Targeting hepatic serine-arginine protein kinase 2 ameliorates alcohol-associated liver disease by alternative splicing control of lipogenesis.

BACKGROUND AND AIMS: Lipid accumulation induced by alcohol consumption is not only an early pathophysiological response but also a prerequisite for the progression of alcohol-associated liver disease (ALD). Alternative splicing regulates gene expression and protein diversity; dysregulation of this process is implicated in human liver diseases. However, how the alternative splicing regulation of lipid metabolism contributes to the pathogenesis of ALD remains undefined. APPROACH AND RESULTS: Serine-arginine-rich protein kinase 2 (SRPK2), a key kinase controlling alternative splicing, is activated in hepatocytes in response to alcohol, in mice with chronic-plus-binge alcohol feeding, and in patients with ALD. Such induction activates sterol regulatory element-binding protein 1 and promotes lipogenesis in ALD. Overexpression of FGF21 in transgenic mice abolishes alcohol-mediated induction of SRPK2 and its associated steatosis, lipotoxicity, and inflammation; these alcohol-induced pathologies are exacerbated in FGF21 knockout mice. Mechanistically, SRPK2 is required for alcohol-mediated impairment of serine-arginine splicing factor 10, which generates exon 7 inclusion in lipin 1 and triggers concurrent induction of lipogenic regulators-lipin 1ß and sterol regulatory element-binding protein 1. FGF21 suppresses alcohol-induced SRPK2 accumulation through mammalian target of rapamycin complex 1 inhibition-dependent degradation of SRPK2. Silencing SRPK2 rescues alcohol-induced splicing dysregulation and liver injury in FGF21 knockout mice. CONCLUSIONS: These studies reveal that (1) the regulation of alternative splicing by SRPK2 is implicated in lipogenesis in humans with ALD; (2) FGF21 is a key hepatokine that ameliorates ALD pathologies largely by inhibiting SRPK2; and (3) targeting SRPK2 signaling by FGF21 may offer potential therapeutic approaches to combat ALD.

Phosphorylated MED1 links transcription recycling and cancer growth.

Mediator activates RNA polymerase II (Pol II) function during transcription, but it remains unclear whether Mediator is able to travel with Pol II and regulate Pol II transcription beyond the initiation and early elongation steps. By using in vitro and in vivo transcription recycling assays, we find that human Mediator 1 (MED1), when phosphorylated at the mammal-specific threonine 1032 by cyclin-dependent kinase 9 (CDK9), dynamically moves along with Pol II throughout the transcribed genes to drive Pol II recycling after the initial round of transcription. Mechanistically, MED31 mediates the recycling of phosphorylated MED1 and Pol II, enhancing mRNA output during the transcription recycling process. Importantly, MED1 phosphorylation increases during prostate cancer progression to the lethal phase, and pharmacological inhibition of CDK9 decreases prostate tumor growth by decreasing MED1 phosphorylation and Pol II recycling. Our results reveal a novel role of MED1 in Pol II transcription and identify phosphorylated MED1 as a targetable driver of dysregulated Pol II recycling in cancer.

Diverse AR-V7 cistromes in castration-resistant prostate cancer are governed by HoxB13.

The constitutively active androgen receptor (AR) splice variant 7 (AR-V7) plays an important role in the progression of castration-resistant prostate cancer (CRPC). Although biomarker studies established the role of AR-V7 in resistance to AR-targeting therapies, how AR-V7 mediates genomic functions in CRPC remains largely unknown. Using a ChIP-exo approach, we show AR-V7 binds to distinct genomic regions and recognizes a full-length androgen-responsive element in CRPC cells and patient tissues. Remarkably, we find dramatic differences in AR-V7 cistromes across diverse CRPC cells and patient tissues, regulating different target gene sets involved in CRPC progression. Surprisingly, we discover that HoxB13 is universally required for and colocalizes with AR-V7 binding to open chromatin across CRPC genomes. HoxB13 pioneers AR-V7 binding through direct physical interaction, and collaborates with AR-V7 to up-regulate target oncogenes. Transcriptional coregulation by HoxB13 and AR-V7 was further supported by their coexpression in tumors and circulating tumor cells from CRPC patients. Importantly, HoxB13 silencing significantly decreases CRPC growth through inhibition of AR-V7 oncogenic function. These results identify HoxB13 as a pivotal upstream regulator of AR-V7-driven transcriptomes that are often cell context-dependent in CRPC, suggesting that HoxB13 may serve as a therapeutic target for AR-V7-driven prostate tumors.

