Cell cycle-dependent phosphorylation of human CDC5 regulates RNA processing

Abstract

CDC5 proteins are components of the pre-mRNA splicing complex and essential for cell cycle progression in yeast, plants and mammals. Human CDC5 is phosphorylated in a mitogen-dependent manner, and its association with the spliceosome is ATP-dependent. Examination of the amino acid sequence suggests that CDC5L may be phosphorylated at up to 28 potential consensus recognition sequences for known kinases, however, the identity of actual phosphorylation sites, their role in regulating CDC5L activity, and the kinases responsible for their phosphorylation have not previously been determined. Using two-dimensional phosphopeptide mapping and nanoelectrospray mass spectrometry, we now show that CDC5L is phosphorylated on at least nine sites in vivo. We demonstrate that while CDC5L is capable of forming homodimers in vitro and in vivo, neither homodimerization nor nuclear localization is dependent on phosphorylation at these sites. Using an in vitro splicing assay, we show that phosphorylation of CDC5L at threonines 411 and 438 within recognition sequences for CDKs are required for CDC5L-mediated pre-mRNA splicing. We also demonstrate that a specific inhibitor of CDK2, CVT-313, inhibits CDC5L phosphorylation in both in vitro kinase assays and in vivo radiolabeling experiments in cycling cells. These studies represent the first demonstration of a regulatory role for phosphorylation of CDC5L, and suggest that targeting these sites or the implicated kinases may provide novel strategies for treating disorders of unguarded cellular proliferation, such as cancer.

Introduction

CDC5 proteins related to Schizosaccharomyces pombe cdc5p and Saccharomyces cerevisiae CEF1 are essential for G₂ progression and mitotic entry.^1–3 Initially, CDC5 proteins were thought to act as DNA binding proteins,^2,4–6 however, CDC5 proteins in yeast and mammals have since been shown to be core components of the spliceosome^7–13 and essential for pre-mRNA splicing.^{7–10,14–17} While much is known about the regulation of cell cycle progression by CDK complexes,¹⁸ the role of pre-mRNA splicing in regulating cell division and G₂/M transit in particular is not completely understood. Data has been reported suggesting that cell division may be regulated by post-transcriptional processing of rate-limiting transcripts, and a specific role for the Saccharomyces cerevisiae TUB1 transcript in CEF1-mediated G₂/M progression has been implicated.¹⁷ The mechanisms by which CDC5 proteins are themselves regulated, however, are not known.

CDC5L is expressed constitutively in proliferating cells,⁴ suggesting that CDC5 proteins are regulated through post-translational modification. We previously showed that CDC5L was phosphorylated in a mitogen-dependent manner,⁴ supporting a role for phosphorylation in regulating its function. Others have demonstrated that CDC5L association with the spliceosome is ATP-dependent,⁹ and that in vitro binding to at least one other spliceosome protein, NIPP1, requires in vitro phosphorylation of the interacting domain of CDC5.¹⁵

Several studies have demonstrated that phosphorylation-dependent mechanisms control the pre-mRNA processing machinery.¹⁹ Dephosphorylation of serine and threonine residues by protein phosphatases 1 and 2A is essential for the catalytic steps of splicing,^20,21 and ATP is crucial for the second catalytic step of pre-mRNA processing.²¹ Furthermore, phosphorylation appears to positively regulate the nuclear import of certain splicing factors, specifically the serine/arginine-rich proteins.²² Phosphorylation also causes the release of these proteins from nuclear speckles and their accumulation at sites of transcription.²³

We now demonstrate that CDC5L, a regulator of G₂/M and component of the pre-mRNA processing complex, is phosphorylated on at least nine sites in vivo, and that CDC5L exists as a nuclear homodimer but that neither homodimerization nor nuclear localization are phosphorylation-dependent. Importantly, we demonstrate that phosphorylation regulates the CDC5L-mediated catalysis of pre-mRNA processing and that the specific sites involved are likely targets of the S-phase-specific CDK2.

Results and Discussion

CDC5L is phosphorylated on nine sites in vivo

We previously reported that CDC5L contains at least 28 potential phosphorylation sites based on amino acid homology to known kinase recognition sequences.⁴ To map those sites at which CDC5L is phosphorylated in vivo, FLAG-tagged CDC5L was transiently transfected into asynchronously dividing COS-7 cells and cultures incubated with ³²P-orthophosphoric acid. Phosphorylated FLAG-CDC5L was isolated by immunoprecipitation, digested with trypsin and tryptic peptides were separated by HPLC. Eleven radiolabeled peptide fractions were identified and numbered corresponding to the HPLC fraction (Fig. 1A). Phosphoamino acid analysis of each peptide fraction revealed that fractions 21, 48 and 90 contained phosphoserine and fractions 17, 44, 55, 72, 81, 83, 86 and 94 contained phosphothreonine.

