and the KRKL sequence to alanine. We stably expressed the wild type and Thr61A mGST-SLBPF
in HeLa cells and then synchronized the cells by a double-thymidine block. Chimeric wild-type mGST-SLBPF
was degraded at G2
in parallel with endogenous SLBP (Fig. , lane 1 and lane 2). mGST-SLBPF
with the Thr61A mutation was stable during G2
phase (Fig. , lanes 3 and 4). This result confirmed that mGST-SLBPF
was degraded by a mechanism similar to that with full-length SLBP, since mutation of Thr61 to alanine in full-length SLBP also prevented SLBP degradation at the end of S phase (37
FIG. 2. A fragment of SLBP fused to GST is sufficient to mimic S/G2 regulation of SLBP. (A) The SLBP fragment (amino acids 51 to 108) sufficient for S/G2 regulation is shown with the regions required for late-S-phase degradation underlined. The Ser/Thr residues (more ...)
We next expressed the GST-SLBP fragment (51 to 108 amino acids) in bacteria and used the recombinant GST-SLBPF protein as a substrate for in vitro kinase assays. We also constructed and expressed similar GST fusion proteins containing TAP or ATP in place of the TTP (amino acids 60 to 62) sequence or a mutant with the KRKL sequence changed to four alanines (KRKL/4A). We prepared an extract from late-S-phase cells (6 h after release from a double-thymidine block) and tested this extract as a source of kinase activity. The extract phosphorylated the wild-type GST-SLBPF and the ATP mutant (Fig. , lanes 1 and 2) but not the TAP mutant or GST protein (Fig. , lanes 3 and 6). The activity was sensitive to roscovitine, demonstrating that this phosphorylation was mediated by Cdk1 or Cdk2 (Fig. , lanes 4 and 5). Moreover, these results suggested that Thr61 was the only site in the GST-SLBP fragment phosphorylated by the kinase(s) in this lysate.
Phosphorylation of Thr61 increases in lysates from late-S, G2-phase cells.
To determine whether the kinase activity responsible for the Thr61 phosphorylation was regulated during the cell cycle, we performed in vitro kinase assays using the GST-SLBPF protein as a substrate and lysates from HeLa cells isolated at different time points in the cell cycle as the kinase source. We tested extracts prepared from cells at different times after release from the thymidine block. The lysates prepared from G1/S (0 h), mid-S phase (4 h), S/G2 (6 h), M/G1 (10 or 11 h), and early and mid-G1 (12 and 14 h) were tested with the wild-type GST-SLBPF substrate, and we detected peak phosphorylation at late S phase (S/G2) overlapping the point where SLBP degradation started (Fig. ). We did not detect significant phosphorylation of this substrate using lysates from G1- or G1/S-phase cells. Note that there is an endogenous protein phosphorylated by kinases in the extract (Fig. ) which is present at similar levels in all lysates (also see Fig. ). Since we have previously shown that in vivo the KRKL region is required for both Thr61 phosphorylation and SLBP degradation at S/G2, we tested the phosphorylation profile of KRKL/4A GST-SLBPF. In the absence of the KRKL sequence, we could not detect significant phosphorylation in extracts from S and S/G2 cells (Fig. , lanes 7 and 8), whereas there was a small amount of phosphorylation detected in the M/G1-phase lysate (Fig. , lane 9). Addition of roscovitine to the in vitro kinase assays abolished the phosphorylation of the SLBP fragment and the residual phosphorylation seen in M/G1-phase extracts on the KRKL/4A mutant, suggesting that all the Thr61 phosphorylation, including the residual phosphorylation of the KRKL/4A mutant in the M/G1-phase extract, was mediated by a cyclin/Cdk (Fig. ).
FIG. 3. Lysates from late-S-phase cells phosphorylate GST-SLBPF in a KRKL-dependent and roscovitine-sensitive manner. (A) In vitro kinase assays were performed in the presence of [γ-32P]ATP, using lysates from HeLa cells collected at indicated time points (more ...)
FIG. 4. Phosphorylation in lysates from late-S-phase cells is on Thr61. (A) In vitro kinase assays with the indicated form of mGST-SLBPF (ATP, TTP, or TAP) or histone H1 protein were performed in the presence of [γ-32P]ATP, using equal amounts of lysate (more ...)
We then performed similar experiments at 2-h intervals after release from the thymidine block, repeating the same kinase assays with different time points throughout S phase. We detected phosphorylation peaking at late S phase (S/G2) on wild-type GST-SLBPF and the ATP mutant but not the TAP mutant (Fig. ), showing that the phosphorylation we detected occurred on Thr61. Interestingly, although phosphorylation on Thr61 significantly increased at S/G2, there was no significant increase in the phosphorylation of H1, indicating that in this experiment the cells had not reached the point in G2 where cyclin B/Cdk1 activity rapidly increases. We analyzed incorporation of 32PO4 into all the extract proteins, and there was similar total phosphorylation activity in the extract from each time point (Fig. ). There were similar total kinase activities in all the lysates, and the patterns of phosphorylated proteins were similar in all the lysates. There was one band which was phosphorylated in parallel with the GST-SLBPF substrate which might be an endogenous substrate for the Thr61 kinase (Fig. , lanes 4 and 5).
