Transcriptomic Clusters Research Articles

Using single cell genomics to change the treatment of lung cancer.

e20563 Background: Lung cancer (LC) is common with a dismal prognosis. Although treatment with immunotherapy (IO) has improved survival outcomes, these therapies remain expensive. Even using biomarker selection, response rates still fall short of 50%. The majority of immune cell profiling done previously uses samples taken from early stage patients focusing on a single immune cell subtype. Here we use single cell RNA sequencing (scRNA-seq) and Cellular Indexing of Transcriptome and Epitopes by sequencing (CITE-seq) to characterise the innate and adaptive immune cell activation state and assess the response to exposure to IO in vitro from patients with advanced LC. Methods: Patients with locally advanced or metastatic LC having an Endobronchial Ultrasound-Guided Transbronchial Needle Aspiration (EBUS-TBNA) have biopsies collected for analysis. Cells are grown in culture +/- nivolumab for 48 hours. scRNA-seq and CITE-seq is conducted using established protocols. Transcriptomic and proteomic data on the innate and adaptive immune cell subsets assess markers of immune activation and/or suppression and the changes after nivolumab are quantified. Results are correlated with clinical outcomes. Results: In a locally advanced/metastatic population, EBUS-TBNA biopsies yield highly cellular samples with a heterogeneous population of immune cells able to be cultured +/- IO using novel, inclusive techniques. Transcriptomic clustering reveals distinct sub-populations within the T-cell, B-cell and innate immune cell compartments. Within these clusters, CITE-seq shows protein expression on individual cells can determine states of exhaustion, cytolytic ability, migratory potential and innate immune activation. Conclusions: Single cell genomics is feasible and informative in patients with advanced LC. This work will form the basis of a functional, real-time assay to assess individualised responses to IO therapy that has the potential to predict IO response.

Expression of the GAF Sensor, Carbohydrate-Active Enzymes, Elicitins, and RXLRs Differs Markedly Between Two Phytophthora cactorum Isolates.

The phytopathogen Phytophthora cactorum infects economically important herbaceous and woody plant species. P. cactorum isolates differ in host specificity; for example, strawberry crown rot is often caused by a specialized pathotype. Here we compared the transcriptomes of two P. cactorum isolates that differ in their virulence to garden strawberry (Pc407: high virulence; Pc440: low virulence). De novo transcriptome assembly and clustering of contigs resulted in 19,372 gene clusters. Two days after inoculation of Fragaria vesca roots, 3,995 genes were differently expressed between the P. cactorum isolates. One of the genes that were highly expressed only in Pc407 encodes a GAF sensor protein potentially involved in membrane trafficking processes. Two days after inoculation, elicitins were highly expressed in Pc407 and lipid catabolism appeared to be more active than in Pc440. Of the carbohydrate-active enzymes, those that degrade pectin were often more highly expressed in Pc440, whereas members of glycosyl hydrolase family 1, potentially involved in the metabolism of glycosylated secondary metabolites, were more highly expressed in Pc407 at the time point studied. Differences were also observed among the RXLR effectors: Pc407 appears to rely on a smaller set of key RXLR effectors, whereas Pc440 expresses a greater number of RXLRs. This study is the first step toward improving understanding of the molecular basis of differences in the virulence of P. cactorum isolates. Identification of the key effectors is important, as it enables effector-assisted breeding strategies toward crown rot-resistant strawberry cultivars.

The transcription factor Duxbl mediates elimination of pre-T cells that fail β-selection.

T cell development is critically dependent on successful rearrangement of antigen-receptor chains. At the β-selection checkpoint, only cells with a functional rearrangement continue in development. However, how nonselected T cells proceed in their dead-end fate is not clear. We identified low CD27 expression to mark pre-T cells that have failed to rearrange their β-chain. Expression profiling and single-cell transcriptome clustering identified a developmental trajectory through β-selection and revealed specific expression of the transcription factor Duxbl at a stage of high recombination activity before β-selection. Conditional transgenic expression of Duxbl resulted in a developmental block at the DN3-to-DN4 transition due to reduced proliferation and enhanced apoptosis, whereas RNA silencing of Duxbl led to a decrease in apoptosis. Transcriptome analysis linked Duxbl to elevated expression of the apoptosis-inducing Oas/RNaseL pathway. RNaseL deficiency or sustained Bcl2 expression led to a partial rescue of cells in Duxbl transgenic mice. These findings identify Duxbl as a regulator of β-selection by inducing apoptosis in cells with a nonfunctional rearrangement.

POH1 contributes to hyperactivation of TGF-β signaling and facilitates hepatocellular carcinoma metastasis through deubiquitinating TGF-β receptors and caveolin-1.