Figure 1.

Identification of CDC5L phosphorylation sites. (A) ³²P-labeled FLAG-CDC5L was immunoprecipitated from cell lysates, digested with trypsin and peptides were separated by HPLC. Fractions containing radiolabeled phosphopeptide were identified by scintillation counting. Subsequent phosphoaminoacid analysis demonstrated that peptides 17, 44, 55, 72, 81, 83, 86 contained phosphothreonine and 21, 48, 90, 94 contained phosphoserine. Fraction numbers are indicated below and phosphorylated residues above each peak. (B) ³²P-labeled CDC5L phosphopeptides were subjected to Edman degradation sequencing. Each graph represents a phosphopeptide (bold number in right upper corner corresponds to HPLC fraction from Fig. 1A), and the peak with maximum radiolabel in each graph represents the phosphorylated amino acid in that peptide (indicated by white number in black box below fraction). (C) ³²P-labeled tryptic peptides of wild type CDC5L and phosphorylation site mutants were separated in two dimensions, and peptides were identified by autoradiography. Mutants analyzed are indicated above each autoradiogram. Phosphorylation sites in CDC5L were determined by comparing 2D profiles of phosphorylation mutants. For example, the T227A and T442A mutants shown identify the ³²P-labeled spots that correlate with T227 and T442, respectively, by their absence from the mutant profiles. Representative examples are shown for five of the nine identified mutants. (D) Collision induced distortion spectra confirming three phosphorylation sites. Fragmentation of the phosphorylated peptide KPALGFYDTSEENYQALDADFRK spanning residues 219–241 indicates an unmodified y14 and phosphorylated y16, which together identify threonine 227 as a phosphorylation site (upper). Similar fragmentation spectrum of the phosphorylated peptide LVLPAPQISDAELQEVVK spanning residues 295–312 demonstrates phosphoserine at position 303 (middle), and of the phosphorylated peptide QVVQTPNTVLSTPFR spanning residues 401–414 identifies phosphothreonine at position 404 (lower). Representative examples are shown for three of the identified phosphorylation sites. (E) Summary of the nine identified phosphorylation sites.

We performed solid phase Edman degradation N-terminal sequencing to determine the phosphorylated position in each peptide (Fig. 1B). For example, ³²P-radioactivity was hydrolyzed after the first cycle of Edman degradation in peptide 17, after four cycles in peptide 21, and after 10 cycles in peptide 44. Using the results of phosphoamino acid analysis and N-terminal degradation of phosphopeptides, we determined several potential phosphorylation sites corresponding to each HPLC peak (Fig. 1A). To resolve these, we performed site directed mutagenesis to generate FLAG-CDC5L mutants in which potential phosphoserine or phosphothreonine sites were replaced by alanine. These mutants were transfected into asynchronously dividing COS-7 cells and labeled with ³²P-ortho-phosphoric acid. Phosphorylated FLAG-CDC5L or its mutant derivatives were isolated by immunoprecipitation, digested with trypsin, and separated by thin-layer electrophoresis in one dimension and cellulose phosphochromatography in the second dimension (Fig. 1C). Comparison of two-dimensional patterns between wild type FLAG-CDC5L and the phosphorylation site mutants demonstrated sites that were phosphorylated in vivo. Comparisons for mutants T227A, T227A/T404A, S303A, T385A and T438A are shown as examples of this approach. The T227A/T404A mutant was also analyzed by HPLC and revealed that peaks corresponding to fractions 81 and 83 comprised Thr404 and peaks corresponding to fractions 86 and 94 comprised Thr227 (Fig. 1A).

We also employed LC-MS/MS mass spectrometry to confirm sites T227, S303 and T404 (Fig. 1D). In total, we identified nine sites on CDC5L that were phosphorylated in vivo using these combined approaches (Fig. 1E). This was consistent with tryptic digest prediction using MS-Digest (http://prospector.ucsf.edu), which had suggested nine possible sites using Scansite.²⁴ All of these sites reside in the CDC5L central domain comprised of basic and proline-rich regions,⁴ a region that has been shown to interact with other proteins in the splicesomal complex.^13,15,16

Phosphorylation does not regulate nuclear localization of CDC5L

To interrogate the role of phosphorylation in CDC5L function, we generated a FLAG-tagged CDC5L mutant containing all nine phosphorylation site mutations, which we designated CDC5LΔS/T (Fig. 2A). We previously had observed that CDC5L could cycle between cytoplasmic and nuclear compartments with mitogenic stimulation.⁴ To determine whether phosphorylation regulates subcellular localization of CDC5L, we transfected proliferating COS-7 cells with FLAG-CDC5L or FLAG-CDC5LΔS/T, and examined subcellular localization by immunostaining with anti-FLAG antibody (Fig. 2B). This demonstrated no difference in nuclear localization with the phosphorylation mutant, suggesting that post-translational modification at these nine sites does not regulate nuclear localization. This is consistent with our previous observation that inhibition of CDC5L phosphorylation by staurosporine did not affect nuclear localization.¹³

Figure 2.

