Cancer Cell, 20

Supplemental Information

A Systematic Screen for CDK4/6 Substrates

Links FOXM1 Phosphorylation

to Senescence Suppression in Cancer Cells

Lars Anders, Nan Ke, Per Hydbring, Yoon J. Choi, Hans R. Widlund, Joel M. Chick, Huili Zhai, Marc Vidal, Stephen P. Gygi, Pascal Braun, and Piotr Sicinski

Inventory of Supplemental Information

Figure S1, related to Figure 1

Table S1, related to Figure 1 (Excel file)

Figure S2, related to Figure 4

Figure S3, related to Figure 5

Figure S4, related to Figure 6

Figure S5, related to Figure 7

Supplemental Experimental Procedures

SUPPLEMENTAL DATA

Figure S1. Relative Phosphorylation Scores (PR-Scores) of all Identified In Vitro Substrates and Their Validation in Human Cells, Related to Figure 1

(A) Potential substrates are categorized here by preference for cyclin D1-CDK4 or cyclin D3-CDK6. Within each table, substrates are arranged from the lowest to the highest degree of phosphorylation.

(B) HEK293 cells were transfected with the indicated substrates plus either empty vector (EV), or cyclin D3-CDK6, or cyclin D1-CDK4. Experimental results obtained from utilizing CDK6 and CDK4 kinase-dead (KM) versions are shown in middle or right panel, respectively. All substrates were GST-tagged at the N-terminus and detected by a GST antibody. Control blots for expression of the transfected cyclin-CDK complexes are shown below.

Table S1. PR-Scores Obtained from In Vitro Phosphorylation of all Successfully Purified Proteins, Related to Figure 1 (Excel File)

Figure S2. Phosphorylation of FOXM1 by CDK4/6, Related to Figure 4

(A) All phosphopeptides from in vivo phosphorylated FOXM1 identified by mass spectrometry are shown. Phosphorylated residues are highlighted in red. The positions of the phosphorylated amino acids are listed in the right column and depicted in the scheme in Figure 4B in the main text.

(B) Phosphopeptides derived from in vitro phosphorylation of N-terminal (N), middle (M) and C-terminal FOXM1 fragments (see also Figure 4H in the main text). Phosphorylated residues are highlighted in red. The positions of the phosphorylated amino acids are listed in the right column and depicted in the scheme in Figure 4H.

(C) Sequences of the identified cyclin D-CDK4/6 motifs in FOXM1 with the phosphorylated sites in red. Basic residues (K, R) downstream of the critical proline are highlighted in blue. Note that the CDK motifs centered on S35, S451 and S704 lack basic residues at the respective positions.

(D and E) Quantification of luciferase activity (/SD) from HeLa cells co-transfected with the 6DB promoter and empty vector (-), wild-type FOXM1 (M1), or FOXM1 RXL1/2/4/5 mutant (RXL1/2/4/5), with or without kinases. The RXL1/2/4/5 mutant contains AXA replacements on all four RXL motifs. 14, cyclin D1 and CDK4 (left panel); 36, cyclin D3-CDK6 (right panel).

(F) Immunoprecipitation (IP) of cyclin D1 from U2OS cells, expressing either empty vector (EV), FOXM1-pcDNA3 or FOXM1-pBABE. Cyclin D1 immunoprecipitates were probed with a FOXM1 antibody (lower panel). Note the absence of a stable FOXM1-cyclin D1 protein interaction. TCL, total cell lysate.

Figure S3. Regulation of FOXM1 Protein Stability, Related to Figure 5

(A) U2OS and SKMEL2 cells were treated with either DMSO, 1M or 5M of the CDK2-specific inhibitor CVT313 (Brooks et al., 1997) for 16 h, and endogenous FOXM1 protein levels were analyzed.