BackgroundHyper-activation of TGF-β signaling is critically involved in progression of hepatocellular carcinoma (HCC). However, the events that contribute to the dysregulation of TGF-β pathway in HCC, especially at the post-translational level, are not well understood.MethodsAssociations of deubiquitinase POH1 with TGF-β signaling activity and the outcomes of HCC patients were examined by data mining of online HCC datasets, immunohistochemistry analyses using human HCC specimens, spearman correlation and survival analyses. The effects of POH1 on the ubiquitination and stability of the TGF-β receptors (TGFBR1 and TGFBR2) and the activation of downstream effectors were tested by western blotting. Primary mouse liver tissues from polyinosinic:polycytidylic acid (poly I:C)- treated Mx-Cre+, poh1f/f mice and control mice were used to detect the TGF-β receptors. The metastatic-related capabilities of HCC cells were studied in vitro and in mice.FindingsHere we show that POH1 is a critical regulator of TGF-β signaling and promotes tumor metastasis. Integrative analyses of HCC subgroups classified with unsupervised transcriptome clustering of the TGF-β response, metastatic potential and outcomes, reveal that POH1 expression positively correlates with activities of TGF-β signaling in tumors and with malignant disease progression. Functionally, POH1 intensifies TGF-β signaling delivery and, as a consequence, promotes HCC cell metastatic properties both in vitro and in vivo. The expression of the TGF-β receptors was severely downregulated in POH1-deficient mouse hepatocytes. Mechanistically, POH1 deubiquitinates the TGF-β receptors and CAV1, therefore negatively regulates lysosome pathway-mediated turnover of TGF-β receptors.ConclusionOur study highlights the pathological significance of aberrantly expressed POH1 in TGF-β signaling hyperactivation and aggressive progression in HCC.

The embryonic transcriptome of Arabidopsis thaliana

Key messageArabidopsis embryos possess unique transcriptomes relative to other plant tissues including somatic embryos, and can be partitioned into four transcriptional phases with characteristic biological processes.Cellular differentiation is associated with changes in transcript populations. Accurate quantification of transcriptomes during development can thus provide global insights into differentiation processes including the fundamental specification and differentiation events operating during plant embryogenesis. However, multiple technical challenges have limited the ability to obtain high-quality early embryonic transcriptomes, namely the low amount of RNA obtainable and contamination from surrounding endosperm and seed-coat tissues. We compared the performance of three low-input mRNA sequencing (mRNA-seq) library preparation kits on 0.1 to 5 nanograms (ng) of total RNA isolated from Arabidopsis thaliana (Arabidopsis) embryos and identified a low-cost method with superior performance. This mRNA-seq method was then used to profile the transcriptomes of Arabidopsis embryos across eight developmental stages. By comprehensively comparing embryonic and post-embryonic transcriptomes, we found that embryonic transcriptomes do not resemble any other plant tissue we analyzed. Moreover, transcriptome clustering analyses revealed the presence of four distinct phases of embryogenesis which are enriched in specific biological processes. We also compared zygotic embryo transcriptomes with publicly available somatic embryo transcriptomes. Strikingly, we found little resemblance between zygotic embryos and somatic embryos derived from late-staged zygotic embryos suggesting that somatic and zygotic embryo transcriptomes are distinct from each other. In addition to the biological insights gained from our systematic characterization of the Arabidopsis embryonic transcriptome, we provide a data-rich resource for the community to explore.

