Search

Figure S1: LNCaP, VCaP and DuCaP cells were grown in charcoal-dextran stripped serum (CSS) for 48 h and treated with increasing concentrations of R1881 or DHT in the presence or absence of Enzalutamide (10 µM) or vehicle (Ctrl) for 48 h. Figure S2: Androgens increased lipid uptake of long-chain fatty acids. LAPC4 cells were grown in CSS for 48 h and treated with 1 nM R1881 or vehicle for 48 h. Figure S3: (A) LNCaP cells were synchronized in G0/G1 by androgen deprivation (CSS for 48 h) followed by treatment with Tunicamycin (1 mg/mL), Hydroxyurea (1 M), or Nocodazole (25 ug/mL) for another 24 h, placing cell cycle blocks in G0/G1, S phase and mitosis, respectively. Figure S4: Oncomine analysis of candidate lipid transporters in (A) Grasso [2], (B) Varambally [3] and (C) LaTulippe datasets [4] comparing gene expression of normal prostate gland versus localized, primary prostate cancer tumor samples. Figure S5: (A) DuCaP (top panel) and VCaP cells (bottom panel) were grown in CSS for 48 h and treated with 10 nM DHT in the absence or presence of Enz (10 µM) or vehicle (Ctrl) for 48 h. Table S1. Primer sequences

Aging is the primary risk factor for most neurodegenerative diseases, and recently coronavirus disease 2019 (COVID-19) has been associated with severe neurological manifestations that can eventually impact neurodegenerative conditions in the long-term. The progressive accumulation of senescent cellsin vivostrongly contributes to brain aging and neurodegenerative co-morbidities but the impact of virus-induced senescence in the aetiology of neuropathologies is unknown. Here, we show that senescent cells accumulate in physiologically aged brain organoids of human origin and that senolytic treatment reduces inflammation and cellular senescence; for which we found that combined treatment with the senolytic drugs dasatinib and quercetin rejuvenates transcriptomic human brain aging clocks. We further interrogated brain frontal cortex regions in postmortem patients who succumbed to severe COVID-19 and observed increased accumulation of senescent cells as compared to age-matched control brains from non-COVID-affected individuals. Moreover, we show that exposure of human brain organoids to SARS-CoV-2 evoked cellular senescence, and that spatial transcriptomic sequencing of virus-induced senescent cells identified a unique SARS-CoV-2 variant-specific inflammatory signature that is different from endogenous naturally-emerging senescent cells. Importantly, following SARS-CoV-2 infection of human brain organoids, treatment with senolytics blocked viral retention and prevented the emergence of senescent corticothalamic and GABAergic neurons. Furthermore, we demonstrate in human ACE2 overexpressing mice that senolytic treatment ameliorates COVID-19 brain pathology following infection with SARS-CoV-2.In vivotreatment with senolytics improved SARS-CoV-2 clinical phenotype and survival, alleviated brain senescence and reactive astrogliosis, promoted survival of dopaminergic neurons, and reduced viral and senescence-associated secretory phenotype gene expression in the brain. Collectively, our findings demonstrate SARS-CoV-2 can trigger cellular senescence in the brain, and that senolytic therapy mitigates senescence-driven brain aging and multiple neuropathological sequelae caused by neurotropic viruses, including SARS-CoV-2.

Background Single-cell RNA sequencing (scRNA-seq) methods have been advantageous for quantifying cell-to-cell variation by profiling the transcriptomes of individual cells. For scRNA-seq data, variability in gene expression reflects the degree of variation in gene expression from one cell to another. Analyses that focus on cell–cell variability therefore are useful for going beyond changes based on average expression and, instead, identifying genes with homogeneous expression versus those that vary widely from cell to cell. Results We present a novel statistical framework, scShapes, for identifying differential distributions in single-cell RNA-sequencing data using generalized linear models. Most approaches for differential gene expression detect shifts in the mean value. However, as single-cell data are driven by overdispersion and dropouts, moving beyond means and using distributions that can handle excess zeros is critical. scShapes quantifies gene-specific cell-to-cell variability by testing for differences in the expression distribution while flexibly adjusting for covariates if required. We demonstrate that scShapes identifies subtle variations that are independent of altered mean expression and detects biologically relevant genes that were not discovered through standard approaches. Conclusions This analysis also draws attention to genes that switch distribution shapes from a unimodal distribution to a zero-inflated distribution and raises open questions about the plausible biological mechanisms that may give rise to this, such as transcriptional bursting. Overall, the results from scShapes help to expand our understanding of the role that gene expression plays in the transcriptional regulation of a specific perturbation or cellular phenotype. Our framework scShapes is incorporated into a Bioconductor R package (https://www.bioconductor.org/packages/release/bioc/html/scShapes.html).

