Feature selection followed by a residuals-based normalization simplifies and improves single-cell gene expression analysis.

Normalization is a critical step in the computational analysis of single-cell RNA-sequencing (scRNA-seq) counts data. The objective is to reduce systematic biases introduced by technical sources that can obscure underlying biological differences. This is typically accomplished by re-scaling the observed counts to reduce the differences in total counts between the cells and then transforming the scaled counts to stabilize the variances. In the standard scRNA-seq workflow, this is followed by feature selection to identify genes that capture most of the biologically meaningful variation across the cells. Here, we propose a simple feature selection method and show that we can perform feature selection before normalization. We also propose a novel residuals-based normalization method that includes a monotonic non-linear transformation to ensure effective variance stabilization of the residuals. We demonstrate significant improvements in downstream clustering analyses through the application of our feature selection and normalization methods to truth-known biological as well as simulated counts data sets. Based on these results, we make the case for a revised scRNA-seq analysis workflow wherein feature selection precedes and in fact informs our residuals-based normalization. This novel workflow has been implemented in an R package called Piccolo.

Therapeutic targeting of ACLY in T-ALL in vivo.

T-cell Acute Lymphoblastic Leukemia (T-ALL) is a hematological malignancy in need of novel therapeutic approaches. Here, we identify the ATP-citrate lyase ACLY as a novel therapeutic target in T-ALL. Our results show that ACLY is overexpressed in T-ALL, and its expression correlates with NOTCH1 activity. To test the effects of ACLY in leukemia progression and the response to NOTCH1 inhibition, we developed an isogenic model of NOTCH1-induced Acly conditional knockout leukemia. Importantly, we observed intrinsic antileukemic effects upon loss of ACLY, which further synergized with NOTCH1 inhibition in vivo . Gene expression profiling analyses showed that the transcriptional signature of ACLY loss very significantly correlates with the signature of NOTCH1 inhibition in vivo , with significantly downregulated pathways related to oxidative phosphorylation, electron transport chain, ribosomal biogenesis and nucleosome biology. Consistently, metabolomic profiling upon ACLY loss revealed a metabolic crisis with accumulation of nucleotide intermediates and reduced levels of several amino acids. Overall, our results identify a link between NOTCH1 and ACLY and unveil ACLY as a novel promising target for T-ALL treatment.

A Therapeutically Targetable NOTCH1-SIRT1-KAT7 Axis in T-cell Leukemia.

T-cell acute lymphoblastic leukemia (T-ALL) is a NOTCH1-driven disease in need of novel therapies. Here, we identify a NOTCH1-SIRT1-KAT7 link as a therapeutic vulnerability in T-ALL, in which the histone deacetylase SIRT1 is overexpressed downstream of a NOTCH1-bound enhancer. SIRT1 loss impaired leukemia generation, whereas SIRT1 overexpression accelerated leukemia and conferred resistance to NOTCH1 inhibition in a deacetylase-dependent manner. Moreover, pharmacologic or genetic inhibition of SIRT1 resulted in significant antileukemic effects. Global acetyl proteomics upon SIRT1 loss uncovered hyperacetylation of KAT7 and BRD1, subunits of a histone acetyltransferase complex targeting H4K12. Metabolic and gene-expression profiling revealed metabolic changes together with a transcriptional signature resembling KAT7 deletion. Consistently, SIRT1 loss resulted in reduced H4K12ac, and overexpression of a nonacetylatable KAT7-mutant partly rescued SIRT1 loss-induced proliferation defects. Overall, our results uncover therapeutic targets in T-ALL and reveal a circular feedback mechanism balancing deacetylase/acetyltransferase activation with potentially broad relevance in cancer. SIGNIFICANCE: We identify a T-ALL axis whereby NOTCH1 activates SIRT1 through an enhancer region, and SIRT1 deacetylates and activates KAT7. Targeting SIRT1 shows antileukemic effects, partly mediated by KAT7 inactivation. Our results reveal T-ALL therapeutic targets and uncover a rheostat mechanism between deacetylase/acetyltransferase activities with potentially broader cancer relevance. This article is highlighted in the In This Issue feature, p. 1.

Cilium Expression Score Predicts Glioma Survival.

