Biology subjects

Donovan, J.

Publications and source records attributed to Donovan, J..

8 recordsLinked to original sources

bioRxiv2026

Penumbria: Advanced 3D cell segmentation for biomedical imaging

Quantitative analysis of three-dimensional cellular architecture is fundamental to understanding tissue organization, disease progression, and drug response. There are effective approaches for 2D segmentation, yet 3D cell segmentation remains a critical bottleneck due to diverse cell morphologies, low signal-to-noise ratios, and data scarcity. We introduce Penumbria, a general-purpose 3D cell segmentation framework that achieves state-of-the-art accuracy across morphologically distinct cell populations and imaging conditions in volumetric microscopy. Penumbria formulates segmentation as a regression problem on distances to cell boundaries, supporting instance reconstruction without shape priors and permitting end-to-end GPU inference. A U-Net-based architecture with xLSTM bottleneck blocks and patch embeddings enables multi-scale feature extraction, long-range modeling of spatial context, and convolutional feature-volume tokenization. The model is extended with two modules: a Global Zernike Phase Layer, which learns Zernike-parameterized phase corrections in the frequency domain to deal with optical aberrations such as defocus and tilt, and a Scaled Geocaps Layer, which samples features at fixed grid locations across multiple spatial scales, routing evidence between them such that a detection is only confident where concordance holds across scales simultaneously. Across four diverse 3D datasets selected to probe the limits of existing methods, Penumbria outperforms Cellpose-SAM on all four and StarDist-3D on three, with comparable accuracy on the fourth, achieving up to a 38% improvement in mean average precision over the second-best method. Penumbrias strong boundary accuracy further supports downstream analyses such as quantifying membrane dynamics or protein localization. We also release a hand-labeled confocal dataset of 1,543 zebrafish neurons across seven volumes, on which Penumbria can be easily and quickly trained.

bioinformatics↗

bioRxiv2025

RNA Sequencing Reveals Divergent Transcriptome Changes in SBP2 and SECISBP2L Depleted Cell Lines

Selenocysteine (Sec), the 21st amino acid, is co-translationally inserted at UGA codons via a specialized machinery requiring SECIS elements, Sec-tRNA^Sec, eEFSec, and SECIS-binding protein 2 (SBP2). While SBP2 is essential for Sec incorporation in vitro and in vivo, the function of its paralog, SECISBP2L, remains incompletely defined. In this study, we investigated the distinct roles of SBP2 and SECISBP2L in the human hepatocellular carcinoma cell line HepG2, which expresses a broad selenoproteome. Using CRISPR-Cas9 genome editing, we generated SBP2 and SECISBP2L edited cell lines. Consistent with previous findings, SBP2 targeting impaired selenoprotein mRNA and protein expression, whereas SECISBP2L targeting did not. However, transcriptomic profiling by RNA-seq revealed that SECISBP2L targeting induced differential expression of over 800 genes, with significant enrichment in pathways related to extracellular matrix organization and cell adhesion. In contrast, SBP2 targeting produced a distinct transcriptomic signature enriched for metabolic and ion transport processes. Notably, only limited overlap in differentially expressed genes was observed between the two knockout models. Mass spectrometry and immunoblot data indicated that CRISPR-targeted SECISBP2L cells produce a truncated protein via internal translation initiation, suggesting that observed gene expression changes may be attributable to loss of a portion of the SECISBP2L N-terminus. These findings support a model in which SECISBP2L plays a noncanonical role in regulating gene expression independent of selenoprotein synthesis. Given prior associations between SECISBP2L downregulation or mutation and cancer progression, our data raise the possibility that SECISBP2L modulates cell adhesion and extracellular matrix gene networks relevant to metastatic potential. This work establishes a foundation for further mechanistic studies into SECISBP2Ls role in gene regulation and disease.

molecular biology↗

bioRxiv2023

Whole blood transcriptional profiles and the pathogenesis of tuberculous meningitis