(B) Effect of PD0332991 on constitutively expressed FOXM1 protein. PD0332991 treatment significantly reduced FOXM1 protein expression from a weak promoter. However, the compounds effect becomes negligible when expression is driven from a very strong promoter. Utilized here is the strong CMV promoter (pcDNA3, middle panel) and the weaker murine leukemia virus (MMuLV) long terminal repeat promoter (pBABE, right panel). U2OS cells were transfected with either empty vector (EV), FOXM1-pcDNA3 or FOXM1-pBABE, and treated with 1 M PD0332991 for 16 h. Tubulin was used as loading control.

Figure S4. Characterization of PD0332991-Induced Senescence in U2OS Cells, Related to Figure 6

(A) Quantification of SA--galactosidase positive U2OS and SKMEL2 cells (/SD) following incubation with either DMSO, 1 M or 5 M CVT313 (Brooks et al., 1997) for 8 days.

(B) Percentage of SA--galactosidase positive U2OS cells (/SD) treated with either DMSO, 500 nM PD0332991, or cultured in serum-free medium for 8 days.

(C) Side scatter (SSC) of cells treated as in B was analyzed by FACS using a BD FACScanTM system.

(D) Cells treated as in B were pulsed with BrdU and stained using the FITC BrdU Flow Kit (BD Pharmingen). Cell cycle parameters were subsequently assessed by FACS analysis. Shown here is the percentage of cells in the indicated cell cycle phases.

(E) U2OS cells stably expressing control (sh-con) or RB1-targeting shRNA (sh-RB1) were treated with either DMSO or 500 nM PD0332991 for 8 days, and the percentage of SA--galactosidase positive cells was quantified (/SD). RB1 knockdown efficiency is shown in Figure 2D in the main text.
(F) FACS analysis of the side scatter (SSC) of DMSO versus PD0332991-treated U2OS cells stably expressing control shRNA (sh-con, upper panel) or RB1-targeting shRNA (sh-RB1, lower panel). Each data set (color) is from 3 independent experiments.

(G) FlowJo analysis of the data obtained in F. Shown here is the relative mean side scatter (SSC; /SD), with the DMSO controls set to 100%.

Figure S5. Effect of PD0332991 Short-Term Treatment on the Cell Cycle in U2OS, Related to Figure 7

CDK4/6 inhibition in U2OS cells for 4 h did not cause broad changes in the cell cycle distribution. U2OS cells were treated with either DMSO (vehicle) or 1 M PD0332991 for 0, 2, 4 or 8 h. Cells were subsequently pulsed with BrdU and stained using the FITC BrdU Flow Kit (BD Pharmingen). Cell cycle parameters were assessed by FACS analysis using a BD FACScanTM system.

SUPPLEMENTAL EXPERIMENTAL PROCEDURES

In Silico Screen of the Human Proteome

The UniProtKB/SWISS-PROT protein database was searched for CDK-motif containing protein sequences using Scansite 2.0 (Obenauer et al., 2003). The database search was run with the provided Cdc2 kinase motif, which represents the canonical phosphorylation target sequence of most cyclin-dependent kinases. The screen was performed with the following considerations: First, Homo sapiens was chosen as “single species”. Second, in the “keyword search” option, the SWISSPROT keyword “nucleus” was specified to enrich for nuclear proteins according to our assumption that active cyclin D-CDK4/6 complexes are localized in the cell nucleus. 2000 putative CDK phosphorylation target sites were obtained, and assigned to 964 human proteins, of which 445 contained at least two consensus sites, and 519 only one site. With Scansite scores ranging from 0.5232 to 0.1986, all obtained CDK consensus sites are defined as medium to high stringency sites, respectively.