Age-associated genes in human mammary gland drive human breast cancer progression.

BACKGROUND: Aging is a comorbidity of breast cancer suggesting that aging-associated transcriptome changes may promote breast cancer progression. However, the mechanism underlying the age effect on breast cancer remains poorly understood. METHOD: We analyzed transcriptomics of the matched normal breast tissues from the 82 breast cancer patients in The Cancer Genome Atlas (TCGA) dataset with linear regression for genes with age-associated expression that are not associated with menopause. We also analyzed differentially expressed genes between the paired tumor and non-tumor breast tissues in TCGA for the identification of age and breast cancer (ABC)-associated genes. A few of these genes were selected for further investigation of their malignancy-regulating activities with in vitro and in vivo assays. RESULTS: We identified 148 upregulated and 189 downregulated genes during aging. Overlapping of tumor-associated genes between normal and tumor tissues with age-dependent genes resulted in 14 upregulated and 24 downregulated genes that were both age and breast cancer associated. These genes are predictive in relapse-free survival, indicative of their potential tumor promoting or suppressive functions, respectively. Knockdown of two upregulated genes (DYNLT3 and P4HA3) or overexpression of the downregulated ALX4 significantly reduced breast cancer cell proliferation, migration, and clonogenicity. Moreover, knockdown of P4HA3 reduced growth and metastasis whereas overexpression of ALX4 inhibited the growth of xenografted breast cancer cells in mice. CONCLUSION: Our study suggests that transcriptome alterations during aging may contribute to breast tumorigenesis. DYNLT3, P4HA3, and ALX4 play significant roles in breast cancer progression.

A novel algorithm for network-based prediction of cancer recurrence.

To develop accurate prognostic models is one of the biggest challenges in "omics"-based cancer research. Here, we propose a novel computational method for identifying dysregulated gene subnetworks as biomarkers to predict cancer recurrence. Applying our method to the DNA methylome of endometrial cancer patients, we identified a subnetwork consisting of differentially methylated (DM) genes, and non-differentially methylated genes, termed Epigenetic Connectors (EC), that are topologically important for connecting the DM genes in a protein-protein interaction network. The ECs are statistically significantly enriched in well-known tumorgenesis and metastasis pathways, and include known epigenetic regulators. Importantly, combining the DMs and ECs as features using a novel random walk procedure, we constructed a support vector machine classifier that significantly improved the prediction accuracy of cancer recurrence and outperformed several alternative methods, demonstrating the effectiveness of our network-based approach.

Agonist and antagonist switch DNA motifs recognized by human androgen receptor in prostate cancer.

Human transcription factors recognize specific DNA sequence motifs to regulate transcription. It is unknown whether a single transcription factor is able to bind to distinctly different motifs on chromatin, and if so, what determines the usage of specific motifs. By using a motif-resolution chromatin immunoprecipitation-exonuclease (ChIP-exo) approach, we find that agonist-liganded human androgen receptor (AR) and antagonist-liganded AR bind to two distinctly different motifs, leading to distinct transcriptional outcomes in prostate cancer cells. Further analysis on clinical prostate tissues reveals that the binding of AR to these two distinct motifs is involved in prostate carcinogenesis. Together, these results suggest that unique ligands may switch DNA motifs recognized by ligand-dependent transcription factors in vivo. Our findings also provide a broad mechanistic foundation for understanding ligand-specific induction of gene expression profiles.

