Optimal alpha reduces error rates in gene expression studies: a meta-analysis approach

Chronic arsenic exposure induces malignant transformation of human HaCaT cells through both deterministic and stochastic changes in transcriptome expression

Biological processes are inherently stochastic, i.e., are partially driven by hard to predict random probabilistic processes. Carcinogenesis is driven both by stochastic and deterministic (predictable non-random) changes. However, very few studies systematically examine the contribution of stochastic events leading to cancer development. In differential gene expression studies, the established data analysis paradigms incentivize expression changes that are uniformly different across the experimental versus control groups, introducing preferential inclusion of deterministic changes at the expense of stochastic processes that might also play a crucial role in the process of carcinogenesis. In this study, we applied simple computational techniques to quantify: (i) The impact of chronic arsenic (iAs) exposure as well as passaging time on stochastic gene expression and (ii) Which genes were expressed deterministically and which were expressed stochastically at each of the three stages of cancer development. Using biological coefficient of variation as an empirical measure of stochasticity we demonstrate that chronic iAs exposure consistently suppressed passaging related stochastic gene expression at multiple time points tested, selecting for a homogenous cell population that undergo transformation. Employing multiple balanced removal of outlier data, we show that chronic iAs exposure induced deterministic and stochastic changes in the expression of unique set of genes, that populate largely unique biological pathways. Together, our data unequivocally demonstrate that both deterministic and stochastic changes in transcriptome-wide expression are critical in driving biological processes, pathways and networks towards clonal selection, carcinogenesis, and tumor heterogeneity.

Mbnl2 loss alters novel context processing and impairs object recognition memory

Patients with myotonic dystrophy type I (DM1) demonstrate visuospatial dysfunction and impaired performance in tasks requiring recognition or memory of figures and objects. In DM1, CUG expansion RNAs inactivate the muscleblind-like (MBNL) proteins. We show that constitutive Mbnl2 inactivation in Mbnl2ΔE2/ΔE2 mice selectively impairs object recognition memory in the novel object recognition test. When exploring the context of a novel arena in which the objects are later encountered, the Mbnl2ΔE2/ΔE2 dorsal hippocampus responds with a lack of enrichment for learning and memory-related pathways, mounting instead transcriptome alterations predicted to impair growth and neuron viability. In Mbnl2ΔE2/ΔE2 mice, saturation effects may prevent deployment of a functionally relevant transcriptome response during novel context exploration. Post-novel context exploration alterations in genes implicated in tauopathy and dementia are observed in the Mbnl2ΔE2/ΔE2 dorsal hippocampus. Thus, MBNL2 inactivation in patients with DM1 may alter novel context processing in the dorsal hippocampus and impair object recognition memory.

Within- and cross-tissue gene regulations were disrupted by PM2.5 nitrate exposure and associated with respiratory functions

BACKGROUND

Pathogenesis of complex diseases often involves multiple organs/tissue-types. To date, the PM_2.5 exposure's toxic effects and induced disease risks were not studied at multi-tissue level.

METHODS

C57BL/6 mice (n = 40) were exposed to PM_2.5 NO₃^- and clean air, respectively, and afterwards assessed respiratory functions and transcriptome in relevant tissues: blood and lung. We constructed within- and cross-tissue gene regulation networks and identified network modules associated with exposure and respiratory functions.

RESULTS

PM_2.5 NO₃^- exposure elevated naïve B cells proportion in blood (p = 0.0028). Among the 6000 highest expressed genes in blood, 18.8 % (1133 genes) were altered by exposure at p ≤ 0.05 level, among which 763 genes were also associated with respiratory function (enrichment folds = 7.63, p = 2.7E-189). The exposure disrupted blood genes were primarily in the immunoregulation pathways. Both within- and cross-tissue gene network modules were perturbed by exposure and associated with respiratory function. An immunodeficiency related cross-tissue module of 555 genes was affected by exposure (p = 0.0023) and strongly correlated with FEV_0.05/FVC (r = 0.61 and p = 3E-5).

