Teng Fei

Integrative genomic analyses reveal clinically relevant long noncoding RNAs in human cancer

Despite growing appreciation of the importance of long noncoding RNAs (lncRNAs) in normal physiology and disease, our knowledge of cancer-related lncRNAs remains limited. By repurposing microarray probes, we constructed expression profiles of 10,207 lncRNA genes in approximately 1,300 tumors over four different cancer types. Through integrative analysis of the lncRNA expression profiles with clinical outcome and somatic copy-number alterations, we identified lncRNAs that are associated with cancer subtypes and clinical prognosis and predicted those that are potential drivers of cancer progression. We validated our predictions by experimentally confirming prostate cancer cell growth dependence on two newly identified lncRNAs. Our analysis provides a resource of clinically relevant lncRNAs for the development of lncRNA biomarkers and the identification of lncRNA therapeutic targets. It also demonstrates the power of integrating publically available genomic data sets and clinical information for discovering disease-associated lncRNAs.

Enhancer RNAs participate in androgen receptor-driven looping that selectively enhances gene activation

Significance We report that enhancer RNAs (eRNAs), a class of long noncoding RNAs, participate in the androgen receptor (AR)-dependent looping complex that enhances spatial communication of distal enhancers and target promoters, leading to transcriptional activation events. Furthermore, our data show that KLK3 eRNA (KLK3e) selectively enhances the gene expression of AR-regulated genes, and provide evidence for a positive regulatory loop in which AR-dependent transcription is modulated by an intermediate eRNA. These findings may translate into improved RNA-based therapy (eRNA suppression) to enhance the durability of androgen deprivation therapy (ADT) and prediction of the efficacy of ADT by measuring the enhancer-derived activity (eRNA expression) in prostate tumors. The androgen receptor (AR) is a key factor that regulates the behavior and fate of prostate cancer cells. The AR-regulated network is activated when AR binds enhancer elements and modulates specific enhancer–promoter looping. Kallikrein-related peptidase 3 (KLK3), which codes for prostate-specific antigen (PSA), is a well-known AR-regulated gene and its upstream enhancers produce bidirectional enhancer RNAs (eRNAs), termed KLK3e. Here, we demonstrate that KLK3e facilitates the spatial interaction of the KLK3 enhancer and the KLK2 promoter and enhances long-distance KLK2 transcriptional activation. KLK3e carries the core enhancer element derived from the androgen response element III (ARE III), which is required for the interaction of AR and Mediator 1 (Med1). Furthermore, we show that KLK3e processes RNA-dependent enhancer activity depending on the integrity of core enhancer elements. The transcription of KLK3e was detectable and its expression is significantly correlated with KLK3 (R2 = 0.6213, P < 5 × 10−11) and KLK2 (R2 = 0.5893, P < 5 × 10−10) in human prostate tissues. Interestingly, RNAi silencing of KLK3e resulted in a modest negative effect on prostate cancer cell proliferation. Accordingly, we report that an androgen-induced eRNA scaffolds the AR-associated protein complex that modulates chromosomal architecture and selectively enhances AR-dependent gene expression.

Refined DNase-seq protocol and data analysis reveals intrinsic bias in transcription factor footprint identification.

Sequencing of DNase I hypersensitive sites (DNase-seq) is a powerful technique for identifying cis-regulatory elements across the genome. We studied the key experimental parameters to optimize performance of DNase-seq. Sequencing short fragments of 50–100 base pairs (bp) that accumulate in long internucleosome linker regions was more efficient for identifying transcription factor binding sites compared to sequencing longer fragments. We also assessed the potential of DNase-seq to predict transcription factor occupancy via generation of nucleotide-resolution transcription factor footprints. In modeling the sequence-specific DNase I cutting bias, we found a strong effect that varied over more than two orders of magnitude. This indicates that the nucleotide-resolution cleavage patterns at many transcription factor binding sites are derived from intrinsic DNase I cleavage bias rather than from specific protein-DNA interactions. In contrast, quantitative comparison of DNase I hypersensitivity between states can predict transcription factor occupancy associated with particular biological perturbations.