Nuclear localization of CDC5L occurs in a phosphorylation-independent manner. (A) Amino acid sequence of the CDC5LΔS/T mutant is shown. Alanine substitutions for serine or threonine are indicated by a. (B) COS-7 cells were transfected with empty vector, FLAG-CDC5L (CDC5L) or FLAG-CDC5LΔS/T (ΔS/T). Subcellular localization of wild type and mutant CDC5L in cycling cells was determined by immunostaining with mouse anti-FLAG and Texas Red-conjugated rabbit anti-mouse (FLAG; red). Nuclei were counterstained with DAPI (blue). Merged images show that both wild type and mutant CDC5L localize to the nucleus in cycling cells.

CDC5L forms homodimers in a phosphorylation-independent manner

CDC5L contains a Myb-like domain that is capable of binding specific nucleotide sequences.^4,5 Others have shown that the Myb-containing telomerase-binding proteins TRF1 and TRF2 in mammals and Taz1p in fission yeast can occur as homodimers.^25,26 This suggested the possibility that CDC5L might also form homodimers. To test this, COS-7 cells were transiently transfected with recombinant FLAG-CDC5L. Cell lysates were cross-linked using glutaraldehyde and then subjected to SDS-PAGE (Fig. 3A, left). Two bands 105 kD and 210 kD in the gluteraldehyde-treated sample suggested that CDC5L was capable of forming homodimers. To verify this, non-cross-linked cell lysates were analyzed by native PAGE (Fig. 3A, right). Similarly, two bands at 105 kD and 210 kD were detected under these conditions, confirming that CDC5L exists as a homodimer in vivo.

Figure 3.

CDC5L homodimerizes independent of phosphorylation. (A) Cell lysates from COS-7 cells transfected with FLAG-CDC5L were treated with gluteraldehyde (GLA), separated by SDS-PAGE, and analyzed by immunoblot with anti-FLAG antibody (left). Cross-linking demonstrated a FLAG-tagged band at 210 kD, consistent with the presence of CDC5L homodimers. Lysates from COS-7 cells transfected with FLAG-CDC5L (CDC5L) or empty vector were separated by native gel electrophoresis and similarly analyzed by immunoblot (right). Under native conditions, bands at both 105 kD and 210 kD were seen, consistent with the presence of CDC5L homodimers in vivo. (B) COS-7 cells were transfected with FLAG- or V5-tagged wild type (CDC5L) or mutant (ΔS/T) CDC5L, and wild type and mutant proteins were immunoprecipitated with anti-FLAG or anti-V5 antibody. Co-precipitated proteins were analyzed by immunoblot. These data demonstrate that CDC5L and CDC5LΔS/T homodimers, as well as CDC5L:CDC5LΔS/T heterodimers, can form under these conditions.

To determine whether phosphorylation regulates CDC5L homodimerization, we co-transfected CDC5L and CDC5LΔS/T, tagged with either FLAG or V5 epitopes, into COS-7 cells. CDC5L or CDC5LΔS/T were then immunoprecipiated using the appropriate antibody, and co-precipitation of CDC5LΔS/T or CDC5L was assessed by immunoblot analysis (Fig. 3B). These studies indicated that both CDC5L and CDC5LΔS/T co-immunoprecipitated both CDC5L and the CDC5LΔS/T mutant. These results suggested that CDC5L homodimerization occurs independent of phosphorylation at the nine sites determined above (Fig. 1).

CDC5L-mediated catalysis of pre-mRNA processing is regulated by CDC5L phosphorylation

CDC5 proteins in yeast and mammals have been shown to participate in the pre-mRNA splicing complex.^{7,8,11,13,14,27,28} We therefore wanted to investigate if phosphorylation of CDC5L regulated assembly of this complex. FLAG-CDC5L or FLAG-CDC5LΔS/T was immunoprecipitated from transiently transfected COS-7, and complexes were resolved on a 4–20% polyacrylamide gradient gel. Compared to previous data demonstrating spliceosomal proteins that associate with CDC5L,¹³ alanine substitutions at the nine identified phosphorylation sites did not appear to affect the ability of CDC5L to interact with previously identified spliceosomal proteins (data not shown). This is consistent with previous observations made by us and others that CDC5L is not required for spliceosome assembly.^9,13