Immunogenomic Landscape of Hematological Malignancies

Understanding factors that shape the immune landscape across hematological malignancies is essential for immunotherapy development. How cancer-cell intrinsic genomic and epigenetic alterations influence immune signatures in hematological malignancies is not known. Here, we integrated over 8,000 transcriptomes of hematologic cancers and multilevel genomic datasets to investigate associations of immune states to cancer molecular subtypes, genetic and epigenetic alterations, and clinical outcomes.We utilized a resource of over 8,000 transcriptomes collected across 36 hematologic malignancies and normal hematopoietic cells (Hemap), together with multi-omics datasets of acute myeloid leukemia (AML) and diffuse large B-cell lymphoma (DLBCL) from The Cancer Genome Atlas and other sources (Figure). In addition to gene expression data, we integrated somatic DNA alterations, methylation data, multiplex immunohistochemistry (mIHC), and flow cytometry to comprehensively map immune-associated features and validate the robustness of the findings. To characterize the composition of the cytolytic immune infiltrate from bulk transcriptomes, we defined a signature of genes most specifically expressed in cytotoxic CD8+ T lymphocytes and natural killer (NK) cells termed cytolytic score.We found significant heterogeneity in the cytotoxic lymphocyte infiltration signature across hematologic malignancies. Highest cytolytic infiltrate was detected in lymphomas and correlated with IFN-γ and myeloid cell infiltration signatures including CXCL9-11 and IDO1, distinguishing the lymphoma microenvironment from leukemias. In addition to transcriptomic microenvironmental properties, specific genetic alterations were associated with cytotoxic lymphocyte infiltration. In DLBCL, driver alterations enriched in the germinal center B-cell like (GCB) molecular subtype including BCL2 translocations and KMT2D were linked to an immune-cold transcriptomic phenotype. In contrast, DTX1 alterations defined immune-infiltrated lymphomas within the GCB molecular subtype. In AML, TP53 mutations and complex karyotype were enriched in a distinct tSNE-based transcriptomic cluster characterized by increased immune infiltration in the bone marrow (BM).Given the importance of effective antigen presentation for adaptive anti-tumor immune responses, we aimed to understand the transcriptional regulation of HLA genes and co-stimulatory and co-inhibitory signaling in subtypes of hematological malignancies. Downregulation of the antigen-presenting HLA II genes was associated with CpG methylation of the promoter region of the HLA class II master regulator CIITA in distinct transcriptomic clusters of AML harboring PML-RARA or NPM1 alterations. Expression of genes encoding immune checkpoint molecules was strongly influenced by the cell-of-origin and microenvironment of each cancer type. We identified novel associations of inhibitory immune checkpoint molecules to disease subtypes, such as VISTA/PD1-H enriched in myeloid malignancies including AML, CML, and MDS, validated by mIHC performed on BM biopsies. Furthermore, variation in the expression of several genes encoding immune checkpoints was associated with somatic mutations (e.g. CD70 in DLBCL), copy-number alterations (e.g. MICB in DLBCL), and DNA methylation (e.g. PDL1 and PDCD1LG2 in AML).Finally, we integrated GTEx gene expression data across tissues to define cancer-germline antigens (CGAs) with an immune privileged tissue expression pattern. CGAs were frequently expressed in multiple myeloma and DLBCL compared to other hematologic malignancies. CGA expression was associated with cytogenetic alterations and increased MYC activity signature in myeloma and CD58 and KLHL6 mutations in DLBCL. In addition, CGA expression in myeloma and DLBCL was linked to reduced antigen gene promoter methylation and decreased survival.In summary, our findings demonstrate that molecular subtypes of hematological malignancies harbor distinct immunological signatures influenced by genetic and epigenetic alterations. Integrating genetic, epigenetic, and transcriptomic data may facilitate the development of precision immune intervention strategies in hematological malignancies. [Display omitted] DisclosuresLeppa:Bayer: Research Funding; Celgene: Consultancy; Roche: Consultancy, Honoraria, Research Funding; Janssen: Consultancy, Research Funding; Takeda: Consultancy, Research Funding. Mustjoki:Bristol-Myers Squibb: Honoraria, Research Funding; Celgene: Honoraria; Pfizer: Honoraria, Research Funding; Ariad: Research Funding; Novartis: Honoraria, Research Funding.

Genetic profiles of transcriptomic clusters of childhood asthma determine specific severe subtype.

Previous studies have defined transcriptomic subtypes of adult asthma using samples of induced sputum and bronchial epithelium; however, those procedures are not readily applicable in the clinic, especially for childhood asthma. We aim to dissect the transcriptomic clusters of childhood asthma using highly variably expressed genes of peripheral blood mononuclear cells (PBMC) among patients. Gene expression of PBMC from 133 asthmatic children and 11 healthy controls was measured with Illumina microarrays. We applied the k-means clustering algorithm of 2048 genes to assign asthmatic children into clusters. Genes with differential expression between asthma clusters and healthy controls were used to investigate whether they could identify severe asthma of children and adults. We identified 3 asthma clusters with distinct inflammatory profiles in peripheral blood. Cluster 1 had the highest eosinophil count. Cluster 2 showed lower counts of both eosinophils and neutrophils. Cluster 3 had the highest neutrophil count and the poorest treatment control. Compared with other patients, Cluster 3 exhibited a unique gene expression pattern which was associated with changes in the glucocorticoid signalling and activation of the T helper 1/T helper 17 (TH 1/TH 17) immune pathways. In the validation studies, an 84-gene signature could identify severe asthma in children on leucocytes, as well as severe asthma in adults on CD8+ T cells. Gene expression profiling of PBMC is useful for the identification of TH 1/TH 17-mediated asthma with poor treatment control. PBMC and CD8+ T cells could be important targets for the investigation and identification of severe asthma.