Genome-wide mapping of SMAD target genes reveals the role of BMP signaling in embryonic stem cell fate determination

Embryonic stem (ES) cells are under precise control of both intrinsic self-renewal gene regulatory network and extrinsic growth factor-triggered signaling cascades. How external signaling pathways connect to core self-renewal transcriptional circuits is largely unknown. To probe this, we chose BMP signaling, which is previously recognized as a master control for both self-renewal and lineage commitment of murine ES cells. Here, we mapped target gene promoter occupancy of SMAD1/5 and SMAD4 on a genome-wide scale and found that they associate with a large group of developmental regulators that are enriched for H3K27 trimethylation and H3K4 trimethylation bivalent marks and are repressed in the self-renewing state, whereas they are rapidly induced upon differentiation. Smad knockdown experiments further indicate that SMAD-mediated BMP signaling is largely required for differentiation-related processes rather than directly influencing self-renewal. Among the SMAD-associated genes, we further identified Dpysl2 (previously known as Crmp2) and the H3K27 demethylase Kdm6b (previously known as Jmjd3) as BMP4-modulated early neural differentiation regulators. Combined with computational analysis, our results suggest that SMAD-mediated BMP signaling balances self-renewal versus differentiation by modulating a set of developmental regulators.

BMP4 Signaling Acts via Dual-Specificity Phosphatase 9 to Control ERK Activity in Mouse Embryonic Stem Cells

Extrinsic BMP and LIF signaling collaboratively maintain mouse embryonic stem cell (ESC) pluripotency, whereas appropriate ERK activity is essential for ESC fate commitment. However, how the extrinsic signals restrain appropriate ERK activity remains elusive. Here, we show that, whereas LIF sustains relatively high ERK activity, BMP4 can steadily attenuate ERK activity by upregulating ERK-specific dual-specificity phosphatase 9 (DUSP9). This upregulation requires Smad1/5 and Smad4 and specifically occurs to DUSP9, but not other DUSPs, and only in ESCs. Through DUSP9-mediated inhibition of ERK activity, BMP signaling reinforces the self-renewal status of mouse ESCs together with LIF. Upon LIF withdrawal, ESCs spontaneously undergo neural differentiation, during which process DUSP9 can partially mediate BMP inhibition on neural commitment. Collectively, our findings identify DUSP9 as a critical mediator of BMP signaling to control appropriate ERK activity critical for ESC fate determination.

Integrative analyses reveal a long noncoding RNA-mediated sponge regulatory network in prostate cancer.

Mounting evidence suggests that long noncoding RNAs (lncRNAs) can function as microRNA sponges and compete for microRNA binding to protein-coding transcripts. However, the prevalence, functional significance and targets of lncRNA-mediated sponge regulation of cancer are mostly unknown. Here we identify a lncRNA-mediated sponge regulatory network that affects the expression of many protein-coding prostate cancer driver genes, by integrating analysis of sequence features and gene expression profiles of both lncRNAs and protein-coding genes in tumours. We confirm the tumour-suppressive function of two lncRNAs (TUG1 and CTB-89H12.4) and their regulation of PTEN expression in prostate cancer. Surprisingly, one of the two lncRNAs, TUG1, was previously known for its function in polycomb repressive complex 2 (PRC2)-mediated transcriptional regulation, suggesting its sub-cellular localization-dependent function. Our findings not only suggest an important role of lncRNA-mediated sponge regulation in cancer, but also underscore the critical influence of cytoplasmic localization on the efficacy of a sponge lncRNA.

