Explainable machine learning identifies multi-omics signatures of muscle response to spaceflight in mice.

The adverse effects of microgravity exposure on mammalian physiology during spaceflight necessitate a deep understanding of the underlying mechanisms to develop effective countermeasures. One such concern is muscle atrophy, which is partly attributed to the dysregulation of calcium levels due to abnormalities in SERCA pump functioning. To identify potential biomarkers for this condition, multi-omics data and physiological data available on the NASA Open Science Data Repository (osdr.nasa.gov) were used, and machine learning methods were employed. Specifically, we used multi-omics (transcriptomic, proteomic, and DNA methylation) data and calcium reuptake data collected from C57BL/6 J mouse soleus and tibialis anterior tissues during several 30+ day-long missions on the international space station. The QLattice symbolic regression algorithm was introduced to generate highly explainable models that predict either experimental conditions or calcium reuptake levels based on multi-omics features. The list of candidate models established by QLattice was used to identify key features contributing to the predictive capability of these models, with Acyp1 and Rps7 proteins found to be the most predictive biomarkers related to the resilience of the tibialis anterior muscle in space. These findings could serve as targets for future interventions aiming to reduce the extent of muscle atrophy during space travel.

Predicting weight loss success on a new Nordic diet: an untargeted multi-platform metabolomics and machine learning approach.

Background and aim: Results from randomized controlled trials indicate that no single diet performs better than other for all people living with obesity. Regardless of the diet plan, there is always large inter-individual variability in weight changes, with some individuals losing weight and some not losing or even gaining weight. This raises the possibility that, for different individuals, the optimal diet for successful weight loss may differ. The current study utilized machine learning to build a predictive model for successful weight loss in subjects with overweight or obesity on a New Nordic Diet (NND). Methods: Ninety-one subjects consumed an NND ad libitum for 26 weeks. Based on their weight loss, individuals were classified as responders (weight loss ≥5%, n = 46) or non-responders (weight loss <2%, n = 24). We used clinical baseline data combined with baseline urine and plasma untargeted metabolomics data from two different analytical platforms, resulting in a data set including 2,766 features, and employed symbolic regression (QLattice) to develop a predictive model for weight loss success. Results: There were no differences in clinical parameters at baseline between responders and non-responders, except age (47 ± 13 vs. 39 ± 11 years, respectively, p = 0.009). The final predictive model for weight loss contained adipic acid and argininic acid from urine (both metabolites were found at lower levels in responders) and generalized from the training (AUC 0.88) to the test set (AUC 0.81). Responders were also able to maintain a weight loss of 4.3% in a 12 month follow-up period. Conclusion: We identified a model containing two metabolites that were able to predict the likelihood of achieving a clinically significant weight loss on an ad libitum NND. This work demonstrates that models based on an untargeted multi-platform metabolomics approach can be used to optimize precision dietary treatment for obesity.

Multi-omic analyses of triptan-treated migraine attacks gives insight into molecular mechanisms.

Migraine is a common, polygenic disorder that is characterized by moderate to severe headache attacks. Migraine attacks are commonly treated with triptans, i.e. serotonin receptor agonists. However, triptans are effective in ~ 60% of the population, and the mechanisms of triptans are debated. Here, we aim to expose the mechanisms of triptan using metabolomics and transcriptomics in spontaneous migraine attacks. We collected temporal multi-omics profiles on 24 migraine patients, using samples collected at a migraine attack, 2 h after treatment with a triptan, when headache-free, and after a cold-pressor test. Differential metabolomic analysis was performed to find metabolites associated with treatment. Their effect was further investigated using correlation analysis and a machine learning approach. We found three differential metabolites: cortisol, sumatriptan and glutamine. The change in sumatriptan levels correlated with a change in GNAI1 and VIPR2 gene expression, both known to regulate cAMP levels. Furthermore, we found fatty acid oxidation to be affected, a mechanism known to be involved in migraine but not previously found in relation to triptans. In conclusion, using an integrative approach we find evidence for a role of glutamine, cAMP regulation, and fatty acid oxidation in the molecular mechanisms of migraine and/or the effect of triptans.