Window of implantation transcriptomic stratification reveals different endometrial subsignatures associated with live birth and biochemical pregnancy

To refine the endometrial window of implantation (WOI) transcriptomic signature by defining new subsignatures associated to live birth and biochemical pregnancy. Retrospective cohort study. University-affiliated invitro fertilization clinic and reproductive genetics laboratory. Healthy fertile oocyte donors (n = 79) and patients with infertility diagnosed by Endometrial Receptivity Analysis (n = 771). None. WOI transcriptomic signatures associated with specific reproductive outcomes. The retrospective cohort study was designed to perform a prediction model based on transcriptomic clusters for endometrial classification (training set, n = 529). The clinical follow-up set in the expected WOI (n = 321) was tested with the transcriptomic predictor to detect WOI variability and the pregnancy outcomes associated with these subsignatures (n = 228). The endometrial receptivity signature was redefined into four WOI transcriptomic profiles. This stratification identified an optimal endometrial receptivity (RR) signature resulting in an ongoing pregnancy rate (OPR) of 80% in terms of live birth, as well as a late receptive-stage (LR) signature with a potential high risk of 50% biochemical pregnancy. Abnormal down-regulation of the cell cycle was the main dysregulated function among the 22 genes associated with biochemical pregnancy. The major differences between the WOI transcriptomic stratification were in the OPR and biochemical pregnancy rate. The OPR ranged from 76.9% and 80% in the late prereceptive (LPR) and RR signatures, respectively, versus 33.3% in the LR. The biochemical pregnancy rate was 7.7% and 6.6% in LPR and RR, respectively, but 50% in LR, which highlights the relevance of endometrial status in the progression of embryonic implantation.

Abstract 4559: Novel T-cell targets for HPV16-associated oropharyngeal cancer immune therapy

Abstract Background: Human papillomavirus subtype 16 (HPV16) is the primary cause of an increasing number of oropharyngeal cancers (HPVOPC) worldwide. Despite effective prophylactic vaccines, low vaccine uptake and high incidence of HPV16 infections provides a strong rationale for the development of novel targeted immune therapies against HPVOPC. We sought to 1) Identify novel CD8+ T-cell (CTL) epitopes from HPV16 antigens; 2) Understand mechanisms of immune resistance in HPVOPC to enhance T-cell based immune therapies. Methods: We developed a comprehensive bioinformatic strategy to predict potential CTL epitopes for 10 HLA class I alleles representing common HLA-supertypes (with >90% global coverage) against HPV16 antigens E2, E6 and E7. T-cell reactivity against candidate epitopes was assessed by short term ex vivo T-cell cultures from healthy donors (n=15) and HPVOPC patients (n=18). We data mined publicly available OPC transcriptomes (University of Michigan (UM-OPC) cohort, n=36) for gene expression analysis. Results: IFN-gamma Elispot assays using 58 candidate epitopes from HPV16 E2, E6 and E7 revealed that >70% of HPVOPC patients had either E2 or E6-specific CTL reactivity. E2-CTL reactivity was >2-fold higher in HPVOPC patients than in healthy controls (mean = 68.75 vs. 24.6 SFUs/106 cells, P = 0.027), while no difference in E6- CTL responses were observed between patients and controls. In contrast, E7-specific CTL responses were low to undetectable. We have further identified 7 novel E2 and 3 novel E6 CTL-epitopes, and confirmed immunogenicity of previously described E2, E6 and E7 epitopes. RNAseq analysis of the UM-OPC dataset indicated that 78% of HPV16+ OPCs expressed E2 (P<0.05), while E6 and E7 was expressed in all HPVOPCs (P<0.01) demonstrating tumor antigen expression. Unsupervised hierarchical clustering of immune transcriptomes revealed that HPVOPCs have higher expression of immune-regulatory ligands than HPV-negative OPC tumors, indicating high immune infiltration and possible immune-editing. We identified high expression of several key immune exhaustion genes such as IDO-1, TIM-3 and LAG-3, previously undescribed in HPVOPC. Conclusion: HPV16-E2 represents a novel antigen target for HPVOPC immune therapy, as demonstrated by T-cell reactivity and tumor antigen expression. Several novel and previously defined E2, E6 and E7 CTL have been confirmed for T-cell reactivity in HPVOPC patients. Transcriptomic analysis of OPCs indicate extensive immune dysfunction in HPVOPCs that are distinct from non-HPV OPCs. The T-cell epitopes identified in this study in combination with blockade of HPVOPC-specific checkpoints may be useful for targeted immunotherapy. Citation Format: Sri Krishna, Falguni Parikh, Peaches Ulrich, Shanshan Yang, Amelia Clark, Seunghee Kim-Schulze, Marshall Posner, Jin Park, Andrew Sikora, Karen S. Anderson. Novel T-cell targets for HPV16-associated oropharyngeal cancer immune therapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 4559. doi:10.1158/1538-7445.AM2017-4559

Abstract LB-257: Estrogen signaling in mature luminal and luminal progenitor cells of BRCA2 carriers and non-carriers