Following its identification as an essential component of the spliceosome, CDC5L was shown to incorporate into the spliceosome in an ATP-dependent manner and to be required for the second catalytic step of pre-mRNA splicing.⁹ This suggested that CDC5L phosphorylation might regulate the splicing process itself. To test this, we obtained a murine embryonic stem cell line, RRC285, known to be heterozygous for targeted disruption of the mouse CDC5 gene (International Gene Trap Consortium; http://www.genetrap.org/).²⁹ We analyzed the growth kinetics of this stem cell line and showed that it grows more slowly (Fig. 4A), and demonstrates a delay in G₂/M (Fig. 4B), consistent with what we previously had observed in cells expressing a CDC5L dominant negative mutant.³

Figure 4.

Phosphorylation of CDC5L regulates its pre-mRNA splicing function. (A) Feeder-independent, mouse embryonic stem cell lines were maintained in log phase growth on gelatin-coated dishes in ES Growth Medium by dilution passaging as they reached 70% confluence. Cells were counted at 24 hr intervals. Data shown are mean ± sem (N = 3). The CDC5 mutant line, RRC285, proliferated at a rate slower than wild type (E14) cells. (B) Asynchronous populations of proliferating cells were fixed, stained for DNA content with propidium iodide, and analyzed by flow cytometry. A representative experiment is shown. An increase in the number of cells with 4N DNA content in the RRC 285 mutant line suggests a delay in G₂/M, similar to the effect seen with a trans-dominant CDC5L mutant.3 (C) Nuclear extracts from CDC5 mutant (RRC285) mouse embryonic stem cells transfected with CDC5LΔS/T (ΔS/T), CDC5L:T411A/T438A (T411A/T438A) and CDC5LΔS/T:A411T/A438T (ΔS/T:A411T/A438T) mutants were assayed for in vitro splicing activity, compared to nuclear extract from untransfected mutant and wild type (E14) cells. While the unprocessed, ³²P-labeled β-globin transcript was seen in all reactions, splicing intermediates and the processed transcript were markedly diminished in the mutant RRC285 cell line, consistent with a CDC5-related splicing defect. The CDC5L mutant containing alanine substitutions at all nine mapped phosphorylation sites (ΔS/T), as well as a mutant containing alanine substitutions at only two sites, T411 and T438 (T411A/T438A), were unable to correct the splicing defect observed in the RRC285 line. However, the phosphorylation mutant in which threonine had been restored at positions 411 and 438 (ΔS/T:A411T/A438T) corrected the splicing defect, as evidenced by the observation of splicing intermediates and processed transcript similar to those seen in wild type cells.

Using an in vitro splicing assay, we then tested whether nuclear extracts from line RRC285 demonstrate a pre-mRNA splicing defect compared to wild type E14 mouse embryonic stem cells. We found that they did, with accumulation of splicing intermediates and significantly less processed mRNA seen (Fig. 4B). We then performed this assay using nuclear extracts from cells transfected with CDC5LΔS/T, which demonstrated that the phosphorylation mutant did not correct the splicing defect in the RRC285 line. Using a series of individual and combination mutants, we determined that alanine substitutions at T411 and T438 (CDC5L:T411A/T438A) were sufficient to abolish the ability of recombinant CDC5L to correct the RRC285 splicing defect, and that CDC5LΔS/T with threonine restored at positions 411 and 438 (CDC5LΔS/T:A411T/A438T) was able to correct the splicing defect in these cells (Fig. 4B). Others previously had reported that in vitro phosphorylation by CDK2-cyclin E was required for CDC5L binding to the nuclear RNA binding protein, NIPP1.¹⁵ Both T411 and T438 are within canonical CDK recognition sequences (S/T-P-X-R/K),³⁰ suggesting that phosphorylation of these two sites may directly couple the cell cycle machinery to CDC5L-mediated splicing activity.

The CDK2 inhibitor, CVT-313, inhibits CDC5L phosphorylation in vitro

To investigate whether CDC5L phosphorylation is accomplished by specific CDKs, we expressed the basic, proline-rich region of CDC5L (amino acid residues 295–795) containing the nine phosphorylation sites as a hexahistidine fusion protein in E. coli (Fig. 5A). Using an in vitro kinase assay, we demonstrated that phosphorylation of this His₆-CDC5L fusion protein was inhibited by the broad-spectrum serine/threonine kinase inhibitor, staurosporine,³¹ but not the tyrosine kinase inhibitor, genistein.³² Furthermore, we demonstrated that while the mitogen-activated protein kinase inhibitor, PD98059,³³ and an inhibitor of protein kinase A activity, Rp-MB-cAMPS,³⁴ had little effect on phosphorylation of the fusion protein, significantly less phosphorylation occurred in the presence of CVT-313,^35,36 a specific inhibitor of CDK2.