More than a Feeling: Dermatological Changes Impacted by Spaceflight.

Spaceflight poses a unique set of challenges to humans and the hostile Spaceflight environment can induce a wide range of increased health risks, including dermatological issues. The biology driving the frequency of skin issues in astronauts is currently not well understood. To address this issue, we used a systems biology approach utilizing NASA's Open Science Data Repository (OSDR) on spaceflown murine transcriptomic datasets focused on the skin, biomedical profiles from fifty NASA astronauts, and confirmation via transcriptomic data from JAXA astronauts, the NASA Twins Study, and the first civilian commercial mission, Inspiration4. Key biological changes related to skin health, DNA damage & repair, and mitochondrial dysregulation were determined to be involved with skin health risks during Spaceflight. Additionally, a machine learning model was utilized to determine key genes driving Spaceflight response in the skin. These results can be used for determining potential countermeasures to mitigate Spaceflight damage to the skin.

Dissection of gastric homeostasis in vivo facilitates permanent capture of isthmus-like stem cells in vitro.

The glandular stomach is composed of two regenerative compartments termed corpus and antrum, and our understanding of the transcriptional networks that maintain these tissues is incomplete. Here we show that cell types with equivalent functional roles in the corpus and antrum share similar transcriptional states including the poorly characterized stem cells of the isthmus region. To further study the isthmus, we developed a monolayer two-dimensional (2D) culture system that is continually maintained by Wnt-responsive isthmus-like cells capable of differentiating into several gastric cell types. Importantly, 2D cultures can be converted into conventional three-dimensional organoids, modelling the plasticity of gastric epithelial cells in vivo. Finally, we utilized the 2D culture system to show that Sox2 is both necessary and sufficient to generate enterochromaffin cells. Together, our data provide important insights into gastric homeostasis, establish a tractable culture system to capture isthmus cells and uncover a role for Sox2 in enterochromaffin cells.

Ductal keratin 15+ luminal progenitors in normal breast exhibit a basal-like breast cancer transcriptomic signature.

Normal breast luminal epithelial progenitors have been implicated as cell of origin in basal-like breast cancer, but their anatomical localization remains understudied. Here, we combine collection under the microscope of organoids from reduction mammoplasties and single-cell mRNA sequencing (scRNA-seq) of FACS-sorted luminal epithelial cells with multicolor imaging to profile ducts and terminal duct lobular units (TDLUs) and compare them with breast cancer subtypes. Unsupervised clustering reveals eleven distinct clusters and a differentiation trajectory starting with keratin 15+ (K15+) progenitors enriched in ducts. Spatial mapping of luminal progenitors is confirmed at the protein level by staining with critical duct markers. Comparison of the gene expression profiles of normal luminal cells with those of breast cancer subtypes suggests a strong correlation between normal breast ductal progenitors and basal-like breast cancer. We propose that K15+ basal-like breast cancers originate in ductal progenitors, which emphasizes the importance of not only lineages but also cellular position within the ductal-lobular tree.

Identifying interactions in omics data for clinical biomarker discovery using symbolic regression.

MOTIVATION: The identification of predictive biomarker signatures from omics and multi-omics data for clinical applications is an active area of research. Recent developments in assay technologies and machine learning (ML) methods have led to significant improvements in predictive performance. However, most high-performing ML methods suffer from complex architectures and lack interpretability. RESULTS: We present the application of a novel symbolic-regression-based algorithm, the QLattice, on a selection of clinical omics datasets. This approach generates parsimonious high-performing models that can both predict disease outcomes and reveal putative disease mechanisms, demonstrating the importance of selecting maximally relevant and minimally redundant features in omics-based machine-learning applications. The simplicity and high-predictive power of these biomarker signatures make them attractive tools for high-stakes applications in areas such as primary care, clinical decision-making and patient stratification. AVAILABILITY AND IMPLEMENTATION: The QLattice is available as part of a python package (feyn), which is available at the Python Package Index (https://pypi.org/project/feyn/) and can be installed via pip. The documentation provides guides, tutorials and the API reference (https://docs.abzu.ai/). All code and data used to generate the models and plots discussed in this work can be found in https://github.com/abzu-ai/QLattice-clinical-omics. SUPPLEMENTARY INFORMATION: Supplementary material is available at Bioinformatics online.