Abstract Background: Carriers of mutations in the breast cancer susceptibility gene BRCA2 have a 50-80% lifetime risk of developing breast cancers, about 70% of which are positive for estrogen receptor (ER), a key oncogenic hormone receptor. Metastatic ER+ tumors also respond to PARP inhibitors associated with their DNA repair defect. There remains much to understand regarding estrogen signaling in BRCA2 carriers and preventing cancers in this high-risk population. Methods: We studied the transcriptomes and in-vitro growth of luminal progenitors and mature luminal cells derived from primary human normal breast tissue obtained from BRCA2 carriers and wild-type patients undergoing prophylactic mastectomy and reduction mammoplasty procedures, respectively. Subsequently, formalin-fixed paraffin-embedded breast cancer tissue samples from twelve case-control matched pairs of BRCA2 carriers and wild-type patients were investigated for ER genomic binding, H3K27Ac enhancer marks, and gene expression changes. Results: Luminal progenitors and mature luminal cells have different gene expression patterns. Unsupervised clustering of transcriptomes from normal tissue of three pairs of BRCA2-mutant and wildtype patients suggested that BRCA2 mutation status can segregate transcriptomes from luminal progenitors but not mature luminal cells. These differential transcriptomes in luminal progenitor cells from BRCA2 carriers were enriched for cell proliferation genes. In agreement, greater proliferative phenotype was observed in BRCA2-mutant luminal progenitor cells in comparison to the wild-type cohort. Interestingly, clustering analyses of twelve pairs of breast tumors indicated no clear differences between ER cistromes, enhancer marks and gene expression between BRCA2 carriers and wildtype patients. Because luminal progenitor cells are more likely to contribute to tumorigenesis, we hypothesize that the identified BRCA2-mutant signature in normal progenitor cells could be retained in tumors obtained from BRCA2 carriers. Conclusions: There are differential gene expression patterns between luminal progenitors and mature luminal cells obtained from normal human breast tissue. The transcriptomes in the luminal progenitor but not mature luminal cells segregate differently based on the BRCA2 mutation status. This identified BRCA2-mutant gene signature in luminal progenitor cells is enriched for pro-proliferation pathways, and BRCA2-mutant luminal progenitor cells exhibit higher cell proliferation, suggesting a potential cell-of-origin in BRCA2-associated breast cancers. Importantly, no significant differences in transcriptomes and ER cistromes are observed in breast tumors from BRCA2 carriers and non-carriers, indicating a need to identify BRCA2-mutant gene signature that may assist in segregating ER+ breast cancers in this high-risk populations. The knowledge from this study may assist in pioneering strategies for cancer prevention in high-risk BRCA2 mutation carriers. Citation Format: Hari Singhal, David Chi, Elgene Lim, Housheng He, Raga Vadhi, Prakash K. Rao, Henry W. Long, Andrea L. Richardson, Judy Garber, Myles Brown. Estrogen signaling in mature luminal and luminal progenitor cells of BRCA2 carriers and non-carriers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr LB-257. doi:10.1158/1538-7445.AM2017-LB-257

MPTH-39. SEX-SPECIFIC ANALYSIS OF GBM TRANSCRIPTOMES REVEALS NEW PROGNOSTICALLY IMPORTANT MOLECULAR SUBTYPES

Molecular subtyping of GBM has advanced basic, translational and clinical neuro-oncology. As there are well-recognized sex differences in GBM incidence and outcome, we hypothesized that analysis of merged male and female datasets had previously masked important molecular features of GBM. To test this hypothesis we applied a JIVE algorithm to decompose male and female complete GBM transcriptome datasets (TCGA) accessed through the UCSC Cancer Genomics Browser, to joint and sex-specific components. We sought to determine whether there are shared or distinct GBM molecular subtypes in males and females. Using consensus hierarchical clustering independently on the sex-specific component from the JIVE, we defined five male and five female transcriptome clusters. Most clusters were a mix of samples previously assigned to the four Verhaak subtypes. While driver events were shared between the male and female clusters, they could be distinguished by their expression profiles, pathway analysis, and by their survival. Most notable were the differences between male and female subsets with the best survival. In female GBM, IDH1 mutation alone defined a single cluster with superior survival. Two male subtypes exhibited better survival, and this could not be completely explained by enrichment for IDH mutant tumors. Pathway analyses of genes differentiating a male or female cluster versus the others indicated that the male clusters with better survival were enriched with expression of collagen genes and extracellular matrix components. No similar phenotype was observed in the female subsets. The correlation between the sex-specific GBM clusters and survival was validated in further analyses of the GSE13041 and 16011 datasets. Together these data indicate that important elements of GBM biology and outcome can be discovered through sex-specific analyses and that distinctions between the biology of male and female GBM may need to be incorporated into precision approaches to treatment.