CONCLUSIONS

This study aims to fill in a major knowledge gap and investigated the effect of PM_2.5 exposure simultaneously in multiple tissues. We provided novel evidence that PM_2.5 NO₃^- exposure profoundly perturbed within- and cross-tissue gene regulations, and highlighted their roles in the etiology of respiratory decline.

How Will the Occupation End? A Real-World Exploration of Hindsight Bias and Retroactive Pessimism Concerning Antigovernment Protests in Bulgaria

We explored hindsight bias and retroactive pessimism related to the occupation of Sofia University in 2013–2014, using a memory design. We found partial evidence for hindsight bias for predictions of whether the occupation would achieve its goal (the government's resignation) but not for specific events related to the occupation, which were deemed unlikely from the start and later perceived not to have occurred. We did not find evidence for retroactive pessimism: Hindsight bias indices for the prediction of the government's resignation were not reliably associated with support for the occupation and disappointment with its outcome. We propose using a recall/reconstruct measure as a more rigorous test for retroactive pessimism, which has so far been demonstrated through reassessment.

Temporal changes in the corpus luteum during early pregnancy reveal regulation of pathways that enhance steroidogenesis and suppress luteolytic mechanisms†

Although rescue of the corpus luteum (CL) is required for pregnancy, luteal function during maternal recognition of pregnancy remains largely unexplored. CL were collected from pregnant cattle on days 14, 17, 20, and 23, to encompass the maternal recognition of pregnancy period. Next-generation sequencing was used to profile mRNA abundance during this time, while tandem mass spectrometry and nanostring technology were used to profile proteins and miRNA, respectively. A total of 1157 mRNA were differentially abundant, while 27 miRNA changed, and 29 proteins tended to change. mRNA that increased were regulators of interferon signaling and DNA repair, while those that decreased were associated with luteolytic processes, such as calcium signaling and matrix metallopeptidase (MMP) signaling, indicating inhibition of these processes. One of these, MMP12, was regulated by prostaglandin F2A in vitro. mRNA that were maximally abundant on day 20 were primarily associated with immune processes. Two of these, C-C motif chemokine ligand 1 and NFKB inhibitor alpha, were regulated by interferon tau in vitro. MiRNA that increased were predicted to inhibit phosphatidylinositol signaling, while those that decreased may be negative regulators of steroidogenesis. One protein that was greater on day 20 than on day 14 was aldehyde dehydrogenase 1 family member A1 (ALDH1A1), which synthesizes retinoic acid. Pharmacological inhibition of this enzyme, or of retinoic acid receptor signaling, led to suppression of progesterone production in vitro. Overall, these data indicate that there are changes in the CL of pregnancy that are important for continued luteal function.

Perspectives on transcriptomics in animal physiology studies

Reductionist approaches in physiology and biochemistry are essential for understanding how animals cope and adapt to their environments. Transcriptomics is no longer restricted to a select few, and accessibility and affordability continue to facilitate its rapid growth as a science. More than 6000 publications (a conservative estimate) over the past decade quantify the response of the transcriptome to a wide breadth of questions in animal physiology. Transcriptomes have been quantified under conditions of hypoxia, climate change, salinity, drought, environmental pollution, and ultraviolet radiation among others; these studies have greatly improved understanding of the molecular machinery required for organismal adaptation. These "snapshots in time" however are never complete as the transcriptome is exquisitely sensitive to an individual's current physiologic state. Animal physiologists new to the field must recognize limitations of transcriptome technologies and consider experimental designs that strengthen physiologic interpretation. Current estimates suggest that a sample size of 6 or more are required for RNA-seq experiments in order to capture the majority of differentially expressed genes confidently. "Outside-the-box" approaches for statistical analyses of data derived from RNA-seq should be explored, as studies continue to point out that high false discoveries rates are pervasive with RNA-seq studies, reminiscent of the early days of microarrays. Incorporating biological variability, rather than reducing it (i.e. pooling strategies), into experimental designs is essential. Moreover, real-time PCR must not be viewed as a "validation step" to justify low samples sizes, but rather an orthogonal method to strengthen biological interpretation. The use of proper experimental controls in transcriptomics studies (i.e. spike-in controls and technical replication) are recommended and there is a pressing need for inter-laboratory tests (round robin experiments) to quantify repeatability and to identify sources of transcriptome variation within the context of animal physiology. Testing the reproducibility of transcriptome experiments in light of physiology in non-model organisms would be a significant contribution to the community. Single cell transcriptomics and multiplexing barcoding strategies such as decode-seq are poised to further advance the reductionist view of animal physiology; researchers are encouraged to consult literature herein and elsewhere for guidance on best practices and limitations of transcriptome technologies when studying the physiology of animals.