Cyclin A/Cdk1 phosphorylates SLBP on Thr61.
In order to determine whether Cdk1 or Cdk2 was the major kinase responsible for the late-S-phase phosphorylation of Thr61, we performed immunodepletion experiments using late-S-phase lysates and Cdk1 or Cdk2 antibodies. In the case of Cdk2 depletion, although most of the Cdk2 was removed (Fig. , lane 1), there was no significant difference in Thr61 phosphorylation. On the other hand, removal of even 40 to 50% of the Cdk1 from the lysate (Fig. , lane 2) significantly reduced the phosphorylation on Thr61, suggesting that Cdk1 was the major kinase that is responsible for the phosphorylation detected in vitro. Mock depletion using the protein A beads alone had no effect on the kinase activity (Fig. , lanes 3 and 5). We further reduced the amounts of Cdk1 in the extract by subjecting the extract to sequential depletion with two treatments with the anti-Cdk1 antibody. This treatment resulted in a further reduction of kinase activity using the GST-SLBPF substrate (Fig. , lane 4) with no decrease in the amount of the Cdk2 protein.
We used antibodies against cyclin A and cyclin B to deplete these proteins and their associated kinases (Fig. ). Depletion of cyclin A resulted in a substantial reduction of kinase activity using the GST-SLBPF substrate (Fig. , lane 1), while depletion of cyclin B (Fig. , lane 2) had no effect. These results suggest that cyclin A/Cdk1 might be the kinase responsible for phosphorylating Thr61 of SLBP.
To determine the abilities of different cyclin/cdk's to phosphorylate SLBP on Thr61, we immunoprecipitated cyclin A and cyclin B complexes and Cdk2 or Cdk1 from late-S-phase lysates and tested their ability to phosphorylate the wild-type TTP, TAP, and KRKL/4A GST-SLBPF proteins using histone H1 as a control substrate. With these substrates, we detected phosphorylation only on GST-SLBPF containing TTP but not on the TAP GST-SLBPF, consistent with the fact we detect phosphorylation of SLBP only on Thr61. In order to determine the cyclin partner, we immunoprecipitated cyclin A/cdk's and cyclin B/cdk's and tested these kinase complexes for their ability to phosphorylate the GST-SLBPF proteins. Only cyclin A/cdk, but not cyclin B/cdk, required the KRKL region to phosphorylate Thr61 (Fig. , lanes 1, 2, 6, and 7), indicating that cyclin B/Cdk1 is not likely to be the kinase responsible for Thr61 phosphorylation in vivo. Note that there is no phosphorylation of KRKL/4A GST-SLBPF detected in the late-S-phase lysates (Fig. ), although there is in the immunoprecipitates (Fig. , lane 7). There is a much higher kinase concentration in the immunoprecipitates than in the lysates, and we presume that at the higher concentration the cyclin B immunoprecipitates can phosphorylate KRKL/4A GST-SLBPF.
FIG. 5. The KRKL region is required for cyclin A/Cdk1 to phosphorylate Thr61. (A) Cyclin A (Cyc A) (lanes 1 to 4) and cyclin B (Cyc B) (lanes 6 to 9) immunoprecipitates (IP) from late-S-phase HeLa cell lysate were used in in vitro kinase assays in the presence (more ...)
We also used antibodies against Cdk2 or Cdk1 to immunoprecipitate the cyclin/Cdk2 complexes and the cyclin/Cdk1 complexes from the same extracts and tested their ability to phosphorylate GST-SLBP on Thr61. As a control, we determined the ability of the immunoprecipitates to phosphorylate histone H1. The Cdk2 immunoprecipitates phosphorylated the SLBP fragment only 1/3 as efficiently as they phosphorylated histone H1 (Fig. , lanes 1 and 3). All of this activity was dependent on the KRKL sequence. In contrast, the Cdk1 immunoprecipitates phosphorylated the SLBP fragment better than they phosphorylated histone H1 (Fig. , lanes 5 and 7). The majority of this activity was dependent on the KRKL sequence (Fig. , lane 6). When we normalized the ability of these Cdk immunoprecipitates to phosphorylate Thr61 to histone H1, we found that the Cdk1 immunoprecipitates phosphorylated GST-SLBPF three to four times more efficiently than the Cdk2 immunoprecipitates, further suggesting that a cyclin/Cdk1 is the kinase that phosphorylates SLBP on Thr61 and triggers its degradation at late S phase.