The accurate classification, prognostication, and treatment of gliomas has been hindered by an existing cellular, genomic, and transcriptomic heterogeneity within individual tumors and their microenvironments. Traditional clustering is limited in its ability to distinguish heterogeneity in gliomas because the clusters are required to be exclusive and exhaustive. In contrast, biclustering can identify groups of co-regulated genes with respect to a subset of samples and vice versa. In this study, we analyzed 1,798 normal and tumor brain samples using an unsupervised biclustering approach. We identified co-regulated gene expression profiles that were linked to proximally located brain regions and detected upregulated genes in subsets of gliomas, associated with their histologic grade and clinical outcome. In particular, we present a cilium-associated signature that when upregulated in tumors is predictive of poor survival. We also introduce a risk score based on expression of 12 cilium-associated genes which is reproducibly informative of survival independent of other prognostic biomarkers. These results highlight the role of cilia in development and progression of gliomas and suggest potential therapeutic vulnerabilities for these highly aggressive tumors.

A novel and highly effective mitochondrial uncoupling drug in T-cell leukemia.

T-cell acute lymphoblastic leukemia (T-ALL) is an aggressive hematologic malignancy. Despite recent advances in treatments with intensified chemotherapy regimens, relapse rates and associated morbidities remain high. In this context, metabolic dependencies have emerged as a druggable opportunity for the treatment of leukemia. Here, we tested the antileukemic effects of MB1-47, a newly developed mitochondrial uncoupling compound. MB1-47 treatment in T-ALL cells robustly inhibited cell proliferation via both cytostatic and cytotoxic effects as a result of compromised mitochondrial energy and metabolite depletion, which severely impaired nucleotide biosynthesis. Mechanistically, acute treatment with MB1-47 in primary leukemias promoted adenosine monophosphate-activated serine/threonine protein kinase (AMPK) activation and downregulation of mammalian target of rapamycin (mTOR) signaling, stalling anabolic pathways that support leukemic cell survival. Indeed, MB1-47 treatment in mice harboring either murine NOTCH1-induced primary leukemias or human T-ALL patient-derived xenografts (PDXs) led to potent antileukemic effects with a significant extension in survival without overlapping toxicities. Overall, our findings demonstrate a critical role for mitochondrial oxidative phosphorylation in T-ALL and uncover MB1-47-driven mitochondrial uncoupling as a novel therapeutic strategy for the treatment of this disease.

A Tumor Suppressor Enhancer of PTEN in T-cell development and leukemia.

Long-range oncogenic enhancers play an important role in cancer. Yet, whether similar regulation of tumor suppressor genes is relevant remains unclear. Loss of expression of PTEN is associated with the pathogenesis of various cancers, including T-cell leukemia (T-ALL). Here, we identify a highly conserved distal enhancer (PE) that interacts with the PTEN promoter in multiple hematopoietic populations, including T-cells, and acts as a hub of relevant transcription factors in T-ALL. Consistently, loss of PE leads to reduced PTEN levels in T-ALL cells. Moreover, PE-null mice show reduced Pten levels in thymocytes and accelerated development of NOTCH1-induced T-ALL. Furthermore, secondary loss of PE in established leukemias leads to accelerated progression and a gene expression signature driven by Pten loss. Finally, we uncovered recurrent deletions encompassing PE in T-ALL, which are associated with decreased PTEN levels. Altogether, our results identify PE as the first long-range tumor suppressor enhancer directly implicated in cancer.

Spleen plays a major role in DLL4-driven acute T-cell lymphoblastic leukemia.

The Notch pathway is highly active in almost all patients with T-cell acute lymphoblastic leukemia (T-ALL), but the implication of Notch ligands in T-ALL remains underexplored. Methods: We used a genetic mouse model of Notch ligand delta like 4 (DLL4)-driven T-ALL and performed thymectomies and splenectomies in those animals. We also used several patient-derived T-ALL (PDTALL) models, including one with DLL4 expression on the membrane and we treated PDTALL cells in vitro and in vivo with demcizumab, a blocking antibody against human DLL4 currently being tested in clinical trials in patients with solid cancer. Results: We show that surgical removal of the spleen abrogated T-ALL development in our preclinical DLL4-driven T-ALL mouse model. Mechanistically, we found that the spleen, and not the thymus, promoted the accumulation of circulating CD4+CD8+ T cells before T-ALL onset, suggesting that DLL4-driven T-ALL derives from these cells. Then, we identified a small subset of T-ALL patients showing higher levels of DLL4 expression. Moreover, in mice xenografted with a DLL4-positive PDTALL model, treatment with demcizumab had the same therapeutic effect as global Notch pathway inhibition using the potent γ-secretase inhibitor dibenzazepine. This result demonstrates that, in this PDTALL model, Notch pathway activity depends on DLL4 signaling, thus validating our preclinical mouse model. Conclusion: DLL4 expression in human leukemic cells can be a source of Notch activity in T-ALL, and the spleen plays a major role in a genetic mouse model of DLL4-driven T-ALL.

