Search

Supplementary table 2 shows Spearman correlation values between drug sensitivity and gene expression for 481 drugs studied in the CTRP dataset.

Supplementary Table 1 shows Spearman correlation values between drug sensitivity and gene expression for 250 drugs studied in the GDSC dataset.

Supplementary Table 6a: Spearman correlation values between drug sensitivity and target expression for 86 drugs common between GDSC and CTRP. Supplementary Table 6b: Correlation matrix comparing gene expression values. There are 558 cell lines and 15,934 genes common between both the datasets. Calculated 558x 558 - Correlation matrix comparing gene expression values for 15,934 genes in one cell line from GDSC compared to 15,934 genes in CTRP. Correlation method used was Spearman. Supplementary Table 6c: Spearman correlation values comparing Area under the curve (AUC) values across common cell lines in the GDSC and CTRP datasets for the 86 drugs common between them. Supplementary Table 6d: List of the cell lines where a better gene expression correlation was observed with a different cell line.

Full listing of supplementary files and their legends

Cancer precision medicine aims to predict the drug likely to yield the best response for a patient. Genomic sequencing of tumors is currently being used to better inform treatment options; however, this approach has had a limited clinical impact due to the paucity of actionable mutations. An alternative to mutation status is the use of gene expression signatures to predict response. Using data from two large-scale studies, The Genomics of Drug Sensitivity of Cancer (GDSC) and The Cancer Therapeutics Response Portal (CTRP), we investigated the relationship between the sensitivity of hundreds of cell lines to hundreds of drugs, and the relative expression levels of the targets these drugs are directed against. For approximately one third of the drugs considered (73/222 in GDSC and 131/360 in CTRP), sensitivity was significantly correlated with the expression of at least one of the known targets. Surprisingly, for 8% of the annotated targets, there was a significant anticorrelation between target expression and sensitivity. For several cases, this corresponded to drugs targeting multiple genes in the same family, with the expression of one target significantly correlated with sensitivity and another significantly anticorrelated suggesting a possible role in resistance. Furthermore, we identified nontarget genes that are significantly correlated or anticorrelated with drug sensitivity, and find literature linking several to sensitization and resistance. Our analyses provide novel and important insights into both potential mechanisms of resistance and relative efficacy of drugs against the same target.

Supplementary Table 3a: Molecular Targets of the Anticancer Drugs from GDSC, CTRP and DrugBank. Spearman correlation and permutation testing results are provided for both the GDSC and CTRP datasets. Supplementary Table 4: List of drugs where we observe both a correlated and an anti-correlated target in GDSC and CRTP datasets. Supplementary Table 5: Fractions of single cells in each cell line expressing both targets from Figure 3. Supplementary Table 7: Transcript isoform aware reanalysis of CTRP-GDSC concordance. Spearman correlation values between drug sensitivity and drug-target transcript isoforms expression for 74 drugs common between GDSC and CTRP. The 541 cell lines common in GDSC and CTRP where drug-target transcript isoform expression and AUC values could be obtained were considered. As GDSC and CTRP provide gene-level expression measurements from different microarray versions (Affymetrix GeneChip HG-U133A in GDSC and HG-U133PLUS2 in CTRP), which potentially detect different isoforms of each gene, we repeated our analysis at the transcript level. Supplementary Table 8: The top 20 strongest correlations and top 20 strongest anti-correlations between drug sensitivity and gene expression for the GDSC and CTRP datasets. Supplementary Table 9: Drug-target pairs that are significantly correlated in cell lines derived from a cancer subtype but not significantly correlated in the pan-cancer analysis.

Supplementary Figure 1: Comparison of correlations for ALK and MET. Blue, red and grey cells denote significant correlations, significant anti-correlations and non-significant correlations respectively. White cells correspond to combinations that were not tested or are not valid drug-target pairs. Supplementary Figure 2: Comparison of correlations for HDAC1, HDAC2, HDAC3, HDAC6 and HDAC8. Blue, red and grey cells denote significant correlations, significant anti-correlations and non-significant correlations respectively. White cells correspond to combinations that were not tested or are not valid drug-target pairs. Supplementary Figure 3: Summary of the degree of overlap between the GDSC and CTRP analyses at the transcript level. As GDSC and CTRP provide gene-level expression measurements from different microarray versions (Affymetrix GeneChip HG-U133A in GDSC and HG-U133PLUS2 in CTRP) which potentially detect different isoforms of each gene, we repeated our concordance analysis at the transcript level. The correlation coefficient was calculated for 74 common drugs (targeting drug-target transcript isoforms). The 541 cell lines common in GDSC and CTRP where drug-target transcript isoform expression and AUC values could be obtained were considered.