BackgroundSingle-cell RNA-sequencing (scRNA-seq) technologies enable the capture of gene expression heterogeneity and consequently cell-to-cell variability at the cell type level. Although different methods have been proposed to quantify cell-to-cell variability, it is unclear what the optimal statistical approach is, especially in light of challenging data structures that are unique to scRNA-seq data like zero inflation.ResultsIn this study, we conducted a systematic evaluation of cell-to-cell gene expression variability using 14 different variability metrics that are commonly applied to transcriptomic data. Performance was evaluated with respect to data-specific features like sparsity and sequencing platform, biological properties like gene length, and the ability to recapitulate true levels of variability based on simulation and known biological gene sets like ribosomal genes and stably expressed genes. scran had the strongest all-round performance, and this metric was then applied to investigate the changes in cell-to-cell variability that occur during ageing. Studying ageing showcases the value of cell-to-cell variability as it is a genetically-regulated program that is influenced by stochastic processes.scRNA-seq datasets from hematopoietic stem cells (HSCs) and B lymphocytes and other cell types from this differentiation lineage were used with scran to identify the genes with consistent patterns of variable and stable expression profiles during differentiation. Furthermore, to understand the regulatory relationship for genes that were differentially-variable in their expression between young and old mice, we constructed networks using transcription factors and their known targets for HSC and B lymphocyte cells. Comparisons of these networks identified a shared TFSfpi1that although was seen to increase in gene expression variability in old mice versus young in both cell types, the corresponding targets were distinct and their gene expression variability had different directions between cell types.ConclusionsThrough these analyses, we highlight the importance of capturing cell-to-cell gene expression variability in a complex biological process like differentiation and ageing, and emphasise the value and specificity of interpreting these findings at the level of individual cell types.

Metastasis is orchestrated by phenotypic and metabolic reprogramming underlying tumor aggressiveness. Redox signaling by mammary tumor knockdown (KD) of the antioxidant glutathione peroxidase 2 (GPx2) enhanced metastasis via dynamic changes in epithelial-to-mesenchymal transition. Single cell RNA sequencing (scRNA-seq) of the control and PyMT/GPx2 KD mammary tumor revealed six luminal and one basal/mesenchymal like (cluster 3) subpopulations. Remarkably, GPx2 KD enhanced the size and basal/mesenchymal gene signature of cluster 3 as well as induced epithelial/mesenchymal (E/M) clusters which expressed markers of oxidative phosphorylation and glycolysis, indicative of hybrid metabolism. These data were validated in human breast cancer xenografts and were supported by pseudotime cell trajectory analysis. Moreover, the E/M and M states were both attenuated by GPx2 gain of function or HIF1α inhibition, leading to metastasis suppression. Collectively, these results demonstrate that redox/HIF1α signaling promotes mesenchymal gene expression, resulting in E/M clusters and a mesenchymal root subpopulation, driving phenotypic and metabolic heterogeneity underlying metastasis.SignificanceBy leveraging single cell RNA analysis, we were able to demonstrate that redox signaling by GPx2 loss in mammary tumors results in HIF1α signaling, which promotes partial and full EMT conversions, represented by distinct tumor cell subpopulations, which in turn express hybrid and binary metabolic states. These data underscore a phenotypic and metabolic co-adaptation in cancer, arguing in favor of the GPx2-HIF1α axis as a therapeutic platform for targeting tumor cell metastasis.

Accurately identifying cell populations is paramount to the quality of downstream analyses and overall interpretations of single-cell RNA-seq (scRNA-seq) datasets but remains a challenge. The quality of single-cell clustering depends on the proximity metric used to generate cell-to-cell distances. Accordingly, proximity metrics have been benchmarked for scRNA-seq clustering, typically with results averaged across datasets to identify a highest performing metric. However, the ‘best-performing’ metric varies between studies, with the performance differing significantly between datasets. This suggests that the unique structural properties of a scRNA-seq dataset, specific to the biological system under study, has a substantial impact on proximity metric performance. Previous benchmarking studies have omitted to factor the structural properties into their evaluations. To address this gap, we developed a framework for the in-depth evaluation of the performance of 17 proximity metrics with respect to core structural properties of scRNA-seq data, including sparsity, dimensionality, cell population distribution and rarity. We find that clustering performance can be improved substantially by the selection of an appropriate proximity metric and neighbourhood size for the structural properties of a dataset, in addition to performing suitable pre-processing and dimensionality reduction. Furthermore, popular metrics such as Euclidean and Manhattan distance performed poorly in comparison to several lessor applied metrics, suggesting the default metric for many scRNA-seq methods should be re-evaluated. Our findings highlight the critical nature of tailoring scRNA-seq analyses pipelines to the system under study and provide practical guidance for researchers looking to optimise cell similarity search for the structural properties of their own data.