Figure 5.

In vitro and in vivo phosphorylation of CDC5L is blocked by the CDK2 inhibitor, CVT-313. (A) Expression of a ~55 kD His₆-CDC5L_295-795 fusion protein in E. coli was induced with IPTG over time, and purified by nickel-affinity chromatography. A representative expression/purification is shown. Bacterial proteins were isolated under native conditions, resolved by SDS-PAGE, and stained with Coomassie R-250. Control (E. coli) and transformed (E. coli + pET-His₆-CDC5L_295-795) bacterial lysates collected 2 hrs post-IPTG induction, and pH 4.0 eluate from nickel column are shown (left). Immunoblot with anti-His-Tag antibody (Novagen) confirmed the identity of the fusion protein (right). (B) Purified His₆-CDC5L_295-795 was phosphorylated directly on the Ni-affinity column using COS-7 nuclear extracts, then analyzed by SDS-PAGE and autoradiography. The ~55 kD purified protein was phosphorylated in vitro in the absence of inhibitors, and this was inhibited by staurosporine but not genistein (left). Use of specific inhibitors showed that phosphorylation of the CDC5L fusion protein was specifically inhibited by CVT-313, but not by PD98059 or Rp-MB-cAMPS. (C) COS-7 cells transfected with FLAG-CDC5L were metabolically labeled with ³²P in the presence of the indicated kinase inhibitors, immunoprecipitated, resolved by SDS-PAGE and quantitated by autoradiography. Relative intensity was determined as the ratio of autoradiographic signal to the amount of immunoprecipitated CDC5L, as assessed by immunoblot analysis, normalized to control sample labeled in the absence of inhibitors. Data shown are mean ± sem (N = 3). Treatment with staurosporine, the CDK inhibitor, purvalanol A, and the CDK2-specific inhibitor, CVT-313, showed similarly decreased levels of CDC5L phosphorylation compared to control (p < 0.001).

In vivo phosphorylation of CDC5L is inhibited by the CDK2 inhibitor, CVT-313

To determine whether similarly specific inhibition of CDC5L phsophorylation occured in vivo with chemical inhibition of CDK2, we transfected proliferating COS-7 cells with FLAG-CDC5L and treated cultures with staurosporine, genistein, CVT-313, PD98059 and Rp-MB-cAMPS, as well as the protein kinase C inhibitor, chelerythrine,³⁷ and the broader spectrum CDK inhibitor, purvalanol A, which inhibits the activities of CDK1, CDK2 and CDK5,³⁸ during metabolic labeling with ³²P-orthophosphoric acid. We then resolved immunoprecipitated CDC5L by SDS-PAGE and analyzed relative intensity of radioactive labeling normalized to amount of immunoprecipitated CDC5L as assessed by immunoblot (Fig. 5C). Again, we observed that inhibition of tyrosine kinases, mitogen-activated protein kinase, and protein kinases A and C did not significantly alter the phosphorylation state of CDC5L. However, broad inhibition of serine/threonine kinases with staurosporine (0.18 ± 0.01 vs 1.00 ± 0.15; p < 0.001), inhibition of CDKs 1, 2 and 5 with purvalanol A (0.11 ± 0.02 vs 1.00 ± 0.15; p < 0.001), and specific inhibition of CDK2 with CVT-313 (0.17 ± 0.02 vs 1.00 ± 0.15; p < 0.001) showed marked, similar reductions in the level of CDC5L phosphorylation in vivo.

It should be noted that CVT-313 has been shown to inhibit other kinases, but at much higher IC₅₀ values, i.e., CDK1 (IC₅₀ = 4.2 μM), CDK4 D1 (IC₅₀ = 215 μM), and mitogen-activated protein kinase/protein kinase A/protein kinase C (IC₅₀ > 1.25 mM), compared to CDK2 (IC₅₀ = 0.5 μM). In addition, CVT-313 has been shown to have profound effects on cell proliferation at concentrations of 5–20 μM, such as arrest of tumor cell growth in vitro and vascular neointima formation in vivo.^35,36 By treating cells with 1 μM CVT-313, we avoided the growth arrest seen at higher concentrations (data not shown), and also assume that the effects seen are most likely due to inhibition of CDK2, however, the possibility exists that other CDKs may have been inhibited as well. A specific role for CDK2 in CDC5L function, however, suggests a plausible model by which CDK2 activity in S-phase of the cell cycle results in phosphorylation/activation of CDC5L and subsequent processing of transcripts necessary for G₂/M progression, as has previously been reported. ¹⁷