High-Throughput Protein Expression in E. coli and Purification

Bacterial protein expression and cell lysis was performed according to Braun et al. (2002) with minor modifications. First, the volume of the expression cultures was increased to 4 ml per sample. Second, 4 h after induction with 1 mM isopropyl -D-thiogalactoside (IPTG), cells were harvested in 170 l resuspension buffer (PBS, 2 mM EDTA, 5 mM 1,4-dithiotreitol, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml Aprotinin, 10 M Pepstatin, 20 M Leupeptin, pH 7.5). Resuspension was achieved by agitating samples on a Labnet shaker for 5 min at 800 rpm at 4C. Following cell lysis, GST fusion proteins were enriched and purified with a ME200 12 channel electronic pipettor equipped with 20 l Glutathione PhyTip columns (Phynexus, San Jose, CA). Proteins were eluted from the resin in 20 l of 100 mM HEPES, 100 mM NaCl, 0.2% Tween 20, 20 mM Glutathione and 2.5 mM 1,4-dithiotreitol. For quality controls, 10 l of every eluate was analyzed on SDS gels, stained with Coumassie Brilliant Blue. The remaining elution volumes were used for in vitro kinase reactions. Every protein was inspected individually for correct size, yield and purity. We were able to successfully purify 75% of the expressed proteins, with yields ranging from 0.2 up to 4 g per sample.

In Vitro Kinase Reactions

10 l of eluted samples were diluted in 10 l of complementation buffer (100 mM HEPES, 20 mM MgCl2, 2 mM EGTA, pH 7.4). Samples were supplemented with 10 Ci -32P-ATP (Perkin Elmer, Waltham, MA), 40 M ATP, followed by addition of either 0.5 l recombinant cyclin D1-CDK4 (Lot 005, ProQinase, Freiburg, Germany), or 0.5 l recombinant cyclin D3-CDK6 (Lot 31211BU, Millipore, EMD). Both kinase complexes phosphorylated the Retinoblastoma protein (RB1) with a comparable efficiency under these assay conditions. Reactions were run at 30C for 20 min and terminated by adding 10 l of Laemmli buffer.

Phosphorylation Quantification and PR-Score Calculation

Kinase reaction products were loaded on SDS gels and transferred to nitrocellulose membranes, followed by staining with Amido black 10B protein stain. Membranes were dried and exposed to x-ray film. Within each panel of kinase reactions, the Retinoblastoma protein (RB1) was included as positive control. This allowed normalization of the phosphorylation levels between the different panels of kinase reactions as well as different membranes. Radioactive signals and respective protein bands were quantified by densitometric analysis using a pdi 420 scanner with Quantity One software (pdi, Huntington Station, NY). For each protein, the relative phosphorylation score (PR-score) was calculated from the ratio of the radioactive signal from the autoradiograph (32P) and the amount of fusion protein from the membrane (AB) according to the following formula: PR-score = 32PS/ABS ABRB1/32PRB1  100%, were S is the substrate, and RB1 the positive control. PR-scores are thus relative rates of phosphorylation, normalized to RB1.

Cell Culture, Transfections, Retroviral Infections and Cell Lysis

All of the cell lines were routinely grown according to the guidelines of the American Type Culture Collection. HEK293 cells were transfected using the calcium phosphate-DNA co-precipitation method. U2OS and HELA cells were transiently transfected with Lipofectamine 2000 (Invitrogen). For retroviral infections, the amphotropic packaging cell line Phoenix A was transfected with respective pBABE plasmids using calcium phosphate-chloroquine. The viral supernatant was collected after 48 h and 72 h, and viral particles were concentrated using the Retro-X concentrator (Stratagene). The concentrate was subsequently used to infect sub-confluent mouse embryonic fibroblasts. Cells were lysed and proteins solubilized as previously reported (Anders et al., 2006).

Antibodies

The following antibodies were used: anti-GST (goat, GE Healthcare), anti-CDK4 (C-22, Santa Cruz), anti-CDK6 (C-21, Santa Cruz), anti-HA (HA.11, Covance), anti-Tubulin (DM1A, Sigma), anti-FOXM1 (K-19, Santa Cruz), anti-RB1 (4H1, Cell Signaling), anti-Cyclin D1 (EPR2241-32, Millipore), anti-Cyclin D3 (C-16, Santa Cruz) and anti-CDH1/fzr (DCS-266, Santa Cruz).