BackgroundMortality and morbidity from tuberculous meningitis (TBM) are frequent and strongly associated with the inflammatory response to Mycobacterium tuberculosis infection. However, the mechanisms driving the associations are uncertain. We sought to identify the gene modules, hubs and pathways associated with the pathogenesis and mortality from TBM, and to identify which best-predicted death. MethodsWe used whole blood RNA sequencing to obtain transcriptional profiles from 281 Vietnamese adults with TBM (207 HIV-negative; 74 HIV-positive), 295 with pulmonary TB (PTB), and 30 healthy controls. The TBM cohort was divided randomly into a discovery cohort (n=142) and a validation cohort (n=139). Weighted gene co-expression network analysis identified clusters of genes (or modules) and hub genes associated with death or disease severity. An overrepresentation analysis identified pathways associated with TBM mortality, with a consensus analysis identifying consensual patterns between HIV-positive and HIV-negative individuals. A multivariate elastic-net Cox regression model selected the candidate predictors of TBM mortality, then model prediction performance using logistic regression and internal bootstrap validation to choose best predictors. ResultsOverall, TBM mortality was associated with increased neutrophil activation and decreased T and B cell activation pathways. Death from TBM was associated with increased angiogenesis in HIV-positive adults, and with activated TNF signaling and down-regulated extracellular matrix organization in HIV-negative adults. PTB and TBM have similar transcriptional profiles compared to healthy controls, although inflammatory genes were more activated in HIV-positive than HIV-negative TBM. The expression of four hub genes - MCEMP1, NELL2, ZNF354C and CD4 - were strongly predictive of death from TBM (AUC 0.80 and 0.86 for HIV-negative and HIV-positive, respectively). ConclusionsWhole blood transcriptional profiling revealed that TBM is associated with a characteristic systemic inflammatory response, similar to that invoked by pulmonary tuberculosis, but with key gene modules, hubs and pathways strongly associated with death. Our analysis suggests a novel 4-gene biomarker for predicting death from TBM, but also opens a new window into TBM pathogenesis that may reveal novel therapeutic targets for this lethal disease.

molecular biology↗

bioRxiv2019

Widespread haploid-biased gene expression in mammalian spermatogenesis associated with frequent selective sweeps and evolutionary conflict

1Mendels first law dictates that alleles segregate randomly during meiosis and are distributed to offspring with equal frequency, requiring sperm to be functionally independent of their genetic payload. Developing mammalian spermatids have been thought to accomplish this by freely sharing RNA from virtually all genes through cytoplasmic bridges, equalizing allelic gene expression across different genotypes. Applying single cell RNA sequencing to developing spermatids, we identify a large class of mammalian genes whose allelic expression ratio is informative of the haploid genotype, which we call genoinformative markers (GIMs). 29% of spermatid-expressed genes in mice and 47% in non-human primates are not uniformly shared, and instead show a confident allelic expression bias of at least 2-fold towards the haploid genotype. This property of GIMs was significantly conserved between individuals and between rodents and primates. Consistent with the interpretation of specific RNA localization resulting in incomplete sharing through cytoplasmic bridges, we observe a strong depletion of GIM transcripts from chromatoid bodies, structures involved in shuttling RNA across cytoplasmic bridges, and an enrichment for 3 UTR motifs involved in RNA localization. If GIMs are translated and functional in the context of fertility, they would be able to violate Mendels first law, leading to selective sweeps through a population. Indeed, we show that GIMs are enriched for signatures of positive selection, accounting for dozens of recent mouse, human, and primate selective sweeps. Intense selection at the sperm level risks evolutionary conflict between germline and somatic function, and GIMs show evidence of avoiding this conflict by exhibiting more testis-specific gene expression, paralogs, and isoforms than expression-matched control genes. The widespread existence of GIMs suggests that selective forces acting at the level of individual mammalian sperm are much more frequent than commonly believed. 2 Authors summaryMendels first law dictates that alleles are distributed to offspring with equal frequency, requiring sperm carrying different genetics to be functionally equivalent. Despite a small number of known exceptions to this, it is widely believed that sharing of gene products through cytoplasmic bridges erases virtually all differences between haploid sperm. Here, we show that a large class of mammalian genes are not completely shared across these bridges, therefore causing sperm phenotype to correspond partly to haploid genotype. We term these genes "genoinformative markers" (GIMs) and show that their identity tends to be conserved from rodents to primates. Because some GIMs can link sperm genotype to function, they can be thought of as selfish genetic elements which lead to natural selection between sperm rather than between organisms, a violation of Mendels first law. We find evidence of this biased inheritance, showing that GIMs are strongly enriched for selective sweeps that spread alleles through mouse and human populations. For genes expressed both in sperm and in somatic tissues, this can cause a conflict because optimizing gene function for sperm may be detrimental to its other functions. We show that there is evolutionary pressure to avoid this conflict, as GIMs are strongly enriched for testis-specific gene expression, testis-specific paralogs, and testis-specific isoforms. Therefore, GIMs and sperm-level natural selection may provide an elegant explanation for the peculiarity of testis gene expression patterns, which are an extreme outlier relative to all other tissues.