Identifying the molecular regulators that control ageing is challenging because the ageing process is influenced by a combination of genetic and environmental factors which makes it difficult to source the contribution of a single gene. Multiple studies have demonstrated that as humans age, increased gene expression heterogeneity results in the dysregulation of key regulators and pathways. Given the dynamic nature of gene expression, it is vital that this data be modelled by statistical approaches that can appropriately account for changes in variability to understand the contribution of heterogeneity during the aging process and properly identify its regulators. This study demonstrates the utility of using mixture models to model biological variability of gene expression occurring during ageing and how novel potential regulators of ageing can be identified.Our mixture modelling approach was applied to gene expression data from the Genotype-Tissue Expression (GTEx) cohort. For every gene, the expression profile was modelled using a mixture model across the cohort where the subset of donors corresponding to each mode was tested for a significant change in age group. The multi-tissue aspect of GTEx was leveraged to find ageing regulators based on this mixture model approach genes that were common across multiple tissues, suggesting that the regulation of ageing may also be controlled through a set of genes that have non-tissue-specific activity. Our approach identified the common ageing regulator mTOR, as significantly associated with a mixture model profile but this gene was not detected through standard differential expression analysis using edgeR. Genes identified by the standard approaches like edgeR and the mixture model-based approach were found to be enriched for similar biological pathways. This suggests that while the specific ageing regulators identified from our approach may be distinct, they generally belong in the same pathways as the genes that are identified by standard approaches. Overall, these results indicate that modelling gene expression variability using mixture models in conjunction with standard differential gene expression can help uncover new regulators that have a potential role for understanding human ageing.

BackgroundMetabolic reprograming, non-mutational epigenetic changes, increased cell plasticity and multidrug tolerance are early hallmarks of therapy resistance in cancer. In this temporary, therapy-tolerant state, cancer cells are highly sensitive to ferroptosis, a form of regulated cell death that is caused by oxidative stress through excess levels of iron-dependent peroxidation of polyunsaturated fatty acids (PUFA). However, mechanisms underpinning therapy-induced ferroptosis hypersensitivity remain to be elucidated.MethodsWe used quantitative single cell imaging of fluorescent metabolic probes, transcriptomics, proteomics and lipidomics to perform a longitudinal analysis of the adaptive response to androgen receptor-targeted therapies (androgen deprivation and enzalutamide) in prostate cancer (PCa).ResultsWe discovered that cessation of cell proliferation and a robust reduction in bioenergetic processes were associated with multidrug tolerance and a strong accumulation of lipids. The gain in lipid biomass was fueled by enhanced lipid uptake through cargo non-selective (macropinocytosis, tunneling nanotubes) and cargo-selective mechanisms (lipid transporters), whereasde novolipid synthesis was strongly reduced. Enzalutamide induced extensive lipid remodeling of all major phospholipid classes at the expense of storage lipids, leading to increased desaturation and acyl chain length of membrane lipids. The rise in membrane PUFA levels enhanced membrane fluidity and lipid peroxidation, causing hypersensitivity to glutathione peroxidase (GPX4) inhibition and ferroptosis. Combination treatments against AR and fatty acid desaturation, lipase activities or growth medium supplementation with antioxidants or PUFAs altered GPX4 dependence. Despite multidrug tolerance, PCa cells displayed an enhanced sensitivity to inhibition of lysosomal processing of exogenous lipids, highlighting an increased dependence on lipid uptake in the therapy-tolerant state.ConclusionsOur work provides mechanistic insight into processes of lipid metabolism that underpin the acquisition of therapy-induced GPX4 dependence and ferroptosis hypersensitivity to standard of care therapies in PCa. It demonstrated novel strategies to suppress the therapy-tolerant state that may have potential to delay and combat resistance to androgen receptor-targeted therapies, a currently unmet clinical challenge of advanced PCa. Since enhanced GPX4 dependence is an adaptive phenotype shared by several types of cancer in response to different therapies, our work might have universal implications for our understanding of metabolic events that underpin resistance to cancer therapies.