Materials and Methods

Cell lines and plasmids

COS-7 cells were obtained from the American Type Culture Collection, maintained in DMEM with 10% calf serum, and transfected using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. E14 and RRC285 feeder-independent, mouse embryonic stem cells were obtained from the International Gene Trap Consortium (http://www.genetrap.org/),²⁹ maintained on gelatin-coated dishes in GMEM (Sigma) with 10% fetal bovine serum, 2 mM L-glutamine, 1 mM sodium pyruvate, non-essential amino acids, LIF (EsGRO; Chemicon), 2-mercaptoethanol, and transfected by electroporation using a Bio-Rad Gene Pulser II with Capacitance Extender set at 0.28 kV, 0.7 kV/cm, 500 μF. Flow cytometric analysis of DNA kinetics was performed as previously described.^3,40

Construction of pBJKS containing the cDNA for wild type FLAG-tagged CDC5L was described previously.¹³ A phosphorylation site mutant of CDC5L containing nine alanine substitutions (CDC5LΔS/T) was constructed using the Quick-Change Multiple Site Directed mutagenesis kit (Stratagene). Primers used were as follows:

• T227A, 5′CCTGCCCTTGGTTTTTATGACGCGTCTG AGGAAAACTACCAAGC3′;

• S303A, 5′GTACTTCCTGCCCCTCAGATTGCAGATG CTGAGCTCCAGGAAGTTGTAAAAGTAGG3′;

• S358A, 5′CTTAGAACACCACGAACACCAGCTGCCC AGGACAGAATTCTGC AGGAAGCCC3′;

• T385A, 5′CACCCCATTGAAAGGTGGACTTAATGCC CCATTGCATGAGAGTGACTTCT CAGG3′;

• T404A, 5′CACAGCGACAAGTTGTACAGGCGCCAAA CACAGTTCTCTCTACTCC3′;

• T411A/S417A/T421A, 5′CACAGTTCTCTCTACT CCATTCAGGGCTCCTGCTAATGGAGCTGAAGGGC TAGCTCCCCGGAGTGGAACAACTCCC AAACC3′;

• T438A, 5′GTTATTAACTCTACTCCGGGTAGAGCTCC TCT TCGCGACAA GTTAAAC3′.

DNA constructs were verified by sequencing. The V5-tag (Invitrogen) was inserted in frame with the CDC5L C-terminus as follows: MluI and BlpI sites were inserted 5′ and 3′, respectively, relative to the FLAG-tag in pBJKS-FLAG-hCDC5,¹³ using site directed mutagenesis, and the FLAG-tag was then replaced by the V5-tag containing MluI- and BlpI-compatible ends.

For metabolic labeling, transfected cells were washed 48 h after transfection and incubated in DMEM without phosphate with 10% dialyzed fetal bovine serum and 1 mCi/ml ³²P-orthophosphoric acid (NEN) for 3hrs at 37°C, 5% CO₂. For inhibitor studies, cells were incubated with 20 μM staurosporine, 10 μM genistein, 10 μM purvalanol A, 1 μM CVT-313, 10 μM Rp-MB-cAMPS, 5 μM chelerythrine or 50 μM PD98059 (all from Calbiochem) during radiolabeling.

Immunoprecipitation of CDC5L

Cells were lysed in 1% NP-40, 1% sodium deoxycholate, 0.1% SDS, 150 mM NaCl, 10 mM sodium phosphate, pH 7.2, 2 mM EDTA, 50 mM NaF, 5 mM sodium orthovanadate, 1 mg/ml okadaic acid, 1:100 phosphatase inhibitor cocktails I and II (Sigma), 1:1000 protease inhibitor cocktail (Sigma). Cells were sonicated and centrifuged at 10,000 × g at 4°C. FLAG- or V5-CDC5L, or mutant derivatives, were immunoprecipitated from supernatants using agarose-conjugated mouse anti-FLAG (Sigma), or mouse anti-V5 (Invitrogen) and protein G-sepharose (Pharmacia). Immunoprecipitates were eluted from agarose with 50 mM Tris-HCl, pH 6.8, 2% SDS, 0.1% bromophenol blue, 10% glycerol, 100 mM dithiothreitol at 95°C, and resolved by electrophoresis in 10% polyacrylamide-SDS gels.

Phosphopeptide and phosphoaminoacid analysis

Following immunoprecipitation and electrophoretic separation, the phosphorylated 105 kD band corresponding to ³²P-labeled CDC5L or its mutant derivatives was excised from the gel and cut into small pieces. These were washed with 25 mM ammonium bicarbonate/50% acetonitrile followed by reduction and alkylation with 1 mM dithiotreitol/25 mM ammonium bicarbonate, then 2 mM iodoacetamide/25 mM ammonium bicarbonate, then 25 mM ammonium bicarbonate/50% acetonitrile. Gel pieces were dried and digested with trypsin in 25 mM ammonium bicarbonate overnight at 37°C. Tryptic peptides were extracted twice by sonication in 50% acetonitrile/0.1% trifluoroacetic acid. Typically 80% of the radioactivity was recovered.