Fusion transcriptome profiling provides insights into alveolar rhabdomyosarcoma

Gene fusions and fusion products were thought to be unique features of neoplasia. However, more and more studies have identified fusion RNAs in normal physiology. Through RNA sequencing of 27 human noncancer tissues, a large number of fusion RNAs were found. By analyzing fusion transcriptome, we observed close clusterings between samples of same or similar tissues, supporting the feasibility of using fusion RNA profiling to reveal connections between biological samples. To put the concept into use, we selected alveolar rhabdomyosarcoma (ARMS), a myogenic pediatric cancer whose exact cell of origin is not clear. PAX3-FOXO1 (paired box gene 3 fused with forkhead box O1) fusion RNA, which is considered a hallmark of ARMS, was recently found during normal muscle cell differentiation. We performed and analyzed RNA sequencing from various time points during myogenesis and uncovered many chimeric fusion RNAs. Interestingly, we found that the fusion RNA profile of RH30, an ARMS cell line, is most similar to the myogenesis time point when PAX3-FOXO1 is expressed. In contrast, full transcriptome clustering analysis failed to uncover this connection. Strikingly, all of the 18 chimeric RNAs in RH30 cells could be detected at the same myogenic time point(s). In addition, the seven chimeric RNAs that follow the exact transient expression pattern as PAX3-FOXO1 are specific to rhabdomyosarcoma cells. Further testing with clinical samples also confirmed their specificity to rhabdomyosarcoma. These results provide further support for the link between at least some ARMSs and the PAX3-FOXO1-expressing myogenic cells and demonstrate that fusion RNA profiling can be used to investigate the etiology of fusion-gene-associated cancers.

Mitochondrial phosphoenolpyruvate carboxykinase (PEPCK-M) and serine biosynthetic pathway genes are co-ordinately increased during anabolic agent-induced skeletal muscle growth.

We aimed to identify novel molecular mechanisms for muscle growth during administration of anabolic agents. Growing pigs (Duroc/(Landrace/Large-White)) were administered Ractopamine (a beta-adrenergic agonist; BA; 20 ppm in feed) or Reporcin (recombinant growth hormone; GH; 10 mg/48 hours injected) and compared to a control cohort (feed only; no injections) over a 27-day time course (1, 3, 7, 13 or 27-days). Longissimus Dorsi muscle gene expression was analyzed using Agilent porcine transcriptome microarrays and clusters of genes displaying similar expression profiles were identified using a modified maSigPro clustering algorithm. Anabolic agents increased carcass (p = 0.002) and muscle weights (Vastus Lateralis: p < 0.001; Semitendinosus: p = 0.075). Skeletal muscle mRNA expression of serine/one-carbon/glycine biosynthesis pathway genes (Phgdh, Psat1 and Psph) and the gluconeogenic enzyme, phosphoenolpyruvate carboxykinase-M (Pck2/PEPCK-M), increased during treatment with BA, and to a lesser extent GH (p < 0.001, treatment x time interaction). Treatment with BA, but not GH, caused a 2-fold increase in phosphoglycerate dehydrogenase (PHGDH) protein expression at days 3 (p < 0.05) and 7 (p < 0.01), and a 2-fold increase in PEPCK-M protein expression at day 7 (p < 0.01). BA treated pigs exhibit a profound increase in expression of PHGDH and PEPCK-M in skeletal muscle, implicating a role for biosynthetic metabolic pathways in muscle growth.

Multi-task consensus clustering of genome-wide transcriptomes from related biological conditions.

Identifying the shared and pathogen-specific components of host transcriptional regulatory programs is important for understanding the principles of regulation of immune response. Recent efforts in systems biology studies of infectious diseases have resulted in a large collection of datasets measuring host transcriptional response to various pathogens. Computational methods to identify and compare gene expression modules across different infections offer a powerful way to identify strain-specific and shared components of the regulatory program. An important challenge is to identify statistically robust gene expression modules as well as to reliably detect genes that change their module memberships between infections. We present MULCCH (MULti-task spectral Consensus Clustering for Hierarchically related tasks), a consensus extension of a multi-task clustering algorithm to infer high-confidence strain-specific host response modules under infections from multiple virus strains. On simulated data, MULCCH more accurately identifies genes exhibiting pathogen-specific patterns compared to non-consensus and nonmulti-task clustering approaches. Application of MULCCH to mammalian transcriptional response to a panel of influenza viruses showed that our method identifies clusters with greater coherence compared to non-consensus methods. Further, MULCCH derived clusters are enriched for several immune system-related processes and regulators. In summary, MULCCH provides a reliable module-based approach to identify molecular pathways and gene sets characterizing commonality and specificity of host response to viruses of different pathogenicities. The source code is available at https://bitbucket.org/roygroup/mulcch sroy@biostat.wisc.edu Supplementary data are available at Bioinformatics online.

