Lipolysis drives expression of the constitutively active receptor GPR3 to induce adipose thermogenesis.

Thermogenic adipocytes possess a therapeutically appealing, energy-expending capacity, which is canonically cold-induced by ligand-dependent activation of ß-adrenergic G protein-coupled receptors (GPCRs). Here, we uncover an alternate paradigm of GPCR-mediated adipose thermogenesis through the constitutively active receptor, GPR3. We show that the N terminus of GPR3 confers intrinsic signaling activity, resulting in continuous Gs-coupling and cAMP production without an exogenous ligand. Thus, transcriptional induction of Gpr3 represents the regulatory parallel to ligand-binding of conventional GPCRs. Consequently, increasing Gpr3 expression in thermogenic adipocytes is alone sufficient to drive energy expenditure and counteract metabolic disease in mice. Gpr3 transcription is cold-stimulated by a lipolytic signal, and dietary fat potentiates GPR3-dependent thermogenesis to amplify the response to caloric excess. Moreover, we find GPR3 to be an essential, adrenergic-independent regulator of human brown adipocytes. Taken together, our findings reveal a noncanonical mechanism of GPCR control and thermogenic activation through the lipolysis-induced expression of constitutively active GPR3.

Brown Adipose Tissue: A Metabolic Regulator in a Hypothalamic Cross Talk?

Since the discovery of functionally competent, energy-consuming brown adipose tissue (BAT) in adult humans, much effort has been devoted to exploring this tissue as a means for increasing energy expenditure to counteract obesity. However, despite promising effects on metabolic rate and insulin sensitivity, no convincing evidence for weight-loss effects of cold-activated human BAT exists to date. Indeed, increasing energy expenditure would naturally induce compensatory feedback mechanisms to defend body weight. Interestingly, BAT is regulated by multiple interactions with the hypothalamus from regions overlapping with centers for feeding behavior and metabolic control. Therefore, in the further exploration of BAT as a potential source of novel drug targets, we discuss the hypothalamic orchestration of BAT activity and the relatively unexplored BAT feedback mechanisms on neuronal regulation. With a holistic view on hypothalamic-BAT interactions, we aim to raise ideas and provide a new perspective on this circuit and highlight its clinical relevance.

Diverse repertoire of human adipocyte subtypes develops from transcriptionally distinct mesenchymal progenitor cells.

Single-cell sequencing technologies have revealed an unexpectedly broad repertoire of cells required to mediate complex functions in multicellular organisms. Despite the multiple roles of adipose tissue in maintaining systemic metabolic homeostasis, adipocytes are thought to be largely homogenous with only 2 major subtypes recognized in humans so far. Here we report the existence and characteristics of 4 distinct human adipocyte subtypes, and of their respective mesenchymal progenitors. The phenotypes of these distinct adipocyte subtypes are differentially associated with key adipose tissue functions, including thermogenesis, lipid storage, and adipokine secretion. The transcriptomic signature of "brite/beige" thermogenic adipocytes reveals mechanisms for iron accumulation and protection from oxidative stress, necessary for mitochondrial biogenesis and respiration upon activation. Importantly, this signature is enriched in human supraclavicular adipose tissue, confirming that these cells comprise thermogenic depots in vivo, and explain previous findings of a rate-limiting role of iron in adipose tissue browning. The mesenchymal progenitors that give rise to beige/brite adipocytes express a unique set of cytokines and transcriptional regulators involved in immune cell modulation of adipose tissue browning. Unexpectedly, we also find adipocyte subtypes specialized for high-level expression of the adipokines adiponectin or leptin, associated with distinct transcription factors previously implicated in adipocyte differentiation. The finding of a broad adipocyte repertoire derived from a distinct set of mesenchymal progenitors, and of the transcriptional regulators that can control their development, provides a framework for understanding human adipose tissue function and role in metabolic disease.

Adenosine activates brown adipose tissue and recruits beige adipocytes via A2A receptors.

