Biology subjects

Petalcorin, M. I. R.

Publications and source records attributed to Petalcorin, M. I. R..

6 recordsLinked to original sources

bioRxiv2026

Computational Synthetic Inner Membrane Reveals Cardiolipin-Leak Control of ATP Output

The inner mitochondrial membrane (IMM) is a densely packed bioenergetic surface where electron transfer, proton pumping, and ATP synthesis are tightly coupled, and where performance is shaped by membrane leak, lipid composition, and higher-order organization. Although experimental reconstruction of oxidative phosphorylation in proteoliposomes has advanced, systematic exploration of design tradeoffs remains challenging because key parameters covary across preparations. Here we present a fully reproducible computational synthetic IMM (syn-IMM) framework that models coupled membrane energization and ATP production, then uses structured perturbations, parameter sweeps, and sensitivity analyses to identify dominant control variables. In a {Delta}{psi}-proxy simulator, we show that ATP output is maximized within a narrow cardiolipin performance window (peak at cardiolipin fraction 0.18 in our benchmarked parameterization), while increasing leak globally suppresses performance, producing a cardiolipin x leak landscape in which coupling integrity is a primary gate. Perturbation experiments separate mechanistic regimes, reduced ATP synthase capacity yields an "energized but unproductive" state with preserved {Delta}{psi} but depressed ATP flux, whereas increased leak reduces usable output despite relatively preserved energization. Monte Carlo sensitivity analysis ranks {Delta}{psi} and respiratory capacity as strongest correlates of ATP output, with ATP synthase capacity contributing positively and leak contributing negatively. A multi-state extension introduces explicit{Delta}{psi} -{Delta}pH partitioning, finite CoQ redox pool dynamics, and cardiolipin-dependent supercomplex fraction S(t), enabling diagnosis of organization kinetics and driving-force partition effects. Together, this syn-IMM platform provides an interpretable bridge between component-level reconstruction and system-level design, offering quantitative acceptance tests and design rules for programmable bioenergetic membranes.

biophysics↗

bioRxiv2025

Quantum-Coherent Identity Preservation and Substrate-Invariant Embodiment: A Theoretical Framework for Sustained Pure-State Dynamics in Complex Biological Systems

Living systems exhibit extraordinary resilience, adaptability, and identity preservation despite continuous atomic turnover. Traditional physics explains this persistence through biochemical stability, but a deeper quantuminformational description remains elusive. Here, we introduce a theoretical framework where a sustaining superoperator ([S]) exactly cancels environmental decoherence ([D]) within the Lindblad formalism, maintaining quantum coherence indefinitely. The resulting sustained pure-state system exhibits vanishing entropy production, stable informational identity, and finite tunneling amplitude under sublinear effective-mass scaling (Meff = m Na) with (0 < < 1). Numerical simulations confirm entropy cancellation, identity invariance under substrate replacement, and anomalous tunneling consistent with coherence-preserving collectivity. These findings propose mathematically consistent conditions for substrate-independent identity persistence and coherent embodiment, connecting concepts from quantum biology, information theory, and open-system thermodynamics.

biophysics↗

bioRxiv2025

Decoding the Energetic Logic of Genetic Systems: A Hybrid Neural-Symbolic Framework for Quantum-Informed Bioenergetics

Life is sustained by the dynamic flow of energy through adenosine triphosphate (ATP), redox carriers such as NADH, and reactive oxygen species (ROS). These molecules not only fuel biochemical reactions but also encode information that regulates gene expression, DNA repair, and replication. Despite decades of biochemical study, the mathematical principles linking cellular energetics to genetic regulation remain unknown. Here we present a hybrid neural-symbolic framework that discovers the governing equations of energy-dependent genetic processes. Using simulated time-series data capturing oscillations in ATP, NADH, and ROS, we trained a neural ordinary differential equation (neural ODE) model to learn the temporal dynamics of gene expression, repair, and replication. The trained model was then analyzed by symbolic regression to extract explicit, interpretable equations describing how energy flow constrains molecular behavior. The resulting system revealed an energetic hierarchy in which transcription dominates during high ATP availability, repair increases under oxidative stress, and replication scales with the NADH/ROS ratio. The symbolic equations recovered exponential and sinusoidal dependencies suggestive of rhythmic, quantum-informed energy coupling. Taken together, these results reveal a mechanistic framework describing how cells distribute energy across competing genetic processes. This work introduces generative bioenergetics, a computational paradigm that unifies machine learning, quantum biology, and mitochondrial systems theory. By translating energy flow into interpretable equations, this approach moves toward a unified model of life as a self-organizing, energy-efficient system where computation and metabolism are fundamentally intertwined.

