Biology subjects

Tran, K.

Publications and source records attributed to Tran, K..

At least 19 recordsLinked to original sources

bioRxiv2026

AN INTEGRATED HEART-ON-A-CHIP NAM RESOLVES DISTINCT CLINICAL PHENOTYPES OF MYOCARDIAL ISCHEMIA-REPERFUSION INJURY

Background: Reperfusion is required for myocardial salvage yet itself injures the heart, and cardioprotective interventions that reduce infarct size experimentally have repeatedly failed to improve outcomes in patients. These interventions are evaluated against viability or infarct size alone, whereas ischemia-reperfusion (I/R) injury produces structural, biochemical, and contractile dysfunction in the same tissue. No in vitro new approach methodology (NAM), however, resolves all three within one system. Methods: A NAM was created within a heart-on-a-chip platform, which was constructed by culturing neonatal rat ventricular myocytes (NRVMs) on muscular thin films. Cardiomyocytes were subjected to normoxia, 2 hours of ischemia, and ischemia followed by 2 hours of reperfusion. Sarcomeric architecture was quantified by immunofluorescence, cardiomyocyte injury by LIVE/DEAD staining and cardiac troponin I release into conditioned media, while contractile stress and spontaneous beat frequency were recorded on the same films before and after each condition. Results: Under ischemia, z-line density decreased while architectural metrics were preserved, indicating cell loss rather than sarcomere disorganization. Furthermore, viability decreased, troponin elevated, and every contractile stress component declined, while spontaneous beat frequency was unchanged. Reperfusion fractured this uniform depression. Cell death and troponin release increased further, yet structure was unchanged from ischemia, so every functional change occurred on a depleted but architecturally intact cardiomyocyte population. Systolic stress recovered partially, whereas diastolic stress exceeded baseline, leaving active stress suppressed despite oxygen recovery. Beat frequency increased with pronounced variability. Conclusions: This heart-on-a-chip NAM resolved structural, biochemical, and contractile injury that global indices conflate, distinguishing ischemic from reperfusion pathophysiology. Interestingly, individual readouts corresponded to distinct clinical syndromes of ischemic injury, positioning this platform for future testing of cardioprotective candidates against the full injury phenotype.

physiology↗

bioRxiv2026

VesSynth: Tubes Are All You Need for Robust Cross-Scale Cross-Modal 3D Vessel Segmentation

The cerebral vasculature is central to brain function, with alterations linked to numerous cerebrovascular and neurological disorders. Yet, no single imaging modality can capture the entire cerebral vascular network in humans. Instead, an array of techniques are sensitized to different spatial scales, while trading off resolution for coverage. Magnetic Resonance Imaging (MRI) typically resolves only large pial vessels, while high-resolution microscopy allows micrometer-scale vessels to be mapped over limited spatial extents. These techniques must therefore be combined to obtain a complete mapping of the cerebral angioarchitecture, which underscores the need for automatic, cross-modal vessel segmentation. Here, we introduce VesSynth, a flexible vessel segmentation framework that achieves state-of-the-art accuracy across multiple modalities and spatial resolutions (MR, optical and X-ray imaging), despite being trained entirely on synthetic data. By enabling consistent vascular mapping across scales, this framework paves the way to comprehensive investigation of cerebrovascular organization and its role in health and disease.

bioengineering↗

bioRxiv2026

Scanning DIA on the ZenoTOF 8600 system enables ultra-sensitive and quantitative proteomics from single cells to post-translational modifications in a compact platform

Mass spectrometry-based proteomics increasingly demands platforms that combine quantitative rigor with the discovery capabilities of accurate mass systems. Here we present the ZenoTOF 8600 system, a compact mass spectrometry system that integrates enhanced ion capture and transmission optics with an optical detection system, Zeno trap-enhanced MS/MS, electron-activated dissociation, and scanning quadrupole data-independent acquisition (ZT Scan DIA). We show that ZT Scan DIA outperforms conventional variable-window DIA (Zeno SWATH DIA) in both identifications and quantitative reproducibility, and demonstrate the platforms versatility across proteomics applications: thousands of protein groups from bulk samples at up to 500 samples per day, single-cell proteomics yielding up to 4,700 proteins, accurate ratio recovery in mixed-species quantitative benchmarks, low-attomole targeted quantitation, and detection of disease-relevant phosphorylation in a Parkinsons disease cellular model using complementary CID and EAD fragmentation. The instruments compact footprint makes it attractive for settings where both analytical breadth and operational robustness are required.