Molecular profiling demonstrates modulation of immune cell function and matrix remodeling during luteal rescue†

The corpus luteum (CL) is essential for maintenance of pregnancy in all mammals and luteal rescue, which occurs around day 16-19 in the cow, is necessary to maintain luteal progesterone production. Transcriptomic and proteomic profiling were performed to compare the day 17 bovine CL of the estrous cycle and pregnancy. Among mRNA and proteins measured, 140 differentially abundant mRNA and 24 differentially abundant proteins were identified. Pathway analysis was performed using four programs. Modulated pathways included T cell receptor signaling, vascular stability, cytokine signaling, and extracellular matrix remodeling. Two mRNA that were less in pregnancy were regulated by prostaglandin F2A in culture, while two mRNA that were greater in pregnancy were regulated by interferon tau. To identify mRNA that could be critical regulators of luteal fate, the mRNA that were differentially abundant during early pregnancy were compared to mRNA that were differentially abundant during luteal regression. Eight mRNA were common to both datasets, including mRNA related to regulation of steroidogenesis and gene transcription. A subset of differentially abundant mRNA and proteins, including those associated with extracellular matrix functions, were predicted targets of differentially abundant microRNA (miRNA). Integration of miRNA and protein data, using miRPath, revealed pathways such as extracellular matrix-receptor interactions, abundance of glutathione, and cellular metabolism and energy balance. Overall, this study has provided a comprehensive profile of molecular changes in the corpus luteum during maternal recognition of pregnancy and has indicated that some of these functions may be miRNA-regulated.

Molecular tumor analysis and liquid biopsy: a feasibility investigation analyzing circulating tumor DNA in patients with central nervous system lymphomas

Background

Central nervous system lymphomas (CNSL) is a devastating disease. Currently, a confirmatory biopsy is required prior to treatment.

Objective

Our investigation aims to prove the feasibility of a minimally-invasive diagnostic approach for the molecular characterization of CNSL.

Methods

Tissue biopsies from 6 patients with suspected CNSL were analyzed using a 649gene next-generation sequencing (NGS) tumor panel (tumor vs. reference tissue (EDTA-blood)). The individual somatic mutation pattern was used as a basis for the digital PCR analyzing circulating tumor DNA (ctDNA) from plasma and cerebrospinal fluid (CSF) samples, identifying one selected tumor mutation during this first step of the feasibility investigation.

Results

NGS-analysis of biopsy tissue revealed a specific somatic mutation pattern in all confirmed lymphoma samples (n = 5, NGS-sensitivity 100%) and none in the sample identified as normal brain tissue (NGS-specificity 100%). cfDNA-extraction was dependent on the extraction-kit used and feasible in 3 samples, in all of which somatic mutations were detectable (100%). Analysis of CSF-derived cfDNA was superior to plasma-derived cfDNA and routine microscopic analysis (lymphoma cells: n = 2, 40%). One patient showed a divergent molecular pattern, typical of Burkitt-Lymphoma (HIV+, serologic evidence of EBV-infection). Lumbar puncture was tolerated without complications, whereas biopsy caused 3 hemorrhages.

Conclusions

Our investigation provides evidence that analysis of cfDNA in central nervous system tumors is feasible using the described protocol. Molecular characterization of CNSL could be achieved by analysis of CSF-derived cfDNA. Knowledge of a tumor’s specific mutation pattern may allow initiation of targeted therapies, treatment surveillance and could lead to minimally-invasive diagnostics in the future.

Electronic supplementary material

The online version of this article (10.1186/s12885-019-5394-x) contains supplementary material, which is available to authorized users.