biophysics↗

bioRxiv2025

Engineering C3.5 Photosynthesis: Coupling Mitochondrial Bioenergetics to Rubisco Efficiency in Arabidopsis

Global food demand is projected to increase by more than 50 percent by 2050, yet most staple crops rely on inefficient C3 photosynthesis. A major limitation arises from Rubisco, the central carbon-fixing enzyme, which catalyzes oxygenation reactions that waste up to 40 percent of fixed carbon through photorespiration. While C4 photosynthesis offers greater efficiency, its anatomical and regulatory complexity has hindered its transfer into C3 crops. Here we propose and evaluate a novel intermediate strategy, termed C3.5 photosynthesis, which reimagines mitochondria as carbon-recycling organelles to enhance Rubisco efficiency. Using Arabidopsis as a model, we integrate computational modeling, simulated phenotyping datasets, and machine learning approaches to benchmark the feasibility of coupling mitochondrial bioenergetics to chloroplast carbon assimilation. We first construct predictive frameworks showing how mitochondrial CO2 release can be recaptured and redirected to chloroplasts through engineered organelle tethers and synthetic transporters. We then simulate the redox and energetic trade-offs of rewiring FoF1 ATP synthase to power bicarbonate transport, providing mechanistic insights into the balance between ATP cost and carbon gain. Our results demonstrate that C3.5 photosynthesis could, in principle, increase net carbon assimilation by 20-50 percent under fluctuating light and heat stress, without the structural reprogramming required for C4 pathways. This work establishes a conceptual and computational foundation for repurposing mitochondria as carbon-recycling hubs, bridging fundamental organelle biology with translational strategies in crop engineering. By combining bioenergetics, organelle engineering, and AI-driven modeling, C3.5 photosynthesis opens a high-risk, high-reward pathway toward climate-resilient agriculture and future food security.

bioengineering↗

bioRxiv2025

Reconstructing the Mitochondrial Proton Motive Force Using Physics-Informed Neural Networks and Surrogate Bioenergetic Signals

The mitochondrial proton motive force (PMF) underlies ATP synthesis, metabolite transport, and energy coupling. Yet, direct measurement of PMF remains technically challenging due to probe invasiveness, calibration drift, and compartmental averaging. Here, we introduce a physics-informed neural network (PINN) framework that reconstructs PMF from surrogate signals including NADH, oxygen, electron flux, proton leak, and reactive oxygen species (ROS). Using a synthetic curriculum dataset derived from biophysical ranges reported in the literature, our model achieved high predictive accuracy (R2 {approx} 0.99, RMSE < 1 mV) under normoxia and hypoxia. SHAP-based interpretability revealed distinct feature contributions: flux and NADH dominated under normoxia, while oxygen and ROS became more influential under hypoxia. Extended analyses demonstrated that PINNs generalize robustly across cross-validation folds, preserve biophysical constraints, and can be adapted to time-series dynamics, capturing PMF decline and recovery during simulated hypoxia-reoxygenation. To our knowledge, this is the first application of PINNs to mitochondrial bioenergetics, bridging machine learning with the chemiosmotic theory. This proof-of-concept establishes a foundation for non-invasive PMF estimation and opens avenues for studying mitochondrial adaptation in physiology and disease.

biophysics↗

bioRxiv2025

AI-Guided Discovery of LDHA Inhibitors Targeting Cancer Metabolism Using Machine Learning and Generative Chemistry: An End-to-End Drug Discovery Pipeline

Targeting cancer metabolism has emerged as a promising therapeutic strategy, particularly through the inhibition of Lactate Dehydrogenase A (LDHA), a key enzyme that supports the Warburg effect in tumor cells. In this study, we present a comprehensive and fully reproducible machine learning (ML) and artificial intelligence (AI)-driven pipeline for the discovery of small-molecule LDHA inhibitors. By integrating bioactivity datasets from ChEMBL and BindingDB, along with natural products from COCONUT and AI-generated compounds from a ChemGPT-based molecular language model, we constructed a diverse and chemically rich screening library. Molecular descriptors were computed using Mordred, followed by feature selection, dataset balancing using SMOTE, and extensive model benchmarking across 11 classifiers. LightGBM was selected as the top-performing model with an AUC of 0.97. SHAP analysis provided model interpretability, revealing key molecular features influencing LDHA inhibition. Additionally, we trained ChemGPT on LDHA-specific SMILES in SELFIES format to generate 1,000 novel molecules, of which over 100 passed stringent drug-likeness, toxicity, and solubility filters. A subset exhibited high LDHA inhibition probabilities (>0.90) and structural novelty. This work highlights the potential of combining predictive modeling and generative chemistry for accelerating the early stages of cancer drug discovery and provides an open-source platform for continued development and validation.

biochemistry↗