GE-37 * INTEGRATED GENOMIC ANALYSIS OF LOW GRADE GLIOMA AND GLIOBLASTOMA REVEALS MOLECULAR STRATIFICATION BY IDH STATUS ACROSS GRADES AND HISTOLOGIES

TCGA has performed comprehensive molecular characterization of glioblastoma (GBM) and diffuse low grade glioma (LGG), which suggested heterogeneity in clinical diagnosis: GBM with isocitrate dehydrogenase (IDH) gene mutations corresponded with a hypermethylator/G-CIMP phenotype and relatively favorable outcome, while IDH-wildtype LGG were associated with GBM-like genomic and methylation features and poor survival. By integrating the GBM and LGG datasets, the TCGA LGG-GBM working group aims to classify them in a clinically relevant manner and to shed light on potential targets for therapy. Unsupervised clustering of 289 LGG and 398 GBM using methylation profiles resulted into six major groups, mainly driven by IDH status and age. G-CIMP/IDH mutant GBM clustered with IDH mutant LGGs and IDHwt LGG grouped with IDHwt GBM. Some but not all LGG with 1p/19q co-deletion, a hallmark of oligodendroglioma, clustered separate from other LGGs and associated with less favorable outcome, suggesting clinical value of methylation based patient stratification. IDH-wildtype LGG and GBM clustered into a classical-mesenchymal class and a proneural-mesenchymal subtype, the latter with normal-like methylation features. Unsupervised transcriptome clustering of 282 LGGs and 152 GBMs was again driven by IDH status, followed by strong associations with mutations in TP53/ATRX. Supervised comparison of IDH wildtype LGG and GBM suggested upregulation of oxidative phosphorylation gene sets in GBM which may provide an intriguing possibility of pathogenetic differences between histologically different but genomically similar gliomas. We evaluated the distribution of genomic alterations in glioma disease pathways and found that G-CIMP GBM uniquely associated with abnormalities in the MYC pathway, revealing a possible explanation for the aggressive behavior of IDH-mutant GBM relative to IDH-mutant LGG. In this first of its kind study, we improve our understanding of gliomagenesis by identifying shared and subtype-unique disease pathways and provide new modalities towards integrated classification of diffuse low and high grade glioma.

Abstract 4799: Identification and characterization of stromal factors with clinical significance in the ovarian tumor microenvironment

Abstract Ovarian cancer is the most lethal gynecologic malignancy in the United States. While the tumor microenvironment has been shown to play important roles in cancer pathogenesis, large-scale transcriptome profiles of the stromal component of ovarian tumors and the identification of stromal prognostic markers and therapeutic targets are lacking. This study seeks to identify secreted stromal factors and evaluate their clinical significance. Transcriptome profiling and clustering analysis on microdissected ovarian cancer associated fibroblasts (CAFs) samples from 83 patients revealed a heterogeneous gene expression pattern. Since CAFs can be derived from mesenchymal stem cells (MSCs), we compared transcriptome profiles of MSC samples obtained from healthy individuals with those from CAFs. Results showed that CAF expression profiles could be classified into two major subtypes. The MSC subtype, which has gene expression pattern highly resembling to that of undifferentiated MSCs, is significantly associated with poorer patient survival when compared to the non-MSC subtype. Further analyses identified two expression clusters within the MSC subtype: 1) the CAF-C cluster which expressed high level of MSC chondrogenic differentiation associated genes including CSPG2, SFRP2 and COMP, and 2) the CAF-O cluster which expressed high level of osteogenic differentiation associated genes including RUNX2 and BGN. Survival analysis showed that patients with the CAF-O expression subtype had overall survival rates comparable to those with the non-MSC subtypes, while the CAF-C subtype is associated with worst clinical outcomes. Pathway analyses revealed strong interactions among these stromal genes. Among them, SFRP2, which has been shown to inhibit osteogenic differentiation of MSCs by inhibiting Wnt signaling and subsequently suppress RUNX2 expression, was selected for further validation studies. Stromal SFRP2 and RUNX2 protein expression levels were evaluated by immunohistochemistry on paraffin tissue sections from 94 patients. Results showed that high levels of stromal SFRP2 were significantly associated with poor patient survival (p &amp;lt; 0.001), while high levels of stromal RUNX2 were significantly associated with improved patient survival (p &amp;lt; 0.001). A significant inverse correlation between SFRP2 and RUNX2 expression levels was also observed (r = -0.279, p = 0.037), suggested that the interactions among these genes may generate different landscapes in the tumor microenvironment, which modulate MSC differentiation and subsequently cancer cell aggressiveness. This study presents a new paradigm of which osteogenic-like and chondrogenic-like CAF signatures are associated with better and poorer patient survival respectively. Understanding the relationship between the expression patterns of stromal factors and MSC differentiation would provide us with new insights into ovarian cancer pathogenesis. Citation Format: Tsz-Lun Yeung, Cecilia S. Leung, Kwong-Kwok Wong, Samuel C. Mok. Identification and characterization of stromal factors with clinical significance in the ovarian tumor microenvironment. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 4799. doi:10.1158/1538-7445.AM2014-4799

