T-cell isolation and stimulation
Normal, non-activated, quiescent human T cells were isolated by negative selection (Dynal, Invitrogen) from single-donor buffy coats (Lea et al, 2003
) and assays to analyze T-cell activation and cell-cycle progression were as described in Lea et al (2003)
Isolation and trypsinization of chromatin/nuclear matrix proteins
Non-activated or stimulated T cells from four individual donors (biological replicates) (usually 3 × 107
cells) were pelletted and resuspended in 20 μl/106
cells ice-cold CSK buffer (10 mM PIPES pH 6.8, 100 mM NaCl, 300 mM sucrose, 1 mM MgCl2
, 1 mM EDTA, 0.1% (v/v) Triton X-100, 0.1 mM ATP, protease and phosphatase inhibitors (Complete, Roche)) as described (Lea et al, 2003
). The insoluble pellet, containing chromatin/nuclear matrix-bound proteins (Krude et al, 1996
), was centrifuged at 1000 g
and washed 3 × with 50 μl PBS. The CSK-soluble proteins (‘free' fraction) were precipitated by adding four volumes of −20°C acetone. Both ‘free' and ‘chromatin' pellets were resuspended in 10 mM Tris–HCl pH 7.8, 10 mM KCl, 1.5 mM MgCl2
. Residual Triton X-100 was removed with the OrgoSol DetergentOUT Detergent removal kit (G-Biosciences). Following detergent removal and resuspension, proteins were denatured by addition of trifluoroethanol (TFE) to 50% (v/v). Proteins were reduced (15 mM DTT at 56°C for 45 min) and alkylated (55 mM iodoacetamide at 25°C for 30 min). Samples were diluted 10-fold with digestion buffer (50 mM Tris–HCl pH 7.8, 2 mM CaCl2
) to decrease the TFE concentration to 5% (v/v). Each sample was treated with 2 μg Proteomics Grade trypsin (Sigma) and incubated at 37°C for 4 h. Trypsin activity was halted by the addition of 1% formic acid, and sample volume was reduced to ~20 μl by SpeedVac centrifugation. Samples were resuspended to 150 μl with Buffer C (95% H2
O, 5% acetonitrile (ACN), 0.1% formic acid). To remove undesirable contaminants, digested samples were cleaned up by reverse phase chromatography on Hypersep C18 Columns (Thermo) and filtered on Microcon 10 kDa Spin Tubes (Centricon) prior to LC-MS/MS.
LC/MS/MS analysis was carried out on an Eksigent nanoflow LC system connected to an LTQ-Orbitrap XL (Thermo). Four injections were carried out per sample. Tryptic peptide mixtures were separated online by reverse phase chromatography on a Zorbax C18 reverse phase column (Agilent) by 5–38% ACN gradient over the course of 230 min. Eluted peptides were directly injected by electrospray ionization into an LTQ-Orbitrap XL hybrid mass spectrometer (Thermo) equipped with a nano-spray ion source for analysis. Full spectra (MS1) were collected at 100 000 resolution. Ion fragmentation spectra (MS2) were collected in a data-dependent manner, with ions required to carry +2 or greater charge for MS2 selection. The top 12 most intense qualifying peaks were selected per round, with peaks selected twice within 30 s excluded from selection for 45 s. Spectra were searched using the Sequest search algorithm against the non-redundant Ensembl 57 Homo sapiens (NCBI36) protein-coding data set. Search results were submitted to PeptideProphet (Keller et al, 2002
) and ProteinProphet (Nesvizhskii et al, 2003
) via TPP (Keller et al, 2005
) and filtered to 1% false positive rate using a target-decoy method with a reverse-concatenated database.
Generation of protein lists
For each biological replicate fraction, four separate LC-MS/MS injections were performed and analyzed independently. Data sets corresponding to the same biological replicate fraction were then merged, with spectral counts summed within each set. All data sets were aligned, and proteins observed in only one biological replicate fraction were removed. Keratins were also removed from the data set. To account for the occurrence of degenerate peptides assigned to multiple proteins, the data set was curated to assign all peptides to a single unique protein or protein group. Protein groups consisted of the smallest set of proteins needed to account for non-unique peptide identifications, whereby all group members shared equal evidence of occurrence.