The quest for an optimal alpha

Researchers who analyze data within the framework of null hypothesis significance testing must choose a critical "alpha" level, α, to use as a cutoff for deciding whether a given set of data demonstrates the presence of a particular effect. In most fields, α = 0.05 has traditionally been used as the standard cutoff. Many researchers have recently argued for a change to a more stringent evidence cutoff such as α = 0.01, 0.005, or 0.001, noting that this change would tend to reduce the rate of false positives, which are of growing concern in many research areas. Other researchers oppose this proposed change, however, because it would correspondingly tend to increase the rate of false negatives. We show how a simple statistical model can be used to explore the quantitative tradeoff between reducing false positives and increasing false negatives. In particular, the model shows how the optimal α level depends on numerous characteristics of the research area, and it reveals that although α = 0.05 would indeed be approximately the optimal value in some realistic situations, the optimal α could actually be substantially larger or smaller in other situations. The importance of the model lies in making it clear what characteristics of the research area have to be specified to make a principled argument for using one α level rather than another, and the model thereby provides a blueprint for researchers seeking to justify a particular α level.

Expression analysis of RNA sequencing data from human neural and glial cell lines depends on technical replication and normalization methods

Background

The potential for astrocyte participation in central nervous system recovery is highlighted by in vitro experiments demonstrating their capacity to transdifferentiate into neurons. Understanding astrocyte plasticity could be advanced by comparing astrocytes with stem cells. RNA sequencing (RNA-seq) is ideal for comparing differences across cell types. However, this novel multi-stage process has the potential to introduce unwanted technical variation at several points in the experimental workflow. Quantitative understanding of the contribution of experimental parameters to technical variation would facilitate the design of robust RNA-Seq experiments.

Results

RNA-Seq was used to achieve biological and technical objectives. The biological aspect compared gene expression between normal human fetal-derived astrocytes and human neural stem cells cultured in identical conditions. When differential expression threshold criteria of |log₂fold change| > 2 were applied to the data, no significant differences were observed. The technical component quantified variation arising from particular steps in the research pathway, and compared the ability of different normalization methods to reduce unwanted variance. To facilitate this objective, a liberal false discovery rate of 10% and a |log₂fold change| > 0.5 were implemented for the differential expression threshold. Data were normalized with RPKM, TMM, and UQS methods using JMP Genomics. The contributions of key replicable experimental parameters (cell lot; library preparation; flow cell) to variance in the data were evaluated using principal variance component analysis. Our analysis showed that, although the variance for every parameter is strongly influenced by the normalization method, the largest contributor to technical variance was library preparation. The ability to detect differentially expressed genes was also affected by normalization; differences were only detected in non-normalized and TMM-normalized data.

Conclusions

The similarity in gene expression between astrocytes and neural stem cells supports the potential for astrocytic transdifferentiation into neurons, and emphasizes the need to evaluate the therapeutic potential of astrocytes for central nervous system damage. The choice of normalization method influences the contributions to experimental variance as well as the outcomes of differential expression analysis. However irrespective of normalization method, our findings illustrate that library preparation contributed the largest component of technical variance.

Electronic supplementary material

The online version of this article (10.1186/s12859-018-2382-0) contains supplementary material, which is available to authorized users.

Hypomethylation of NANOG promoter in colonic mucosal cells of obese patients: a possible role of NF-κB

Obesity and particularly central obesity are the main risk factors of colon cancer. All intestinal cell populations including stem cells, their progenitors and differentiated colonocytes seem to be the origin of colorectal cancer. However, recent data support the role of differentiated cells as cancer origin especially during inflammation. Based on Yamanaka's seminal work, re-expression of few transcription factors in terminally differentiated cells creates stemness properti'es. Although these transcription factors are involved in tumorigenesis, they are epigenetically repressed in adult tissues. We proposed that obesity might regulate methylation of stemness genes in colonocytes via inflammatory signalling. Obesity-associated inflammation was analysed using Western blot analysis of phospho-IκB (inhibitor of NF-κB). Methylation-sensitive high-resolution melting analysis was performed on colonic mucosal samples of twenty obese and twenty normal-weight men to analyse promoter methylation of POU5F1 (OCT4), NANOG, MYC and CDKN2A. TNF-treated HT-29 cells were used to recapitulate the effect of NF-κB activation on stemness genes methylation. Our results showed that colonic phosphorylation of IκB, as a signal of NF-κB activation, was higher in obese subjects compared with their normal-weight counterparts. Moreover, promoter methylation of NANOG was likely to be lower in obese subjects and correlated with central obesity. HT-29 cells incubated by TNF-α showed hypomethylation of POU5F1 and MYC genes in addition to the NANOG. These results suggest that obesity-induced inflammation might be involved in the regulation of DNA methylation of oncogenic genes such as NANOG in differentiated colonocytes and thus predispose them to later oncogenic alterations.