biophysics↗

bioRxiv2026

Grounding olfactory perception in language: Benchmarks and models for generating natural language odor descriptions

Recent advances in deep learning have enabled prediction of odorant perception from molecular structure, opening new avenues for odor classification. However, most existing models are limited to predicting percepts from fixed vocabularies and fail to capture the full richness of olfactory experience. Progress is further limited by the scarcity of large-scale olfactory datasets and the lack of standardized metrics for evaluating free-form natural-language odor descriptions. To address these challenges, we introduce Odor Description and Inference Evaluation Understudy (ODIEU), a benchmark which includes perceptual descriptions of over 10,000 molecules paired with a model-based metric for evaluating free-form odor text descriptions. The model-based metric uses Sentence-BERT (SBERT) models which are fine-tuned on olfactory descriptions to allow better evaluation of human-generated odor texts. Using the fine-tuned SBERT models, we show that free-form text odor descriptions contain additional perceptual information in their syntactic structure compared to semantic labels. We further introduce CIRANO (Chemical Information Recognition and Annotation Network for Odors), a transformer-based model that generates free-form odor descriptions directly from molecular structure, thus implementing the molecular structure-to-text (S2T) prediction. CIRANO achieves performance comparable to humans. Finally, we generate human-like descriptions from mouse olfactory bulb neural data using an invertible SBERT model, yielding neural-to-text (N2T) predictions highly aligned with human descriptions. Together, CIRANO and ODIEU establish a standardized framework for generating natural language olfactory descriptions and evaluating their alignment with human perception. Code is available at https://github.com/KoulakovLab/ODIEU

animal behavior and cognition↗

bioRxiv2026

A Scalable Design for Proximity-Inducing Molecules

Chimeric molecules, which bring together an effector enzyme and a protein-of-interest (POI) to add/remove post-translational modifications (PTMs), are furnishing transformative modalities (e.g., PROTACs). However, these chimeras scalability is limited as they employ rare, non-inhibitory binders of effectors. We report GRoup-transfer chimeras for Inducing Proximity (GRIPs) that employ abundantly available effectors inhibitors to append POI binder on the effector using group-transfer handles. To demonstrate scalability, we develop 6 GRIPs classes for 3 PTMs utilizing diverse inhibitor, spanning 16 effector-POI pairs. Furthermore, we report a toolbox of 42 tunable group-transfer handles for Cys/Lys residues and [~]5000 inhibitor-residue pairs for diverse effectors. Using global proteomics, we confirm the specificity for group transfer and PTM editing. GRIPs endowed new functionalities to POI drugs, including preventing rebound signaling upon drug withdrawal, a more potent/persistent inhibition, and inhibitor-induced pathway activation in 4 fully-endogenous systems. In diverse hemi-endogenous systems (tagged POI), GRIPs induced condensate formation with reduced off-targets, cleared pathogenic PTMs, and initiated PTM crosstalk. Overall, GRIPs provide a scalable and versatile platform for PTM editing. Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=184 SRC="FIGDIR/small/706349v1_ufig1.gif" ALT="Figure 1"> View larger version (41K): org.highwire.dtl.DTLVardef@132f70dorg.highwire.dtl.DTLVardef@f9aa62org.highwire.dtl.DTLVardef@12143edorg.highwire.dtl.DTLVardef@f296d4_HPS_FORMAT_FIGEXP M_FIG C_FIG

synthetic biology↗

bioRxiv2025

Orthogonal resistance mechanisms of classical- and induced-proximity inhibitors

Resistance development is an inevitable failure mode of many drugs, pointing to the need to develop agents with orthogonal resistance mechanisms. Induced-proximity modalities, an emergent class of therapeutics, operate by forming a ternary complex with the protein-of-interest (POI) and effectors, unlike classical inhibitors that form binary complexes with the POI. Using KRAS as a model system, we employed base editor tiling mutagenesis screening to show that induced-proximity inhibitors exhibit orthogonal resistance mechanisms to classical inhibitors despite overlapping binding sites, offering an opportunity to circumvent resistance mechanisms of classical inhibitors. These findings highlight the use of base editor mutagenesis screens to prioritize inhibitors with orthogonal resistance mechanisms and the potential of induced-proximity inhibitors to overcome the drug resistance of classical inhibitors.