Cancer precision medicine aims to predict the drug likely to yield the best response for a patient. Genomic sequencing of tumors is currently being used to better inform treatment options; however, this approach has had a limited clinical impact due to the paucity of actionable mutations. An alternative to mutation status is the use of gene expression signatures to predict response. Using data from two large-scale studies, The Genomics of Drug Sensitivity of Cancer (GDSC) and The Cancer Therapeutics Response Portal (CTRP), we investigated the relationship between the sensitivity of hundreds of cell lines to hundreds of drugs, and the relative expression levels of the targets these drugs are directed against. For approximately one third of the drugs considered (73/222 in GDSC and 131/360 in CTRP), sensitivity was significantly correlated with the expression of at least one of the known targets. Surprisingly, for 8% of the annotated targets, there was a significant anticorrelation between target expression and sensitivity. For several cases, this corresponded to drugs targeting multiple genes in the same family, with the expression of one target significantly correlated with sensitivity and another significantly anticorrelated suggesting a possible role in resistance. Furthermore, we identified nontarget genes that are significantly correlated or anticorrelated with drug sensitivity, and find literature linking several to sensitization and resistance. Our analyses provide novel and important insights into both potential mechanisms of resistance and relative efficacy of drugs against the same target.

Diagnostic imaging of aggressive cancer with a high stroma content may benefit from the use of imaging contrast agents targeted with peptides that have high binding affinity to the extracellular matrix (ECM). In this study, we report the use of superparamagnetic iron-oxide nanoparticles (IO-NP) conjugated to a nonapeptide, CSGRRSSKC (CSG), which specifically binds to the laminin-nidogen-1 complex in tumours. We show that CSG-IO-NP accumulate in tumours, predominantly in the tumour ECM, following intravenous injection into a murine model of pancreatic neuroendocrine tumour (PNET). In contrast, a control untargeted IO-NP consistently show poor tumour uptake, and IO-NP conjugated to a pentapeptide. CREKA that bind fibrin clots in blood vessels show restricted uptake in the angiogenic vessels of the tumours. CSG-IO-NP show three-fold higher intratumoral accumulation compared to CREKA-IO-NP. Magnetic resonance imaging (MRI) T2-weighted scans and T2 relaxation times indicate significant uptake of CSG-IO-NP irrespective of tumour size, whereas the uptake of CREKA-IO-NP is only consistent in small tumours of less than 3 mm in diameter. Larger tumours with significantly reduced tumour blood vessels show a lack of CREKA-IO-NP uptake. Our data suggest CSG-IO-NP are particularly useful for detecting stroma in early and advanced solid tumours.

Uncaging promoter and enhancer dynamics In order to understand cellular differentiation, it is important to understand the timing of the regulation of gene expression. Arner et al. used cap analysis of gene expression (CAGE) to analyze gene enhancer and promoter activities in a number of human and mouse cell types. The RNA of enhancers was transcribed first, followed by that of transcription factors, and finally by genes that are not transcription factors. Science , this issue p. 1010

Scientific Data, 4, ISSN:2052-4463

Myeloid leukemia factor 1 (MLF1) is an oncoprotein associated with hemopoietic lineage commitment and acute myeloid leukemia. Here we show that Mlf1 associated with a novel binding partner, Mlf1-associated nuclear protein (Manp), a new heterogeneous nuclear ribonucleoprotein (hnRNP) family member, related to hnRNP-U. Manp localized exclusively in the nucleus and could redirect Mlf1 from the cytoplasm into the nucleus. The nuclear content of Mlf1 was also regulated by 14-3-3 binding to a canonical 14-3-3 binding motif within the N terminus of Mlf1. Significantly Mlf1 contains a functional nuclear export signal and localized primarily to the nuclei of hemopoietic cells. Mlf1 was capable of binding DNA, and microarray analysis revealed that it affected the expression of several genes, including transcription factors. In summary, this study reveals that Mlf1 translocates between nucleus and cytoplasm, associates with a novel hnRNP, and influences gene expression.

Myeloid leukemia factor 1 (MLF1) is a novel oncoprotein involved in translocations associated with acute myeloid leukemia (AML), especially erythroleukemias. In this study, we demonstrate that ectopic expression of Mlf1 prevented J2E erythroleukemic cells from undergoing biological and morphological maturation in response to erythropoietin (Epo). We show that Mlf1 inhibited Epo-induced cell cycle exit and suppressed a rise in the cell cycle inhibitor p27(Kip1). Unlike differentiating J2E cells, Mlf1-expressing cells did not downregulate Cul1 and Skp2, components of the ubiquitin E3 ligase complex SCF(Skp2) involved in the proteasomal degradation of p27(Kip1). In contrast, Mlf1 did not interfere with increases in p27(Kip1) and terminal differentiation initiated by thyroid hormone withdrawal from erythroid cells, or cytokine-stimulated maturation of myeloid cells. These data demonstrate that Mlf1 interferes with an Epo-responsive pathway involving p27(Kip1) accumulation, which inhibits cell cycle arrest essential for erythroid terminal differentiation.

