Biology subjects

Petrova, K.

Publications and source records attributed to Petrova, K..

6 recordsLinked to original sources

bioRxiv2026

Quantitative imaging of native tissue architecture combined with deep multi-omics reveals the spatial biology of kidney aging

Aging alters tissue architecture and metabolism in spatially heterogeneous ways, yet the connections between these measurable structural and compositional phenotypes and underlying molecular changes are largely unknown. Here, we introduce NORIGAMI (NoRI-Guided AI for Multimodal Integration), a pipeline combining Normalized Raman Imaging (NoRI), deep learning, quantitative morphology, deep proteomics, and metabolomics to investigate renal aging. Unlike conventional histology, NoRI provides label-free, scattering-corrected measurements of absolute total protein and lipid mass concentrations (mg/mL). In 32 male rats aged 9, 18, 21 and 25 months, NORIGAMI predicted the chronological age of held-out rats from NoRI images (mean absolute error 0.78 months; R^2 = 0.97), and whole-tissue images outperformed models trained on isolated tubules. Quantitative imaging identified 41 age-associated morphological and compositional features (FDR < 0.05), revealing anatomically heterogeneous and non-linear changes in lipid droplet concentration, droplet size, cytoplasmic lipid concentration and tubular architecture. Deep molecular profiling quantified 11,485 proteins and 269 metabolites, revealing coordinated declines in differentiated proximal tubule (PT) identity and oxidative metabolism, selective increases in basement-membrane collagen-IV chains, and increasing molecular heterogeneity with age. A protein clock, built mainly from extracellular-matrix and heparan-sulfate proteins, predicted age with a mean absolute error of 1.6 months. Old kidneys diverged along a dominant injury-remodeling axis, seen separately in the proteome, in kidney RNA from 54 other rats and in the metabolome, in which tubular-injury and immune proteins, including the NAD+ consuming enzyme CD38, rise as differentiated-tubule proteins and NAD+ fall. NORIGAMI subsequently linked NAD+ associated molecular changes to quantitative features of tubular geometry and protein-lipid organization, while complementary KIM-1-centered analyses identified overlapping morphological associations involving epithelial injury and lipid handling and yielded additional molecular candidates for mechanistic validation. Collectively, these findings suggest that declining tubular metabolic function, altered lipid organization and injury-associated molecular programs represent interconnected features of renal aging. NORIGAMI establishes a framework for identifying the quantitative structural manifestations of molecular aging and generating spatially informed, experimentally testable hypotheses about tissue dysfunction.

physiology↗

bioRxiv2026

Whole-Embryo 3D Quantification Reveals Conserved Topological Design and Scaling of Germ Layers in Xenopus

How embryos with markedly different absolute sizes and cell numbers establish comparable tissue organization during development remains a fundamental question in developmental biology. To address this question, we compared two closely related Xenopus species that differ substantially in embryonic size, Xenopus laevis and Xenopus tropicalis. We generated a whole-embryo quantitative 3D atlas of cell allocation, spatial organization, and mitotic dynamics at key time points between gastrulation to tailbud stages. Using tissue clearing, and 3D imaging we tracked single-nucleus coordinates across developmental milestones to resolve how body plans adapt to organismal scale. We show that embryonic scaling is not achieved through simple proportional changes in cell number. Instead, the smaller X. tropicalis embryo is characterized by a distinct high-density tissue organization associated with a persistently higher mitotic index (~1.4-fold higher than in X. laevis at both gastrula and tailbud stages). Across development, this is accompanied by a near-doubling of cell number in X. tropicalis without a proportional increase in embryo volume. We quantify tissue organization and find species-specific cellular architectures during gastrulation that largely converge by the tailbud stage. At this stage, homologous tissues display broadly similar structural profiles despite persistent differences in embryo size, cell number, and density. Together, our findings reveal that closely related vertebrate embryos can follow distinct cellular organization trajectories while converging toward comparable tissue architecture, providing a quantitative framework for understanding robust body plan formation across divergent physical scales.

Developmental Biology↗

bioRxiv2025

Normalized Raman Imaging for Studies of Tissue Physiology of the Kidney

Conventional histological relies on fixation, embedding, sectioning, and staining methods that distort cellular architecture, extract lipids, and introduces variability, limiting reproducibility. We present Normalized Stimulated Raman Imaging (NoRI) that enables quantitative, label-free protein and lipid measurements at high-spatial resolution. NoRI computationally corrects protein, lipid, and water Raman signals, allowing quantitative biomass measurements while preserving tissue architecture, facilitating analysis by convolutional neural networks. Applied to mouse kidney, NoRI accurately classified tubule types (F1-[harmonic mean of precision and recall against manual annotation]=0.93), anatomical regions (F1=0.91), and biological sex (F1=0.97), revealing greater cytoplasmic lipid (+6.9mg/mL; p=0.028) and nuclear protein (+26.3mg/mL; p<0.001) in female tubules. In acute kidney injury, NoRI captured dynamic lipid and protein organization and quantified brush border remodelling and lipid droplet changes over 25 days. Spatial lipid quantification was central for feature classification in AKI (F1=1.0). These results establish NoRI as a reproducible, quantitative method for diagnostics, histopathology, and feature discovery.

physiology↗

bioRxiv2024

Single Cell Transcriptome Defines Cell Type Repertoire of Adult Daphnia magna.