biochemistry↗

bioRxiv2025

Phosphorylation-inducing molecules for regulating dynamic cellular processes

Dynamic cellular processes often employ protein phosphorylation for rapid information transfer within and between cells. Phosphorylation-inducing chimeric small molecules (PHICS) have been developed for targeted protein phosphorylation by on-demand inducing a kinase-protein pairing. However, widespread application of PHICS has been limited as previously reported PHICS that recruited AMP-activated protein kinase (AMPK) required serum starvation and target protein overexpression, recruited only a few of the potential AMPK complexes, and exhibited poor dose- and temporal control. Herein, we report an AMPK PHICS platform that operates under physiological conditions (i.e., no serum starvation or target overexpression), recruits multiple AMPK complexes, and can induce target protein phosphorylation with dose and temporal control. We demonstrated the utility of this platform for controlling phosphorylations that underlie two dynamic cellular processes, namely oncogenic signaling and phase separation. An AMPK PHICS directed against Brutons Tyrosine Kinase (BTK), which is a driver of several B cell malignancies, was effective at inducing the death of drug-resistant cancer cells. Here, PHICS induced inhibitory phosphorylations on BTK and attenuated its oncogenic pathway. Phosphorylation of Liprin-3 induces phase separation in neurons and is critical for neurotransmitter release. Using AMPK PHICS that phosphorylated Liprin-3, we induced Liprin-3 phase separation with dose and temporal control. We envision this PHICS platform to find utility in inducing and controlling protein phosphorylations for basic research and biomedicine. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=102 SRC="FIGDIR/small/652895v1_ufig1.gif" ALT="Figure 1"> View larger version (24K): org.highwire.dtl.DTLVardef@16bbb20org.highwire.dtl.DTLVardef@1ca5267org.highwire.dtl.DTLVardef@1e805d5org.highwire.dtl.DTLVardef@1d6ec22_HPS_FORMAT_FIGEXP M_FIG C_FIG

biochemistry↗

bioRxiv2025

Niche Macrophages Recycle Iron to Tumor Cells and Foster Erythroblast Mimicry to Promote Bone Metastasis and Anemia

Bone marrow is both a primary site for blood cell production and a fertile niche for metastatic cancer cell growth, notably in breast cancer. Although anemia is common among patients with bone metastasis, the mechanistic link between metastatic colonization and disrupted erythropoiesis remains poorly understood. Using in vivo niche labeling and single-cell RNA sequencing, we identified a specialized population of VCAM1+CD163+CCR3+ macrophages enriched in the bone metastatic niche. These macrophages, typically essential for erythropoiesis in healthy bone marrow, are co- opted by tumor cells to support their growth through iron acquisition. The hijacking of these macrophages by tumor cells reduces iron availability for erythroblasts, impairing erythropoiesis and contributing to anemia. With increased iron supply, tumor cells further adapt by mimicking erythroblasts, producing hemoglobin under GATA1 regulation in response to hypoxic stress. Notably, macrophages with similar iron- regulating features were found in human bone metastases across multiple cancer types, and elevated HBB expression in breast cancer correlates with increased risk of bone metastasis. These findings establish iron-recycling macrophages as essential regulators within the metastatic bone niche, revealing novel insights into the interplay between immune modulation, metal metabolism and tumor cell plasticity in driving metastatic progression and anemia.

cancer biology↗

bioRxiv2025

Metabolite-sensitive cross-bridge models of human atria reveal the impact of diabetes on muscle mechanoenergetics