Genome-wide analysis reveals positional-nucleosome-oriented binding pattern of pioneer factor FOXA1.

The compaction of nucleosomal structures creates a barrier for DNA-binding transcription factors (TFs) to access their cognate cis-regulatory elements. Pioneer factors (PFs) such as FOXA1 are able to directly access these cis-targets within compact chromatin. However, how these PFs interplay with nucleosomes remains to be elucidated, and is critical for us to understand the underlying mechanism of gene regulation. Here, we have conducted a computational analysis on a strand-specific paired-end ChIP-exo (termed as ChIP-ePENS) data of FOXA1 in LNCaP cells by our novel algorithm ePEST. We find that FOXA1 chromatin binding occurs via four distinct border modes (or footprint boundary patterns), with a preferential footprint boundary patterns relative to FOXA1 motif orientation. In addition, from this analysis three fundamental nucleotide positions (oG, oS and oH) emerged as major determinants for blocking exo-digestion and forming these four distinct border modes. By integrating histone MNase-seq data, we found an astonishingly consistent, 'well-positioned' configuration occurs between FOXA1 motifs and dyads of nucleosomes genome-wide. We further performed ChIP-seq of eight chromatin remodelers and found an increased occupancy of these remodelers on FOXA1 motifs for all four border modes (or footprint boundary patterns), indicating the full occupancy of FOXA1 complex on the three blocking sites (oG, oS and oH) likely produces an active regulatory status with well-positioned phasing for protein binding events. Together, our results suggest a positional-nucleosome-oriented accessing model for PFs seeking target motifs, in which FOXA1 can examine each underlying DNA nucleotide and is able to sense all potential motifs regardless of whether they face inward or outward from histone octamers along the DNA helix axis.

Integrative analysis identifies targetable CREB1/FoxA1 transcriptional co-regulation as a predictor of prostate cancer recurrence.

Identifying prostate cancer-driving transcription factors (TFs) in addition to the androgen receptor promises to improve our ability to effectively diagnose and treat this disease. We employed an integrative genomics analysis of master TFs CREB1 and FoxA1 in androgen-dependent prostate cancer (ADPC) and castration-resistant prostate cancer (CRPC) cell lines, primary prostate cancer tissues and circulating tumor cells (CTCs) to investigate their role in defining prostate cancer gene expression profiles. Combining genome-wide binding site and gene expression profiles we define CREB1 as a critical driver of pro-survival, cell cycle and metabolic transcription programs. We show that CREB1 and FoxA1 co-localize and mutually influence each other's binding to define disease-driving transcription profiles associated with advanced prostate cancer. Gene expression analysis in human prostate cancer samples found that CREB1/FoxA1 target gene panels predict prostate cancer recurrence. Finally, we showed that this signaling pathway is sensitive to compounds that inhibit the transcription co-regulatory factor MED1. These findings not only reveal a novel, global transcriptional co-regulatory function of CREB1 and FoxA1, but also suggest CREB1/FoxA1 signaling is a targetable driver of prostate cancer progression and serves as a biomarker of poor clinical outcomes.

BRCA1 Interacting Protein COBRA1 Facilitates Adaptation to Castrate-Resistant Growth Conditions.