Coordinated maintenance of H3K36/K27 methylation by histone demethylases preserves germ cell identity and immortality.

Germ cells have evolved unique mechanisms to ensure the transmission of genetically and nongenetically encoded information, whose alteration compromises germ cell immortality. Chromatin factors play fundamental roles in these mechanisms. H3K36 and H3K27 methyltransferases shape and propagate a pattern of histone methylation essential for C. elegans germ cell maintenance, but the role of respective histone demethylases remains unexplored. Here, we show that jmjd-5 regulates H3K36me2 and H3K27me3 levels, preserves germline immortality, and protects germ cell identity by controlling gene expression. The transcriptional and biological effects of jmjd-5 loss can be hindered by the removal of H3K27demethylases, indicating that H3K36/K27 demethylases act in a transcriptional framework and promote the balance between H3K36 and H3K27 methylation required for germ cell immortality. Furthermore, we find that in wild-type, but not in jmjd-5 mutants, alterations of H3K36 methylation and transcription occur at high temperature, suggesting a role for jmjd-5 in adaptation to environmental changes.

A large-scale genome-wide gene expression analysis in peripheral blood identifies very few differentially expressed genes related to antidepressant treatment and response in patients with major depressive disorder.

A better understanding of the biological factors underlying antidepressant treatment in patients with major depressive disorder (MDD) is needed. We perform gene expression analyses and explore sources of variability in peripheral blood related to antidepressant treatment and treatment response in patients suffering from recurrent MDD at baseline and after 8 weeks of treatment. The study includes 281 patients, which were randomized to 8 weeks of treatment with vortioxetine (N = 184) or placebo (N = 97). To our knowledge, this is the largest dataset including both gene expression in blood and placebo-controlled treatment response measured by a clinical scale in a randomized clinical trial. We identified three novel genes whose RNA expression levels at baseline and week 8 are significantly (FDR < 0.05) associated with treatment response after 8 weeks of treatment. Among these genes were SOCS3 (FDR = 0.0039) and PROK2 (FDR = 0.0028), which have previously both been linked to depression. Downregulation of these genes was associated with poorer treatment response. We did not identify any genes that were differentially expressed between placebo and vortioxetine groups at week 8 or between baseline and week 8 of treatment. Nor did we replicate any genes identified in previous peripheral blood gene expression studies examining treatment response. Analysis of genome-wide expression variability showed that type of treatment and treatment response explains very little of the variance, a median of <0.0001% and 0.05% in gene expression across all genes, respectively. Given the relatively large size of the study, the limited findings suggest that peripheral blood gene expression might not be the best approach to explore the biological factors underlying antidepressant treatment.

Author Correction: Identification of epilepsy-associated neuronal subtypes and gene expression underlying epileptogenesis.

A Correction to this paper has been published: https://doi.org/10.1038/s41467-020-19869-5.

Identification of epilepsy-associated neuronal subtypes and gene expression underlying epileptogenesis.