Type 2 diabetes is associated with a range of adverse health outcomes, including metabolic dysfunction and increased risk of heart failure. Although the interactions between diabetes and heart disease are complex and incompletely understood, they can be more effectively investigated using multiscale, biophysically-based models of the underlying physiological processes. In this study, experimental data from non-diabetic and diabetic human atrial muscles were used to develop metabolite-sensitive cross-bridge models representing each group. The parameter-isation of these cross-bridge models revealed that reduced muscle stress development and a leftward shift of the complex modulus measured in the diabetic muscles could be attributed to reduced cross-bridge stiffness and slower cross-bridge detachment rates, respectively. These cross-bridge models were also integrated into muscle models to investigate the effects of diabetic cross-bridge function, Ca2+ handling and altered metabolite availability on isometric and physiological work-loop contractions. The diabetic model produced isometric twitches with lower amplitude and prolonged duration and, in response to lowered ATP concentration, the diastolic stress increased notably. In work-loop simulations, the diabetic model exhibited slower shortening, reduced work output and lower power of shortening. However, it was also more efficient and had a less pronounced negative response to increases in Pi concentration. These simulations demonstrate that while experimentally measured differences in diabetic cardiac tissues can lead to impaired function during physiological contractions, they may also offer compensatory advantages. The insights of this study offer clear mechanisms of mechanoenergetic dysfunction in diabetic heart muscle, identifying potential therapeutic targets to improve cardiac outcomes for individuals with diabetes.

physiology↗

bioRxiv2024

Weight Transfer in the Reinforcement Learning Model of Songbird Acquisition

Song acquisition behavior observed in the songbird system provides a notable example of learning through trial- and-error which parallels human speech acquisition. Studying songbird vocal learning can offer insights into mechanisms underlying human language. We present a computational model of song learning that integrates reinforcement learning (RL) and Hebbian learning and agrees with known songbird circuitry. The song circuit outputs activity from nucleus RA, which receives two primary inputs: timing information from area HVC and stochastic activity from nucleus LMAN. Additionally, song learning relies on Area X, a basal ganglia area that receives dopaminergic inputs from VTA. In our model, song is first acquired in the HVC-to-Area X connectivity, employing an RL mechanism that involves node perturbation. This information is then consolidated into HVC-to-RA synapses through a Hebbian mechanism. The transfer of weights from Area X to RA takes place via the thalamus, utilizing a specific form of spike-timing-dependent plasticity (STDP). Thus, we present a computational model grounded in songbird circuitry in which the optimal policy is initially guided by RL and subsequently transferred to another circuit through Hebbian plasticity.

neuroscience↗

bioRxiv2024

Pharmacological characterisation of allosteric modulators at human mGlu5

The metabotropic glutamate receptor 5 (mGlu5) is a Class C G protein-coupled receptor, ubiquitously expressed throughout the CNS. With major roles in cognition, learning and memory, mGlu5 dysfunction is linked with numerous neurodegenerative and neuropsychiatric disorders, presenting a viable therapeutic target. Allosteric modulators bind topographically distinct sites from glutamate and other orthosteric agonists and enhance (positive allosteric modulators, PAMs), inhibit (negative allosteric modulators, NAMs) or do not effect (neutral allosteric ligands, NALs) mGlu5 function. While mGlu5 modulators have efficacy in in vivo rodent models of CNS disorders, none have made it to the clinic. We hypothesise this lack of translatability is arises from preclinical optimisation using on non-human pharmacological data, as functional studies are predominantly performed using rat mGlu5 and non-human brain neuronal cultures. Here we assess and quantify the impact of eleven chemically and pharmacologically diverse mGlu5 PAMs, NAMs and NALs on human mGlu5 activity using radioligand binding, iCa2+ mobilisation and IP1 accumulation assays. By comparing to published and newly generated data for rat mGlu5 we show that while modulator pharmacology is relatively consistent across species, ligand dependent species differences in allosteric modulator affinity, cooperativity and probe dependence are evident. Additionally, we report PAM-dependent effects on orthosteric agonist kinetic profiles at human mGlu5. Together, these data highlight the importance of systematic evaluation of mGlu5 allosteric ligand activity at human mGlu5 to improve drug design and overcome potential barriers to translatability to clinical settings.