Cellular senescence is characterised by a state of permanent cell cycle arrest. It is accompanied by often variable release of the so-called senescence-associated secretory phenotype (SASP) factors, and occurs in response to a variety of triggers such as persistent DNA damage, telomere dysfunction, or oncogene activation. While cellular senescence is a recognised driver of organismal ageing, the extent of heterogeneity within and between different senescent cell populations remains largely unclear. Elucidating the drivers and extent of variability in cellular senescence states is important for discovering novel targeted seno-therapeutics and for overcoming cell expansion constraints in the cell therapy industry. Here we combine cell biological and single cell RNA-sequencing approaches to investigate heterogeneity of replicative senescence in human ESC-derived mesenchymal stem cells (esMSCs) as MSCs are the cell type of choice for the majority of current stem cell therapies and senescence of MSC is a recognized driver of organismal ageing. Our data identify three senescent subpopulations in the senescing esMSC population that differ in SASP, oncogene expression, and escape from senescence. Uncovering and defining this heterogeneity of senescence states in cultured human esMSCs allowed us to identify potential drug targets that may delay the emergence of senescent MSCsin vitroand perhapsin vivoin the future.

In search of redox mechanisms in breast cancer, we uncovered a striking role for glutathione peroxidase 2 (GPx2) in oncogenic signaling and patient survival. GPx2 loss stimulates malignant progression due to reactive oxygen species/hypoxia inducible factor-α (HIF1α)/VEGFA (vascular endothelial growth factor A) signaling, causing poor perfusion and hypoxia, which were reversed by GPx2 reexpression or HIF1α inhibition. Ingenuity Pathway Analysis revealed a link between GPx2 loss, tumor angiogenesis, metabolic modulation, and HIF1α signaling. Single-cell RNA analysis and bioenergetic profiling revealed that GPx2 loss stimulated the Warburg effect in most tumor cell subpopulations, except for one cluster, which was capable of oxidative phosphorylation and glycolysis, as confirmed by coexpression of phosphorylated-AMPK and GLUT1. These findings underscore a unique role for redox signaling by GPx2 dysregulation in breast cancer, underlying tumor heterogeneity, leading to metabolic plasticity and malignant progression.

De novo lipogenesis is a well-described androgen receptor (AR)–regulated metabolic pathway that supports prostate cancer tumor growth by providing fuel, membrane material, and steroid hormone precursor. In contrast, our current understanding of lipid supply from uptake of exogenous lipids and its regulation by AR is limited, and exogenous lipids may play a much more significant role in prostate cancer and disease progression than previously thought. By applying advanced automated quantitative fluorescence microscopy, we provide the most comprehensive functional analysis of lipid uptake in cancer cells to date and demonstrate that treatment of AR-positive prostate cancer cell lines with androgens results in significantly increased cellular uptake of fatty acids, cholesterol, and low-density lipoprotein particles. Consistent with a direct, regulatory role of AR in this process, androgen-enhanced lipid uptake can be blocked by the AR-antagonist enzalutamide, but is independent of proliferation and cell-cycle progression. This work for the first time comprehensively delineates the lipid transporter landscape in prostate cancer cell lines and patient samples by analysis of transcriptomics and proteomics data, including the plasma membrane proteome. We show that androgen exposure or deprivation regulates the expression of multiple lipid transporters in prostate cancer cell lines and tumor xenografts and that mRNA and protein expression of lipid transporters is enhanced in bone metastatic disease when compared with primary, localized prostate cancer. Our findings provide a strong rationale to investigate lipid uptake as a therapeutic cotarget in the fight against advanced prostate cancer in combination with inhibitors of lipogenesis to delay disease progression and metastasis. Implications: Prostate cancer exhibits metabolic plasticity in acquiring lipids from uptake and lipogenesis at different disease stages, indicating potential therapeutic benefit by cotargeting lipid supply.