Brown adipose tissue (BAT) is specialized in energy expenditure, making it a potential target for anti-obesity therapies. Following exposure to cold, BAT is activated by the sympathetic nervous system with concomitant release of catecholamines and activation of ß-adrenergic receptors. Because BAT therapies based on cold exposure or ß-adrenergic agonists are clinically not feasible, alternative strategies must be explored. Purinergic co-transmission might be involved in sympathetic control of BAT and previous studies reported inhibitory effects of the purinergic transmitter adenosine in BAT from hamster or rat. However, the role of adenosine in human BAT is unknown. Here we show that adenosine activates human and murine brown adipocytes at low nanomolar concentrations. Adenosine is released in BAT during stimulation of sympathetic nerves as well as from brown adipocytes. The adenosine A2A receptor is the most abundant adenosine receptor in human and murine BAT. Pharmacological blockade or genetic loss of A2A receptors in mice causes a decrease in BAT-dependent thermogenesis, whereas treatment with A2A agonists significantly increases energy expenditure. Moreover, pharmacological stimulation of A2A receptors or injection of lentiviral vectors expressing the A2A receptor into white fat induces brown-like cells-so-called beige adipocytes. Importantly, mice fed a high-fat diet and treated with an A2A agonist are leaner with improved glucose tolerance. Taken together, our results demonstrate that adenosine-A2A signalling plays an unexpected physiological role in sympathetic BAT activation and protects mice from diet-induced obesity. Those findings reveal new possibilities for developing novel obesity therapies.

Epigenome- and Transcriptome-wide Changes in Muscle Stem Cells from Low Birth Weight Men.

Background: Being born with low birth weight (LBW) is a risk factor for muscle insulin resistance and type 2 diabetes (T2D), which may be mediated by epigenetic mechanisms programmed by the intrauterine environment. Epigenetic mechanisms exert their prime effects in developing cells. We hypothesized that muscle insulin resistance in LBW subjects may be due to early differential epigenomic and transcriptomic alterations in their immature muscle progenitor cells.Results: Muscle progenitor cells were obtained from 23 healthy young adult men born at term with LBW, and 15 BMI-matched normal birth weight (NBW) controls. The cells were subsequently cultured and differentiated into myotubes. DNA and RNA were harvested before and after differentiation for genome-wide DNA methylation and RNA expression measurements.After correcting for multiple comparisons (q ≤ 0.05), 56 CpG sites were found to be significantly, differentially methylated in myoblasts from LBW compared with NBW men, of which the top five gene-annotated CpG sites (SKI, ARMCX3, NR5A2, NEUROG, ESRRG) previously have been associated to regulation of cholesterol, fatty acid and glucose metabolism and muscle development or hypertrophy. LBW men displayed markedly decreased myotube gene expression levels of the AMPK-repressing tyrosine kinase gene FYN and the histone deacetylase gene HDAC7. Silencing of FYN and HDAC7 was associated with impaired myotube formation, which for HDAC7 reduced muscle glucose uptake.Conclusions: The data provides evidence of impaired muscle development predisposing LBW individuals to T2D is linked to and potentially caused by distinct DNA methylation and transcriptional changes including down regulation of HDAC7 and FYN in their immature myoblast stem cells.

Adipogenesis in Primary Cell Culture.

Obesity involves a contrasting expansion of the energy-storing white fat and loss of functionally competent brown fat, an energy-consuming thermogenic adipose. Leveraging our understanding of white and brown adipocyte recruitment and investigating factors that regulate these processes might reveal novel targets for counteracting obesity. In vitro differentiation of primary preadipocytes mimics many of the morphological and transcriptional events occurring during adipogenesis in vivo. Moreover, preadipocytes isolated from a specific depot maintain features of their originating niche. This makes in vitro adipogenesis a valuable model for identifying differential regulation patterns between brown and white adipogenesis. In this chapter, we describe step-by-step how to isolate brown and white preadipocytes from human tissue biopsies and how to culture and differentiate them in vitro. We discuss this process, what to consider, and how this in vitro system can be used to model in vivo adipogenesis.