pharmacology and toxicology↗

bioRxiv2024

Comprehensive proteolytic profiling of Aedes aegypti mosquito midgut extracts: Unraveling the blood meal protein digestion system

To sustain the gonotrophic cycle, the Aedes aegypti mosquito must acquire a blood meal from a human or other vertebrate host. However, in the process of blood feeding, the mosquito may facilitate the transmission of several bloodborne viral pathogens (e.g., dengue, Zika, and chikungunya). The blood meal is essential as it contains proteins that are digested into polypeptides and amino acid nutrients that are eventually used for egg production. These proteins are digested by several midgut proteolytic enzymes. As such, the female mosquitos reliance on blood may serve as a potential target for vector and viral transmission control. However, this strategy may prove to be challenging since midgut proteolytic activity is a complex process dependent on several exo- and endo-proteases. Therefore, to understand the complexity of Ae. aegypti blood meal digestion, we used Multiplex Substrate Profiling by Mass Spectrometry (MSP-MS) to generate global proteolytic profiles of sugar- and blood-fed midgut tissue extracts, along with substrate profiles of recombinantly expressed midgut proteases. Our results reveal a shift from high exoproteolytic activity in sugar-fed mosquitoes to an expressive increase in endoproteolytic activity in blood-fed mosquitoes. This approach allowed for the identification of 146 cleaved peptide bonds (by the combined 6 h and 24 h blood-fed samples) in the MSP-MS substrate library, and of these 146, 99 (68%) were cleaved by the five recombinant proteases evaluated. These reveal the individual contribution of each recombinant midgut protease to the overall blood meal digestion process of the Ae. aegypti mosquito. Further, our molecular docking simulations support the substrate specificity of each recombinant protease. Therefore, the present study provides key information of midgut proteases and the blood meal digestion process in mosquitoes, which may be exploited for the development of potential inhibitor targets for vector and viral transmission control strategies. Author SummaryThe Aedes aegypti mosquito is a vector of viral pathogens that can be transmitted directly to humans. For instance, the transmission of dengue, Zika, or chikungunya viruses may happen during the Ae. aegypti acquisition of an infected blood meal. This blood meal is important for the anautogenous mosquito because without the digestion of blood proteins the mosquito will not obtain the necessary nutrients needed for egg production. After imbibing a blood meal, midgut digestive enzymes (proteases) are expressed and secreted into the lumen. To fully understand their roles in blood meal digestion, we used a special technique called Multiplex Substrate Profiling by Mass Spectrometry (MSP-MS). This method allows us to generate global proteolytic activity profiles of Ae. aegypti midgut tissue extracts that were fed with sugar or blood. In addition, we generated substrate cleavage profiles of recombinantly expressed midgut proteases allowing us to understand the enzyme preferences for blood proteins. Therefore, utilizing this approach, we found the contribution of each individual recombinant protease tested relative to the global activity profile of blood-fed midgut tissue extracts. This may be a starting point for the validation of midgut protease inhibition and the development of a new potential vector control strategy.

biochemistry↗

bioRxiv2024

Unbiased multiplex antigen screening of Cerebrospinal Fluid detects microbial and autoantigenic epitopes associated with Multiple Sclerosis

To comprehensively investigate the intrathecal antibody profile of multiple sclerosis (MS), we examined the cerebrospinal fluid of 195 patients (92 MS and 103 non-MS) for antibodies using a multiplex unbiased bacteria peptide library. We first tested against Epstein-Barr nuclear antigen 1 (EBNA1) for epitope sites enriched in MS and found a significant enrichment at position 407-419. We then used the data to perform a high-throughput screen against a library of 129 viruses known to infect humans. We discovered several additional epitopes from viruses such as Hantaan virus, Human Herpesvirus 6A and Human respiratory syncytial virus B associated with MS. Besides viral epitopes, we also screened for potential autoantigens of the central nervous system (CNS). We discovered several autoantigenic epitopes in proteins such as ADRB3, HTR3A and MPO that were significantly enriched for MS. Because of previous associations of Toxoplasma gondii infection with MS, we also performed a Toxoplasma gondii specific analysis and discovered additional epitopes enriched for MS. We further assessed epitope-epitope correlations within the patient samples and found distinct patterns of association between these microbial and autoantigenic epitopes. Finally, we performed machine-learning to determine if these epitopes are predictive for MS and found that the model incorporating all the epitopes could most effectively discriminate between MS and non-MS (ROC-AUC score = 0.91). Our results demonstrate the effectiveness of multiplex unbiased screens to detect the identity of potentially cross-reactive antibodies targeting MS CNS epitopes and they can also be used as effective biomarkers for MS. One Sentence SummaryWe performed an unbiased multiplex bacteria peptide antibody library screen on cerebrospinal fluid samples of patients with multiple sclerosis (MS) as well as non-MS controls and detected multiple viral and autoantigenic epitopes that are significantly enriched in MS patient samples.