The processes regulating tumor metastasis are multivariate and complex. Redox regulation of the tumor phenotype by GPx2 knockdown (KD) in breast cancer led us to uncover dramatic effects on spontaneous metastasis. Analysis of single cell RNA sequencing (scRNAseq) data from GPx2 KD tumor and control tumor, revealed that both tumors were comprised of several luminal-like tumor cell clusters and one mesenchymal-like cell cluster (cluster 3). Notably, GPx2 KD promoted a significant increase in the size of mesenchymal cells (cluster 3) relative to control, which might be due to the stimulation of epithelial-to-mesenchymal transition (EMT) in response to GPx2 loss. In support of this view, GPx2 KD stimulated an increase in mRNA expression of basal/mesenchymal (KRT5, KRT14, KRT17, Vimentin, Twist1, Twist2, CDH2) genes and a decrease in mRNA expression of epithelial/luminal (Cldn7 and Epcam) genes, especially in cluster 3. Moreover, GPx2 KD upregulated mRNA expression of basal/mesenchymal (Twist2, CDH2, and KRT14) genes in most luminal-like clusters expressing epithelial/luminal (Epcam, Cldn3/7, CDH1, KRT8/18) genes, implying these clusters may be undergoing EMT transition in a hybrid epithelial/mesenchymal state in response to GPx2 loss. Validation of these data in cell lines and tumors showed that GPx2 KD dramatically enhanced EMT via activation of ROS/HIF1α-mediated signaling. Importantly, these effects were reversed by GPx2 re-expression or HIF1α inhibition, which was capable of suppressing EMT and metastasis. Collectively, these results indicate that GPx2 loss promotes breast cancer metastasis by stimulating EMT due to HIF1α signaling, highlighting the impact of GPx2 and HIF1 on therapeutic intervention in metastasis. Citation Format: Zuen Ren, Huizhi Liang, Malindrie Dharmaratne, Miriam Morales-Gallego, Atefeh Taherian Fard, Jessica Mar, Kimita Suyama, Outhiriaradjou Benard, Michael B. Prystowsky, Larry Norton, Rachel B. Hazan. Loss of glutathione peroxidase 2 promotes epithelial to mesenchymal transition and breast cancer metastasis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 968.

Hyperleptinaemia is a well-established therapeutic side effect of drugs inhibiting the androgen axis in prostate cancer (PCa), including main stay androgen deprivation therapy (ADT) and androgen targeted therapies (ATT). Given significant crossover between the adipokine hormone signalling of leptin and multiple cancer-promoting hallmark pathways, including growth, proliferation, migration, angiogenesis, metabolism and inflammation, targeting the leptin axis is therapeutically appealing, especially in advanced PCa where current therapies fail to be curative. In this study, we uncover leptin as a novel universal target in PCa and are the first to highlight increased intratumoural leptin and leptin receptor (LEPR) expression in PCa cells and patients' tumours exposed to androgen deprivation, as is observed in patients' tumours of metastatic and castrate resistant (CRPC) PCa. We also reveal the world-first preclinical evidence that demonstrates marked efficacy of targeted leptin-signalling blockade, using Allo-aca, a potent, specific, and safe LEPR peptide antagonist. Allo-aca-suppressed tumour growth and delayed progression to CRPC in mice bearing LNCaP xenografts, with reduced tumour vascularity and altered pathways of apoptosis, transcription/translation, and energetics in tumours determined as potential mechanisms underpinning anti-tumour efficacy. We highlight LEPR blockade in combination with androgen axis inhibition represents a promising new therapeutic strategy vital in advanced PCa treatment.

Background Basal cell carcinoma (BCC) of the skin is the most common form of human cancer, with more than 90% of tumours presenting with clear genetic activation of the Hedgehog pathway. However, polygenic risk factors affecting mechanisms such as DNA repair and cell cycle checkpoints or which modulate the tumour microenvironment or host immune system play significant roles in determining whether genetic mutations culminate in BCC development. We set out to define background genetic factors that play a role in influencing BCC susceptibility via promoting or suppressing the effects of oncogenic drivers of BCC. Methods We performed genome-wide association studies (GWAS) on 17,416 cases and 375,455 controls. We subsequently performed statistical analysis by integrating data from population-based genetic studies of multi-omics data, including blood- and skin-specific expression quantitative trait loci and methylation quantitative trait loci, thereby defining a list of functionally relevant candidate BCC susceptibility genes from our GWAS loci. We also constructed a local GWAS functional interaction network (consisting of GWAS nearest genes) and another functional interaction network, consisting specifically of candidate BCC susceptibility genes. Results A total of 71 GWAS loci and 46 functional candidate BCC susceptibility genes were identified. Increased risk of BCC was associated with the decreased expression of 26 susceptibility genes and increased expression of 20 susceptibility genes. Pathway analysis of the functional candidate gene regulatory network revealed strong enrichment for cell cycle, cell death, and immune regulation processes, with a global enrichment of genes and proteins linked to TReg cell biology. Conclusions Our genome-wide association analyses and functional interaction network analysis reveal an enrichment of risk variants that function in an immunosuppressive regulatory network, likely hindering cancer immune surveillance and effective antitumour immunity.