Proteins were quantified by comparison of spectral counts, as described in Lu et al (2007)
and Vogel and Marcotte (2008)
. In summary, frequencies of observation were calculated from spectral counts for each protein in each biological replicate fraction, and individual Z
-scores were calculated by comparing frequencies at 0 and 40 h time points to determine the statistical significance of differences in relative abundance (Lu et al, 2007
; Vogel and Marcotte, 2008
). A z
-score of ±1.96, corresponding to a P
-value <0.05, was used as a determinant of significance. Composite Z
-scores for free and bound fractions were then calculated from the individual Z
-scores for each biological replicate. Fold-change was estimated as a ratio of frequencies of observation for each protein at 0 and 40 h time points. For each protein, a pseudo-count of 1 was added to each spectral count total prior to frequency calculations in order to adjust for proteins with a zero-count.
siRNA and transfection
Pools of SF3B2, SF3B4 and control siRNA (SmartPool, Dharmacon), or each of four individual siRNA (siGenome, Dharmacon) or stabilized siRNA (Stealth, Invitrogen) were transfected into quiescent primary human T cells by electroporation (Nucleofection, Amaxa) (Orr et al, 2010
). Sequences of the siRNA used are shown on the next page (antisense sequences).
Cells transfected with siRNA were cultured overnight without stimulus to recover, stimulated with PMA/ionomycin and samples were typically taken 72 h later for further analyses. Samples were also analyzed at other time points up to 128 h (5 days) later, as indicated in the text. For eIF6, the siRNA-transfected cells were left in culture for 3 days to deplete low levels of endogenous protein present in quiescent cells before stimulation. The level of SF3B2 or eIF6 protein expressed after siRNA transfection was determined by western blotting using 2 × 105 cells per well and the signal obtained was compared with incremental amounts of the sample transfected with control siRNA (4 × 103–2 × 105 cells, that is, 2–100% of the SF3B2 siRNA-transfected cells) that was run on the same blot.
Western blotting was carried out with 4–12% (w/v) polyacrylamide Bis-Tris gels (Invitrogen-Novex). Antibodies used were Apobec 3C (C8-1, Professor Neuberger, MRC, Cambridge, UK); BAF53 (Dr Michael Cole, Dartmouth College, USA); HnRNP-K (C-terminus, Professor Bomsztyk, U Washington, USA); p68 (DDX5) (PAb204, Upstate; from Professor Frances Fuller-Pace, Dundee, Scotland); Prohibitin (Ab-1, Neomarkers; from Professor Eric Lam, Imperial College, London); RUNX3 (Professor Paul Farrell, Imperial College, London); SAFB1 (Dr S Osterreich, MD Anderson, Houston, TX, USA); Topoisomerase I (C21, BD-Pharmingen; from Dr Andrew Porter, Imperial College, London); MeCP2 (ab3752), HP1γ (ab10480), Histone H3 (ab1791) SF3B4 (ab11803) (Abcam); cyclin A2 (BF683), NF-κB-p50 (K10-895.12.50), PARP-1 (C2-10), pRb (PMG3245), STAT3 (Ab-84) (BD-Pharmingen); p54nrb (611278), BAF190 (610390), Mcm2 (BM28) (BD Transduction Laboratories); eIF6 (#3263; from Professor Alan Warren, University of Cambridge); p44/42 MAP kinase; phospho-mTOR (S2448), phospho-eIF4EBP1 (T37/46), phospho-eIF4E (S209), RPS6 (#2317) (Cell Signaling Technology); GAPDH (MAB374), PP1-α (AB4082, Chemicon); phospho-pRb(S807/811) (New England BioLabs, Hertfordshire, UK); Aly/THOC4 (sc-32311), cdc2 (Cdk1) (17), Cdc6 (C19), cdk6 (C21), cyclin D3 (C16), E2F-1 (C20), HELLS/Lsh (H-4, sc-46665), Lamin A/C (N-18; sc-6215), SAM68 (C20), Xin1/Xirp1 (D8, T-20) (Santa Cruz Biotechnology, Inc., Santa Cruz, CA); hnRNP-A2/B1 (DP3B3), Mcm7 (DCS 141) (Sigma); LSP-1 (Ab-16) (Transduction Laboratories). In the course of this study, western blot membranes were either routinely cut into sections and each section probed with a different antibody or the blots were re-probed sequentially with different antibodies, which recognize proteins of different molecular weights (e.g., Mcm2 and Mcm7). Sections of the films that showed the bands corresponding to each antibody (determined by size comparison with molecular weight markers) were selected in the scanner software (Epson Scan 3.04), scanned individually as separate image files and saved as JPEGs. In some cases, changes in brightness and/or contrast were applied linearly equally across the entire image. These are available as Source Data.