immunology↗

bioRxiv2024

Molecular Basis for RNA Cytidine Acetylation by NAT10

Human NAT10 acetylates the N4 position of cytidine in RNA, predominantly on rRNA and tRNA, to facilitate ribosome biogenesis and protein translation. NAT10 has been proposed as a therapeutic target in cancers as well as aging-associated pathologies such as Hutchinson-Gilford Progeria Syndrome (HGPS). The [~]120 kDa NAT10 protein uses its acetyl-CoA-dependent acetyltransferase, ATP-dependent helicase, and RNA binding domains in concert to mediate RNA-specific N4-cytidine acetylation. While the biochemical activity of NAT10 is well known, the molecular basis for catalysis of eukaryotic RNA acetylation remains relatively undefined. To provide molecular insights into the RNA-specific acetylation by NAT10, we determined the single particle cryo-EM structures of Chaetomium thermophilum NAT10 (CtNAT10) bound to a bisubstrate cytidine-CoA probe with and without ADP. The structures reveal that NAT10 forms a symmetrical heart-shaped dimer with conserved functional domains surrounding the acetyltransferase active sites harboring the cytidine-CoA probe. Structure-based mutagenesis with analysis of mutants in vitro supports the catalytic role of two conserved active site residues (His548 and Tyr549 in CtNAT10), and two basic patches, both proximal and distal to the active site for RNA-specific acetylation. Yeast complementation analyses and senescence assays in human cells also implicates NAT10 catalytic activity in yeast thermoadaptation and cellular senescence. Comparison of the NAT10 structure to protein lysine and N-terminal acetyltransferase enzymes reveals an unusually open active site suggesting that these enzymes have been evolutionarily tailored for RNA recognition and cytidine-specific acetylation.

biophysics↗

bioRxiv2024

Development of a high-throughput, quantitative platform using human cerebral organoids to study virus-induced neuroinflammation in Alzheimer's disease

Neuroinflammation is a central process in the pathogenesis of several neurodegenerative diseases such as Alzheimers disease (AD), and there are active efforts to target pathways involved in neuroinflammation for molecular biomarker discovery and therapeutic development in neurodegenerative diseases. It was also proposed that there may be an infectious etiology in AD that is associated with viruses such as herpes simplex virus (HSV-1) and influenza A virus (IAV), leading to neuroinflammation-induced AD pathogenesis or disease progression. We sought to develop high-throughput, quantitative molecular biomarker assays using dissociated cells from human cerebral organoids (dcOrgs), that can used for screening compounds to reverse AD-associated neuroinflammation. We found that HSV-1 infection, but not IAV infection, in dcOrgs led to increased intracellular A{beta}42 and phosphorylated Tau-Thr212 (pTau-212) expression, lower ratios of secreted A{beta}42/40, as well as neuronal loss, and increased proportions of astrocytes and microglia, which are hallmarks of AD. Among the glia cell-type markers, Iba1 (microglia) and GFAP (astrocyte) expression were most strongly correlated with HSV-1 expression, which further supported that these biomarkers are perturbed by glia-mediated neuroinflammation. By performing large-scale RNA sequencing, we observed that differentially expressed transcripts in HSV-1 infected dcOrgs were specifically enriched for AD-associated GWAS genes, but not for genes associated with other common neurodegenerative, neuropsychiatric or autoimmune diseases. Immediate treatment of HSV-1 infected dcOrgs with anti-herpetic drug acyclovir (ACV) rescued most of the cellular and transcriptomic biomarkers in a dosage-dependent manner, indicating that it is possible to use our high-throughput platform to identify compounds or target genes that can reverse these neuroinflammation-induced biomarkers associated with AD.