COBRA1 (co-factor of BRCA1) is one of the four subunits of the negative elongation factor originally identified as a BRCA1-interacting protein. Here, we provide first-time evidence for the oncogenic role of COBRA1 in prostate pathogenesis. COBRA1 is aberrantly expressed in prostate tumors. It positively influences androgen receptor (AR) target gene expression and promoter activity. Depletion of COBRA1 leads to decreased cell viability, proliferation, and anchorage-independent growth in prostate cancer cell lines. Conversely, overexpression of COBRA1 significantly increases cell viability, proliferation, and anchorage-independent growth over the higher basal levels. Remarkably, AR-positive androgen dependent (LNCaP) cells overexpressing COBRA1 survive under androgen-deprivation conditions. Remarkably, treatment of prostate cancer cells with well-studied antitumorigenic agent, 2-methoxyestradiol (2-ME2), caused significant DNA methylation changes in 3255 genes including COBRA1. Furthermore, treatment of prostate cancer cells with 2-ME2 downregulates COBRA1 and inhibition of prostate tumors in TRAMP (transgenic adenocarcinomas of mouse prostate) animals with 2-ME2 was also associated with decreased COBRA1 levels. These observations implicate a novel role for COBRA1 in progression to CRPC and suggest that COBRA1 downregulation has therapeutic potential.

Cholesterol Regulates Monocyte Rolling through CD44 Distribution.

Cholesterol is an important risk factor of atherosclerosis, due to its active uptake by monocytes/macrophages. Monocyte recruitment from flowing blood to atherosclerotic foci is the key first step in the development of atherosclerosis. Cholesterol content alters cell membrane stiffness, and lateral lipid and protein diffusion. We hypothesized that cholesterol content will modulate the recruitment of monocytes to inflamed endothelial surface by altering the dynamics of adhesion receptors. We depleted or enriched the cellular cholesterol levels using methyl-ß-cyclodextran in freshly isolated human monocytes. We investigated the effect of these changes on the mechanics of monocyte rolling on E-selectin surfaces at 1 dyn/cm2 in microchannels. Using imaging flow cytometry and atomic force microscopy, we characterized the distribution of lipid rafts and the E-selectin counterreceptor CD44 on the monocyte surface. We observed that lower levels of cholesterol resulted in the uniform, CD44-mediated rolling of monocytes on the E-selectin-coated surfaces. We also observed that cells depleted of cholesterol had higher membrane fluidity, and more uniform distribution of CD44 counterreceptor, which resulted in smooth motion of the cells compared to cells enriched with cholesterol. This work demonstrates that cholesterol can modulate monocyte adhesion by regulating the receptor mobility, and our results provide insights into the biophysical regulation of inflammation for the better understanding of diseases like atherosclerosis and hypercholesterolemia.

Deep sequencing of HPV E6/E7 genes reveals loss of genotypic diversity and gain of clonal dominance in high-grade intraepithelial lesions of the cervix.

BACKGROUND: Human papillomavirus (HPV) is the carcinogen of almost all invasive cervical cancer and a major cause of oral and other anogenital malignancies. HPV genotyping by dideoxy (Sanger) sequencing is currently the reference method of choice for clinical diagnostics. However, for samples with multiple HPV infections, genotype identification is singular and occasionally imprecise or indeterminable due to overlapping chromatograms. Our aim was to explore and compare HPV metagenomes in abnormal cervical cytology by deep sequencing for correlation with disease states. RESULTS: Low- and high-grade intraepithelial lesion (LSIL and HSIL) cytology samples were DNA extracted for PCR-amplification of the HPV E6/E7 genes. HPV+ samples were sequenced by dideoxy and deep methods. Deep sequencing revealed ~60% of all samples (n = 72) were multi-HPV infected. Among LSIL samples (n = 43), 27 different genotypes were found. The 3 dominant (most abundant) genotypes were: HPV-39, 11/43 (26%); -16, 9/43 (21%); and -35, 4/43 (9%). Among HSIL (n = 29), 17 HPV genotypes were identified; the 3 dominant genotypes were: HPV-16, 21/29 (72%); -35, 4/29 (14%); and -39, 3/29 (10%). Phylogenetically, type-specific E6/E7 genetic distances correlated with carcinogenic potential. Species diversity analysis between LSIL and HSIL revealed loss of HPV diversity and domination by HPV-16 in HSIL samples. CONCLUSIONS: Deep sequencing resolves HPV genotype composition within multi-infected cervical cytology. Biodiversity analysis reveals loss of diversity and gain of dominance by carcinogenic genotypes in high-grade cytology. Metagenomic profiles may therefore serve as a biomarker of disease severity and a population surveillance tool for emerging genotypes.