Abnormal epigenetic changes during differentiation of human skeletal muscle stem cells from obese subjects.

BACKGROUND: Human skeletal muscle stem cells are important for muscle regeneration. However, the combined genome-wide DNA methylation and expression changes taking place during adult myogenesis have not been described in detail and novel myogenic factors may be discovered. Additionally, obesity is associated with low relative muscle mass and diminished metabolism. Epigenetic alterations taking place during myogenesis might contribute to these defects. METHODS: We used Infinium HumanMethylation450 BeadChip Kit (Illumina) and HumanHT-12 Expression BeadChip (Illumina) to analyze genome-wide DNA methylation and transcription before versus after differentiation of primary human myoblasts from 14 non-obese and 14 obese individuals. Functional follow-up experiments were performed using siRNA mediated gene silencing in primary human myoblasts and a transgenic mouse model. RESULTS: We observed genome-wide changes in DNA methylation and expression patterns during differentiation of primary human muscle stem cells (myoblasts). We identified epigenetic and transcriptional changes of myogenic transcription factors (MYOD1, MYOG, MYF5, MYF6, PAX7, MEF2A, MEF2C, and MEF2D), cell cycle regulators, metabolic enzymes and genes previously not linked to myogenesis, including IL32, metallothioneins, and pregnancy-specific beta-1-glycoproteins. Functional studies demonstrated IL-32 as a novel target that regulates human myogenesis, insulin sensitivity and ATP levels in muscle cells. Furthermore, IL32 transgenic mice had reduced insulin response and muscle weight. Remarkably, approximately 3.7 times more methylation changes (147,161 versus 39,572) were observed during differentiation of myoblasts from obese versus non-obese subjects. In accordance, DNMT1 expression increased during myogenesis only in obese subjects. Interestingly, numerous genes implicated in metabolic diseases and epigenetic regulation showed differential methylation and expression during differentiation only in obese subjects. CONCLUSIONS: Our study identifies IL-32 as a novel myogenic regulator, provides a comprehensive map of the dynamic epigenome during differentiation of human muscle stem cells and reveals abnormal epigenetic changes in obesity.

Epigenetic programming of adipose-derived stem cells in low birthweight individuals.

AIMS/HYPOTHESIS: Low birthweight (LBW) is associated with dysfunctions of adipose tissue and metabolic disease in adult life. We hypothesised that altered epigenetic and transcriptional regulation of adipose-derived stem cells (ADSCs) could play a role in programming adipose tissue dysfunction in LBW individuals. METHODS: ADSCs were isolated from the subcutaneous adipose tissue of 13 normal birthweight (NBW) and 13 LBW adult men. The adipocytes were cultured in vitro, and genome-wide differences in RNA expression and DNA methylation profiles were analysed in ADSCs and differentiated adipocytes. RESULTS: We demonstrated that ADSCs from LBW individuals exhibit multiple expression changes as well as genome-wide alterations in methylation pattern. Reduced expression of the transcription factor cyclin T2 encoded by CCNT2 may play a key role in orchestrating several of the gene expression changes in ADSCs from LBW individuals. Indeed, silencing of CCNT2 in human adipocytes decreased leptin secretion as well as the mRNA expression of several genes involved in adipogenesis, including MGLL, LIPE, PPARG, LEP and ADIPOQ. Only subtle genome-wide mRNA expression and DNA methylation changes were seen in mature cultured adipocytes from LBW individuals. CONCLUSIONS/INTERPRETATION: Epigenetic and transcriptional changes in LBW individuals are most pronounced in immature ADSCs that in turn may programme physiological characteristics of the mature adipocytes that influence the risk of metabolic diseases. Reduced expression of CCNT2 may play a key role in the developmental programming of adipose tissue.