Gene expression of adipokines and adipokine receptors in the tumor microenvironment: associations of lower expression with more aggressive breast tumor features.

PURPOSE: Limited epidemiologic data are available on the expression of adipokines leptin (LEP) and adiponectin (ADIPOQ) and adipokine receptors (LEPR, ADIPOR1, ADIPOR2) in the breast tumor microenvironment (TME). The associations of gene expression of these biomarkers with tumor clinicopathology are not well understood. METHODS: NanoString multiplexed assays were used to assess the gene expression levels of LEP, LEPR, ADIPOQ, ADIPOR1, and ADIPOR2 within tumor tissues among 162 Black and 55 White women with newly diagnosed breast cancer. Multivariate mixed effects models were used to estimate associations of gene expression with breast tumor clinicopathology (overall and separately among Blacks). RESULTS: Black race was associated with lower gene expression of LEPR (P = 0.002) and ADIPOR1 (P = 0.01). Lower LEP, LEPR, and ADIPOQ gene expression were associated with higher tumor grade (P = 0.0007, P < 0.0001, and P < 0.0001, respectively) and larger tumor size (P < 0.0001, P = 0.0005, and P < 0.0001, respectively). Lower ADIPOQ expression was associated with ER- status (P = 0.0005), and HER2-enriched (HER2-E; P = 0.0003) and triple-negative (TN; P = 0.002) subtypes. Lower ADIPOR2 expression was associated with Ki67+ status (P = 0.0002), ER- status (P < 0.0001), PR- status (P < 0.0001), and TN subtype (P = 0.0002). Associations of lower adipokine and adipokine receptor gene expression with ER-, HER2-E, and TN subtypes were confirmed using data from The Cancer Genome Atlas (P-values < 0.005). CONCLUSION: These findings suggest that lower expression of ADIPOQ, ADIPOR2, LEP, and LEPR in the breast TME might be indicators of more aggressive breast cancer phenotypes. Validation of these findings are warranted to elucidate the role of the adipokines and adipokine receptors in long-term breast cancer prognosis.

Tissue- and development-stage-specific mRNA and heterogeneous CNV signatures of human ribosomal proteins in normal and cancer samples.

We give results from a detailed analysis of human Ribosomal Protein (RP) levels in normal and cancer samples and cell lines from large mRNA, copy number variation and ribosome profiling datasets. After normalizing total RP mRNA levels per sample, we find highly consistent tissue specific RP mRNA signatures in normal and tumor samples. Multiple RP mRNA-subtypes exist in several cancers, with significant survival and genomic differences. Some RP mRNA variations among subtypes correlate with copy number loss of RP genes. In kidney cancer, RP subtypes map to molecular subtypes related to cell-of-origin. Pan-cancer analysis of TCGA data showed widespread single/double copy loss of RP genes, without significantly affecting survival. In several cancer cell lines, CRISPR-Cas9 knockout of RP genes did not affect cell viability. Matched RP ribosome profiling and mRNA data in humans and rodents stratified by tissue and development stage and were strongly correlated, showing that RP translation rates were proportional to mRNA levels. In a small dataset of human adult and fetal tissues, RP protein levels showed development stage and tissue specific heterogeneity of RP levels. Our results suggest that heterogeneous RP levels play a significant functional role in cellular physiology, in both normal and disease states.

TuBA: Tunable biclustering algorithm reveals clinically relevant tumor transcriptional profiles in breast cancer.