START: a system for flexible analysis of hundreds of genomic signal tracks in few lines of SQL-like queries.

BACKGROUND: A genomic signal track is a set of genomic intervals associated with values of various types, such as measurements from high-throughput experiments. Analysis of signal tracks requires complex computational methods, which often make the analysts focus too much on the detailed computational steps rather than on their biological questions. RESULTS: Here we propose Signal Track Query Language (STQL) for simple analysis of signal tracks. It is a Structured Query Language (SQL)-like declarative language, which means one only specifies what computations need to be done but not how these computations are to be carried out. STQL provides a rich set of constructs for manipulating genomic intervals and their values. To run STQL queries, we have developed the Signal Track Analytical Research Tool (START, http://yiplab.cse.cuhk.edu.hk/start/ ), a system that includes a Web-based user interface and a back-end execution system. The user interface helps users select data from our database of around 10,000 commonly-used public signal tracks, manage their own tracks, and construct, store and share STQL queries. The back-end system automatically translates STQL queries into optimized low-level programs and runs them on a computer cluster in parallel. We use STQL to perform 14 representative analytical tasks. By repeating these analyses using bedtools, Galaxy and custom Python scripts, we show that the STQL solution is usually the simplest, and the parallel execution achieves significant speed-up with large data files. Finally, we describe how a biologist with minimal formal training in computer programming self-learned STQL to analyze DNA methylation data we produced from 60 pairs of hepatocellular carcinoma (HCC) samples. CONCLUSIONS: Overall, STQL and START provide a generic way for analyzing a large number of genomic signal tracks in parallel easily.

JAK2V617F influences epigenomic changes in myeloproliferative neoplasms.

Negative valine (V) to phenylalanine (F) switch at the Janus kinase (JAK2) 617 codon (V617F) is the dominant driver mutation in patients with myeloproliferative neoplasms (MPNs). JAK2V617F was proved to be sufficient for cell transformation; however, independent mutations might influence the following epigenomic modifications. To assess the JAK2V617F-induced downstream epigenomic changes without interferences, we profiled the epigenomic changes in ectopically expressed JAK2V617F in Ba/F3 cells. Antibodies against phosphorylated signal transducer and activator of transcription 3 (pSTAT3) and enhancer of zeste homolog 2 (EZH2) were used for chromatin-immunoprecipitation sequencing (ChIP-seq) to detect the downstream epigenomic targets in the JAK2-STAT3 signaling pathway. To confirm the JAK2V617F-induced epigenetic changes in vivo, DNA methylation changes in the target loci in patients with MPNs were detected through methylation-specific polymerase chain reaction and were clustered against the changes within controls. We found that ectopically expressed JAK2V617F in Ba/F3 cells reduced the binding specificity; it was associated with cis-regulatory elements and recognized DNA motifs in both pSTAT3-downstream and EZH2-associated targets. Overlapping target loci between the control and JAK2V617F were <3% and 0.4%, respectively, as identified through pSTAT3 and EZH2 ChIP-seq. Furthermore, the methylation changes in the direct target loci (FOXH1, HOXC9, and SRF) were clustered independently from the control locus (L1TD1) and other mutation genes (HMGA2 and Lin28A) in the analyzed MPN samples. Therefore, JAK2V617F influences target binding in both pSTAT3 and EZH2. Without mutations in epigenetic regulators, JAK2V617F can induce downstream epigenomic modifications. Thus, epigenetic changes in JAK2 downstream targets might be trackable in vivo.

Methylation of the Tumor Suppressor Genes HIC1 and RassF1A Clusters Independently From the Methylation of Polycomb Target Genes in Colon Cancer.