The Cdk2 complexes present in the cell are cyclin A/Cdk2 and cyclin E/Cdk2. Since there is a constant high level of cyclin A/Cdk2 activity during S phase and there is little Thr61 phosphorylation activity in the corresponding extracts (Fig. ), it is unlikely that cyclin A/Cdk2 is responsible for the phosphorylation of Thr61. The levels of cyclin E and cyclin E/Cdk2 activity are maximal in early S phase and low at late S phase (9
), suggesting that this kinase complex is not responsible for the Thr61 kinase activity. The Cdk1 complexes that could be present in the cell are cyclin A/Cdk1 and cyclin B/Cdk1, and cyclin B/Cdk1 is the only known cyclin B/cdk complex.
We tested the abilities of recombinant cyclin A/Cdk2, cyclin A/Cdk1, and cyclin B/Cdk1 in amounts that showed similar histone H1 kinase activity (Fig. , lanes 7 to 9) to phosphorylate the GST-SLBPF fusion protein (Fig. ). Cyclin A/Cdk1 phosphorylated the GST-SLBPF proteins on Thr61 (Fig. , lanes 1 and 4), while cyclin A/Cdk2 had very little activity toward these substrates (Fig. , lanes 2 and 5). Cyclin B/Cdk1 also phosphorylated the GST-SLBPF substrate on Thr61 (Fig. , lanes 3 and 6), although not as effectively as cyclin A/Cdk1.
We then tested whether the activities of the recombinant kinases were dependent on the KRKL sequence, which is essential for SLBP degradation (37
). Recombinant cyclin A/Cdk1 phosphorylated GST-SLBPF
six times better than it phosphorylated the GST-SLBPF
KRKL mutant (Fig. , lanes 1 and 2). In contrast, cyclin B/Cdk1 phosphorylated both GST-SPBP fusion proteins equally well. Thus, all the data are consistent with cyclin A/Cdk1 being the kinase responsible for phosphorylating SLBP Thr61 at the end of S phase and triggering SLBP degradation.
RNAi knockdown of Cdk1 results in stabilization of SLBP.
To further examine the role of Cdk1, we knocked down Cdk1 expression in synchronized HeLa cells using Cdk1-specific RNA interference (RNAi) (Fig. ). Using the strategy shown in Fig. , where siRNA treatment was combined with a double-thymidine block, we have been able to partially knock down Cdk1 while maintaining the ability to synchronize the cells. Control cells were treated with the same siRNA protocol using a control, C2 siRNA (C2) (31
). The levels of the Cdk1 protein were reduced by approximately 50% by the RNAi treatment (Fig. , cf. lanes 1 to 3 with lanes 4 to 6). At 6 h after release from the thymidine block, the cells were in late S phase. There was less SLBP degradation at late S phase when Cdk1 was knocked down compared to results for the control cells (Fig. , lanes 3 and 6). The amount of SLBP remaining was normalized to a loading control, and the results of two experiments are averaged in Fig. . FACS analysis of these cells showed that cells progressed through S phase and the cell cycle distribution of the cells was comparable in the treated and control cells (Fig. , right).
FIG. 6. Cdk1 knockdown inhibits late-S-phase SLBP degradation. (A) Outline of experimental setup for combining RNAi treatment with the synchronization procedure. (B) HeLa cells were transfected with Cdk1 siRNA (lanes 1 to 3) or a control siRNA (Cont.) (lanes (more ...) Timing of SLBP degradation does not depend on Thr60 phosphorylation.
Previously we showed that changing either Thr60 or Thr61 to alanine stabilized SLBP at G2
). We previously tested some of the possible changes of Thr60 and Thr61 to acidic amino acids, but all of the changes we tested stabilized SLBP (37
). Specifically, we could not mimic the phosphorylation on Thr61 by substitution with an acidic amino acid, and the SFEEP mutant was also stable (37
). We constructed additional mutants in an attempt to mimic the phosphorylation of Thr60 and created the SFEDP, SFDTP, and SFETP mutants. We then selected populations of cells stably expressing the mutant SLBP proteins at a level similar to that for wild-type SLBP. The SFDTP mutant was stable at the end of S phase, indicating that the aspartic acid at position 60 did not mimic the phosphothreonine (Fig. , bottom). In contrast, the SFETP mutant was regulated properly (Fig. , top) and was degraded in parallel with wild-type SLBP. This was the only phosphorylation mutant that mimicked the wild-type protein (Fig. ). Treatment of the cells at 4 h after release with the proteosome inhibitor MG132 stabilized endogenous SLBP as well as the SFETP SLBP mutant protein (Fig. , lane 4). This result demonstrated that the glutamic acid at position 60 was capable of mimicking this phosphothreonine and must be able to target SLBP to the proteosome when Thr61 is also phosphorylated. Since SFETP SLBP was expressed at normal levels during S phase and then degraded, we conclude that phosphorylation of Thr61 by cyclin A/Cdk1 is essential for proper timing of the degradation of SLBP. In support of this, it has been shown that cyclin A/Cdk1 is activated at the end of S phase (21
). Thus, the mutagenesis data taken together with our biochemical results implicate cyclin A/Cdk1 as the kinase that phosphorylates Thr61 of SLBP, and this phosphorylation triggers degradation of SLBP at the end of S phase.