Glucose tolerance is associated with differential expression of microRNAs in skeletal muscle: results from studies of twins with and without type 2 diabetes.

AIMS/HYPOTHESIS: We aimed to identify microRNAs (miRNAs) associated with type 2 diabetes and risk of developing the disease in skeletal muscle biopsies from phenotypically well-characterised twins. METHODS: We measured muscle miRNA levels in monozygotic (MZ) twins discordant for type 2 diabetes using arrays. Further investigations of selected miRNAs included target prediction, pathway analysis, silencing in cells and association analyses in a separate cohort of 164 non-diabetic MZ and dizygotic twins. The effects of elevated glucose and insulin levels on miRNA expression were examined, and the effect of low birthweight (LBW) was studied in rats. RESULTS: We identified 20 miRNAs that were downregulated in MZ twins with diabetes compared with their non-diabetic co-twins. Differences for members of the miR-15 family (miR-15b and miR-16) were the most statistically significant, and these miRNAs were predicted to influence insulin signalling. Indeed, miR-15b and miR-16 levels were associated with levels of key insulin signalling proteins, miR-15b was associated with the insulin receptor in non-diabetic twins and knockdown of miR-15b/miR-16 in myocytes changed the levels of insulin signalling proteins. LBW in twins and undernutrition during pregnancy in rats were, in contrast to overt type 2 diabetes, associated with increased expression of miR-15b and/or miR-16. Elevated glucose and insulin suppressed miR-16 expression in vitro. CONCLUSIONS: Type 2 diabetes is associated with non-genetic downregulation of several miRNAs in skeletal muscle including miR-15b and miR-16, potentially targeting insulin signalling. The paradoxical findings in twins with overt diabetes and twins at increased risk of the disease underscore the complexity of the regulation of muscle insulin signalling in glucose homeostasis.

Integrating a Multi-label Deep Learning Approach with Protein Information to Compare Bioactive Peptides in Brain and Plasma.

Peptide therapeutics is gaining momentum. Advances in the field of peptidomics have enabled researchers to harvest vital information from various organisms and tissue types concerning peptide existence, expression and function. The development of mass spectrometry techniques for high-throughput peptide quantitation has paved the way for the identification and discovery of numerous known and novel peptides. Though much has been achieved, scientists are still facing difficulties when it comes to reducing the search space of the large mass spectrometry-generated peptidomics datasets and focusing on the subset of functionally relevant peptides. Moreover, there is currently no straightforward way to analytically compare the distributions of bioactive peptides in distinct biological samples, which may reveal much useful information when seeking to characterize tissue- or fluid-specific peptidomes. In this chapter, we demonstrate how to identify, rank, and compare predicted bioactive peptides and bioactivity distributions from extensive peptidomics datasets. To aid this task, we utilize MultiPep, a multi-label deep learning approach designed for classifying peptide bioactivities, to identify bioactive peptides. The predicted bioactivities are synergistically combined with protein information from the UniProt database, which assist in navigating through the jungle of putative therapeutic peptides and relevant peptide leads.

The novel adrenergic agonist ATR-127 targets skeletal muscle and brown adipose tissue to tackle diabesity and steatohepatitis.