BACKGROUND: Methylation changes within tumor suppressor (TS) genes or polycomb group target (PcG) genes alter cell fates. Chromatin associated with PcG targets is bivalent in stem cells, while TS genes are not normally bivalent. PcG target methylation changes have been identified in tumor stem cells, and abnormal methylation is found in TS genes in cancers. If the epigenetic states of genes influence DNA methylation, then methylation of PcG targets and TS genes may evolve differently during cancer development. More importantly, methylation changes may be part of a sequence in tumorigenesis. METHODS: Chromatin and methylation states of 4 PcG targets and 2 TS genes were determined in colon cancer cells. The methylation states were also detected in 100 pairs of colon cancer samples. Principle component analysis (PCA) was used to reveal whether TS methylation or PcG methylation was the main methylation change associated with colon cancers. RESULTS: Chromatin and methylation states differ in colon cancer cell lines. The methylation states within PcG targets clustered independently from the methylation states in TS genes, a finding we previously reported in liver cancers. PCA in colon cancers revealed the strongest association with methylation changes in 2 TS genes, HIC1 and RassF1A. Loss of HIC1 methylation correlated with decreased tumor migration. CONCLUSIONS: PcG and TS methylation states cluster independently from each other. The deduced principle component correlated better with TS methylation than PcG methylation in colon cancer. Abnormal methylation changes may represent a sequential biomarker profile to identify developing colon cancer.

Targeting RPL39 and MLF2 reduces tumor initiation and metastasis in breast cancer by inhibiting nitric oxide synthase signaling.

We previously described a gene signature for breast cancer stem cells (BCSCs) derived from patient biopsies. Selective shRNA knockdown identified ribosomal protein L39 (RPL39) and myeloid leukemia factor 2 (MLF2) as the top candidates that affect BCSC self-renewal. Knockdown of RPL39 and MLF2 by specific siRNA nanoparticles in patient-derived and human cancer xenografts reduced tumor volume and lung metastases with a concomitant decrease in BCSCs. RNA deep sequencing identified damaging mutations in both genes. These mutations were confirmed in patient lung metastases (n = 53) and were statistically associated with shorter median time to pulmonary metastasis. Both genes affect the nitric oxide synthase pathway and are altered by hypoxia. These findings support that extensive tumor heterogeneity exists within primary cancers; distinct subpopulations associated with stem-like properties have increased metastatic potential.

Computational analysis reveals a correlation of exon-skipping events with splicing, transcription and epigenetic factors.

Alternative splicing (AS), in higher eukaryotes, is one of the mechanisms of post-transcriptional regulation that generate multiple transcripts from the same gene. One particular mode of AS is the skipping event where an exon may be alternatively excluded or constitutively included in the resulting mature mRNA. Both transcript isoforms from this skipping event site, i.e. in which the exon is either included (inclusion isoform) or excluded (skipping isoform), are typically present in one cell, and maintain a subtle balance that is vital to cellular function and dynamics. However, how the prevailing conditions dictate which isoform is expressed and what biological factors might influence the regulation of this process remain areas requiring further exploration. In this study, we have developed a novel computational method, graph-based exon-skipping scanner (GESS), for de novo detection of skipping event sites from raw RNA-seq reads without prior knowledge of gene annotations, as well as for determining the dominant isoform generated from such sites. We have applied our method to publicly available RNA-seq data in GM12878 and K562 cells from the ENCODE consortium and experimentally validated several skipping site predictions by RT-PCR. Furthermore, we integrated other sequencing-based genomic data to investigate the impact of splicing activities, transcription factors (TFs) and epigenetic histone modifications on splicing outcomes. Our computational analysis found that splice sites within the skipping-isoform-dominated group (SIDG) tended to exhibit weaker MaxEntScan-calculated splice site strength around middle, 'skipping', exons compared to those in the inclusion-isoform-dominated group (IIDG). We further showed the positional preference pattern of splicing factors, characterized by enrichment in the intronic splice sites immediately bordering middle exons. Finally, our analysis suggested that different epigenetic factors may introduce a variable obstacle in the process of exon-intron boundary establishment leading to skipping events.