genetics↗

bioRxiv2019

The effect of sample size on polygenic hazard models for prostate cancer

We aimed to determine the effect of sample size on performance of polygenic hazard score (PHS) models in predicting the age at onset of prostate cancer. Age and genotypes were obtained for 40,861 men from the PRACTICAL consortium. The dataset included 201,590 SNPs per subject, and was split into training (34,444 samples) and testing (6,417 samples) sets. Two PHS model-building strategies were investigated. Established-SNP model considered 65 SNPs that had been associated with prostate cancer in the literature. A stepwise SNP selection was used to develop Discovery-SNP models. The performance of each PHS model was calculated for random sizes of the training set (1 to 30 thousand). The performance of a representative Established-SNP model was estimated for random sizes of the testing set (0.5 to 6 thousand). Mean HR98/50 (hazard ratio of top 2% to the average in the test set) of the Established-SNP model increased from 1.73[95%CI: 1.69-1.77] to 2.41[2.40-2.43] when the number of training samples was increased from 1 to 30 thousand. The corresponding HR98/50 of the Discovery-SNP model increased from 1.05[0.93-1.18] to 2.19[2.16-2.23]. HR98/50 of a representative Established-SNP model using testing set sample sizes of 0.6 and 6 thousand observations were 1.78[1.70-1.85] and 1.73[1.71-1.76], respectively. We estimate that a study population of 20 to 30 thousand men is required to develop Discovery-SNP PHS models for prostate cancer. The required sample size could be reduced to 10 thousand samples, if a set of SNPs associated with the disease has already been established.\n\nAuthor summaryPolygenic hazard scores represent a recent advancement in polygenic prediction to model the age of onset of various diseases, such as Alzheimers disease or prostate cancer. These scores accumulate small effect sizes from several tens of genetic variants and can be used to establish an individuals risk of experiencing an event relative to a control population across time. The largest barrier to the development of polygenic hazard scores is the large number of study subjects needed to develop the underlying models. We sought to understand the effect of varying the total number of samples on the performance of a polygenic hazard score in the context of prostate cancer. We found that the performance of the score did not appreciably change beyond 20 to 30 thousand observations when developing the model from scratch. However, when the discovery of the genetic variants can be borrowed from those already identified in the literature to be associated with the disease, the required number of samples is reduced to 10 thousand with no appreciable detriment in performance. We hope that these results can guide the design of future studies of polygenic scores in other diseases and demonstrate the importance of genome-wide association studies.

genetics↗

bioRxiv2018

Concerted 2-5A-Mediated mRNA Decay and Transcription Reprogram Protein Synthesis in dsRNA Response