GSK3β mediates suppression of cyclin D2 expression by tumor suppressor PTEN

PTEN, encoding a lipid phosphatase, is a tumor suppressor gene and is mutated in various types of cancers. It is reported to regulate G1 to S phase transition of the cell cycle by influencing the expression, protein stability and subcellular location of cyclin D1. Here, we provide evidence that PTEN modulates the transcription and protein stability of cyclin D2. Targeted deletion of Pten in mouse embryonic fibroblasts (MEFs) endowed cells with greater potential to overcome G1 arrest than wild-type MEFs and led to the elevated expression of cyclin D2, which was suppressed by the introduction of PTEN. We further defined a pathway involving GSK3β and β-catenin/TCF in PTEN-mediated suppression of cyclin D2 transcription. LiCl, an inhibitor of GSK3β, abolished inhibitory effect of PTEN on cyclin D2 expression, and TCF members could directly bind to the promoter of cyclin D2 and regulate its transcription in a CREB-dependent manner. Our results indicate that the downregulation of cyclin D2 expression by PTEN is mediated by the GSK3β/β-catenin/TCF pathway in cooperation with CREB, and suggest a convergence from the PI-3 kinase/PTEN pathway and the Wnt pathway in modulation of cyclin D2 expression.

Smad2 mediates Activin/Nodal signaling in mesendoderm differentiation of mouse embryonic stem cells

Although Activin/Nodal signaling regulates pluripotency of human embryonic stem (ES) cells, how this signaling acts in mouse ES cells remains largely unclear. To investigate this, we confirmed that mouse ES cells possess active Smad2-mediated Activin/Nodal signaling and found that Smad2-mediated Activin/Nodal signaling is dispensable for self-renewal maintenance but is required for proper differentiation toward the mesendoderm lineage. To gain insights into the underlying mechanisms, Smad2-associated genes were identified by genome-wide chromatin immunoprecipitation-chip analysis. The results showed that there is a transcriptional correlation between Smad2 binding and Activin/Nodal signaling modulation, and that the development-related genes were enriched among the Smad2-bound targets. We further identified Tapbp as a key player in mesendoderm differentiation of mouse ES cells acting downstream of the Activin/Nodal-Smad2 pathway. Taken together, our findings suggest that Smad2-mediated Activin/Nodal signaling orchestrates mesendoderm lineage commitment of mouse ES cells through direct modulation of corresponding developmental regulator expression.

Amplitude modulation of androgen signaling by c-MYC

Androgen-stimulated growth of the molecular apocrine breast cancer subtype is mediated by an androgen receptor (AR)-regulated transcriptional program. However, the molecular details of this AR-centered regulatory network and the roles of other transcription factors that cooperate with AR in the network remain elusive. Here we report a positive feed-forward loop that enhances breast cancer growth involving AR, AR coregulators, and downstream target genes. In the absence of an androgen signal, TCF7L2 interacts with FOXA1 at AR-binding sites and represses the basal expression of AR target genes, including MYC. Direct AR regulation of MYC cooperates with AR-mediated activation of HER2/HER3 signaling. HER2/HER3 signaling increases the transcriptional activity of MYC through phosphorylation of MAD1, leading to increased levels of MYC/MAX heterodimers. MYC in turn reinforces the transcriptional activation of androgen-responsive genes. These results reveal a novel regulatory network in molecular apocrine breast cancers regulated by androgen and AR in which MYC plays a central role as both a key target and a cooperating transcription factor to drive oncogenic growth.

Modulation of long noncoding RNAs by risk SNPs underlying genetic predispositions to prostate cancer

Long noncoding RNAs (lncRNAs) represent an attractive class of candidates to mediate cancer risk. Through integrative analysis of the lncRNA transcriptome with genomic data and SNP data from prostate cancer genome-wide association studies (GWAS), we identified 45 candidate lncRNAs associated with risk to prostate cancer. We further evaluated the mechanism underlying the top hit, PCAT1, and found that a risk-associated variant at rs7463708 increases binding of ONECUT2, a novel androgen receptor (AR)-interacting transcription factor, at a distal enhancer that loops to the PCAT1 promoter, resulting in upregulation of PCAT1 upon prolonged androgen treatment. In addition, PCAT1 interacts with AR and LSD1 and is required for their recruitment to the enhancers of GNMT and DHCR24, two androgen late-response genes implicated in prostate cancer development and progression. PCAT1 promotes prostate cancer cell proliferation and tumor growth in vitro and in vivo. These findings suggest that modulating lncRNA expression is an important mechanism for risk-associated SNPs in promoting prostate transformation.