Institute of Physiology

Physiologie-Gebäude am Brühlplatz

The Institute provides undergraduate and post-graduate education for students in medicine and life sciences. It carries out research mainly in the field of heart- and neurophysiology.

To the Institute’s website

Managing Director

Prof. Christian Soeller

Highlights 2025

The images show large clusters of the cardiac ryanodine receptor in cardiac myocytes from mice. The punctate signal clearly reveals receptors and their constituent subunits in the sarcoplasmic reticulum of these large cells that make up the walls of the chambers of the heart.

First optical resolution of subunits of the cardiac ryanodine receptor in myocytes

The cardiac ryanodine receptor (RyR2) constitutes the molecular basis of the process of calcium-induced calcium release which is essential for the forceful contraction of the heart during every beat. In a study published in the journal Nature Communications we present purely optical data of RyR2 distribution with sub-molecular resolution by applying 3D MINFLUX microscopy. The study, led by first author Dr. Alexander Clowsley and Prof. Christian Soeller, uses single-domain antibodies and DNA-PAINT to determine the location of individual RyR2 subunits with high precision (~3 nm) and resolve the 3D orientations of RyR2s in-situ. We measured labeling efficiencies of ~50%, implying RyR2 tetramer detection probability approaching 95%. Ventricular myocytes from mice contained large clusters containing many tens of close-packed RyR2s, resolving apparent discrepancies between electron microscopy and previous super-resolution microscopy data. This study showcases the recent MINFLUX technology installed at the Institute using a state-of-the-art microscope supported by the SNSF, the University of Bern large equipment investment fund and the MIC infrastructure.

Clowsley et al., Nat Commun. 2025

The figure summarizes the approach and the effects of KCNH2 SupRep gene therapy. Abbbreviations: AAV9, adeno-associated virus serotype 9; APD90, action potential duration at 90% repolarization; KCNH2-SupRep, KCNH2-specific suppression-and-replacement; KCNH2-shIMM, shRNA-immune KCNH2; IKr, rapid delayed rectifier K+ current; SQT1, type 1 short QT syndrome; UT, untreated; WT, wildtype.

First successful gene therapy correction of live-threatening cardiac arrythmias in an animal model

Type 1 short QT syndrome (SQT1) is a genetic channelopathy caused by gain-of-function variants in KCNH2, resulting in shortened cardiac repolarization and QT intervals, which predispose patients to ventricular arrhythmias and sudden cardiac death. In our study published in the European Heart Journal, we investigated the therapeutic efficacy of a novel KCNH2-specific suppression-and-replacement (KCNH2- SupRep) gene therapy in our transgenic rabbit model of SQT1. We could demonstrate that KCNH2-SupRep corrects the disease phenotype on all levels, e.g., normalized the QT without increasing repolarization heterogeneity in vivo, resolves apicobasal APD heterogeneity in the whole heart, suggesting an antiarrhythmic effect confirmed by reduced re-entry-mediated arrhythmogenesis in silico. At the cellular levels, it prolongs APD, partially normalizes the ion current IKr, and restores electro-mechanical interplay. In sum, this study is the first to demonstrate the efficacy of gene therapy in restoring the physiological phenotype in a medium-sized animal model of SQT1, highlighting its potential for future clinical application.

Nimani et al., Eur Heart J. 2025

The figure shows the experimental paradigm for sensory stimulation and mismatch generation (upper row), examples of evoked calcium transients in individual (middle) and all (lower) stimulus-responsive neurons.

Top-down modulation of sensory processing and mismatch in the mouse posterior parietal cortex

An important function of the neocortex is to compare sensory stimuli with internal predictions of the outside world. In our publication in Nature Communications we showed that deviations evoke neuronal mismatch responses, which allow expectations to be updated. We elucidated the mechanisms behind sensory feedback mismatch and prediction formation by creating a learned association of an auditory-tactile stimulus sequence in awake head-fixed mice, where a sound predicted an up-coming whisker stimulus and introduced mismatches by omitting or altering the whisker stimulus intensity. We found that layer 2/3 posterior parietal cortex (PPC) neurons could report stimulus sequence mismatches, as well as displaying neural correlates of expectation. Inhibition of PPC-projecting secondary motor cortex (M2) neurons suppressed these correlates, along with population mismatch responses. Hence, M2 can influence sensory processing in the PPC and potentially contribute to the prediction of sensory feedback from learned relationships within sequences of sensory stimuli. This work contributes to a better understanding of information processing in the brain and the underlying neuronal mechanisms that can be impaired in neurological disorders.

Raltschev et al., Nat Commun. 2025

Top left: cortical microcircuitry for real-time, phase-free and fully local learning in the GLE framework. Bottom: under GLE, cortical hierarchies can learn complex spatio-temporal convolutions on both static and dynamical input data (images, movies, sounds etc.). This allows them to attend to relevant parts of the input patterns. Top right: Starting from scratch, without pre-engineered structures, GLE-based cortical networks can compete with state-of-the-art, but biologically implausible machine learning solutions.

Backpropagation through space, time and the brain

Efficient learning in physical neuronal networks, bound by spatio-temporal locality constraints, remains a topic of intense debate. State-of-the-art AI algorithms rely on assumptions that violate this locality in various ways. In our publication in Nature Communications, we introduced Generalized Latent Equilibrium (GLE), a framework for fully local learning in physical, dynamical neuronal networks. In the spirit of fundamental physics, we have constructed an energy function as a complete description of a neuronal network, from which everything else can be derived. In particular, we derived neuronal dynamics from energy conservation and synapse dynamics as gradient descent. The result is an online approximation of backpropagation through space and time in deep networks of cortical microcircuits with continuously active, local synaptic plasticity. GLE exploits dendritic morphology to enable complex information storage and processing in single neurons, as well as their ability to react in anticipation of their future input. This provides a rigorous solution to the problem of learning complex dynamical patterns in cortical networks, thereby also lending itself to applications in neuromorphic computing.

Ellenberger et al., Nat Commun. 2025