Epilepsy is one of the most common neurological disorders, yet its pathophysiology is poorly understood due to the high complexity of affected neuronal circuits. To identify dysfunctional neuronal subtypes underlying seizure activity in the human brain, we have performed single-nucleus transcriptomics analysis of >110,000 neuronal transcriptomes derived from temporal cortex samples of multiple temporal lobe epilepsy and non-epileptic subjects. We found that the largest transcriptomic changes occur in distinct neuronal subtypes from several families of principal neurons (L5-6_Fezf2 and L2-3_Cux2) and GABAergic interneurons (Sst and Pvalb), whereas other subtypes in the same families were less affected. Furthermore, the subtypes with the largest epilepsy-related transcriptomic changes may belong to the same circuit, since we observed coordinated transcriptomic shifts across these subtypes. Glutamate signaling exhibited one of the strongest dysregulations in epilepsy, highlighted by layer-wise transcriptional changes in multiple glutamate receptor genes and strong upregulation of genes coding for AMPA receptor auxiliary subunits. Overall, our data reveal a neuronal subtype-specific molecular phenotype of epilepsy.

Complex IV subunit isoform COX6A2 protects fast-spiking interneurons from oxidative stress and supports their function.

Parvalbumin-positive (PV+ ) fast-spiking interneurons are essential to control the firing activity of principal neuron ensembles, thereby regulating cognitive processes. The high firing frequency activity of PV+ interneurons imposes high-energy demands on their metabolism that must be supplied by distinctive machinery for energy generation. Exploring single-cell transcriptomic data for the mouse cortex, we identified a metabolism-associated gene with highly restricted expression to PV+ interneurons: Cox6a2, which codes for an isoform of a cytochrome c oxidase subunit. Cox6a2 deletion in mice disrupts perineuronal nets and enhances oxidative stress in PV+ interneurons, which in turn impairs the maturation of their morphological and functional properties. Such dramatic effects were likely due to an essential role of COX6A2 in energy balance of PV+ interneurons, underscored by a decrease in the ATP-to-ADP ratio in Cox6a2-/- PV+ interneurons. Energy disbalance and aberrant maturation likely hinder the integration of PV+ interneurons into cortical neuronal circuits, leading to behavioral alterations in mice. Additionally, in a human patient bearing mutations in COX6A2, we found a potential association of the mutations with mental/neurological abnormalities.

Joint analysis of heterogeneous single-cell RNA-seq dataset collections.

Single-cell RNA sequencing is often applied in study designs that include multiple individuals, conditions or tissues. To identify recurrent cell subpopulations in such heterogeneous collections, we developed Conos, an approach that relies on multiple plausible inter-sample mappings to construct a global graph connecting all measured cells. The graph enables identification of recurrent cell clusters and propagation of information between datasets in multi-sample or atlas-scale collections.

HLA-DM Stabilizes the Empty MHCII Binding Groove: A Model Using Customized Natural Move Monte Carlo.

MHC class II molecules bind peptides derived from extracellular proteins that have been ingested by antigen-presenting cells and display them to the immune system. Peptide loading occurs within the antigen-presenting cell and is facilitated by HLA-DM. HLA-DM stabilizes the open conformation of the MHCII binding groove when no peptide is bound. While a structure of the MHCII/HLA-DM complex exists, the mechanism of stabilization is still largely unknown. Here, we applied customized Natural Move Monte Carlo to investigate this interaction. We found a possible long-range mechanism that implicates the configuration of the membrane-proximal globular domains in stabilizing the open state of the empty MHCII binding groove.

Tracing the origin of adult intestinal stem cells.

Adult intestinal stem cells are located at the bottom of crypts of Lieberkühn, where they express markers such as LGR51,2 and fuel the constant replenishment of the intestinal epithelium1. Although fetal LGR5-expressing cells can give rise to adult intestinal stem cells3,4, it remains unclear whether this population in the patterned epithelium represents unique intestinal stem-cell precursors. Here we show, using unbiased quantitative lineage-tracing approaches, biophysical modelling and intestinal transplantation, that all cells of the mouse intestinal epithelium-irrespective of their location and pattern of LGR5 expression in the fetal gut tube-contribute actively to the adult intestinal stem cell pool. Using 3D imaging, we find that during fetal development the villus undergoes gross remodelling and fission. This brings epithelial cells from the non-proliferative villus into the proliferative intervillus region, which enables them to contribute to the adult stem-cell niche. Our results demonstrate that large-scale remodelling of the intestinal wall and cell-fate specification are closely linked. Moreover, these findings provide a direct link between the observed plasticity and cellular reprogramming of differentiating cells in adult tissues following damage5-9, revealing that stem-cell identity is an induced rather than a hardwired property.