Deciphering the Combined Effects of Environmental Stressors on Gene Transcription: A Conceptual Approach

The use of classical mixture toxicity models to predict the combined effects of environmental stressors based on toxicogenomics (OMICS) data is still in its infancy. Although several studies have made attempts to implement mixture modeling in OMICS analysis to understand the low-dose interactions of stressors, it is not clear how interactions occur at the molecular level and how results generated from such approaches can be better used to inform future studies and cumulative hazard assessment of multiple stressors. The present work was therefore conducted to propose a conceptual approach for combined effect assessment using global gene expression data, as illustrated by a case study on assessment of combined effects of gamma radiation and depleted uranium (DU) on Atlantic salmon ( Salmo salar). Implementation of the independent action (IA) model in reanalysis of a previously published microarray gene expression dataset was performed to describe gene expression patterns of combined effects and identify key gene sets and pathways that were relevant for understanding the interactive effects of these stressors. By using this approach, 3120 differentially expressed genes (DEGs) were found to display additive effects, whereas 279 (273 synergistic, 6 antagonistic) were found to deviate from additivity. Functional analysis further revealed that multiple toxicity pathways, such as oxidative stress responses, cell cycle regulation, lipid metabolism, and immune responses were enriched by DEGs showing synergistic gene expression. A key toxicity pathway of DNA damage leading to enhanced tumorigenesis signaling is highlighted and discussed in detail as an example of how to take advantage of the approach. Furthermore, a conceptual workflow describing the integration of combined effect modeling, OMICS analysis, and bioinformatics is proposed. The present study presents a conceptual framework for utilizing OMICS data in combined effect assessment and may provide novel strategies for dealing with data analysis and interpretation of molecular responses of multiple stressors.

Are we closer to the vision? A proposed framework for incorporating omics into environmental assessments

Environmental science has benefited a great deal from omics-based technologies. High-throughput toxicology has defined adverse outcome pathways (AOPs), prioritized chemicals of concern, and identified novel actions of environmental chemicals. While many of these approaches are conducted under rigorous laboratory conditions, a significant challenge has been the interpretation of omics data in "real-world" exposure scenarios. Clarity in the interpretation of these data limits their use in environmental monitoring programs. In recent years, one overarching objective of many has been to address fundamental questions concerning experimental design and the robustness of data collected under the broad umbrella of environmental genomics. These questions include: (1) the likelihood that molecular profiles return to a predefined baseline level following remediation efforts, (2) how reference site selection in an urban environment influences interpretation of omics data and (3) what is the most appropriate species to monitor in the environment from an omics point of view. In addition, inter-genomics studies have been conducted to assess transcriptome reproducibility in toxicology studies. One lesson learned from inter-genomics studies is that there are core molecular networks that can be identified by multiple laboratories using the same platform. This supports the idea that "omics-networks" defined a priori may be a viable approach moving forward for evaluating environmental impacts over time. Both spatial and temporal variability in ecosystem structure is expected to influence molecular responses to environmental stressors, and it is important to recognize how these variables, as well as individual factor (i.e. sex, age, maturation), may confound interpretation of network responses to chemicals. This mini-review synthesizes the progress made towards adopting these tools into environmental monitoring and identifies future challenges to be addressed, as we move into the next era of high throughput sequencing. A conceptual framework for validating and incorporating molecular networks into environmental monitoring programs is proposed. As AOPs become more defined and their potential in environmental monitoring assessments becomes more recognized, the AOP framework may prove to be the conduit between omics and penultimate ecological responses for environmental risk assessments.