OBJECTIVE: Simultaneous activation of ß2- and ß3-adrenoceptors (ARs) improves whole-body metabolism via beneficial effects in skeletal muscle and brown adipose tissue (BAT). Nevertheless, high-efficacy agonists simultaneously targeting these receptors whilst limiting activation of ß1-ARs - and thus inducing cardiovascular complications - are currently non-existent. Therefore, we here developed and evaluated the therapeutic potential of a novel ß2-and ß3-AR, named ATR-127, for the treatment of obesity and its associated metabolic perturbations in preclinical models. METHODS: In the developmental phase, we assessed the impact of ATR-127's on cAMP accumulation in relation to the non-selective ß-AR agonist isoprenaline across various rodent ß-AR subtypes, including neonatal rat cardiomyocytes. Following these experiments, L6 muscle cells were stimulated with ATR-127 to assess the impact on GLUT4-mediated glucose uptake and intramyocellular cAMP accumulation. Additionally, in vitro, and in vivo assessments are conducted to measure ATR-127's effects on BAT glucose uptake and thermogenesis. Finally, diet-induced obese mice were treated with 5 mg/kg ATR-127 for 21 days to investigate the effects on glucose homeostasis, body weight, fat mass, skeletal muscle glucose uptake, BAT thermogenesis and hepatic steatosis. RESULTS: Exposure of L6 muscle cells to ATR-127 robustly enhanced GLUT4-mediated glucose uptake despite low intramyocellular cAMP accumulation. Similarly, ATR-127 markedly increased BAT glucose uptake and thermogenesis both in vitro and in vivo. Prolonged treatment of diet-induced obese mice with ATR-127 dramatically improved glucose homeostasis, an effect accompanied by decreases in body weight and fat mass. These effects were paralleled by an enhanced skeletal muscle glucose uptake, BAT thermogenesis, and improvements in hepatic steatosis. CONCLUSIONS: Our results demonstrate that ATR-127 is a highly effective, novel ß2- and ß3-ARs agonist holding great therapeutic promise for the treatment of obesity and its comorbidities, whilst potentially limiting cardiovascular complications. As such, the therapeutic effects of ATR-127 should be investigated in more detail in clinical studies.

Adipogenic and SWAT cells separate from a common progenitor in human brown and white adipose depots.

Adipocyte function is a major determinant of metabolic disease, warranting investigations of regulating mechanisms. We show at single-cell resolution that progenitor cells from four human brown and white adipose depots separate into two main cell fates, an adipogenic and a structural branch, developing from a common progenitor. The adipogenic gene signature contains mitochondrial activity genes, and associates with genome-wide association study traits for fat distribution. Based on an extracellular matrix and developmental gene signature, we name the structural branch of cells structural Wnt-regulated adipose tissue-resident (SWAT) cells. When stripped from adipogenic cells, SWAT cells display a multipotent phenotype by reverting towards progenitor state or differentiating into new adipogenic cells, dependent on media. Label transfer algorithms recapitulate the cell types in human adipose tissue datasets. In conclusion, we provide a differentiation map of human adipocytes and define the multipotent SWAT cell, providing a new perspective on adipose tissue regulation.

Deficient leukemia inhibitory factor signaling in muscle precursor cells from patients with type 2 diabetes.

The cytokine leukemia-inhibitory factor (LIF) is expressed by skeletal muscle and induces proliferation of muscle precursor cells, an important feature of skeletal muscle maintenance and repair. We hypothesized that muscle precursor cells from patients with type 2 diabetes had a deficient response to LIF. The mRNA and protein expressions of LIF and its receptor (LIFR) were measured in skeletal muscle biopsies from healthy individuals and patients with type 2 diabetes by use of qPCR and Western blot. LIF signaling and response were studied following administration of recombinant LIF and siRNA knockdown of suppressor of cytokine signaling (SOCS)3 in myoblast cultures established from healthy individuals and patients with type 2 diabetes. Myoblast proliferation rate was assessed by bromodeoxyuridine incorporation. LIF and LIFR proteins were increased in both muscle tissue and cultured myoblasts from diabetic patients. Nonetheless, in the diabetic myoblasts, LIF-induced phosphorylation of signal transducer and activator of transcription (STAT)1 and STAT3 was impaired. The deficient response to LIF administration in the diabetic myoblasts was further emphasized by a lack of increase in LIF-stimulated cell proliferation and a decreased LIF-stimulated induction of the proliferation-promoting factors cyclin D1, JunB, and c-myc. SOCS3 protein was upregulated in diabetic myoblasts, and knockdown of SOCS3 rescued LIF-induced gene expression in diabetic myoblasts, whereas neither STAT1 or STAT3 signaling nor proliferation rate was affected. In conclusion, although LIF and LIFR proteins were increased in muscle tissue and myoblasts from diabetic patients, LIF signaling and LIF-stimulated cell proliferation were impaired in diabetic myoblasts, suggesting a novel mechanism by which muscle function is compromised in diabetes.