Neuroserpin expression during human brain development and in adult brain revealed by immunohistochemistry and single cell RNA sequencing.

Neuroserpin is a serine-protease inhibitor mainly expressed in the CNS and involved in the inhibition of the proteolytic cascade. Animal models confirmed its neuroprotective role in perinatal hypoxia-ischaemia and adult stroke. Although neuroserpin may be a potential therapeutic target in the treatment of the aforementioned conditions, there is still no information in the literature on its distribution during human brain development. The present study provides a detailed description of the changing spatiotemporal patterns of neuroserpin focusing on physiological human brain development. Five stages were distinguished within our examined age range which spanned from the 7th gestational week until adulthood. In particular, subplate and deep cortical plate neurons were identified as the main sources of neuroserpin production between the 25th gestational week and the first postnatal month. Our immunohistochemical findings were substantiated by single cell RNA sequencing data showing specific neuronal and glial cell types expressing neuroserpin. The characterization of neuroserpin expression during physiological human brain development is essential for forthcoming studies which will explore its involvement in pathological conditions, such as perinatal hypoxia-ischaemia and adult stroke in human.

In silico structural modeling of multiple epigenetic marks on DNA.

Availability and implementation: The code together with examples and tutorials are available from http://www.cs.ox.ac.uk/mosaics. Contact: peter.minary@cs.ox.ac.uk. Supplementary information: Supplementary data are available at Bioinformatics online.

Modeling Functional Motions of Biological Systems by Customized Natural Moves.

Simulating the functional motions of biomolecular systems requires large computational resources. We introduce a computationally inexpensive protocol for the systematic testing of hypotheses regarding the dynamic behavior of proteins and nucleic acids. The protocol is based on natural move Monte Carlo, a highly efficient conformational sampling method with built-in customization capabilities that allows researchers to design and perform a large number of simulations to investigate functional motions in biological systems. We demonstrate the use of this protocol on both a protein and a DNA case study. Firstly, we investigate the plasticity of a class II major histocompatibility complex in the absence of a bound peptide. Secondly, we study the effects of the epigenetic mark 5-hydroxymethyl on cytosine on the structure of the Dickerson-Drew dodecamer. We show how our customized natural moves protocol can be used to investigate causal relationships of functional motions in biological systems.

Exploring peptide/MHC detachment processes using hierarchical natural move Monte Carlo.

MOTIVATION: The binding between a peptide and a major histocompatibility complex (MHC) is one of the most important processes for the induction of an adaptive immune response. Many algorithms have been developed to predict peptide/MHC (pMHC) binding. However, no approach has yet been able to give structural insight into how peptides detach from the MHC. RESULTS: In this study, we used a combination of coarse graining, hierarchical natural move Monte Carlo and stochastic conformational optimization to explore the detachment processes of 32 different peptides from HLA-A*02:01. We performed 100 independent repeats of each stochastic simulation and found that the presence of experimentally known anchor amino acids affects the detachment trajectories of our peptides. Comparison with experimental binding affinity data indicates the reliability of our approach (area under the receiver operating characteristic curve 0.85). We also compared to a 1000 ns molecular dynamics simulation of a non-binding peptide (AAAKTPVIV) and HLA-A*02:01. Even in this simulation, the longest published for pMHC, the peptide does not fully detach. Our approach is orders of magnitude faster and as such allows us to explore pMHC detachment processes in a way not possible with all-atom molecular dynamics simulations. AVAILABILITY AND IMPLEMENTATION: The source code is freely available for download at http://www.cs.ox.ac.uk/mosaics/. CONTACT: bernhard.knapp@stats.ox.ac.uk SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.