Global transcriptome analyses of human and murine terminal erythroid differentiation

AbstractWe recently developed fluorescence-activated cell sorting (FACS)-based methods to purify morphologically and functionally discrete populations of cells, each representing specific stages of terminal erythroid differentiation. We used these techniques to obtain pure populations of both human and murine erythroblasts at distinct developmental stages. RNA was prepared from these cells and subjected to RNA sequencing analyses, creating unbiased, stage-specific transcriptomes. Tight clustering of transcriptomes from differing stages, even between biologically different replicates, validated the utility of the FACS-based assays. Bioinformatic analyses revealed that there were marked differences between differentiation stages, with both shared and dissimilar gene expression profiles defining each stage within transcriptional space. There were vast temporal changes in gene expression across the differentiation stages, with each stage exhibiting unique transcriptomes. Clustering and network analyses revealed that varying stage-specific patterns of expression observed across differentiation were enriched for genes of differing function. Numerous differences were present between human and murine transcriptomes, with significant variation in the global patterns of gene expression. These data provide a significant resource for studies of normal and perturbed erythropoiesis, allowing a deeper understanding of mechanisms of erythroid development in various inherited and acquired erythroid disorders.

Transcriptome deep-sequencing and clustering of expressed isoforms from Favia corals.

BackgroundGenomic and transcriptomic sequence data are essential tools for tackling ecological problems. Using an approach that combines next-generation sequencing, de novo transcriptome assembly, gene annotation and synthetic gene construction, we identify and cluster the protein families from Favia corals from the northern Red Sea.ResultsWe obtained 80 million 75 bp paired-end cDNA reads from two Favia adult samples collected at 65 m (Fav1, Fav2) on the Illumina GA platform, and generated two de novo assemblies using ABySS and CAP3. After removing redundancy and filtering out low quality reads, our transcriptome datasets contained 58,268 (Fav1) and 62,469 (Fav2) contigs longer than 100 bp, with N50 values of 1,665 bp and 1,439 bp, respectively. Using the proteome of the sea anemone Nematostella vectensis as a reference, we were able to annotate almost 20% of each dataset using reciprocal homology searches. Homologous clustering of these annotated transcripts allowed us to divide them into 7,186 (Fav1) and 6,862 (Fav2) homologous transcript clusters (E-value ≤ 2e-30). Functional annotation categories were assigned to homologous clusters using the functional annotation of Nematostella vectensis. General annotation of the assembled transcripts was improved 1-3% using the Acropora digitifera proteome. In addition, we screened these transcript isoform clusters for fluorescent proteins (FPs) homologs and identified seven potential FP homologs in Fav1, and four in Fav2. These transcripts were validated as bona fide FP transcripts via robust fluorescence heterologous expression. Annotation of the assembled contigs revealed that 1.34% and 1.61% (in Fav1 and Fav2, respectively) of the total assembled contigs likely originated from the corals’ algal symbiont, Symbiodinium spp.ConclusionsHere we present a study to identify the homologous transcript isoform clusters from the transcriptome of Favia corals using a far-related reference proteome. Furthermore, the symbiont-derived transcripts were isolated from the datasets and their contribution quantified. This is the first annotated transcriptome of the genus Favia, a major increase in genomics resources available in this important family of corals.

Structural and functional organization of RNA regulons in the post-transcriptional regulatory network of yeast

Post-transcriptional control of mRNA transcript processing by RNA binding proteins (RBPs) is an important step in the regulation of gene expression and protein production. The post-transcriptional regulatory network is similar in complexity to the transcriptional regulatory network and is thought to be organized in RNA regulons, coherent sets of functionally related mRNAs combinatorially regulated by common RBPs. We integrated genome-wide transcriptional and translational expression data in yeast with large-scale regulatory networks of transcription factor and RBP binding interactions to analyze the functional organization of post-transcriptional regulation and RNA regulons at a system level. We found that post-transcriptional feedback loops and mixed bifan motifs are overrepresented in the integrated regulatory network and control the coordinated translation of RNA regulons, manifested as clusters of functionally related mRNAs which are strongly coexpressed in the translatome data. These translatome clusters are more functionally coherent than transcriptome clusters and are expressed with higher mRNA and protein levels and less noise. Our results show how the post-transcriptional network is intertwined with the transcriptional network to regulate gene expression in a coordinated way and that the integration of heterogeneous genome-wide datasets allows to relate structure to function in regulatory networks at a system level.