Can we target obesity using a single-cell atlas of adipose tissue?

Adipose tissue is a multicellular tissue with depot-dependent functions including safe energy storage and release, immune defense, thermogenesis, and organ padding. Emont et al. zoom in on white adipose tissue at a single-cell resolution and provide an extensive resource for future insights in how to target obesity.1.

Isolation and Characterization of Human Brown Adipocytes.

Brown adipose tissue (BAT) is a thermoregulatory fat with energy-consuming properties. The location and heterogeneity of this tissue makes it complicated to sample before and after interventions in humans, and an in vitro model for mechanistic and molecular studies is therefore of great value. We here describe a protocol for isolation of progenitors from the stromal vascular fraction of BAT biopsies obtained surgically from adult humans. We further present how these cells are differentiated in vitro and finally how they are characterized for thermogenic capacity. Methods for characterization described here include norepinephrine-induced thermogenic gene expression using qPCR; norepinephrine-induced mitochondrial uncoupling using the Seahorse XFe96 Analyzer, and norepinephrine-induced expression of UCP1 using the RNAscope® Technology.

The human batokine EPDR1 regulates ß-cell metabolism and function.

OBJECTIVE: Ependymin-Related Protein 1 (EPDR1) was recently identified as a secreted human batokine regulating mitochondrial respiration linked to thermogenesis in brown fat. Despite that EPDR1 is expressed in human pancreatic ß-cells and that glucose-stimulated mitochondrial metabolism is critical for stimulus-secretion coupling in ß-cells, the role of EPDR1 in ß-cell metabolism and function has not been investigated. METHODS: EPDR1 mRNA levels in human pancreatic islets from non-diabetic (ND) and type 2 diabetes (T2D) subjects were assessed. Human islets, EndoC-ßH1 and INS1 832/13 cells were transfected with scramble (control) and EPDR1 siRNAs (EPDR1-KD) or treated with human EPDR1 protein, and glucose-stimulated insulin secretion (GSIS) assessed by ELISA. Mitochondrial metabolism was investigated by extracellular flux analyzer, confocal microscopy and mass spectrometry-based metabolomics analysis. RESULTS: EPDR1 mRNA expression was upregulated in human islets from T2D and obese donors and positively correlated to BMI of donors. In T2D donors, EPDR1 mRNA levels negatively correlated with HbA1c and positively correlated with GSIS. EPDR1 silencing in human islets and ß-cell lines reduced GSIS whereas treatment with human EPDR1 protein increased GSIS. Epdr1 silencing in INS1 832/13 cells reduced glucose- and pyruvate- but not K+-stimulated insulin secretion. Metabolomics analysis in Epdr1-KD INS1 832/13 cells suggests diversion of glucose-derived pyruvate to lactate production and decreased malate-aspartate shuttle and the tricarboxylic acid (TCA) cycle activity. The glucose-stimulated rise in mitochondrial respiration and ATP/ADP-ratio was impaired in Epdr1-deficient cells. CONCLUSION: These results suggests that to maintain glucose homeostasis in obese people, upregulation of EPDR1 may improve ß-cell function via channelling glycolysis-derived pyruvate to the mitochondrial TCA cycle.

Human visceral and subcutaneous adipose stem and progenitor cells retain depot-specific adipogenic properties during obesity.