Following tryptic digestion, phosphopeptide separation was done using previously described methods.⁴¹ Briefly, tryptic peptides were separated by HPLC on a Sephasil Peptide C8 column (Pharmacia). A gradient of 0–70% buffer B (50% acetonitrile/0.057% trifluoroacetic aid) was run against buffer A (0.065% trifluoroacetic acid) over 30 column volumes followed by 5 volumes of 100% buffer B to elute the peptides. Fractions containing radiolabeled phosphopeptides were identified by scintillation counting.

Phosphopeptide separation for two-dimensional phosphopeptide mapping was done using standard methods.⁴² In brief, peptides were separated in the first dimension on cellulose plates (Merck) using a Hunter Thin Layer Chromatography peptide mapping system (C.B.S. Scientific) for 25 min at 1000 V in 25% formic acid 88%, 7.8% acetic acid, pH 1.9. Second dimension chromatography was carried out overnight in 37.5% n-butanol, 25% pyridine, 7.5% acetic acid.

For phosphoaminoacid analysis, peptides were hydrolyzed in 6 N HCl at 95°C for 1 hr, then separated on cellulose plates using a Hunter Thin Layer Chromatography peptide mapping system.⁴² First dimension separation was done at pH 1.9 for 35 min, followed by second dimension separation at pH 3.5 for 25 min. Standards were stained with ninhydrin (Sigma) and plates were analyzed on a Storm 860 phosphorimager (Molecular Dynamics).

The site of phosphorylation of each phosphopeptide obtained by HPLC-separation was determined by solid phase Edman degradation. The peptides were coupled to Sequelon-AA membranes (Millipore) and processed using a Procise N-terminal sequencer (Applied Biosystems) as described previously.⁴³

Mass spectrometry

Nanoelectrospray mass spectrometry was performed as described⁴⁴ at the UCSF Mass Spectrometry Facility. Following immunoprecipitation and electrophoretic separation, the phosphorylated 105 kD band corresponding to ³²P-labeled CDC5L was excised from the gel. The protein was reduced with dithiothreitol, and cysteines were alkylated with iodoacetamide. The protein was subjected to tryptic digestion overnight at 37°C and peptides were subsequently extracted with 50% acetonitrile, 2% formic acid, then vacuum centrifuged to dryness, resuspended in 0.1% formic acid, and analyzed by LC-MS/MS. Nanoflow HPLC was performed using an Ultimate HPLC system with a Famos Autosampler (Dionex). A 75 mm I.D. × 150 mm pepmap column (Dionex) was used for HPLC separation, employing a gradient of 5–40% acetonitrile/0.1% formic acid over 50 min followed by electrospray mass spectrometry using a QSTAR Pulsar (Applied Biosystems). When a peptide was observed above a minimum threshold intensity in the mass spectrum, it was automatically selected for fragmentation by collision-induced dissociation. An inclusion list corresponding to masses of tryptic peptides containing predicted phosphorylation sites was also employed, i.e., if the mass of a predicted phosphopeptide was observed then it was automatically selected for MS/MS even if other more intense peaks were also present in the spectrum. Fragmentation spectra were initially analyzed against the NCBInr database using Mascot (http://www.matrixscience.com). Matches were then confirmed manually and any unidentified spectra were manually interpreted using MS-Homology (http://prospector.ucsf.edu).

Immunofluorescence microscopy

Immunofluorescence staining was performed as previously described.^4,13 Cells were grown on gelatin-coated cover slips, fixed in 3% paraformaldehyde, blocked with 150 mM sodium acetate, pH 7.0, 0.1% skim milk in PBS, and permeabilized using 0.5% Triton X-100 in blocking solution. After washing with 15 mM sodium acetate, pH 7.0, 0.15% skim milk in PBS, cells were incubated with 2.5 mg/ml mouse-anti-FLAG antibody (Sigma), rinsed with washing solution, and incubated with 2 mg/ml Texas Red-conjugated goat anti-mouse and 4′,6-diamidino-2-phenylindole (DAPI) to stain nuclei. Cover slips were mounted onto glass slides using the Slow Fade Anti Fade Kit (Molecular Probes) and examined using a Nikon Microphot FXA microscope and RT-Slider SPOT digital camera with SPOT software (Diagnostics Instruments Inc.,).