RNA degradation by RNase L during 2-5A-mediated decay (2-5AMD) is a conserved mammalian stress response to viral and endogenous double-stranded RNA (dsRNA). 2-5AMD onsets rapidly and facilitates a switch of protein synthesis from homeostasis to production of interferons (IFNs). To understand the mechanism of this protein synthesis reprogramming, we examined 2-5AMD in human cells. 2-5AMD triggers polysome collapse characteristic of a translation initiation defect, but translation initiation complexes and ribosomes purified from the translation-arrested cells remain functional. Using spike-in RNA-seq we found that basal messenger RNAs (mRNAs) rapidly decay, while mRNAs encoding IFNs and IFN-stimulated genes evade 2-5AMD and accumulate. The IFN evasion results from the combined effect of better mRNA stability and positive feedback amplification in the IFN response. Therefore, 2-5AMD and transcription act in concert to revamp the cellular mRNA composition. The resulting preferential accumulation of innate immune mRNAs establishes \"prioritized\" synthesis of defense proteins.

cell biology↗

bioRxiv2018

Realtime 2-5A kinetics suggests interferons β and {lambda} evade global arrest of translation by RNase L

Cells of all mammals recognize double-stranded RNA (dsRNA) as a foreign material. In response, they release interferons (IFNs) and activate a ubiquitously expressed pseudokinase/endoribonuclease RNase L. RNase L executes regulated RNA decay and halts global translation. Here we developed a biosensor for 2,5-oligoadenylate (2-5A), the natural activator of RNase L. We found that 2-5A was acutely synthesized by cells in response to dsRNA sensing, which immediately triggered cellular RNA cleavage by RNase L and arrested host protein synthesis. However, translation-arrested cells still transcribed IFN-stimulated genes (ISGs) and secreted IFNs of types I and III (IFN-{beta} and IFN-{lambda}). Our data suggests that IFNs escape from the action of RNase L on translation. We propose that 2-5A/RNase L pathway serves to rapidly and accurately suppress basal protein synthesis, preserving privileged production of defense proteins of the innate immune system.\n\nSignificanceRNase L is a mammalian enzyme that can stop global protein synthesis during interferon response. Cells must balance the need to make interferons (which are proteins) with the risk to lose cell-wide translation due to RNase L. This balance can most simply be achieved if RNase L was activated late in the interferon response. However, we show by engineering a biosensor for the RNase L pathway, that on the contrary, RNase L activation precedes interferon synthesis. Further, translation of interferons evades the action of RNase L. Our data suggest that RNase L facilitates a switch of protein synthesis from homeostasis to specific needs of innate immune signaling.

cell biology↗

bioRxiv2016

Y-RNA and tRNA Cleavage by RNase L Mediates Terminal dsRNA Response

Double-stranded RNA (dsRNA) is a danger signal that triggers endonucleolytic degradation of RNA inside infected and stressed mammalian cells. This mechanism inhibits growth and ultimately removes problematic cells via apoptosis. To elucidate the molecular functions of this program and understand the connection between RNA cleavage and programmed cell death, we visualized dsRNA-induced degradation of human small RNAs using RtcB ligase-assisted RNA sequencing (RtcB RNA-seq). RtcB RNA-seq revealed strong cleavage of select transfer RNAs (tRNAs) and autoantigenic Y-RNAs, and identified the innate immune receptor RNase L as the responsible endoribonuclease. RNase L cleaves the non-coding RNA (ncRNA) targets site-specifically, releasing abundant ncRNA fragments, and downregulating full-length tRNAs and Y-RNAs. The depletion of a single Y-RNA, RNY1, appears particularly important and the loss of this Y-RNA is sufficient to initiate apoptosis. Site-specific cleavage of small ncRNA by RNase L thus emerges as an important terminal step in dsRNA surveillance.

immunology↗