Promoter Constructs and Reporter Gene Assays

The CCNG2, CCNE2, SKP2, IGFBP-1 and 6DB promoter plasmids were previously described (Chen et al., 2006;Geng et al., 1996;Vigo et al., 1999;Furuyama et al., 2000). All other promoters were cloned by nested PCR from genomic DNA, which was extracted from U2OS cells using the DNeasy Blood & Tissue kit (Qiagen). Two-step nested PCR reactions were performed with the PfuUltra II Hotstart PCR Master Mix (Stratagene) as follows: ‘Nestout’ primers were utilized in the first PCR reaction (30 cycles), and 1 l of each reaction was subsequently used as template for a second PCR (30 cycles), using the respective ‘Nestin’ primers. Amplified promoter sequences were digested with the restriction enzymes indicated below and ligated into the pGL3 basic vector (Promega):

Gene / bp* / Nestout Fw primer / Nestout Rv primer / Nestin Fw primer / Nestin Rv primer / Digest
TSC22D1 / 2753 / CTAGCCATTTGTGCCACGGACATT / AAACTCCTAGATCCATCGCCACTG / ACCACGCGTAAGGAATCTAGCATATTTATGTGA / TGGAGATCTGCAATTGCAGCCAAAAACACCCTC / Mlu1/
Bgl2
DTL / 2448 / CATTGCTGGGAGGAGTGCAAATGA / TTACCCTTGCCATGGGAGAACTCA / ACCACGCGTTCTGTGTCTAAGATTGACACCTGA / TGGAGATCTCAGGGTCGGAGGAAAAGCCTCAGC / Mlu1/
Bgl2
MSH6 / 2424 / ACAGTAAGCCAAGATCACGCCACT / AGGCCTTGTTGGCATCACTCA / ACCCTCGAGTCCAGCCTCTGTGACAGAGCCAGA / TGGAAGCTTACCGACAGCCGGCAAGGCCCAACC / Xho1/
Hind3
PPIL5 / 2305 / ATCCATGCTTCAACAACCCAAGCC / TCCAAGAAGAACGCGAGACTGACA / ACCAGATCTCAACCCAAGCCTAAGTAACACCAA / TGGAAGCTTCTCGCCCAACGGCCACAACCACGT / Bgl2/
Hind3
CDCA5 / 2283 / TGTCAGTCTTCTTGGCTACCACA / AGCGAGAAGATTCCCAAACAAGCC / ACCACGCGTTCTAAGTCCAGGCAGGTGCTACCA / TGGAGATCTAACTTAGGCTCCGTCTCGAGCTCC / Mlu1/
Bgl2
XRCC2 / 2260 / CTCATGCCTGTAATCCCAGCACTT / AGCCCTATGGAAGGCACTACACAC / ACCACGCGTCTTGAGCCCAGGAGTTGTAAACCA / TGGAGATCTCGCCCCGAAGGCTCGGCGCAGGAG / Mlu1/
Bgl2

*Length (bp) of the cloned human promoter sequences, including 5’UTRs.

U2OS or HELA cells were seeded and transfected in 24-well plates. Cells were lysed 30 h post-transfection and reporter assays performed with the Dual Luciferase Reporter Assay System (Promega).