Detailed knowledge of transcriptional responses to environmental and developmental cues is impossible without single cell (SC) resolution data. We performed two SC RNAseq experiments surveying transcriptional profiles of females and males of D. magna, a freshwater plankton crustacean which is both a classic and emerging new model for eco-physiology, toxicology, and evolutionary genomics. We were able to identify over 30 distinct cell types about half of which could be functionally annotated. First, we identified ovaries- and testis-related cell types by focusing on female- and male-specific clusters. Second, we compared markers between SC clusters and bulk RNAseq data on transcriptional profiles of early embryos, circulating hemocytes, midgut, heads (containing brain, eyes, muscles and hepatic caeca), antennae II, and carapace. Finally, we compared transcriptional profiles of Daphnia cell clusters with orthologous markers of 250+ cell types annotated in Drosophila cell atlas. This allowed us to recognize striated muscle cells, gut enterocytes, cuticular cells, as well as 5 different neuron types, including photoreceptors and 3 ovaries-related clusters, one of which tentatively identified as the germ line cells. One well-defined cluster showed a significant enrichment in markers of both hemocytes and fat body of Drosophila, but not with bulk RNAseq data from circulating hemocytes, allowing us to hypothesize the existence of non-circulating, fat body-associated population of hemocytes in Daphnia. On the other hand, the circulating hemocytes express numerous cuticular proteins suggesting their role, in addition to macrophagy, in wound repair. At the same time numerous cell types remain unidentified, including those that map to FCA groups ambiguously or are characterized by Daphnia-specific markers with no clear orthology in the fruitfly. Likewise, many known or presumed cell types or tissues in Daphnia have not been identified to SC clusters. A detailed in-situ hybridization study would be necessary to match not yet annotated SC clusters to functional cell groups. HighlightsO_LIFirst single-cell transcriptomic atlas for Daphnia magna, identifies > 30 distinct cell types. C_LIO_LINovel cell type representing circulating hemocytes may play a role in cuticle regeneration. C_LIO_LIEvidence for non-circulating hemocyte-like cells associated with the fat body in Daphnia. C_LIO_LICuticle/epithelial cells expressing photoreceptors, suggesting light-sensing capabilities. C_LIO_LISubfunctionalization of divergent paralogs across cell types for ecological versatility. C_LI

cell biology↗

bioRxiv2024

A revised single-cell transcriptomic atlas of Xenopus embryo reveals new differentiation dynamics.

This paper introduces an updated single-cell atlas for pivotal developmental stages in Xenopus, encompassing gastrulation, neurulation, and early tailbud. Notably surpassing its predecessor, the new atlas enhances gene mapping, read counts, and gene/cell type nomenclature. Leveraging the latest Xenopus tropicalis genome version, alongside advanced alignment pipelines and machine learning for cell type assignment, this release maintains consistency with previous cell type annotations while rectifying nomenclature issues. Employing an unbiased approach for cell type assignment proves especially apt for embryonic contexts, given the considerable number of non-terminally differentiated cell types. An alternative cell type attribution here adopts a fuzzy, non-deterministic stance, capturing the transient nature of early embryo progenitor cells by presenting an ensemble of types in superposition. The value of the new resource is emphasized through numerous examples, with a focus on previously unexplored germ cell populations where we uncover novel transcription onset features. Offering interactive exploration via a user-friendly web portal and facilitating complete data downloads, this atlas serves as a comprehensive and accessible reference.

developmental biology↗

bioRxiv2021

Using deep convolutional neural networks to forecast spatial patterns of Amazonian deforestation

AO_SCPLOWBSTRACTC_SCPLOWO_LITropical forests are subject to diverse deforestation pressures but their conservation is essential to achieve global climate goals. Predicting the location of deforestation is challenging due to the complexity of the natural and human systems involved but accurate and timely forecasts could enable effective planning and on-the-ground enforcement practices to curb deforestation rates. New computer vision technologies based on deep learning can be applied to the increasing volume of Earth observation data to generate novel insights and make predictions with unprecedented accuracy. C_LIO_LIHere, we demonstrate the ability of deep convolutional neural networks to learn spatiotemporal patterns of deforestation from a limited set of freely available global data layers, including multispectral satellite imagery, the Hansen maps of historic deforestation (2001-2020) and the ALOS JAXA digital surface model, to forecast future deforestation (2021). We designed four original deep learning model architectures, based on 2D Convolutional Neural Networks (2DCNN), 3D Convolutional Neural Networks (3DCNN), and Long Short-Term Memory (LSTM) Recurrent Neural Networks (RNN) to produce spatial maps that indicate the risk to each forested pixel (~30 m) in the landscape of becoming deforested within the next year. They were trained and tested on data from two ~80,000 km2 tropical forest regions in the Southern Peruvian Amazon. C_LIO_LIWe found that the networks could predict the likely location of future deforestation to a high degree of accuracy. Our best performing model - a 3DCNN - had the highest pixel-wise accuracy (80-90%) when validated on 2020 deforestation based 2014-2019 training. Visual examination of the forecasts indicated that the 3DCNN network could automatically discern the drivers of forest loss from the input data. For example, pixels around new access routes (e.g. roads) were assigned high risk whereas this was not the case for recent, concentrated natural loss events (e.g. remote landslides). C_LIO_LICNNs can harness limited time-series data to predict near-future deforestation patterns, an important step in using the growing volume of satellite remote sensing data to curb global deforestation. The modelling framework can be readily applied to any tropical forest location and used by governments and conservation organisations to prevent deforestation and plan protected areas. C_LI

ecology↗