Additional file 1: Supplementary Figures. Figure S1. Gender and age demographics/distribution of UK Biobank derived BCC cases and controls. Figure S2. Regional plots of MC1R. Figure S3. FI network for the protein-coding ‘nearest-genes’ identified by GWAS analysis.

Additional file 3: Supplementary Table S1.

Additional file 1. Supplementary Info.

Additional file 2. Supplementary Figures.

Additional file 7: Supplementary Table S5.

The propensity of cancer cells to transition between epithelial and mesenchymal phenotypic states via the epithelial–mesenchymal transition (EMT) program can regulate metastatic processes, cancer progression, and treatment resistance. Transcriptional investigations using reversible models of EMT, revealed the mesenchymal-to-epithelial reverting transition (MErT) to be enriched in clinical samples of metastatic castrate resistant prostate cancer (mCRPC). From this enrichment, a metastasis-derived gene signature was identified that predicted more rapid cancer relapse and reduced survival across multiple human carcinoma types. Additionally, the transcriptional profile of MErT is not a simple mirror image of EMT as tumour cells retain a transcriptional “memory” following a reversible EMT. This memory was also enriched in mCRPC samples. Cumulatively, our studies reveal the transcriptional profile of epithelial–mesenchymal plasticity and highlight the unique transcriptional properties of MErT. Furthermore, our findings provide evidence to support the association of epithelial plasticity with poor clinical outcomes in multiple human carcinoma types.

C.H. Waddington introduced the epigenetic landscape as a metaphor to represent cellular decision-making during development. Like a population of balls rolling down a rough hillside, developing cells follow specific trajectories (valleys) and eventually come to rest in one or another low-energy state that represents a mature cell type. Waddington depicted the topography of this landscape as determined by interactions among gene products, thereby connecting genotype to phenotype. In modern terms, each point on the landscape represents a state of the underlying genetic regulatory network, which in turn is described by a gene expression profile. In this chapter we demonstrate how the mathematical formalism of Hopfield networks can be used to model this epigenetic landscape. Hopfield networks are auto-associative artificial neural networks; input patterns are stored as attractors of the network and can be recalled from noisy or incomplete inputs. The resulting models capture the temporal dynamics of a gene regulatory network, yielding quantitative insight into cellular development and phenotype.

C.H. Waddington introduced the epigenetic landscape as a metaphor to represent cellular decision-making during development. Like a population of balls rolling down a rough hillside, developing cells follow specific trajectories (valleys) and eventually come to rest in one or another low-energy state that represents a mature cell type. Waddington depicted the topography of this landscape as determined by interactions among gene products, thereby connecting genotype to phenotype. In modern terms, each point on the landscape represents a state of the underlying genetic regulatory network, which in turn is described by a gene expression profile. In this chapter we demonstrate how the mathematical formalism of Hopfield networks can be used to model this epigenetic landscape. Hopfield networks are auto-associative artificial neural networks; input patterns are stored as attractors of the network and can be recalled from noisy or incomplete inputs. The resulting models capture the temporal dynamics of a gene regulatory network, yielding quantitative insight into cellular development and phenotype.