Destination Vectors, ORFs, Mutagenesis and FOXM1 Fragments

For protein expression in E. coli, cDNAs from the human ORFeome v3.1 collection (Rual et al., 2005; Lamesch et al., 2007) were gateway-cloned into a modified, gateway compatible version of pGEX-2tk (pGEX-2tk-DEST)(Braun et al., 2002). For mobility shift experiments in HEK293 cells, we used our in-house generated expression vector pLA-DEST. Stratagene’s QuikChange site-directed mutagenesis kit was used for all amino acid replacements, as well as for deletion of the FOXM1 transactivation domain to generate TAD, comprising FOXM1 amino acid sequence positions 1 to 437. FOXM1c was used as the prototypical FOXM1 isoform throughout this study. FOXM1 phosphorylation site mutants were obtained by replacing the phosphoacceptor residues (serine or threonine) with alanine. Catalytically inactive CDK4 and CDK6 were obtained by replacing the ATP binding lysine residues 35 and 43, respectively, with methionine. FOXM1 N-terminal (FOXM1/N), middle (FOXM1/M) and C-terminal (FOXM1/C) regions were PCR amplified, digested and inserted via BsrG1/EcoR1 into pGEX-2tk-DEST. Primer sequences and the protein sequence coverage (start and end positions) of the amplified FOXM1 fragments are described below; unpaired primer overhangs are emphasized in italic font, and restriction sites are marked in bold:

Fragment* / Position (aa) / Fw primer / Rv primer
FOXM1/N / 2 - 353 / TAGCTGTAC AAAACTAGCCCCCGTCGGCCACT / GACTGAAT TCATGTTCCGGCGGAGCTCTGGAT
FOXM1/M / 345 - 537 / TAGCTGTAC AATCCAGAGCTCCGCCGGAACAT / GACTGAAT TCTCCCTGTGTTGAATCACAAGCA
FOXM1/C / 529 - 763 / TAGCTGTAC ATGCTTGTGATTCAACACAGGGA / GACTGAATTC CTGTAGCTCAGGAATAAACTGG

*FOXM1/N, N-terminal fragment; FOXM1/M, middle fragment; FOXM1/C, C-terminal fragment; aa, amino acid.

Analysis of In Vitro Phosphorylation Sites by Mass Spectrometry

Phosphorylated FOXM1 fragments were run on an SDS gel, and protein-containing gel slices reduced with 20mM Tris (2-Carboxyethyl)-phosphine and alkylated with 1% iodoethanol, followed by overnight digestion with trypsin/chymotrypsin. The resulting peptides were extracted from the gel using 5% formic acid/45% acetonitrile in water, and dried. Samples were re-constituted in 0.1% formic acid and analyzed by Liquid Chromatograph Tandem Mass Spectrometry (LC-MS/MS)using a Linear Trap Quadrupole Fourier Transform (LTQ-FT) or Linear Trap Quadrupole (LTQ) Orbitrap Mass Spectrometer (Thermo Fisher Scientific Inc., San Jose, CA) hyphenated with an Agilent HPLC system (Agilent, Andover, MA). The resulting MS/MS fragmentation data were analyzed with MASCOT (Matrix Science, version 2.2). All identifications were manually inspected for correctness.

Quantification of In Vivo Phosphorylation Sites by Mass Spectrometry

FLAG-FOXM1 was immunoprecipitated from HeLa cells and resolved on an SDS gel. Gel bands corresponding to FOXM1 were cut from the gel and destained using 200 ml of 50% acetonitrile (ACN) and 50 mM ammonium bicarbonate at 37°C. The solution was removed and the process was repeated twice. Gel pieces were then resuspended in 200 ml of 50 mM ammonium bicarbonate and digested using LysC (Wako, Japan) at a 1:100 enzyme-to-protein ratio overnight at room temperature. Peptides were extracted from the gel using 200 ml of 5% formic acid and 50% ACN. Eluates were dried using a vacuum centrifuge, and the pellets resuspended in 50 ml of 200 mM Na-HEPES. Each sample was divided in two, followed by addition of 10 mg of amino reactive tandem mass tag reagents (TMT). Samples were assigned with different mass tags: reagents 126 and 129 (FOXM1 overexpression alone), reagents 127 and 130 (FOXM1 plus cyclin D3-CDK6) and reagents 128 and 131 (FOXM1 plus cyclin D3-CDK6KM kinase-inactive mutant). Samples were incubated for 1 h at room temperature with TMT reagent. Reactions were quenched with 4 ml of 5% hydroxylamine (v/v) for 15 minutes at room temperature, and acidified with 10 ml of 10% formic acid, combined and dried by vacuum centrifugation. Samples were resuspended in 1% formic acid and desalted using a stage tip (C18) (Rappsilber et al., 2003).