BACKGROUND: Traditional clustering approaches for gene expression data are not well adapted to address the complexity and heterogeneity of tumors, where small sets of genes may be aberrantly co-expressed in specific subsets of tumors. Biclustering algorithms that perform local clustering on subsets of genes and conditions help address this problem. We propose a graph-based Tunable Biclustering Algorithm (TuBA) based on a novel pairwise proximity measure, examining the relationship of samples at the extremes of genes' expression profiles to identify similarly altered signatures. RESULTS: TuBA's predictions are consistent in 3,940 breast invasive carcinoma samples from 3 independent sources, using different technologies for measuring gene expression (RNA sequencing and Microarray). More than 60% of biclusters identified independently in each dataset had significant agreement in their gene sets, as well as similar clinical implications. Approximately 50% of biclusters were enriched in the estrogen receptor-negative/HER2-negative (or basal-like) subtype, while >50% were associated with transcriptionally active copy number changes. Biclusters representing gene co-expression patterns in stromal tissue were also identified in tumor specimens. CONCLUSIONS: TuBA offers a simple biclustering method that can identify biologically relevant gene co-expression signatures not captured by traditional unsupervised clustering approaches. It complements biclustering approaches that are designed to identify constant or coherent submatrices in gene expression datasets, and outperforms them in identifying a multitude of altered transcriptional profiles that are associated with observed genomic heterogeneity of diseased states in breast cancer, both within and across tumor subtypes, a promising step in understanding disease heterogeneity, and a necessary first step in individualized therapy.

Gamma Secretase Inhibition by BMS-906024 Enhances Efficacy of Paclitaxel in Lung Adenocarcinoma.

Notch signaling is aberrantly activated in approximately one third of non-small cell lung cancers (NSCLC). We characterized the interaction between BMS-906024, a clinically relevant Notch gamma secretase inhibitor, and front-line chemotherapy in preclinical models of NSCLC. Chemosensitivity assays were performed on 14 human NSCLC cell lines. There was significantly greater synergy between BMS-906024 and paclitaxel than BMS-906024 and cisplatin [mean combination index (CI) value, 0.54 and 0.85, respectively, P = 0.01]. On an extended panel of 31 NSCLC cell lines, 25 of which were adenocarcinoma, the synergy between BMS-906024 and paclitaxel was significantly greater in KRAS- and BRAF-wildtype than KRAS- or BRAF-mutant cells (mean CI, 0.43 vs. 0.90, respectively; P = 0.003). Paclitaxel-induced Notch1 activation was associated with synergy between BMS-906024 and paclitaxel in the KRAS- or BRAF-mutant group. Knockdown of mutant KRAS increased the synergy between BMS-906024 and paclitaxel in heterozygous KRAS-mutant cell lines. Among KRAS- or BRAF-mutant NSCLC, there was a significant correlation between synergy and mutant or null TP53 status, as well as between synergy and a low H2O2 pathway signature. Exogenous overexpression of activated Notch1 or Notch3 had no effect on the enhanced sensitivity of NSCLC to paclitaxel by BMS-906024. In vivo studies with cell line- and patient-derived lung adenocarcinoma xenografts confirmed enhanced antitumor activity for BMS-906024 plus paclitaxel versus either drug alone via decreased cell proliferation and increased apoptosis. These results show that BMS-906024 sensitizes NSCLC to paclitaxel and that wild-type KRAS and BRAF status may predict better patient response to the combination therapy. Mol Cancer Ther; 16(12); 2759-69. ©2017 AACR.

The Modular Adaptive Ribosome.

The ribosome is an ancient machine, performing the same function across organisms. Although functionally unitary, recent experiments suggest specialized roles for some ribosomal proteins. Our central thesis is that ribosomal proteins function in a modular fashion to decode genetic information in a context dependent manner. We show through large data analyses that although many ribosomal proteins are essential with consistent effect on growth in different conditions in yeast and similar expression across cell and tissue types in mice and humans, some ribosomal proteins are used in an environment specific manner. The latter set of variable ribosomal proteins further function in a coordinated manner forming modules, which are adapted to different environmental cues in different organisms. We show that these environment specific modules of ribosomal proteins in yeast have differential genetic interactions with other pathways and their 5'UTRs show differential signatures of selection in yeast strains, presumably to facilitate adaptation. Similarly, we show that in higher metazoans such as mice and humans, different modules of ribosomal proteins are expressed in different cell types and tissues. A clear example is nervous tissue that uses a ribosomal protein module distinct from the rest of the tissues in both mice and humans. Our results suggest a novel stratification of ribosomal proteins that could have played a role in adaptation, presumably to optimize translation for adaptation to diverse ecological niches and tissue microenvironments.