FIG. 7. Changing Thr60 to Glu mimics phosphorylation of Thr60. (A) The first threonine, T60, in the SFTTP motif was changed to either glutamic acid (top) or aspartic acid (bottom) and stably expressed in HeLa cells as His-tagged SLBP (HisSLBP). Cells were synchronized (more ...) CK2 phosphorylates Thr60, and phosphorylation is primed by phosphorylation of Thr61.
Analysis of the sequence around the SFTTPE sequence in SLBP (http://cbs.dtu.dk/services/NetPhosK
) suggested that CK2 might phosphorylate T60. We tested the involvement of CK2 as a possible Thr60 kinase by using two different specific inhibitors of CK2, 4,5,6,7-tetrabromo-2-azabenzimidazole (TBB) and 2-dimethylamino-4,5,6,7-tetrabromo-1H-benzimidazole (DMAT). Treatment of cells with either TBB or DMAT at late S phase resulted in stabilization of SLBP in G2
, and the degree of stabilization was dependent on the inhibitor concentration (Fig. ). Since CK2 phosphorylates serines or threonines which are often adjacent to a negative charge (acidic amino acids or a phosphorylated amino acid), we tested two synthetic peptide substrates that contained amino acids 54 to 68 of SLBP, one of which had a phosphothreonine at the equivalent of Thr61 and one which had a threonine at this position. The unphosphorylated substrate was not phosphorylated by CK2, but the phosphorylated substrate was actively phosphorylated (Fig. ), suggesting that phosphorylation of Thr61 primed the ability of CK2 to phosphorylate Thr60. We subjected the phosphorylated peptide to mass spectrometry analysis using a MALDI-time-of-flight spectrometer, and about 40% of the phosphorylated peptide was phosphorylated in a second site in 30 min (Fig. , right). We have confirmed that a significant amount of the detected phosphorylation is on Thr60 by analysis of phosphoamino acids by thin-layer chromatography (data not shown).
FIG. 8. CK2 phosphorylates Thr60 of SLBP. (A) Western blot analysis of SLBP levels in HeLa cells which were treated 4 h after release from a double thymidine block (at late S phase) and collected 4 h later at G2 with either DMSO (lanes 2 and 4) or different concentrations (more ...)
To demonstrate that cyclin A/Cdk1 could prime phosphorylation by CK2 on Thr60 in the context of the region of SLBP that targets SLBP degradation, we carried out sequential phosphorylation reactions using our GST-SLBPF fusion proteins as substrates. The first reaction was carried out in the presence of unlabeled ATP in the presence or absence of recombinant cyclin A/Cdk1, and then a second incubation was carried out with radiolabeled ATP in the presence or absence of recombinant CK2. After 20 h of incubation at 30°C, cyclin A/Cdk1 was inactivated (Fig. , lane 2), and incubation of the substrate only with CK2 and radiolabeled ATP resulted in no significant phosphorylation (Fig. , lane 3). There was substantial phosphorylation of GST-SLBPF that was dependent on the addition of CK2 and prior incubation with cyclin A/Cdk1 (Fig. , lane 1). Phosphorylation by CK2 was dependent on both the cyclin binding site in GST-SPBPF (Fig. , lane 5) and the presence of Thr61 in the substrate (Fig. , lane 8).
We confirmed these results by analysis of the GST-SLBP fragments by mass spectrometry. Using ESI (electrospray ionization) to determine the molecular weight of the GST-SLBP fragments, we showed that incubation with cyclin A/Cdk1 resulted in addition of one phosphate to GST-SLBP and subsequent incubation with CK2 resulted in addition of a second phosphate. The phosphorylation sites were confirmed as T61 and T60 by analysis by mass spectrometry on a MALDI TOF/TOF spectrometer(data not shown).
We have identified the two kinases responsible for phosphorylation of Thr60 and Thr61 that result in degradation of SLBP. Cyclin A/Cdk1 phosphorylates Thr61 at the end of S phase, and this phosphorylation is necessary for subsequent phosphorylation by CK2 on Thr60. Thus, activation of cyclin A/Cdk1 is responsible for the timing of SLBP degradation at the end of S phase.