Search bioRxiv⌕ Search
Home/Archive/bioRxiv

bioRxiv · 10.1101/855890

Mechano-calcium and mechano-electric feedbacks in the human cardiomyocyte analyzed in a mathematical model

Balakina-Vikulova, N. A.·Panfilov, A.·Solovyova, O.·Katsnelson, L. B.
View original source ↗Export metadata ↓

Abstract

Experiments on animal hearts (rat, rabbit, guinea pig, etc.) have demonstrated that mechano-calcium feedback (MCF) and mechano-electric feedback (MEF) are very important for myocardial self-regulation because they adjust the cardiomyocyte contractile function to various mechanical loads and to mechanical interactions between heterogeneous myocardial segments in the ventricle walls. In in vitro experiments on these animals, MCF and MEF manifested themselves in several basic classical phenomena (e.g. load dependence, length dependence of isometric twitches, etc.), and in the respective responses of calcium transients and action potentials. However, it is extremely difficult to study simultaneously the electrical, calcium, and mechanical activities of the human heart muscle in vitro. Mathematical modeling is a useful tool for exploring these phenomena. We have developed a novel model to describe electromechanical coupling and mechano-electric feedbacks in the human cardiomyocyte. It combines the ten Tusscher - Panfilov electrophysiological model of the human cardiomyocyte with our module of myocardium mechanical activity taken from the Ekaterinburg - Oxford model and adjusted to human data. Using it, we simulated isometric and afterloaded twitches and effects of MCF and MEF on excitation-contraction coupling. MCF and MEF were found to affect significantly the duration of the calcium transient and action potential in the human cardiomyocyte model in response to both smaller afterloads as compared to bigger ones and various mechanical interventions applied during isometric and afterloaded twitches.

Source connections

Explore related subjects

biophysicsPreprint

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Balakina-Vikulova, N. A., Panfilov, A., Solovyova, O., Katsnelson, L. B.. 2019-11-26. Mechano-calcium and mechano-electric feedbacks in the human cardiomyocyte analyzed in a mathematical model. https://doi.org/10.1101/855890

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2026

Functional Screening of Kv7.5 Channel Variants Associated with Neuronal Dysfunction

The KCNQ5 gene encodes Kv7.5 channels, which contribute to the regulation of neuronal excitability. Kv7.5 genetic variants associated with neurological dysfunction have been reported, but functional characterization of Kv7.5 variants is limited and causal links to pathogenicity are lacking. We have conducted the first automated patch clamp functional screen of a panel of heterozygous Kv7.5/Kv7.3 channel variants. We tested nine mutations located in diverse locations throughout the Kv7.5 channel and identified some variants with gain-of-function effects (L341I+/-, G347S+/-, P369R+/-, R497C+/-), some with loss-of-function effects (V145G+/-, W303C+/-), and others with no apparent effect (A131V+/-, R359H+/-, R491H+/-). Variants produced a gain of function by left-shifting the voltage-dependence of channel activation alone or in combination with increasing current density. Loss of function occurred as a result of reduced current density. Using these findings, we initiated the first z-score classification to guide understanding of Kv7.5 variant pathogenicity, which can be refined as more variants are functionally characterized and new variants are discovered. Lastly, we observed higher sensitivity to channel block by XE991 in L341I+/-, G347S+/-, P369R+/- gain-of-function variants indicating potential for targeted pharmacogenetic approaches to restore normal Kv7.5 channel function.

biophysics↗
bioRxiv2026

Learning inhibitor interactions from protein deep mutational scanning data and deep learning models

Recent advances in protein language models, geometric deep learning, and structure prediction frameworks have the potential to accelerate computational drug discovery. However, it remains unclear how well these methods generalize beyond their training sets. Mutation-dependent drug responses present a challenge to these methods because there are often large effects from relatively small perturbations. Here, we systematically benchmark diverse mutational effect and drug-target affinity prediction models using deep mutational scanning datasets spanning 11 inhibitor conditions for the MET kinase domain. Across both mutational effect prediction and drug-target affinity tasks, current AI models produce largely ligand-invariant predictions despite experimentally distinct inhibitor-dependent fitness landscapes. Ligand-aware geometric models, including LigandMPNN and LASErMPNN, partially recover experimentally validated resistance-associated hotspots, highlighting the importance of explicitly modeling protein-ligand interactions and local structural flexibility. In contrast, deep learning affinity prediction models underestimate mutation-induced changes in binding affinity. This deficiency likely reflects mutation-invariant structural priors derived from co-folded structure prediction frameworks and the absence of mutation-conditioned supervision in current affinity datasets. Finally, we further extended our analysis to several kinase-inhibitor deep mutational scanning datasets and confirm the overall trends. Our findings suggest that a primary limitation in improving performance on this task lies in the scarcity of mutation-conditioned protein-ligand training data rather than model architecture alone.