genomics↗

bioRxiv2023

Chemoproteomics identifies proteoform-selective caspase-2 inhibitors

Caspases are a highly conserved family of cysteine-aspartyl proteases known for their essential roles in regulating apoptosis, inflammation, cell differentiation, and proliferation. Complementary to genetic approaches, small-molecule probes have emerged as useful tools for modulating caspase activity. However, due to the high sequence and structure homology of all twelve human caspases, achieving selectivity remains a central challenge for caspase-directed small-molecule inhibitor development efforts. Here, using mass spectrometry-based chemoproteomics, we first identify a highly reactive non-catalytic cysteine that is unique to caspase-2. By combining both gel-based activity-based protein profiling (ABPP) and a tobacco etch virus (TEV) protease activation assay, we then identify covalent lead compounds that react preferentially with this cysteine and afford a complete blockade of caspase-2 activity. Inhibitory activity is restricted to the zymogen or precursor form of monomeric caspase-2. Focused analogue synthesis combined with chemoproteomic target engagement analysis in cellular lysates and in cells yielded both pan-caspase reactive molecules and caspase-2 selective lead compounds together with a structurally matched inactive control. Application of this focused set of tool compounds to stratify caspase contributions to initiation of intrinsic apoptosis, supports compensatory caspase-9 activity in the context of caspase-2 inactivation. More broadly, our study highlights future opportunities for the development of proteoform-selective caspase inhibitors that target non-conserved and non-catalytic cysteine residues.

biochemistry↗

bioRxiv2023

CodonBERT: Large Language Models for mRNA Design and Optimization

AO_SCPLOWBSTRACTC_SCPLOWmRNA based vaccines and therapeutics are gaining popularity and usage across a wide range of conditions. One of the critical issues when designing such mRNAs is sequence optimization. Even small proteins or peptides can be encoded by an enormously large number of mRNAs. The actual mRNA sequence can have a large impact on several properties including expression, stability, immunogenicity, and more. To enable the selection of an optimal sequence, we developed CodonBERT, a large language model (LLM) for mRNAs. Unlike prior models, CodonBERT uses codons as inputs which enables it to learn better representations. CodonBERT was trained using more than 10 million mRNA sequences from a diverse set of organisms. The resulting model captures important biological concepts. CodonBERT can also be extended to perform prediction tasks for various mRNA properties. CodonBERT outperforms previous mRNA prediction methods including on a new flu vaccine dataset.

bioinformatics↗

bioRxiv2023

Supraclavicular brown adipocytes originate from Tbx1+ myoprogenitors

Brown adipose tissue (BAT) dissipates energy as heat, contributing to temperature control, energy expenditure, and systemic homeostasis. In adult humans, BAT mainly exists in supraclavicular areas and its prevalence is associated with cardiometabolic health. However, the developmental origin of supraclavicular BAT remains unknown. Here, using genetic fate mapping in mice, we demonstrate that supraclavicular brown adipocytes do not develop from the Pax3+/Myf5+ epaxial dermomyotome that gives rises to interscapular BAT. Instead, the Tbx1+ lineage that specifies the pharyngeal mesoderm marks the majority of supraclavicular brown adipocytes. Tbx1Cre-mediated ablation of peroxisome proliferator-activated receptor gamma (PPAR{gamma}) or PR/SET Domain 16 (PRDM16), components of the transcriptional complex for brown fat determination, leads to supraclavicular BAT paucity or dysfunction, thus rendering mice more sensitive to cold exposure. Moreover, human deep neck BAT expresses higher levels of the TBX1 gene than subcutaneous neck white adipocytes. Taken together, our observations reveal location-specific developmental origins of BAT depots and call attention to Tbx1+lineage cells when investigating human relevant supraclavicular BAT.

developmental biology↗