Supplementary Information Supplementary Figure Legends Figure S1. Characterization of Human Histone H1 Variant-specific Antibodies

and analysis of the H1 variants pattern in T47D cells. (A) Alignment of N-terminal sequences of human H1 variants underlying peptide sequences used for production of polyclonal antibodies. (B) Newly-generated antibodies against H1.1, 2, 3 and 5 recognize specifically its corresponding variant in a Western blot with recombinant human H1 variants produced in yeast (courtesy of N. Happel and D. Doenecke). H1.4 antibody does not recognize its histone. A commercial H1.0 antibody (Abcam ab11079) is also highly specific. Commercial histone H1 (phospho T146) antibody (Abcam ab3596) uniquely recognizes recombinant H1.4 variant produced in yeast. (C) Western blot with the H1 phospho-T146 antibody on recombinant H1.4 produced in bacteria (lane 1), total H1 extracted from T47D cells untreated (lane 2) or treated with colcemid (50 ng/ml) overnight (lane 3).


Customized microarray hybridization and data analysis
A cDNA microarray platform containing 826 cDNA clones that were selected for its involvement in breast cancer was generated (B. Miñana, L. Sumoy, M. Beato, A. Jordan, C. Ballare, M. Melia; GEO accession number GPL5953; http://www.ncbi.nlm.nih.gov/projects/geo/index.cgi). cDNA inserts were PCR amplified and spotted on Corning UltraGAPS amino-modified glass slides. mRNA samples were processed for first and second strand cDNA synthesis and in vitro transcription with T7 RNA polymerase, basically as reported elsewhere [1,2].
Universal reference RNA was obtained from Statagene. RNA was directly labeled with Cy3-or Cy5-dUTP (Amersham) and hybridized to spotted slides as described [3]. After washing, fluorescent images were obtained using a G2565BA Microarray Scanner System (Agilent) and TIFF images were quantified using GenePix 6.0 (Molecular Devices) software.
Raw data was processed using MARGE, an in house developed web implementation of LIMMA, a microarray statistical analysis package of Bioconductor (http://www.bioconductor.org) that is run in the R programming environment [4][5][6]. Gene intensities were background subtracted (taking mean of channel intensities and median of background). Spots with intensities <2 times the local background in either or both dye filter channels (Cy3 or Cy5) as well as controls were excluded from normalization, and were referred as "not reliable". An intensity dependent normalization algorithm (global lowess) was applied using a smoothing factor f=0.2 for all experiments. Normalized Log 2 Ratios (Intensity Cy5/Intensity Cy3) were scaled so that they all had the same median absolute standard deviation across all the arrays, to give the same weight to each gene, and not only due to the magnitude of the expression ratio [7]. The computed B statistic rank value from all replicate hybridizations was used to determine the genes with significant changes. We considered genes that showed a 1.4-fold gene up or down-regulation relative to control sample with a B-rank value above the 90 th percentile as significant. The value of fold change or copy number relative change Sancho M. et al.

5
was calculated as 2 Log2Ratio , if the value of the ratio was >0, or 2 -1/Log2Ratio , if it was <0. In order to do the statistical analysis of the data, we have used the opensource, freely available software package for microarray data management and analysis TM4 obtained from TIGR (http://www.tigr.org/software/) [8] that applies the Significance Analyses of Microarrays (SAM) method. The raw data was summarized per probe using BeadStudio software Gene Expression module and the summary data file was processed using the PILLA web einterface tool (Lozano et al, unpublished), an implementation of the Lumi package [9] developed within the Bioconductor project in the R statistical programming environment [5]. Data were normalized using the rsn method and vst as the variance stabilization method. The log2 intensities were median centered and log ratios were computed as differences in log2 intensities for each probe. The SAM (significance analysis of microarrays) two class unpaired comparison test was applied with 100 permutations to detect statistically significant differences in gene expression between treated and control conditions [10].

Illumina microarray hybridization and data analysis
All microarray hybridizations were performed at the Microarrays unit of the Centre de Regulació Genòmica (CRG).