biophysics↗
bioRxiv2026

Scalable Contact-Gated Stochastic Operators and Universal Reduction Laws for Macromolecular Protein Dynamics, Allostery, and Kinetics

Constructing continuous, thermodynamically reversible reduced equations of motion from atomistic molecular dynamics (MD) that scale from fast-folding miniproteins to multi-domain allosteric enzymes in explicit water remains a foundational problem in computational biophysics. Across 15 molecular systems (890,000 SHA-256-verified frames, 0.89 us cumulative sampling) spanning 13 structurally diverse proteins from Chignolin (d=18) to explicit-solvent Adenylate Kinase (d=426, 40,110 particles) and Maltose-Binding Protein (d=738, 49,215 particles), plus two analytical verification peptides (d=2, 4), we formulate a contact-gated, GENERIC-regularized stochastic operator on the backbone torus T^d. We show that collective slow modes compress sub-linearly across two orders of magnitude in chain length - with spectral participation ratio scaling as PR(d)=1.061 d^0.494 (95% CI [0.332, 0.666], Delta AICc=-18.15) and short-time diffusion rank as r_D^90(d)=0.799 d^0.930 (Delta AICc=-14.39) - whereas cumulative 90% equilibrium variance in large native-basin enzymes transitions to extensive proportionality (10.0% blind extrapolation error vs. 35.7% for small-protein power laws). Consequently, macromolecular tractability above d>200 requires off-diagonal sparsity rather than global PCA truncation alone: universal peptide-plane crankshaft anti-correlation (mean R(psi_i, phi_i+1)=-0.603 +/- 0.054) and sparse 3D tertiary contact gating (mean k_eff=5.97, reaching 99.3% sparsity at d=738) enable a sparse directed acyclic contact-graph Cholesky estimator scaling empirically as O(d^0.92) in time and O(d^1.06) in memory (65.8x memory compression, 99.6x faster than Graphical LASSO, and higher blind held-out log-likelihood +0.303/d vs. dense +0.205/d on MBP). Biophysically, relative solvent-accessible surface area (rSASA_i) unifies dry linear equipartition in protein cores (alpha_core=1.00, 12.33% blind thermal error) with viscosity-coupled loops (alpha_loop=1.28) and peptides (alpha_peptide=1.57); parameter-free kinematic projection reproduces solution NMR Lipari-Szabo S^2 order parameters (mean r_s=+0.483 across four BMRB enzymes, Ubiquitin Pearson r=+0.921) and X-ray B-factors (mean r_s=+0.276); and 13-fold zero-shot Leave-One-Protein-Out transfer predicts unseen protein operators directly from static PDB coordinates (r_s=+0.706, R^2=+0.359 ridge; r_s=+0.721, R^2=+0.410 GBDT). Applied to functional discovery, explicit-solvent apo toroidal operators localize mechanical domain-closure hinges in AdK (PR-AUC=0.348) and MBP (AUROC=0.690 vs. 0.446 ANM), resolve distal Met20 <-> FG catalytic loop coupling in E. coli DHFR (71.6% recovery at 93.2% sparsity) and beta1-beta2 <-> Ile44 routing in Ubiquitin (7.1x amplification) on blind trajectories, and quantify sub-diffusive Mori-Zwanzig reaction-coordinate memory (mean gamma_M=0.593 +/- 0.126). A validation-tuned head-to-head benchmark against Markov State Models, TICA, Smoluchowski diffusion, Elastic Network Models, and unconstrained SDEs across all 13 proteins maps the physical regimes where continuous contact-gated GENERIC operators (lowest Frobenius error 0.288, exact 0.000 spurious entropy vs. median 65.79 kB/ns/d, IQR 53.76-104.66, mean 346.17) and discrete or harmonic baselines each excel.

biophysics↗