Samples were analyzed withLiquid Chromatograph Tandem Mass Spectrometry (LC-MS/MS)using an LTQ-Orbitrap Velos hybrid mass spectrometer (MS) (Thermo Fisher, San Jose, CA) equipped with a Famos autosampler (LC Packings, Sunnyvale, CA) and an Accela 600 pump (Thermo Fisher). 4 µL of sample was loaded onto a pulled fused silica microcapillary column (125 µm, 20 cm bed volume) packed with C18reverse-phase resin (Magic C18 AQ; 3-µm particles; 200-Å pore size; Michrom Bioresources, Auburn, CA) using a Famos autosampler (LC Packings, San Francisco, CA). The peptides were separated using the Accela pump across a 70 min linear gradient of 6−27% buffer B (97% ACN, 0.125% FA) with an in-column flow rate of 0.5-1ul/minute. Electrospray ionization was performed using a 1.8 kV through a PEEK junction inlet of the microcapillary column. In each data collection cycle, one full MS scan (300–1500m/z) was acquired in the Orbitrap (6 × 104resolution setting, automatic gain control (AGC) target of 3 x 106) and the top 10 most abundant ions were selected for isolation and fragmentation in the ion-trap by collision-induced dissociation (CID)-MS2. Ions were selected for isolation when their intensity reached a threshold of 500 counts. CID was performed using a 2 m/z isolation window, an AGC setting of 2 x 103, a maximum ion accumulation time of 150 ms and a wide band activation. Previously selected ions were dynamically excluded for 90 s. After each MS2 acquisition, the most intense ions within the m/z range between 110-160% of the precursor m/z were selected for high-energy collision dissociation (HCD)-MS3. Fragment ion isolation used a 4 m/z window, the AGC setting was set to 2 x 104 and the maximum ion time was 250 ms. Normalized collision energies were set to 35% and 60% with an activation time of 10 ms and 20 ms for MS2 and MS3 methods, respectively.

An in-house suite of software tools was used to convert mass spectrometric data from raw file to mzmxl format. Erroneous peptide ion charge state and monoisotopic m/z was corrected (Huttlin et al., 2010). MS/MS spectra assignments were made using the Sequest algorithm (Yates et al., 1995) using the entire mouse IPI database (version 3.6). Sequest searches were performed using a target-decoy strategy (Elias and Gygi, 2007) with the mouse IPI database in correct orientation (forward database) and the same database but with all sequences in reverse orientation (reverse database). The mouse IPI database contained 113,477 entries. The data was searched using a precursor ion tolerance of 10 ppm, considering LysC specificity and allowing two missed cleavages. A static modification was assigned for TMT labels on the N-termini and lysine residues with a mass of 229.162932 Da. In addition, a dynamic modification on methionine was also considered (+15.99492 Da). A peptide level false discovery (FDR) rate of less than 1% was used as a threshold for peptide identifications using the target decoy strategy. Addition filtering was achieved using a linear discriminant analysis, which combined several parameters into a single probability for each peptide and these probabilities were used to achieve a less than 1% FDR. The parameters used for linear discriminant analysis were Xcorr, DCn, peptide mass accuracy and charge state, and peptide length.

TMT reporter ion intensities were determined as follows:for reporter ion quantification, a 0.06 m/z window centered on the theoretical m/z value for each reporter ion was monitored and the signal intensity closest to the theoretical mass was recorded. Reporter ion intensities were normalized to the accumulation time of each MS3 spectrum. Peptide quantification was determined using the average reporter intensity for each replicate for each sample condition.