A yeast two-hybrid screen was conducted to identify binding partners of Mlf1, an oncoprotein recently identified in a translocation with nucleophosmin that causes acute myeloid leukemia. Two proteins isolated in this screen were 14-3-3zeta and a novel adaptor, Madm. Mlf1 contains a classic RSXSXP sequence for 14-3-3 binding and is associated with 14-3-3zeta via this phosphorylated motif. Madm co-immunoprecipitated with Mlf1 and co-localized in the cytoplasm. In addition, Madm recruited a serine kinase, which phosphorylated both Madm and Mlf1 including the RSXSXP motif. In contrast to wild-type Mlf1, the oncogenic fusion protein nucleophosmin (NPM)-MLF1 did not bind 14-3-3zeta, had altered Madm binding, and localized exclusively in the nucleus. Ectopic expression of Madm in M1 myeloid cells suppressed cytokine-induced differentiation unlike Mlf1, which promotes maturation. Because the Mlf1 binding region of Madm and its own dimerization domain overlapped, the levels of Madm and Mlf1 may affect complex formation and regulate differentiation. In summary, this study has identified two partner proteins of Mlf1 that may influence its subcellular localization and biological function.

Regulated transcription controls the diversity, developmental pathways and spatial organization of the hundreds of cell types that make up a mammal. Using single-molecule cDNA sequencing, we mapped transcription start sites (TSSs) and their usage in human and mouse primary cells, cell lines and tissues to produce a comprehensive overview of mammalian gene expression across the human body. We find that few genes are truly 'housekeeping', whereas many mammalian promoters are composite entities composed of several closely separated TSSs, with independent cell-type-specific expression profiles. TSSs specific to different cell types evolve at different rates, whereas promoters of broadly expressed genes are the most conserved. Promoter-based expression analysis reveals key transcription factors defining cell states and links them to binding-site motifs. The functions of identified novel transcripts can be predicted by coexpression and sample ontology enrichment analyses. The functional annotation of the mammalian genome 5 (FANTOM5) project provides comprehensive expression profiles and functional annotation of mammalian cell-type-specific transcriptomes with wide applications in biomedical research.