Abdominal obesity associates with cardiometabolic disease and an accumulation of lipids in the visceral adipose depot, whereas lipid accumulation in the subcutaneous depot is more benign. We aimed to further investigate whether the adipogenic properties where cell-intrinsic, or dependent on a depot-specific or obesity-produced microenvironment. We obtained visceral and subcutaneous biopsies from non-obese women (n = 14) or women living with morbid obesity (n = 14) and isolated adipose stem and progenitor cells (ASPCs) from the stromal vascular fraction of non-obese (n = 13) and obese (n = 13). Following in vitro differentiation into mature adipocytes, we observed a contrasting pattern with a lower gene expression of adipogenic markers and a higher gene expression of immunogenic markers in the visceral compared to the subcutaneous adipocytes. We identified the immunogenic factor BST2 as a marker for visceral ASPCs. The effect of obesity and insulin resistance on adipogenic and immunogenic markers in the in vitro differentiated cells was minor. In contrast, differentiation with exogenous Tumor necrosis factor resulted in increased immunogenic signatures, including increased expression of BST2, and decreased adipogenic signatures in cells from both depots. Our data, from 26 women, underscore the intrinsic differences between human visceral and subcutaneous adipose stem and progenitor cells, suggest that dysregulation of adipocytes in obesity mainly occurs at a post-progenitor stage, and highlight an inflammatory microenvironment as a major constraint of human adipogenesis.

Coagulation factors promote brown adipose tissue dysfunction and abnormal systemic metabolism in obesity.

Brown adipose tissue (BAT) has a role in maintaining systemic metabolic health in rodents and humans. Here, we show that metabolic stress induces BAT to produce coagulation factors, which then-together with molecules derived from the circulation-promote BAT dysfunction and systemic glucose intolerance. When mice were fed a high-fat diet (HFD), the levels of tissue factor, coagulation Factor VII (FVII), activated coagulation Factor X (FXa), and protease-activated receptor 1 (PAR1) expression increased significantly in BAT. Genetic or pharmacological suppression of coagulation factor-PAR1 signaling in BAT ameliorated its whitening and improved thermogenic response and systemic glucose intolerance in mice with dietary obesity. Conversely, the activation of coagulation factor-PAR1 signaling in BAT caused mitochondrial dysfunction in brown adipocytes and systemic glucose intolerance in mice fed normal chow. These results indicate that BAT produces endogenous coagulation factors that mediate pleiotropic effects via PAR1 signaling under metabolic stress.

Perspectives on the role of brown adipose tissue in human body temperature and metabolism.

In this Backstory, Camilla Scheele shares the journey of her group's research on human brown adipose tissue leading up to the study of winter swimmers' thermoregulation and energy expenditure (https://www.cell.com/cell-reports-medicine/fulltext/S2666-3791(21)00266-4) appearing in this issue of Cell Reports Medicine.

MultiPep: a hierarchical deep learning approach for multi-label classification of peptide bioactivities.

Peptide-based therapeutics are here to stay and will prosper in the future. A key step in identifying novel peptide-drugs is the determination of their bioactivities. Recent advances in peptidomics screening approaches hold promise as a strategy for identifying novel drug targets. However, these screenings typically generate an immense number of peptides and tools for ranking these peptides prior to planning functional studies are warranted. Whereas a couple of tools in the literature predict multiple classes, these are constructed using multiple binary classifiers. We here aimed to use an innovative deep learning approach to generate an improved peptide bioactivity classifier with capacity of distinguishing between multiple classes. We present MultiPep: a deep learning multi-label classifier that assigns peptides to zero or more of 20 bioactivity classes. We train and test MultiPep on data from several publically available databases. The same data are used for a hierarchical clustering, whose dendrogram shapes the architecture of MultiPep. We test a new loss function that combines a customized version of Matthews correlation coefficient with binary cross entropy (BCE), and show that this is better than using class-weighted BCE as loss function. Further, we show that MultiPep surpasses state-of-the-art peptide bioactivity classifiers and that it predicts known and novel bioactivities of FDA-approved therapeutic peptides. In conclusion, we present innovative machine learning techniques used to produce a peptide prediction tool to aid peptide-based therapy development and hypothesis generation.