Homodimerization assay

COS-7 cells were transfected with FLAG- or V5-tagged CDC5L or CDC5LΔS/T and nuclei were isolated and extracted using previously published methods.¹³ For gluteraldehyde cross-linking studies, lysates were treated with 1 mM gluteraldehyde.⁴⁵ FLAG- or V5-CDC5L and mutants were immunoprecipitated as detailed above, and proteins were suspended in 0.1% bromophenol blue, 10% glycerol for electrophoretic separation in 4–20% polyacrylamide-Tris-HCl gels under native conditions, or in 50 mM Tris-HCl, pH 6.8, 2% SDS, 0.1% bromophenol blue, 10% glycerol, 100 mM dithiothreitol for electrophoresis in 8% polyacrylamide-SDS gels.

In vitro splicing assay

Pre-mRNA processing was analyzed using an in vitro splicing assay according to published methods.⁴⁶ Nuclear extracts were prepared using previously described methods¹³ from mouse embryonic stem cells transfected with FLAG-tagged CDC5L, CDC5LΔS/T, CDC5L:T411A/T438A or CDC5LΔS/T:A411T/A438T. Labeled, capped pre-mRNA for β-globin was transcribed from pBluescript-β-globin⁴⁷ in 40 mM Tris-HCl, pH 7.9, 8 mM M MgCl₂, 2 mM spermidine, 50 mM NaCl, 10 μ dithiothreitol, 50 μM each NTP, 1 mM 5′,5′GpppG, 50 μCi [α-³²P] UTP, 1 unit/μl RNAsin, 0.5 units/μl T3 RNA polymerase (Stratagene). Splicing reactions were separated on a 10% polyacrylamide, 8 M urea denaturing gel and splicing products identified by autoradiography.

Protein expression in E. coli and in vitro kinase assay

Nucleotides 885–2385 of the CDC5L cDNA (Genbank U86753;⁴) encoding amino acids 295–795 were amplified by PCR and cloned into pET-29c(+) (Novagen) for expression in E. coli. The cDNA sequence encoding a His₆-tag was engineered into the 3′ end of the reverse PCR primer such that a CDC5L(295–795)-His₆ fusion protein would be encoded. The fusion protein was expressed and purified as previously described.⁵

For kinase assays, purified CDC5L(295–795)-His₆ was mixed with [γ-³²P]ATP, COS-7 cell extract, and incubated in 100 μl 20 mM HEPES, pH 7.5, 50 mM NaCl, 2 mM MnCl₂, 10 mM MgCl₂, 0.5% NP-40, 0.5 mM PMSF, 5 mM benzamidine hydrochloride, 5 mM NaF, 1 mM NaVO₃ and the specific inhibitor at 30°C for 10 minutes.⁴⁸ Cell extract as a source of kinase activity was prepared from subconfluent, serum-stimulated COS-7 cells lysed in 20 mM HEPES-NaOH, pH 7.5, 50 mM NaCl, 1% Triton X-100, 10% glycerol, protease and phosphotase inhibitors.⁴⁹ Phosphorylated proteins were separated by electrophoresis in 15% polyacrylamide-SDS gels. Specific inhibitors included 20 μM staurosporine, 10 μM genistein, 1 μM CVT-313, 10 μM Rp-MB-cAMPS and 50 μM PD98059 (all from Calbiochem).

Conclusion

CDC5L, a component of the pre-mRNA processing complex, is an important regulator of G₂ progression and mitotic entry in yeast as well as in plants and mammals.^2,3,39 While known to be a phosphoprotein in cycling cells,⁴ a specific role for CDC5L phosphorylation in vivo had not previously been demonstrated. Our studies now show that CDC5L is phosphorylated in vivo on at least nine sites, and that phosphorylation at these sites does not affect CDC5L nuclear localization or spliceosomal assembly. We also show that CDC5L exists as a homodimer in mammalian cells, similar to other Myb-like proteins, but that homodimerization occurs in a phosphorylation-independent manner. More importantly, however, we demonstrate that phosphorylation of CDC5L at T411 and T438 have a positive regulatory effect on pre-mRNA splicing efficiency, that these sites are contained within CDK consensus recognition sequences, and that a specific inhibitor of CDK2 blocks CDC5L phosphorylation both in vitro and in vivo. This constitutes the first demonstration of a regulatory role for phosphorylation in CDC5L function, and also couples the cell cycle machinery to pre-mRNA processing through CDC5L phosphorylation. Manipulation of these sites on CDC5L through chemical means may provide new strategies for regulating cell division in disease states such as cancer and other disorders of unguarded cellular proliferation.

Abbreviations

CDC5L

human CDC5

PBS

phosphate-buffered saline

CDK

cyclin-dependent kinase