Abstract 253 Hemopoietic lineage commitment is controlled, in part, by transcription factors that regulate specific genes required for the formation of mature blood cells. Differentiation along particular hemopoietic lineages is dependant not only on the presence of particular transcription factors, but also on appropriate concentrations - altering transcription factor levels can force cells into different hemopoietic pathways. Transcription factors undergo numerous post-translational modifications and are controlled spatially via sub-cellular localisation. De-regulation of transcription factors can result in leukemias, or other blood disorders. GATA-1 is an example of a key lineage-determining gene, essential for erythropoiesis. Increasing GATA-1 levels promotes maturation along the erythroid pathway, whereas reducing GATA-1 concentrations favours myelopoiesis. GATA-1 regulation occurs at multiple levels including transcription, translation and post-translational modifications such as phosphorylation, acetylation, ubiquitination and sumoylation. Although GATA-1 ubiquitination modifies the protein for proteasomal degradation, the effect of adding small ubiquitin-like modier (Sumo) to GATA-1 is unclear. Several examples of hemopoietic differentiation plasticity have been observed. We reported a lineage switch by erythroleukemic J2E cells which spontaneously developed a monoblastoid phenotype. Two genes (Hls5 and Hls7/Mlf1) were isolated from this lineage switch with potential lineage-determining features. Hls5 is a member of the RBCC (Ring finger, B-box, Coiled-coil) family of proteins, which includes PML. Ectopic expression of Hls5 impedes erythroid differentiation by reducing GATA-1 levels, and suppressing hemoglobin synthesis. Significantly, Hls5 relocates from the cytoplasm to associate with GATA-1 in the nucleus, where it interferes with DNA binding and transactivation of GATA-1. Several members of the RBCC family are ubiquitin E3 ligases, catalysing the final step in the ubiquitination process - these molecules play a vital role in regulating the levels of target proteins. Here we show that Hls5 is a bona fide ubiquitin E3 ligase, in partnership with several ubiquitin E2 enzymes. The Ring finger is critical for Hls5 ligase activity as mutation of key residues within the Ring finger ablates catalytic activity. Interestingly, a yeast 2 hybrid screen for Hls5 interactors identified Ubc9 and Pias1, which act as E2 and E3 enzymes in the sumoylation cascade. Co-immunoprecipitation, BRET and co-localization experiments confirmed the Hls5 association with Ubc9 and Pias1. Moreover, Hls5 binds Sumo-1 (but not Sumo-2 or 3), and co-localizes with Sumo-1 in discrete nuclear bodies. Thus, Hls5 interacts with several components of the intracellular sumoylation machinery. Hls5 can also reduce sumoylated proteins globally, indicating it may target these modified proteins for degradation. Recently, a new family of ubiquitin E3 ligases has been described which specifically mark sumoylated proteins for degradation. These Sumo-targeted ubiquitin ligases (STUbL) are found primarily in yeast, and only one mammalian STUbL has been identified. We postulated that Hls5 may be a STUbL, capable of regulating sumoylated GATA-1. Our data demonstrate that while Hls5 is able to bind GATA-1 via the B-box and Coiled-coil domains, it preferentially associates with sumoylated GATA-1 through a canonical Sumo interacting motif (SIM). This results in increased GATA-1 ubiquitination and, as a consequence, levels of sumoylated GATA-1 are reduced substantially. Since mutation of the lysine necessary for Sumo attachment does not affect GATA-1 transactivation, sumoylation may act as a prelude to ubiquitination and protein turn-over. We propose, therefore, that GATA-1 mediates transcription of target genes, and is subsequently sumoylated by Pias1 and Ubc9 – addition of Sumo moieties to GATA-1 enhance binding to Hls5, which in turn impedes GATA-1 DNA binding, and promotes ubiquitination for proteasomal degradation. This model is consistent with decreased levels of GATA-1 in erythroid cells ectopically expressing Hls5, and with the original isolation of Hls5 as a potential lineage-determining gene involved with the erythroid to monoblastoid lineage switch. Thus, Hls5 is a novel STUbL which plays a role in hemopoietic lineage commitment by modulating GATA-1 activity and content. Disclosures: No relevant conflicts of interest to declare.

The J2E erythroblastoid cell line responds to erythropoietin (Epo) by morphological maturation and hemoglobin synthesis. However, on rare occasions, these cells have undergone a spontaneous lineage switch and display features of monoblastoid cells which do not respond to Epo. Amongst the genes up-regulated in the monoblastoid variants were Hemopoietic lineage switch (Hls) 5 and 7. Hls5 is a recently identified member of the RING finger, B Box, Coiled coil (RBCC) or tripartite motif (TRIM) family, which includes PML, a gene involved in acute promyelocytic leukaemia. Hls7 is the murine orthologue of Myeloid leukaemia factor 1 (Mlf1), a gene involved in a t(3;5), associated with acute myeloid leukaemia. We have shown previously that Hls7/Mlf1 imposes a dramatic phenotypic change upon the erythroid cells, rendering them monoblastoid (Williams, J. et al EMBO 1999). We have studied the role of Hls5 and Mlf1 in erythroid commitment and differentiation. Ectopic expression of Hls5 inhibits globin production in erythroid cells and suppresses development of B-FUE and C-FUE. A yeast-two-hybrid screen identified FOG-1 as a binding partner of Hls5. Significantly, FOG-1 is a transcriptional co-regulator of GATA-1, which controls globin gene expression. While Hls5 is able to enhance the repression of GATA-1 activity imposed by FOG-1, it is also able to repress GATA-1 transcriptional activity in the absence of FOG-1. Using electrophoretic mobility shift assay we have shown that Hls5 is able to reduce GATA-1 binding to DNA in a dose dependant manner. This observation that Hls5 reduces GATA-1 binding to promoter elements is mirrored by chromatin immunoprecipitation assays. Expression of MLF1 is highest in CD34+ cells, but is markedly down regulated during erythroid differentiation. Microarray analysis identified a number of known transcriptional regulators differentially expressed in the presence of Mlf1 including ets1, Myc intron binding protein and Tbr2. Mlf1 is able to bind DNA and luciferase reporter assays demonstrated that Mlf1 is able to affect transcription. In addition, Mlf1 interacts with a novel member of the hnRNP family viz Mlf1 associated nuclear protein (Manp). Manp binds to DNA, is able to influence the subcellular localisation of Mlf1 by translocating Mlf1 from the cytoplasm to the nucleus. Importantly, Manp also has an affect on transcription. These data demonstrate that both Hls5 and Mlf1 affect transcription of genes associated with erythroid differentiation.