Tumor heterogeneity is known to drive breast cancer progression and metastasis. Current therapies target mostly inter-tumor heterogeneity (different patients) without taking into consideration intratumor heterogeneity (individual tumor). Hypoxia is a hallmark of aggressive tumors that is known to select for intratumor cells that are prone for metastasis. Here we show that knockdown of a ROS scavenging enzyme, Glutathione Peroxidase 2, in the PyMT mammary tumor model results in highly hypoxic and metastatic tumors. Single cell RNA-seq on both GPx2 knockdown and control tumors revealed a high degree of intratumor heterogeneity resulting in several clusters that were either of luminal/epithelial or basal-like/mesenchymal type shared by both tumors. Interestingly, GPx2 loss caused increases in a mesenchymal-like/quiescent and a highly proliferative luminal population, implying that GPx2 loss might drive a cancer stem-like population that is conducive of metastasis. Metascape pathway analysis of differentially expressed genes identified pathways that were significantly overrepresented in GPx2 knockdown tumors involving mitochondrial oxidative phosphorylation and response to hypoxia, consistent with the high levels of reactive oxygen species in these tumors. Finally, differential gene expression analysis on mesenchymal and luminal subpopulations predicts novel candidate gene signatures stimulated by GPx2 loss, that might be used to identify or target metastatic cells in tumors. In sum, our data imply that oxidative stress conveys a metastatic phenotype via effects on tumor cell heterogeneity. Citation Format: Zuen Ren, Malindrie Dharmaratne, Ameya Kulkarni, Atefeh Taherian Fard, Jessica Mar, Phillip M. Galbo, Yang Liu, Deyou Zheng, Huizhi Liang, Outhiriaradjou Benard, Kimita Suyama, Michael B. Prystowsky, Larry Norton, Rachel B. Hazan. Redox regulation of tumor cell heterogeneity in the PyMT mammary tumor model revealed by single cell RNA sequencing [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 4426.

Genome-wide regulatory networks enable cells to function, develop, and survive. Perturbation of these networks can lead to appearance of a disease phenotype. Inspired by Conrad Waddington's epigenetic landscape of cell development, we use a Hopfield network formalism to construct an attractor landscape model of disease progression based on protein- or gene-correlation networks of Parkinson's disease, glioma, and colorectal cancer. Attractors in this landscape correspond to normal and disease states of the cell. We introduce approaches to estimate the size and robustness of these attractors, and take a network-based approach to study their biological features such as the key genes and their functions associated with the attractors. Our results show that the attractor of cancer cells is wider than the attractor of normal cells, suggesting a heterogeneous nature of cancer. Perturbation analysis shows that robustness depends on characteristics of the input data (number of samples per time-point, and the fraction which converge to an attractor). We identify unique gene interactions at each stage, which reflect the temporal rewiring of the gene regulatory network (GRN) with disease progression. Our model of the attractor landscape, constructed from large-scale gene expression profiles of individual patients, captures snapshots of disease progression and identifies gene interactions specific to different stages, opening the way for development of stage-specific therapeutic strategies.

Accurate subtyping or classification of breast cancer is important for ensuring proper treatment of patients and also for understanding the molecular mechanisms driving this disease. While there have been several gene signatures proposed in the literature to classify breast tumours, these signatures show very low overlaps, different classification performance, and not much relevance to the underlying biology of these tumours. Here we evaluate DNA-damage response (DDR) and cell cycle pathways, which are critical pathways implicated in a considerable proportion of breast tumours, for their usefulness and ability in breast tumour subtyping. We think that subtyping breast tumours based on these two pathways could lead to vital insights into molecular mechanisms driving these tumours. Here, we performed a systematic evaluation of DDR and cell-cycle pathways for subtyping of breast tumours into the five known intrinsic subtypes. Homologous Recombination (HR) pathway showed the best performance in subtyping breast tumours, indicating that HR genes are strongly involved in all breast tumours. Comparisons of pathway based signatures and two standard gene signatures supported the use of known pathways for breast tumour subtyping. Further, the evaluation of these standard gene signatures showed that breast tumour subtyping, prognosis and survival estimation are all closely related. Finally, we constructed an all-inclusive super-signature by combining (union of) all genes and performing a stringent feature selection, and found it to be reasonably accurate and robust in classification as well as prognostic value. Adopting DDR and cell cycle pathways for breast tumour subtyping achieved robust and accurate breast tumour subtyping, and constructing a super-signature which contains feature selected mix of genes from these molecular pathways as well as clinical aspects is valuable in clinical practice.

The propensity of cancer cells to transition between epithelial and mesenchymal phenotypic states via the epithelial–mesenchymal transition (EMT) program can regulate metastatic processes, cancer progression, and treatment resistance. Transcriptional investigations using reversible models of EMT, revealed the mesenchymal-to-epithelial reverting transition (MErT) to be enriched in clinical samples of metastatic castrate resistant prostate cancer (mCRPC). From this enrichment, a metastasis-derived gene signature was identified that predicted more rapid cancer relapse and reduced survival across multiple human carcinoma types. Additionally, the transcriptional profile of MErT is not a simple mirror image of EMT as tumour cells retain a transcriptional “memory” following a reversible EMT. This memory was also enriched in mCRPC samples. Cumulatively, our studies reveal the transcriptional profile of epithelial–mesenchymal plasticity and highlight the unique transcriptional properties of MErT. Furthermore, our findings provide evidence to support the association of epithelial plasticity with poor clinical outcomes in multiple human carcinoma types.