Three-tiered role of the pioneer factor GATA2 in promoting androgen-dependent gene expression in prostate cancer.

In prostate cancer, androgen receptor (AR) binding and androgen-responsive gene expression are defined by hormone-independent binding patterns of the pioneer factors FoxA1 and GATA2. Insufficient evidence of the mechanisms by which GATA2 contributes to this process precludes complete understanding of a key determinant of tissue-specific AR activity. Our observations suggest that GATA2 facilitates androgen-responsive gene expression by three distinct modes of action. By occupying novel binding sites within the AR gene locus, GATA2 positively regulates AR expression before and after androgen stimulation. Additionally, GATA2 engages AR target gene enhancers prior to hormone stimulation, producing an active and accessible chromatin environment via recruitment of the histone acetyltransferase p300. Finally, GATA2 functions in establishing and/or sustaining basal locus looping by recruiting the Mediator subunit MED1 in the absence of androgen. These mechanisms may contribute to the generally positive role of GATA2 in defining AR genome-wide binding patterns that determine androgen-responsive gene expression profiles. We also find that GATA2 and FoxA1 exhibit both independent and codependent co-occupancy of AR target gene enhancers. Identifying these determinants of AR transcriptional activity may provide a foundation for the development of future prostate cancer therapeutics that target pioneer factor function.

A personalized committee classification approach to improving prediction of breast cancer metastasis.

MOTIVATION: Metastasis prediction is a well-known problem in breast cancer research. As breast cancer is a complex and heterogeneous disease with many molecular subtypes, predictive models trained for one cohort often perform poorly on other cohorts, and a combined model may be suboptimal for individual patients. Furthermore, attempting to develop subtype-specific models is hindered by the ambiguity and stereotypical definitions of subtypes. RESULTS: Here, we propose a personalized approach by relaxing the definition of breast cancer subtypes. We assume that each patient belongs to a distinct subtype, defined implicitly by a set of patients with similar molecular characteristics, and construct a different predictive model for each patient, using as training data, only the patients defining the subtype. To increase robustness, we also develop a committee-based prediction method by pooling together multiple personalized models. Using both intra- and inter-dataset validations, we show that our approach can significantly improve the prediction accuracy of breast cancer metastasis compared with several popular approaches, especially on those hard-to-learn cases. Furthermore, we find that breast cancer patients belonging to different canonical subtypes tend to have different predictive models and gene signatures, suggesting that metastasis in different canonical subtypes are likely governed by different molecular mechanisms. AVAILABILITY AND IMPLEMENTATION: Source code implemented in MATLAB and Java available at www.cs.utsa.edu/∼jruan/PCC/.

Integrated analysis of genome-wide DNA methylation and gene expression profiles in molecular subtypes of breast cancer.

Aberrant DNA methylation of CpG islands, CpG island shores and first exons is known to play a key role in the altered gene expression patterns in all human cancers. To date, a systematic study on the effect of DNA methylation on gene expression using high resolution data has not been reported. In this study, we conducted an integrated analysis of MethylCap-sequencing data and Affymetrix gene expression microarray data for 30 breast cancer cell lines representing different breast tumor phenotypes. As well-developed methods for the integrated analysis do not currently exist, we created a series of four different analysis methods. On the computational side, our goal is to develop methylome data analysis protocols for the integrated analysis of DNA methylation and gene expression data on the genome scale. On the cancer biology side, we present comprehensive genome-wide methylome analysis results for differentially methylated regions and their potential effect on gene expression in 30 breast cancer cell lines representing three molecular phenotypes, luminal, basal A and basal B. Our integrated analysis demonstrates that methylation status of different genomic regions may play a key role in establishing transcriptional patterns in molecular subtypes of human breast cancer.