Cytospin preparations of 40 000 T cells were made on poly-lysine-coated microscope slides (Shandon Cytospin 4 (Thermo)). Slides were fixed in 2% (w/v) paraformaldehyde for 30 min at room temperature, washed three times with PBS and stored immersed in PBS at 4°C. Cells were permeabilized with 0.2% (v/v) Triton X-100 in PBS for 10 min at room temperature, washed three times in PBS, blocked in PBS/10% (w/v) BSA and washed three times with PBS. Slides were incubated overnight with 1/100 Xirp1 antibody in blocking solution at 4°C, washed three times with PBS, incubated for 1 h in 1:200 TRITC-conjugated donkey anti-goat secondary antibody (Jackson Laboratories) in blocking solution, then washed as above. Cells were counterstained with a DNA stain (DAPI), rinsed in PBS, mounted using Antifade and analyzed with an Olympus (Provis AX70, Olympus) fluorescent microscope with DAPI/FITC/rhodamine triple pass filters. Confocal microscopy and Z-stacks of 50 optical slices was performed using a Leica DMIRE2 instrument (with thanks to Dr M-J Bijlmakers, Immunology Department, KCL).
Gene expression analysis
Gene expression arrays (U133A, Affymetrix) were probed with RNA isolated from at least four biological replicates of T cells in G0
and at time points corresponding to cells during the G0
transition (1–8 h), mid/late G1
(24 and 32 h) and later in the first cell cycle (48 h). The procedures were according to the manufacturer's protocols (one-round amplification; Affymetrix). Data were normalized using the VSN algorithm (version 3.2.1) (Huber et al, 2002
) implemented in Bioconductor (Biobase/Affymetrix versions 1.16.2/1.16.0) (Gentleman et al, 2004
) and R version 2.6.1 (Dessau and Pipper, 2008
). The probe sets were mapped on the ENSEMBL (Hubbard et al, 2009
), NCBI 36 assembly, allowing for one alignment error between the probe and transcripts sequences. Probe sets containing probes matching more than one gene, or which matches contained alignment errors, were ignored for further analysis. The final expression values were averaged over replicated time points, and over probe sets reporting the same gene.
Protein interaction network analysis
A network consisting of human protein–protein interaction pairs was generated using HumanNet (http://www.functionalnet.org/humannet/
) (Lee et al, 2011a
). HumanNet is a probabilistic functional gene network derived from the integration of large-scale ‘omics' data sets across multiple species. Proteins showing significant changes in abundance (1.96 Z
2) in the bound and unbound fractions were mapped onto the network and visualized using Cytoscape (http://www.cytoscape.org
Biological replicate experiments and statistical analysis
Each experiment was carried out with cells isolated from at least three independent donors (n=X indicates X independent donors). Results are expressed as mean±standard error of the mean for the n=X experiments. The significance of differences between conditions was assessed by the Students ‘t'-test.
Archiving mass spectrometry and microarray data
All mass spectrometry data described in this paper are publically available through PeptideAtlas (http://www.peptideatlas.org/PASS/PASS00016
). Microarray data have been deposited in GEO with the accession number GSE32607