The epigenetic landscape was introduced by Conrad Waddington as a metaphor of cellular development. Like a ball rolling down a hillside is channelled through a succession of valleys until it reaches the bottom, cells follow specific trajectories from a pluripotent state to a committed state. Transcription factors (TFs) interacting as a network (the gene regulatory network (GRN)) orchestrate this developmental process within each cell. Here, we quantitatively model the epigenetic landscape using a kind of artificial neural network called the Hopfield network (HN). An HN is composed of nodes (genes/TFs) and weighted undirected edges, resulting in a weight matrix (W) that stores interactions among the nodes over the entire network. We used gene co-expression to compute the edge weights. Through W, we then associate an energy score (E) to each input pattern (pattern of co-expression for a specific developmental stage) such that each pattern has a specific E. We propose that, based on the co-expression values stored in W, HN associates lower E values to stable phenotypic states and higher E to transient states. We validate our model using time course gene-expression data sets representing stages of development across 12 biological processes including differentiation of human embryonic stem cells into specialized cells, differentiation of THP1 monocytes to macrophages during immune response and trans-differentiation of epithelial to mesenchymal cells in cancer. We observe that transient states have higher energy than the stable phenotypic states, yielding an arc-shaped trajectory. This relationship was confirmed by perturbation analysis. HNs offer an attractive framework for quantitative modelling of cell differentiation (as a landscape) from empirical data. Using HNs, we identify genes and TFs that drive cell-fate transitions, and gain insight into the global dynamics of GRNs.

We conducted an integrated study of cell growth parameters, product formation, and the dynamics of intracellular metabolite concentrations using Escherichia coli with genes knocked out in the glycolytic and oxidative pentose phosphate pathway (PPP) for glucose catabolism. We investigated the same characteristics in the wild-type strain, using acetate or pyruvate as the sole carbon source. Dramatic effects on growth parameters and extracellular and intracellular metabolite concentrations were observed after blocking either glycolytic breakdown of glucose by inactivation of phosphoglucose isomerase (disruption of pgi gene) or pentose phosphate breakdown of glucose by inactivation of glucose-6-phosphate dehydrogenase (disruption of zwf gene). Reducing power (NADPH) was mainly produced through PPP when the pgi gene was knocked out, while NADPH was produced through the tricarboxylic acid (TCA) cycle by isocitrate dehydrogenase or NADP-linked malic enzyme when the zwf gene was knocked out. As expected, when the pgi gene was knocked out, intracellular concentrations of PPP metabolites were high and glycolytic and concentrations of TCA cycle pathway metabolites were low. In the zwf gene knockout, concentrations of PPP metabolites were low and concentrations of intracellular glycolytic and TCA cycle metabolites were high.

Different types of organophosphorous compounds constitute most potent pesticides. These chemicals attack the nervous system of living organisms causing death. Different organisms produce enzymes to degrade these chemicals. These enzymes are present in simple microorganisms from archaea, bacteria to complex eukaryotes like humans. A comparison of representative eight shortlisted enzymes involved in the degradation and inactivation of organophosphates from a wide range of organisms was performed to infer the basis of their common functionality. There is little sequence homology in these enzymes which results in divergent tertiary structures. The only feature that these enzymes seem to share is their amino acid composition. However, structural analysis has shown no significant similarities among this functionally similar group of organophosphate degrading enzymes.

Malarial parasite has long been a subject of research for a large community of scientists and has yet to be conquered. One of the main obstacles to effectively control this disease is rapidly evolving genetic structure of Plasmodium parasite itself. In this study, we focused on chromosome 4 of the Plasmodium falciparum and Plasmodium vivax species and carried out comparative studies of genes that are responsible for antigenic variation in respective species. Comparative analysis of genes responsible for antigenic variation (var and vir genes in P. falciparum and P. vivax, respectively) showed significant difference in their respective nucleotide sequence lengths as well as amino acid composition. The possible association of exon's length on pathogenecity of respective Plasmodium species was also investigated, and analysis of gene structure showed that on the whole, exon lengths in P. falciparum are larger compared to P. vivax. Analysis of tandem repeats across the genome has shown that the size of repetitive sequences has a direct effect on chromosomes length, which can also be a potential reason for P. falciparum's greater variability and hence pathogenecity than P. vivax.